Aligned helps revenue teams win and onboard customers faster by collaborating more effectively with customers

Aligned

aligned

This is Part 6 of a 7-part series of the research paper “Towards Usability Guidelines for the Design of Effective Arabic Websites”. Use the table of links below to navigate to the other parts. Table of Links Abstract and I. Introduction II. Previous Works III. Research Methodology IV. Font and Image Inspection Studies V. Results VI. Discussion VII. Conclusions, Future Research Directions, and References VI. DISCUSSION Our comprehensive inspection of current practices concerning text formatting and image usage within the top visited Arabic websites in Saudi Arabia revealed various interesting findings. The usability inspection was performed to gather insights about the font types and sizes used to style Arabic text, especially that they could directly influence the legibility of text on websites. Let us review and attempt to answer the research questions posited in the introduction. What font types and font sizes are used by major Arabic websites? On the positive side, the selection of the Arabic websites included a large pool of websites, which were selected objectively by relying on Alexa Internet latest ranking. In effect, the inspection covered a total of 73 top viewed websites (52 local and 21 international) in Saudi Arabia. The websites' ranking is based on a set of metrics, primarily the average viewing time for the last three months. Typically, a font has several characteristics including type, size, width, weight, and serifs [37], [46]. However, for the purpose of this research, we directed our focus towards two key font characteristics namely type and size. On the web, font type usually refers to the font family that represents a certain typeface, including all sizes and styles, for example, Arial and Verdana. To style a particular family type, the size of text may be increased or decreased using the font size attribute. There are different ways to resize the font including Ems (em), pixels (px), points (pt), and percent (%). In our font size analysis, we used pixels (px) as the sole measurement for the font since it is the main measurement unit for media shown on computer screens. Moreover, web designers usually use pixels as their key measurement unit for font size. Points, however, are used to measure font size for print media. It is noted that more than half of the Arabic websites breached the principle of consistency when applying font types to the text. This means that multiple font types were used on the same web pages. Previous research, however, has already emphasized the role of consistency of user interface elements, such as font, in the overall usability of websites [35], [36]. On that basis, it is recommended that designers should adhere to the same font family to create attractive and consistent Arabic web designs. Moreover, there was no evident consensus amongst the websites in favoring a particular font type in order to display the Arabic text. More than 35 font types were used by the selected 73 websites. There are various potential theories that could explain this staggering variability; firstly, the lack of research and thereby guidelines with respect to styling Arabic text to guide the effective design of Arabic websites; secondly, the weak conformance to international design practices and recommendations; thirdly, the lack of awareness that text formatting might ultimately impact text legibility and thereby readers' mental load and overall satisfaction. Nonetheless, a pattern of font usage emerged within the Arabic websites; six font types recurred several times in the formatting of titles/headline, menus, and paragraphs. These font types were used differently according to the type of text element. For titles and menus, Droid Arabic Kufi, Tahoma, Helvetica, and Arial, emerged as favorites. It is no surprise that Droid Arabic Kufi came out as a winner for it is a unique font that reflects the Arabic calligraphy. This font gives a true feeling of the traditional Arabic scripts (Kufic scripts) that date back to the 7th century. For paragraphs, Tahoma, Helvetica, Arial and Times New Roman were used more frequently. International recommendations suggest using font type Verdana since it is one of the most legible fonts as rated by online readers of English websites [38]. Nonetheless, this font did not appear frequently in the Arabic websites that we tested. Moreover, other famous fonts that were found earlier to be legible or attractive by English readers, such as Georgia [18] and Courier [10], did not rank well in the Arabic websites. However, Arial emerged as one of the top six font types used by the Arabic websites. Previous research confirmed that font type Arial was read significantly faster and was perceived as legible [10]. Design guidelines have indicated that font size 12 points (~ 16 pixels) is the most suited size for displaying text on English websites [39]. This is somewhat disputable as subsequent research suggested the use of font size 14 points (~19 pixels) instead. For displaying the Arabic text, recommendations are inconclusive yet. For example, [17] proposed to use font 14 points (~19 pixels) whereas [4] argued that font size 16 (22 pixels) to 18 points (~24 pixels) must be used for better readability. These suggestions are in strike contrast with the current practices of the Arabic web design. More than 60% of the inspected websites used font size ranging between 12 pixels to 16 pixels (i.e. 9 points to 12 points) to style the paragraphs. Similarly, menu items are sized from 13 pixels to 14 pixels, which is still considerably illegible [17]. In respect to using images in web design, we aimed at answering the following question. What types of images are incorporated within the Saudi websites given the Saudi culture? All Arabic websites included images as part of their content. Authors in [40] and [41] suggested that graphics improve the overall attractiveness of websites. Moreover, 90% of the Arabic websites we tested included human images. This is known to improve user engagement, trust, and satisfaction [21]. However, 80% of the Arabic websites used conservative images that respect the Islamic identity and cultural values of the Saudi people. Research studies agree that web designs should adhere to the cultural values of its users to achieve success and acceptance [42], [43], [44]. Two-thirds of the websites used images of both genders. The remaining third, however, used images of men only. This could be explained by the masculine dimension of the Saudi society as indicated by Hofstede scores [45]. More than 80% of the websites included images containing groups of people. Again, this can be justified by the collectivism dimension of the Saudi society [45]. Finally, there was high usage of animations, possibly, in an attempt to engage web visitors and improve the aesthetics of the websites. Evidently, the high presence of images in these websites contributed to lower performance scores and increased site loading time. Optimization of images‟ size, using the appropriate image compression techniques, is therefore recommended as a result of our analysis. Authors:
(1) Abdallah Namoun, Department of Information Systems, Faculty of Computer and Information Systems, Madinah, Saudi Arabia;
(2) Ahmad B. Alkhodre, Department of Information Technology, Faculty of Computer and Information Systems, Madinah, Saudi Arabia. This paper is available on arxiv under CC BY 4.0 DEED license. This is Part 6 of a 7-part series of the research paper “Towards Usability Guidelines for the Design of Effective Arabic Websites”. Use the table of links below to navigate to the other parts. This is Part 6 of a 7-part series of the research paper “ Towards Usability Guidelines for the Design of Effective Arabic Websites ”. Use the table of links below to navigate to the other parts. Part 6 Towards Usability Guidelines for the Design of Effective Arabic Websites Table of Links Abstract and I. Introduction Abstract and I. Introduction II. Previous Works II. Previous Works III. Research Methodology III. Research Methodology IV. Font and Image Inspection Studies IV. Font and Image Inspection Studies V. Results V. Results VI. Discussion VI. Discussion VII. Conclusions, Future Research Directions, and References VII. Conclusions, Future Research Directions, and References VI. DISCUSSION Our comprehensive inspection of current practices concerning text formatting and image usage within the top visited Arabic websites in Saudi Arabia revealed various interesting findings. The usability inspection was performed to gather insights about the font types and sizes used to style Arabic text, especially that they could directly influence the legibility of text on websites. Let us review and attempt to answer the research questions posited in the introduction. What font types and font sizes are used by major Arabic websites? On the positive side, the selection of the Arabic websites included a large pool of websites, which were selected objectively by relying on Alexa Internet latest ranking. In effect, the inspection covered a total of 73 top viewed websites (52 local and 21 international) in Saudi Arabia. The websites' ranking is based on a set of metrics, primarily the average viewing time for the last three months. Typically, a font has several characteristics including type, size, width, weight, and serifs [37], [46]. However, for the purpose of this research, we directed our focus towards two key font characteristics namely type and size. On the web, font type usually refers to the font family that represents a certain typeface, including all sizes and styles, for example, Arial and Verdana. To style a particular family type, the size of text may be increased or decreased using the font size attribute. There are different ways to resize the font including Ems (em), pixels (px), points (pt), and percent (%). In our font size analysis, we used pixels (px) as the sole measurement for the font since it is the main measurement unit for media shown on computer screens. Moreover, web designers usually use pixels as their key measurement unit for font size. Points, however, are used to measure font size for print media. It is noted that more than half of the Arabic websites breached the principle of consistency when applying font types to the text. This means that multiple font types were used on the same web pages. Previous research, however, has already emphasized the role of consistency of user interface elements, such as font, in the overall usability of websites [35], [36]. On that basis, it is recommended that designers should adhere to the same font family to create attractive and consistent Arabic web designs. Moreover, there was no evident consensus amongst the websites in favoring a particular font type in order to display the Arabic text. More than 35 font types were used by the selected 73 websites. There are various potential theories that could explain this staggering variability; firstly, the lack of research and thereby guidelines with respect to styling Arabic text to guide the effective design of Arabic websites; secondly, the weak conformance to international design practices and recommendations; thirdly, the lack of awareness that text formatting might ultimately impact text legibility and thereby readers' mental load and overall satisfaction. Nonetheless, a pattern of font usage emerged within the Arabic websites; six font types recurred several times in the formatting of titles/headline, menus, and paragraphs. These font types were used differently according to the type of text element. For titles and menus, Droid Arabic Kufi, Tahoma, Helvetica, and Arial, emerged as favorites. It is no surprise that Droid Arabic Kufi came out as a winner for it is a unique font that reflects the Arabic calligraphy. This font gives a true feeling of the traditional Arabic scripts (Kufic scripts) that date back to the 7th century. For paragraphs, Tahoma, Helvetica, Arial and Times New Roman were used more frequently. International recommendations suggest using font type Verdana since it is one of the most legible fonts as rated by online readers of English websites [38]. Nonetheless, this font did not appear frequently in the Arabic websites that we tested. Moreover, other famous fonts that were found earlier to be legible or attractive by English readers, such as Georgia [18] and Courier [10], did not rank well in the Arabic websites. However, Arial emerged as one of the top six font types used by the Arabic websites. Previous research confirmed that font type Arial was read significantly faster and was perceived as legible [10]. Design guidelines have indicated that font size 12 points (~ 16 pixels) is the most suited size for displaying text on English websites [39]. This is somewhat disputable as subsequent research suggested the use of font size 14 points (~19 pixels) instead. For displaying the Arabic text, recommendations are inconclusive yet. For example, [17] proposed to use font 14 points (~19 pixels) whereas [4] argued that font size 16 (22 pixels) to 18 points (~24 pixels) must be used for better readability. These suggestions are in strike contrast with the current practices of the Arabic web design. More than 60% of the inspected websites used font size ranging between 12 pixels to 16 pixels (i.e. 9 points to 12 points) to style the paragraphs. Similarly, menu items are sized from 13 pixels to 14 pixels, which is still considerably illegible [17]. In respect to using images in web design, we aimed at answering the following question. What types of images are incorporated within the Saudi websites given the Saudi culture? All Arabic websites included images as part of their content. Authors in [40] and [41] suggested that graphics improve the overall attractiveness of websites. Moreover, 90% of the Arabic websites we tested included human images. This is known to improve user engagement, trust, and satisfaction [21]. However, 80% of the Arabic websites used conservative images that respect the Islamic identity and cultural values of the Saudi people. Research studies agree that web designs should adhere to the cultural values of its users to achieve success and acceptance [42], [43], [44]. Two-thirds of the websites used images of both genders. The remaining third, however, used images of men only. This could be explained by the masculine dimension of the Saudi society as indicated by Hofstede scores [45]. More than 80% of the websites included images containing groups of people. Again, this can be justified by the collectivism dimension of the Saudi society [45]. Finally, there was high usage of animations, possibly, in an attempt to engage web visitors and improve the aesthetics of the websites. Evidently, the high presence of images in these websites contributed to lower performance scores and increased site loading time. Optimization of images‟ size, using the appropriate image compression techniques, is therefore recommended as a result of our analysis. Authors: (1) Abdallah Namoun, Department of Information Systems, Faculty of Computer and Information Systems, Madinah, Saudi Arabia; (2) Ahmad B. Alkhodre, Department of Information Technology, Faculty of Computer and Information Systems, Madinah, Saudi Arabia. Authors : Authors (1) Abdallah Namoun, Department of Information Systems, Faculty of Computer and Information Systems, Madinah, Saudi Arabia; (2) Ahmad B. Alkhodre, Department of Information Technology, Faculty of Computer and Information Systems, Madinah, Saudi Arabia. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv available on arxiv

Design Arabic Websites with Consistent Fonts, Optimal Sizes, and Culturally Aligned Images

RateX Goes Live on Solana Mainnet on 17th Oct 2024. RateX, the world’s first leveraged yield exchange, has officially launched on the Solana mainnet, marking a significant milestone in the decentralized finance (DeFi) space. Built with innovation at its core, RateX enables users to trade synthetic yield tokens (YT) of various yield-bearing assets (YBA) with leverage of up to 10x. This powerful platform introduces groundbreaking solutions to maximize capital efficiency while addressing common DeFi challenges, such as impermanent loss and limited yield-trading opportunities. RateX, a proud recipient of a grant from the Solana Foundation and the first-place winner of the Solana Renaissance Hackathon (MCM), aims to push the boundaries of decentralized finance. The platform’s unique value proposition comes from its use of synthetic yield tokens, which track the yield of different assets and allow traders to take positions with leveraged exposure. This gives users more flexibility and opportunities to capitalize on fluctuations in yields, empowering them to generate higher returns on their investments. The introduction of RateX also brings a cutting-edge Time-Decaying Automated Market Maker (AMM) to the forefront. This AMM is designed to improve capital efficiency while reducing the risks traditionally associated with liquidity provisioning, particularly in expiring contracts. The mechanism minimizes the potential for impermanent loss—a common issue for liquidity providers (LPs) in volatile markets—and aims to eventually eliminate it altogether. As a result, RateX offers a more robust and rewarding environment for both traders and liquidity providers. In addition to its leveraged yield trading capabilities, RateX caters to users looking for stability by providing fixed-yield options. These options allow users to lock in specific yields over a predetermined period, offering more predictable returns compared to the highly variable DeFi market. Whether traders are seeking high-risk, high-reward opportunities or more stable, consistent gains, RateX provides an ecosystem that accommodates diverse trading strategies. Key Partnerships with Jupiter and Jito Further bolstering its platform, RateX has announced strategic partnerships with two prominent players in the Solana ecosystem—Jupiter and Jito. These partnerships not only strengthen RateX’s technological infrastructure but also extend its reach within the DeFi and staking communities. Jupiter, one of the largest decentralized trading platforms on Solana, is known for its high liquidity and strong governance framework. With an extensive community and a focus on enabling smooth cross-asset trading, Jupiter plays a pivotal role in bringing liquidity to the RateX platform. This partnership ensures that RateX users will benefit from deep liquidity pools, efficient price discovery, and seamless execution of trades. Jupiter’s active governance community also opens the door for collaborative development efforts, allowing RateX to evolve in line with community-driven innovations. On the staking side, RateX has joined forces with Jito, the leading liquid staking platform on Solana. Jito’s expertise in liquid staking will enhance RateX’s offerings by providing users with access to staking-derived yield-bearing assets. By leveraging Jito’s technology, RateX can integrate liquid staking into its synthetic yield ecosystem, offering even more diverse options for traders to earn yield. With this partnership, users can expect enhanced staking efficiency, better capital utilization, and a broader range of yield-bearing assets available for leveraged trading. A New Era for DeFi: RateX as a Pillar of Infrastructure The launch of RateX comes at a time when the DeFi sector is evolving rapidly, and the demand for more sophisticated financial products is higher than ever. By pioneering leveraged yield trading and forming key partnerships with Jupiter and Jito, RateX is positioning itself as a crucial pillar of the Solana ecosystem and DeFi infrastructure at large. RateX’s mission is to redefine how yield movements are traded, making the entire process more efficient, transparent, and accessible to both seasoned traders and new entrants to the space. As the world’s first leveraged yield exchange, RateX is primed to play a central role in shaping the next phase of decentralized finance, often referred to as the “next DeFi summer.” With its innovative platform, strong partnerships, and focus on capital efficiency, RateX is setting the stage for a new era of financial freedom and opportunities for DeFi enthusiasts worldwide. Learn more about RateX and join the future of decentralized finance: Website: rate-x.io Twitter: RateX_Dex Discord: RateX Community RateX Goes Live on Solana Mainnet on 17th Oct 2024. RateX, the world’s first leveraged yield exchange, has officially launched on the Solana mainnet, marking a significant milestone in the decentralized finance (DeFi) space. Built with innovation at its core, RateX enables users to trade synthetic yield tokens (YT) of various yield-bearing assets (YBA) with leverage of up to 10x. This powerful platform introduces groundbreaking solutions to maximize capital efficiency while addressing common DeFi challenges, such as impermanent loss and limited yield-trading opportunities. RateX Goes Live on Solana Mainnet on 17th Oct 2024. RateX, a proud recipient of a grant from the Solana Foundation and the first-place winner of the Solana Renaissance Hackathon (MCM), aims to push the boundaries of decentralized finance. The platform’s unique value proposition comes from its use of synthetic yield tokens, which track the yield of different assets and allow traders to take positions with leveraged exposure. This gives users more flexibility and opportunities to capitalize on fluctuations in yields, empowering them to generate higher returns on their investments. The introduction of RateX also brings a cutting-edge Time-Decaying Automated Market Maker (AMM) to the forefront. This AMM is designed to improve capital efficiency while reducing the risks traditionally associated with liquidity provisioning, particularly in expiring contracts. The mechanism minimizes the potential for impermanent loss—a common issue for liquidity providers (LPs) in volatile markets—and aims to eventually eliminate it altogether. As a result, RateX offers a more robust and rewarding environment for both traders and liquidity providers. Time-Decaying Automated Market Maker (AMM) In addition to its leveraged yield trading capabilities, RateX caters to users looking for stability by providing fixed-yield options. These options allow users to lock in specific yields over a predetermined period, offering more predictable returns compared to the highly variable DeFi market. Whether traders are seeking high-risk, high-reward opportunities or more stable, consistent gains, RateX provides an ecosystem that accommodates diverse trading strategies. Key Partnerships with Jupiter and Jito Key Partnerships with Jupiter and Jito Further bolstering its platform, RateX has announced strategic partnerships with two prominent players in the Solana ecosystem—Jupiter and Jito. These partnerships not only strengthen RateX’s technological infrastructure but also extend its reach within the DeFi and staking communities. Jupiter, one of the largest decentralized trading platforms on Solana, is known for its high liquidity and strong governance framework. With an extensive community and a focus on enabling smooth cross-asset trading, Jupiter plays a pivotal role in bringing liquidity to the RateX platform. This partnership ensures that RateX users will benefit from deep liquidity pools, efficient price discovery, and seamless execution of trades. Jupiter’s active governance community also opens the door for collaborative development efforts, allowing RateX to evolve in line with community-driven innovations. On the staking side, RateX has joined forces with Jito, the leading liquid staking platform on Solana. Jito’s expertise in liquid staking will enhance RateX’s offerings by providing users with access to staking-derived yield-bearing assets. By leveraging Jito’s technology, RateX can integrate liquid staking into its synthetic yield ecosystem, offering even more diverse options for traders to earn yield. With this partnership, users can expect enhanced staking efficiency, better capital utilization, and a broader range of yield-bearing assets available for leveraged trading. A New Era for DeFi: RateX as a Pillar of Infrastructure A New Era for DeFi: RateX as a Pillar of Infrastructure The launch of RateX comes at a time when the DeFi sector is evolving rapidly, and the demand for more sophisticated financial products is higher than ever. By pioneering leveraged yield trading and forming key partnerships with Jupiter and Jito, RateX is positioning itself as a crucial pillar of the Solana ecosystem and DeFi infrastructure at large. RateX’s mission is to redefine how yield movements are traded, making the entire process more efficient, transparent, and accessible to both seasoned traders and new entrants to the space. As the world’s first leveraged yield exchange, RateX is primed to play a central role in shaping the next phase of decentralized finance, often referred to as the “next DeFi summer.” With its innovative platform, strong partnerships, and focus on capital efficiency, RateX is setting the stage for a new era of financial freedom and opportunities for DeFi enthusiasts worldwide. Learn more about RateX and join the future of decentralized finance: Learn more about RateX and join the future of decentralized finance: Website: rate-x.io Website: rate-x.io Twitter: RateX_Dex Twitter: RateX_Dex Discord: RateX Community Discord: RateX Community

RateX Goes Live on Solana: The First Leveraged Yield Exchange is Now Jupiter & Jito Aligned 

RateX geht auf Solana live: Die erste Leveraged Yield Exchange ist jetzt auf Jupiter und Jito ausgerichtet

RateX se lanza en línea en Solana: el primer exchange con rendimiento apalancado ahora está alineado con Jupiter y Jito

RateX در Solana فعال می شود: اولین تبادل بازدهی اهرمی اکنون مشتری و جیتو هم تراز شده است 

RateX-ը գործարկվում է Solana-ում. Առաջին լծակային եկամտաբերության փոխանակումն այժմ հավասարեցված է Յուպիտերին և Ջիտոնին 

RateX Hadir di Solana: Bursa Hasil Leverage Pertama Kini Selaras dengan Jupiter & Jito

RateX va in diretta su Solana: il primo exchange di rendimento con leva finanziaria è ora allineato con Jupiter e Jito

RateXがSolanaで稼働開始: 初のレバレッジ利回り取引所がJupiterとJitoと提携

RateX Solana дээр ажиллаж байна: Анхны хөшүүрэгтэй өгөөжийн бирж нь одоо Бархасбадь ба Житотой нийцэж байна 

RateX Bersiaran Langsung di Solana: Pertukaran Hasil Leveraj Pertama Kini Diselaraskan dengan Musytari & Jito 

RateX gaat live op Solana: de eerste leveraged yield exchange is nu Jupiter & Jito Aligned

RateX Solanapi kawsan: Ñawpaq kaq Intercambio de Rendimiento Apalancado Kunanqa Jupiter & Jito Alineado kachkan 

RateX Goes Live on Solana: Ang Unang Leveraged Yield Exchange ay Jupiter at Jito Aligned na Ngayon 

A lot of remote teams start using Slack as a watering hole for getting everyone together to discuss what's going on and what'll happen next. Having all of these discussions happening in the #general channel is way too noisy. So teams start creating a bunch of random channels devoted to a specific purpose. Here are some of the channels we have at Hacker Noon. - We talk about stats, KPIs and really anything we can easily quantify that correlates to the success of HN. #analytics - Easy way for us to gather discussions about HN from any random corner of the internet. #aroundtheweb - Talk about how we build and maintain our infrastructure. #backend - Where we talk about supporting and developing our . #community community forum - Discussion about how we can incorporate more blindingly green pixels into our work without actually blinding users. #design - How do we promote our author's stories to give them a broader reader base? #distribution - Discussion about how our editorial team can help take story submissions and release them into the wild. #editorial - Discussion about newsletters and emails we're sending to users based on their activity. #emails - Talk about how we can iterate on our product to make it more usable and enjoyable. #frontend - Raw feed of new releases. #git - Hiring related discussion goes here. #jobs - Where we keep the team updated on what's new. #progress - Space where we share tips and tricks. #protips - Where we talk about the product direction and challenges we're facing. #product - Where we fix bugs. #techsupport - Where we monitor the status of our tests. #testing The above list is really only a sample of our channels. As you might notice, it's natural to organize channels based on a function within the organization. If a few people start working together on something, they'll either DM or start a new channel to keep organized. Overall, I think this type of setup is pretty natural and I don't really have a problem with it. But I do feel that something is missing from this setup and I'd like to talk about the problem and a potential solution. When we organize our communication by function, people  become fixated on their immediate responsibilities, but detached from how their work fits into the bigger picture. This detachment tends to get people thinking about how they can solve problems limited to the domain they have authority over. The result is often a prioritization of developer/designer problems over user problems. When user problems aren't a priority, teams tend to invest way too much energy scaling up things that aren't needed. It's like building a massive dog house with every amenity your dog could ever need...then finding out your dog prefers to sleep on the porch. What a colossal waste of time and money. You could have just bought a comfortable mat and Rover would have been much happier. It's absolutely critical to spend time daily talking about what your users actually need. Otherwise, your team is going to be stressed out building dog houses for no good reason. In the above list of Slack channels, we're not 100% silo'ing ourselves by how we organize our communication by function. A few of the channels, such as , are setup to talk about the big picture. But these channels are often detached from the actual work being done and are often neglected. #product So we're going to try reorganizing a few of our Slack channels to hopefully inspire more user-centric discussion and hopefully break down a few of our self imposed silos. 1. We're removing a few channels such as , , and . No more watering holes for people to organize purely based on their function. #design #frontend #backend 2. We'll introduce and phase out channels based on bigger cross-functional milestones. For example, we're working on annotations functionality and have a temporary channel. This brings together designers, developers, marketers, and whoever else is interested in contributing. This gets people talking about the problem we're trying to solve. #annotations 3. We're changing up how we think about the channel to make it more of a central focus for everyone on the team. To achieve this, we're introducing a few new product channels that more closely map to our development process. 👇👇👇 #product Our development process is a funnel, not a roadmap Instead of building a timeline of things we plan to build, we've got a funnel. At the top, we've got a list of things we're thinking about building. The middle is for various stages of projects in progress. And the bottom is for deploying projects into the wild. Here are the current stages and the associated Slack channels: At this stage, we're just talking about the product. Discussion should inspire ideas about things we can do to better solve problems for our users. Discussion Stage: #product When we've developed a theory about something that has the potential to improve Hacker Noon, We post and pin an idea in this channel for further discussion. At this stage, 3 things can happen. Idea Stage: #product-ideas We immediately decide to build it. This usually happens because the value is clear and we know what needs to happen to make it a reality. In this case we fast track it past the next stage and jumps to the . Delivery Stage We decide the idea isn't worth investing in right now. We think the idea is valuable but needs to be validated before we invest energy building it. In this case, it moves to the . Discovery Stage This is for ideas we're actively exploring. Here we'll usually prototype concepts to share with the team or users. Once we're confident that a design or prototype is worth building, we'll created a pinned post in the channel. This thread should include all the deliverables needed to move onto the next stage. Discovery Stage: #product-opportunities #product-opportunities This is where active development work is happening to either introduce new features or make changes to existing features. Once a change is complete and in production, we'll add a post to the channel so everyone is aware of what is changing. Delivery Stage: #product-progress #product-progress Once we deliver a feature/change that we want to track, we should create a pinned thread in that details what we'd like to learn from the change. Maybe we're interested in seeing how the change impacts engagement, we might want to monitor cost/performance, or maybe we'd like to get qualitative feedback from users. Results Stage: #product-results #product-results Conclusion A lot of people view Slack as a distraction that just gets in the way of real work from getting done. It certainly can have that effect if you aren't careful. Slack can also be a tool that reinforces silos and discourages collaboration. Slack can also be an invaluable tool that helps people work together to bring ideas to life. It's just not possible to get the same level of communication through emails, meetings, and other traditional means. A little organization along with a little discipline can go a long way.

Keeping Our Ducks Aligned With Slack

Intro Nearly all markets at some point go through a bubble. But not all bubbles are the same, and some of them tend to get more attention due to various events, hype or the level of a potential impact on the global economy. Just for a perspective, at some point of their development, the following stocks had a significant drop in value: GOPRO (GoPro): -95%, FIT (FitBit): -92%; LC (LendingClub): -91%; SNAP (most known as a creator of Snapchat): -83%; ZNGA (Zynga): -77%; TWTR (Twitter): -63%; SPOT (Spotify): -48%; DOCU (Docusign): -45%; FB (Facebook): -42%. Perhaps the most conspicuous analogy of an infamous crypto crash is the dot-com bubble. Crypto market cap fell by 80% to its high, while Nasdaq Composite Index’s fell by 78% during the dot-com crash. Other similarities include explosive growth and grandiose expectations of the capabilities of underlying technologies. Does this mean that crypto market is following the pattern of the dot-com bubble? Indeed, the dot-com bubble and the hypothetical crypto bubble share many striking similarities. Introduction Dot-com In the early 1990s, the internet was fast emerging and by mid-1990s a rapid increase in the households computer ownership rose from 15% to 35%. The sense of the “next big thing” incentivized the launch of internet-based companies met with significant injections of investments thanks to the low interest rates. Crypto In the early 2010s, bitcoin and its underlying technology blockchain were fast emerging. A few years later, increasing publicity, a rush to make a quick fortune, and the idea of a decentralized and free-of-the-government-involvement financial system incentivized the launch of many blockchain-based companies, met with significant injections of investments thanks to the economic stability in some countries and the lack of thereof in others. Investment FOMO Dot-com Adding .com to a company’s name significantly increased the companies’ valuations. Most of these companies didn’t have profits, but pure speculation and a lack of a robust fundamental basis led to their significant overvaluation and the interest of the investors sensing the ‘next big thing’. Crypto Adding ‘crypto’ or ‘blockchain’ words to a company’s name significantly the companies’ valuations. Most of these companies didn’t have a product, robust business model or profits. However, pure speculation and a lack of a robust fundamental basis led to their significant overvaluation and the interest of the investors sensing the ‘next big thing’. increased IPO and ICO IPO: Dot-com As the bubble grew, Investment banks got involved, profiting greatly from initial public offerings (IPO) and subsequently, touting further investment. ICO: Crypto As the bubble grew, the coin issuance companies got involved, profiting greatly from initial coin offerings (ICO) that resulted in 8.7bn in total funding. Peak Dot-com Many people had started trading full time and investors were turning into millionaires overnight. Between 1995 and 2000, the Nasdaq composite stock market index rose 400 to an all-time reaching the value of over in March of 2000. % $6 trillion Crypto Between 2016 and 2017, the price of bitcoin and many other coins kept rising, and by the end of 2017 the Bitcoin price rose to $19k. Many investors started trading full time and turning into millionaires overnight. Bubble Burst Dot-com Expectations were set high, the market was too excited, while many of the dot-coms were unable to come up with sustainable business models to help them prove at least some correlation between companies’ valuations and the invested funds. The dot-com bubble burst happened in March 2000, shortly after the Nasdaq index posted its high of 5048 points. This was followed by sell-offs of the stocks by large tech firms such as Intel, Dell, and Cisco causing a public panic. Crypto Expectations were set high, the market was too excited, while many of the blockchain companies were unable to develop sustainable business models which would help prove at least some correlation between companies’ valuations and the invested funds. The crypto bubble burst in December 2017, shortly after the Bitcoin price surpassed $19k. This was the breaking point for selling off the bitcoin and other crypto holdings by many investors, causing even more panic. Sequel Dot-com Within the 2-year period after the burst, the Nasdaq composite dropped from 5048.62 points to 1114.11 points, equal to 77% and a disappeared $5 trillion worth of investments. Hundreds of companies liquidated what little capital they had left, and closed by 2004. 48% of companies managed to continue their operations. Some of those that managed to stay afloat eventually surpassed their original market caps. Among those are Amazon and Google. Crypto Hundreds of companies liquidated what little capital they had left, and closed by 2019. Only some of the companies managed to continue their operations. Some of them were met with regulatory accusations. Companies that managed to stay afloat could potentially surpass their original market caps. New players are entering the industry. What’s common: dot-com & Crypto According to Morgan Stanley report, cryptocurrency price chart and Nasdaq index chart indeed correlate. During both of these historical events, the eagerness to make quick monetary rewards and the excitement for new disruptive technology fuelled a FOMO (fear of missing out)-based investments and the eventual high drops by 80%. Post-Bubble Dot-com In December 1999, Amazon‘s stock price hit its first and the only one for the upcoming decade, peak of $107, followed by a massive drop of ~90% to ~$6 in about a year. It took almost a decade to reach its previous highs. Today, AMZN’s price is at $1,950, which is 32000% growth from its lows $6. Crypto Following the crash and till now, the hype has fallen down and speculation got to the normal levels. Meanwhile, the companies are developing the technology and launching new products while regulators are working out the regulatory mechanisms of the industry. All these advancements lead to a slow but sure adoption. A crypto market leader Bitcoin (BTC) is currently trading at just over $5,000, down -75% from its all-time high of around $19,665 in December 2017. If we go ahead and draw a straight parallel line between the performance of these market leaders, we can predict an astonishing future for Bitcoin. Assuming we take the Bitcoin’s lowest point after all highs at $3,288.45, the AMZN’s ~32000% growth would take us to around $1,052,304 per Bitcoin. Regardless of how realistic this price point is, this dot-com comparison offers a point of view that we should consider. However, it is important to note that blindly comparing the price of one of the dot-com market leaders to Bitcoin would be immature. Is This Important? While this comparison offers an interesting and a potentially possible point of view on the future of Bitcoin and the blockchain industry, an output of any endeavor, be it new technology, a new type of currency or a completely new industry, and its value have to eventually align. Among success factors are efficiency/ technological robustness, market adoption, network effects, the scale of its impact on economies and industries, team, and a community. The scope of a provided value will determine Amazons and Googles of the blockchain industry. In the meantime, while Bitcoin is going through continuous improvements, we may very well see new market players entering the gameboard with the only desire to get the crown.

How Irrational Exuberance aligned Dot-Com and Crypto bubbles.

Authors:
(1) VINÍCIUS YU OKUBO, Dept. Electronic Systems Engineering, Polytechnic School, University of São Paulo, Brazil;
(2) KOTARO SHIMIZU, Department of Applied Physics, The University of Tokyo, Tokyo 113-8656, Japan;
(3) B. S. SHIVARAM, Department of Physics, University of Virginia, Charlottesville, Virginia 22904, USA;
(4) HAE YONG KIM, Dept. Electronic Systems Engineering, Polytechnic School, University of São Paulo, Brazil. Authors: Authors: (1) VINÍCIUS YU OKUBO, Dept. Electronic Systems Engineering, Polytechnic School, University of São Paulo, Brazil; (2) KOTARO SHIMIZU, Department of Applied Physics, The University of Tokyo, Tokyo 113-8656, Japan; (3) B. S. SHIVARAM, Department of Physics, University of Virginia, Charlottesville, Virginia 22904, USA; (4) HAE YONG KIM, Dept. Electronic Systems Engineering, Polytechnic School, University of São Paulo, Brazil. Table of Links Abstract and I. Introduction Abstract and I. Introduction II. Related Work II. Related Work III. Methodology III. Methodology IV. Experiments and Results IV. Experiments and Results V. Conclusion and References V. Conclusion and References IV. EXPERIMENTS AND RESULTS A. PERFORMANCE EVALUATION A. PERFORMANCE EVALUATION A. PERFORMANCE EVALUATION Evaluating TM-CNN poses a certain challenge because there are no benchmarks or published results for direct comparison. So, we defined that junction and terminal detections reviewed by human observers as the gold standard and compared the algorithms against this standard. For testing purposes, we annotated 5 independent images from varying experiments, using the same procedure as the training images (subsection III-E). False detections were excluded since they do not contribute to the detection evaluation. In total, the test set contained 3,551 junctions and 3,653 terminals. B. RESULTS FROM A PHYSICS PERSPECTIVE B. RESULTS FROM A PHYSICS PERSPECTIVE B. RESULTS FROM A PHYSICS PERSPECTIVE In Bi:YIG films, it is known that a perfect stripe pattern is the most energetically favorable and stable configuration of magnetic moments [4]. In this structure, the magnetic moments exhibit stripes aligning straight in one direction throughout the entire structure, resulting in large spatial coherence. Meanwhile, the experimentally obtained structures often manifest labyrinthine patterns, as illustrated by dark and bright stripes in Fig. 1. In these labyrinthine patterns, stripes are aligned within small domains but propagate in different directions between different domains, and thus the spatial coherence of stripes is smaller than the perfect stripe. Notably, junctions and terminals are defects emerging in the formation of these labyrinthine patterns, intricately linked to the growth of spatial coherence in the system. Introducing such defects into the perfect stripe structure leads to the bending and branching of the initially aligned stripes, eventually reducing spatial coherence accompanied by the transformation to labyrinthine patterns. Indeed, a plethora of defects are observed in the labyrinthine patterns shown in Fig. 2. Conversely, the elimination of these defects plays a crucial role in the transition from a quenched state to an annealed state, where the system increases the spatial coherence. Each of these defects cannot be eliminated alone by continuous deformation of the stripes due to their topological properties. They can be removed through the pair annihilation. For instance, in Fig. 2, terminals are branched from junctions and they can be removed pair wisely by reducing the branching distance. These defects, known as topological defects, are widely recognized for playing a crucial role in the formation of periodic structures. By detecting the positions of defects, the evolution of the labyrinthine patterns would be systematically quantified. Specifically in the context of magnetic labyrinthine patterns, a much smaller dataset was used to detect these defects manually [4]. Recently, persistent homology has been used to extract the topological features of the labyrinthine patterns in a systematic way [38]. However, it is still difficult to detect the positions of defects and classify them into junctions and terminals. In this article, we achieved the systematic detection. To quantify the evolution of the labyrinthine patterns, we applied the TM-CNN technique to the 444 images and investigated how the number of defects and their locations changes step by step. Fig. 6 illustrates the step dependence of the number of junctions and terminals during the demagnetization process. Results from 6 different experimental runs were averaged to estimate the number of defects. In the initial stages of the demagnetization process, there are approximately 800 defects. The number of defects sharply decreases around step 5 and remains nearly unchanged from step 10 onward. Consequently, the transition to the quenched state is inferred to start before step 5 and be completed by step 10. The reduction in the number of defects during the demagnetization process is compatible with the naive expectation based on the physical argument of a decrease in the number of defects associated with the increased spatial coherence of the stripes. Furthermore, the number of junctions and terminals are close, which is consistent with the topological argument that junctions and terminals are paired. Further quantification of the evolution of the labyrinthine patterns based on the spatial distribution of the defects (and not only on their number) is discussed in another article [6]. V. CONCLUSION In this work, we presented a new algorithm named TMCNN to detect defects in magnetic labyrinthine patterns, contributing to a pioneering analysis in material science. Our study characterized the evolution of junctions and terminals in magnetic stripes during demagnetization procedures, aiming at better understanding defect arrangement in magnetic materials [6]. TM-CNN employs a two-stage detection procedure, combining template matching for initial detection and a convolutional network classifier for refining misdetections. This approach ensures a high detection accuracy and facilitates dataset annotation through a semi-automatic procedure. In our experiments, TM-CNN exhibited performance superior to other techniques, achieving an impressive F1 score of 0.988. This high performance is mainly due to TMCNN’s ability to locate small and clustered objects. TM-CNN achieves almost 100% accuracy with a simple CNN classifier with less than half a million parameters and can be used even on computers without GPUs. While TM-CNN was developed for defect detection in labyrinthine magnetic patterns, its potential applications are not limited to this field. Future research could explore the use of TM-CNN in other domains, such as identifying bifurcations in blood vessels or adapting it to other structures that can be modeled using templates. REFERENCES [1] Masaya Uchida, Yoshinori Onose, Yoshio Matsui, and Yoshinori Tokura. Real-space observation of helical spin order. Science, 311(5759):359–361, 2006. [2] T. Garel and S. Doniach. Phase transitions with spontaneous modulation the dipolar ising ferromagnet. Phys. Rev. B, 26:325–329, Jul 1982. [3] P. Molho, J. L. Porteseil, Y. Souche, J. Gouzerh, and J. C. S. Levy. Irreversible evolution in the topology of magnetic domains (invited). Journal of Applied Physics, 61(8):4188–4193, 04 1987. [4] M. Seul, L. R. Monar, L. O’Gorman, and R. Wolfe. Morphology and local structure in labyrinthine stripe domain phase. Science, 254(5038):1616– 1618, 1991. [5] Naoto Nagaosa and Yoshinori Tokura. Topological properties and dynamics of magnetic skyrmions. Nature Nanotechnology, 8(12):899–911, Dec 2013. [6] Kotaro Shimizu, Vinicius Yu Okubo, Rose Knight, Ziyuan Wang, Joseph Burton, Hae Yong Kim, Gia-Wei Chern, and B. S. Shivaram. Machine Learning Assisted Characterization of Labyrinthine Pattern Transitions. arXiv:2311.10558, 2023. [7] Kai Briechle and Uwe D. Hanebeck. Template matching using fast normalized cross correlation. In David P. Casasent and Tien-Hsin Chao, editors, Optical Pattern Recognition XII, volume 4387, pages 95 – 102. International Society for Optics and Photonics, SPIE, 2001. [8] P. Viola and M. Jones. Rapid object detection using a boosted cascade of simple features. In Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001, volume 1, pages I–I, 2001. [9] N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), volume 1, pages 886–893 vol. 1, 2005. [10] Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural networks. 25, 2012. [11] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition, 2015. [12] Ross Girshick, Jeff Donahue, Trevor Darrell, and Jitendra Malik. Rich feature hierarchies for accurate object detection and semantic segmentation, 2014. [13] M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, and A. Zisserman. The pascal visual object classes (voc) challenge. International Journal of Computer Vision, 88(2):303–338, June 2010. [14] Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region proposal networks, 2016. [15] Joseph Redmon, Santosh Divvala, Ross Girshick, and Ali Farhadi. You only look once: Unified, real-time object detection, 2016. [16] Yang Liu, Peng Sun, Nickolas Wergeles, and Yi Shang. A survey and performance evaluation of deep learning methods for small object detection. Expert Systems with Applications, 172:114602, 2021. [17] Ricardo H Maruta, Hae Yong Kim, Danilo R Huanca, and Walter J Salcedo. A new correlation-based granulometry algorithm with application in characterizing porous silicon nanomaterials. ECS Transactions, 31(1):273, 2010. [18] Hae Yong Kim, Ricardo Hitoshi Maruta, Danilo Roque Huanca, and Walter Jaimes Salcedo. Correlation-based multi-shape granulometry with application in porous silicon nanomaterial characterization. Journal of Porous Materials, 20:375–385, 2013. [19] Sidnei Alves De Araújo, Jorge Henrique Pessota, and Hae Yong Kim. Beans quality inspection using correlation-based granulometry. Engineering Applications of Artificial Intelligence, 40:84–94, 2015. [20] Gui-Song Xia, Julie Delon, and Yann Gousseau. Accurate junction detection and characterization in natural images. International Journal of Computer Vision, 106(1):31–56, Jan 2014. [21] Harry Pratt, Bryan M. Williams, Jae Yee Ku, Charles Vas, Emma McCann, Baidaa Al-Bander, Yitian Zhao, Frans Coenen, and Yalin Zheng. Automatic detection and distinction of retinal vessel bifurcations and crossings in colour fundus photography. Journal of Imaging, 4(1), 2018. [22] Sheng He, Marco Wiering, and Lambert Schomaker. Junction detection in handwritten documents and its application to writer identification. Pattern Recognition, 48(12):4036–4048, 2015. [23] M Elena Martinez-Perez, Alun D Hughes, Alice V Stanton, Simon A Thom, Neil Chapman, Anil A Bharath, and Kim H Parker. Retinal vascular tree morphology: a semi-automatic quantification. IEEE Trans Biomed Eng, 49(8):912–917, August 2002. [24] K. Liu, Y.S. Huang, and C.Y. Suen. Identification of fork points on the skeletons of handwritten chinese characters. IEEE Transactions on Pattern Analysis and Machine Intelligence, 21(10):1095–1100, 1999. [25] Chungnan Lee and Bohom Wu. A chinese-character-stroke-extraction algorithm based on contour information. Pattern Recognition, 31(6):651– 663, 1998. [26] Michael Maire, Pablo Arbelaez, Charless Fowlkes, and Jitendra Malik. Using contours to detect and localize junctions in natural images. In 2008 IEEE Conference on Computer Vision and Pattern Recognition, pages 1–8, 2008. [27] Ran Su, Changming Sun, and Tuan D. Pham. Junction detection for linear structures based on hessian, correlation and shape information. Pattern Recognition, 45(10):3695–3706, 2012. [28] R. Deriche and T. Blaszka. Recovering and characterizing image features using an efficient model based approach. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 530–535, 1993. [29] He Zhao, Yun Sun, and Huiqi Li. Retinal vascular junction detection and classification via deep neural networks. Computer Methods and Programs in Biomedicine, 183:105096, 2020. [30] Kaiming He, Georgia Gkioxari, Piotr Dollár, and Ross B. Girshick. Mask R-CNN. CoRR, abs/1703.06870, 2017. [31] I B Puchalska, G A Jones, and H Jouve. A new aspect on the observation of domain structure in garnet epilayers. J. Phys. D: Appl. Phys., 11(15):L175, oct 1978. [32] Hae Yong Kim and Sidnei Alves de Araújo. Grayscale template-matching invariant to rotation, scale, translation, brightness and contrast. In Domingo Mery and Luis Rueda, editors, Advances in Image and Video Technology, pages 100–113, Berlin, Heidelberg, 2007. Springer Berlin Heidelberg. [33] J. P. Lewis. Fast normalized cross-correlation. Vision Interface, 95:120, 1995. [34] Sidnei Alves de Araújo and Hae Yong Kim. Ciratefi: An rst-invariant template matching with extension to color images. Integr. Comput.-Aided Eng., 18(1):75–90, jan 2011. [35] Hae Yong Kim. Rotation-discriminating template matching based on fourier coefficients of radial projections with robustness to scaling and partial occlusion. Pattern Recognition, 43(3):859–872, 2010. [36] Li Deng. The mnist database of handwritten digit images for machine learning research [best of the web]. IEEE Signal Processing Magazine, 29(6):141–142, 2012. [37] Christian Eggert, Stephan Brehm, Anton Winschel, Dan Zecha, and Rainer Lienhart. A closer look: Small object detection in faster r-cnn. In 2017 IEEE International Conference on Multimedia and Expo (ICME), pages 421–426, 2017. [38] T. Yamada, Y. Suzuki, C. Mitsumata, K. Ono, T. Ueno, I. Obayashi, Y. Hiraoka, and M. Kotsugi. Visualization of topological defect in labyrinth magnetic domain by using persistent homology. Vacuum and Surface Science, 62(3):153–160, 3 2019. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

What Can Defects Tell Us About Magnetic Materials? TM-CNN's Defect Whisperer Approach

The pace of technology adoption and innovation continues to reshape industries, workflows, and expectations. From advancements in DevOps and artificial intelligence (AI) to increasing security concerns, organizations are under pressure to adapt to and take advantage of these technology opportunities to be successful in this dynamic environment. Incremental Gains and the Role of AI in DevOps Incremental Gains and the Role of AI in DevOps Progress is rarely achieved in a single flip of a switch. Rather, success is built through consistent improvements made over time. The cycle times of those compounding improvements are shortening with key technology advancements in areas like AI. But the teams and businesses having the most success with AI are applying this same incremental improvement methodology. This year, one focus area across DevOps teams will be the growing emphasis on incremental productivity gains. More specifically, regular interdepartmental progress updates – whether on a monthly or quarterly basis – will be essential to build momentum and ensure that teams stay aligned with wider business objectives. As a result, agile principles with focuses on continuous improvement are becoming more integral to DevOps workflows. In this context, technological advancements can help play a role by enabling smarter, more efficient processes that deliver more value, faster. In fact, research found that DevOps teams are actively leveraging AI to do just that – some of the key challenges this technology is being used to address include skills gaps, cost reduction, and software quality. In terms of team efficiency, AI-augmented DevOps tools could help save teams over 40 hours per month – equivalent to an entire workweek – which frees up time for teams to focus on longer-term strategy and other focal areas of their work. research research Throughout this year, AI will continue to play a key role in helping organizations make consistent gains, while also maintaining scalability and confidence in their systems. Security Concerns in Generative AI-Driven Systems Security Concerns in Generative AI-Driven Systems While generative AI offers transformative business opportunities, it also requires careful oversight. Given that these systems create outputs based on learned patterns and probabilities, unexpected responses are somewhat inevitable. The responsible adoption of generative AI hinges on robust risk management, in which testing plays a key role. It is clear that companies not embracing generative AI tools in any capacity will be left behind. However, this has led to many companies and their employees leveraging AI without first ensuring the proper knowledge and safeguards have been established, potentially resulting in irresponsible and unsafe usage. As this technology continues to be integrated across enterprise operations, it is expected that security concerns will become more prominent. Cybersecurity risks, including the potential for internal corporate information to be leaked through public models, are pressing issues that demand attention. The broader issue of information leakage across digital ecosystems also warrants attention. Outdated documentation and accessible URLs will remain vulnerabilities, highlighting the need for thorough data management and ongoing vigilance in securing sensitive information. Beyond technical checkpoints, responsible AI adoption requires a continuous cycle of risk assessment, testing, and mitigation that includes societal and ethical considerations – ranging from data bias to compliance. AI Regulations – Legislation Remains Top of Mind Despite Recent Vetoes AI Regulations – Legislation Remains Top of Mind Despite Recent Vetoes While the U.S. still does not have comprehensive legislation that directly regulates AI, including generative AI, last year saw steps taken in states across the country that could ultimately change the landscape of the technology's development and deployment. California made headlines when legislatures issued an AI safety bill with the goal of holding AI companies accountable for any harm their models may cause. Yet, critics were quick to note the bill could actually stifle AI innovation, a concern that ultimately resulted in the governor shutting down the bill. The veto of California’s AI safety bill uncovered the complexities of balancing innovation with accountability when it comes to this technology—and we still don’t know its full potential. AI safety bill AI safety bill California’s AI safety bill California’s AI safety bill So, what can companies looking to continue innovating with AI do in the meantime? In today’s dynamic, AI-driven environments, manual testing cannot keep pace. AI-powered testing is no longer optional; it is essential for ensuring confidence in every release. So when it comes to navigating this evolving technology landscape, business leaders must embrace intelligent, adaptive testing solutions to automate complex scenarios.​ By automating complex testing scenarios and leveraging predictive insights, organizations can scale quality practices to match the speed of development. In shifting how software is built – embedding continuous quality from start to finish – businesses can ensure every line of code is reliable and every release meets customer expectations. With this speed comes complexity, though. Therefore, to ensure these current – and future – generative AI systems remain responsible and secure, deployment methods will rely on models that prioritize security, data privacy, and compliance at every stage. Transition to Cloud Picks up Speed Transition to Cloud Picks up Speed A continued focus area for today’s businesses is the acceleration of application migration to the cloud. Insight from Gartner reveals that end-user cloud spending is set to increase 22 percent globally, largely driven by rapid enterprise AI adoption. As a result, organizations will continue to weather the pressures to adopt faster release cycles that are driven by competition and the demands of modern environments. Gartner Gartner For SaaS vendors, this acceleration should come as no surprise. They must continue to prioritize frequent updates, leveraging automation to help manage the complexity of cloud applications. For internal teams, the challenge lies in maintaining agility and efficiency while keeping pace with these demands. AI – when used properly – will be an instrumental tool in addressing these challenges, offering scalable solutions to streamline processes and reduce operational burdens. Welcoming the Road Ahead Welcoming the Road Ahead As organizations advance further into 2025, advancements in technology and ongoing regulations will continue to define not only the year’s key challenges, but also its opportunities. DevOps teams that continue to embrace AI, while also prioritizing incremental gains and efficiency, will be able to navigate this complex landscape with confidence. And with a strategic focus on continuous improvement, the months ahead offer a road to long-term growth and resilience.

AI Won’t Save Your Team—unless You Embrace Incremental Change

Authors:
(1) Frank Palma Gomez, from Boston University and the work done by him and Ramon during their internship in Google Research and Google DeepMind respectively;
(2) Ramon Sanabria, The University of Edinburgh and the work done by him and Frank during their internship in Google Research and Google DeepMind respectively;
(3) Yun-hsuan Sung, Google Research;
(4) Daniel Cer, Google Research;
(5) Siddharth Dalmia, Google DeepMind and Equal Advising Contributions;
(6) Gustavo Hernandez Abrego, Google Research and Equal Advising Contributions. Authors: Authors: (1) Frank Palma Gomez, from Boston University and the work done by him and Ramon during their internship in Google Research and Google DeepMind respectively; (2) Ramon Sanabria, The University of Edinburgh and the work done by him and Frank during their internship in Google Research and Google DeepMind respectively; (3) Yun-hsuan Sung, Google Research; (4) Daniel Cer, Google Research; (5) Siddharth Dalmia, Google DeepMind and Equal Advising Contributions; (6) Gustavo Hernandez Abrego, Google Research and Equal Advising Contributions. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Method 2 Method 3 Data and Tasks 3 Data and Tasks 4 Model 4 Model 5 Experiments 5 Experiments 6 Related Work 6 Related Work 7 Conclusion 7 Conclusion 8 Acknowledgements and References 8 Acknowledgements and References A Appendix A Appendix Abstract Large language models (LLMs) are trained on text-only data that go far beyond the languages with paired speech and text data. At the same time, Dual Encoder (DE) based retrieval systems project queries and documents into the same embedding space and have demonstrated their success in retrieval and bi-text mining. To match speech and text in many languages, we propose using LLMs to initialize multi-modal DE retrieval systems. Unlike traditional methods, our system doesn’t require speech data during LLM pre-training and can exploit LLM’s multilingual text understanding capabilities to match speech and text in languages unseen during retrieval training. Our multi-modal LLMbased retrieval system is capable of matching speech and text in 102 languages despite only training on 21 languages. Our system outperforms previous systems trained explicitly on all 102 languages. We achieve a 10% absolute improvement in Recall@1 averaged across these languages. Additionally, our model demonstrates cross-lingual speech and text matching, which is further enhanced by readily available machine translation data. 1 Introduction LLMs have demonstrated their effectiveness in modelling textual sequences to tackle various downstream tasks (Brown et al., 2020; Hoffmann et al., 2022; Chowdhery et al., 2023). This effectiveness has led to the development of powerful LLMs capable of modelling text in a wide range of languages. The abundance of textual data in different languages across the internet has fueled the progress of multi-lingual models (Johnson et al., 2017; Xue et al., 2020; Siddhant et al., 2022). On the other hand, speech technologies are prevalent in smartphones and personal assistants, but their language availability is relatively limited compared to the languages that LLMs support (Baevski et al., 2020; Radford et al., 2023). Various efforts have explored solutions to the speech-text data scarcity problem (Duquenne et al., 2021; Ardila et al., 2019; Wang et al., 2020). Works such as SpeechMatrix (Duquenne et al., 2022) use separate speech and text encoders to mine semantically similar utterances that are neighbors in an embedding space. However, these approaches are limiting because they require speech and text encoders that have aligned representation spaces. We posit that we can retrieve speech and text utterances by aligning both modalities within the embedding space built from a single pre-trained LLM. We take inspiration from previous works that use pre-trained LLMs to perform automatic speech recognition (ASR) and automatic speech translation (AST) (Rubenstein et al., 2023; Wang et al., 2023; Hassid et al., 2023). Our intuition is that we can perform the speech and text alignment leveraging the capabilities of text-only LLMs without requiring two separate models. In this paper, we propose converting LLMs into speech and text DE retrieval systems without requiring speech pre-training and outperform previous methods with significantly less data. By discretizing speech into acoustic units (Hsu et al., 2021), we extend our LLMs embedding layer and treat the acoustic units as ordinary text tokens. Consequently, we transform our LLM into a retrieval system via a contrastive loss allowing us to match speech and text utterances in various languages. Our contributions are the following: We build a speech-to-text symmetric DE from a pre-trained LLM. We show that our retrieval system is effective matching speech and text in 102 languages of FLEURS (Conneau et al., 2023 despite only training on 21 languages.


We show that our model exhibits cross-lingual speech and text matching without training on this type of data. At the same time, we find that cross-lingual speech and text matching is furthered improved by training on readily available machine translation data. We build a speech-to-text symmetric DE from a pre-trained LLM. We show that our retrieval system is effective matching speech and text in 102 languages of FLEURS (Conneau et al., 2023 despite only training on 21 languages. We build a speech-to-text symmetric DE from a pre-trained LLM. We show that our retrieval system is effective matching speech and text in 102 languages of FLEURS (Conneau et al., 2023 despite only training on 21 languages. We show that our model exhibits cross-lingual speech and text matching without training on this type of data. At the same time, we find that cross-lingual speech and text matching is furthered improved by training on readily available machine translation data. We show that our model exhibits cross-lingual speech and text matching without training on this type of data. At the same time, we find that cross-lingual speech and text matching is furthered improved by training on readily available machine translation data. 2 Method We train a transformer-based DE model that encodes speech and text given a dataset D = {(xi , yi)}, where xi is a speech utterance and yi is its transcription. We denote the speech and text embeddings as xi = E(xi) and yi = E(yi), respectively, where E is a transformer-based DE that encodes speech and text. 2.1 Generating Audio Tokens We convert raw speech into discrete tokens using the process in Lakhotia et al. (2021); Borsos et al. (2023). The process converts a speech query xi into an embedding using a pre-trained speech encoder. The output embedding is then discretized into a set of tokens using k-means clustering. We refer to the resulting tokens as audio tokens. We use the 2B variant of the Universal Speech Model (USM) encoder (Zhang et al., 2023) as the speech encoder and take the middle layer as the embedding for xi . Additionally, we generate audio tokens at 25Hz using k-means clustering, resulting in a set of 1024 possible audio tokens. We will refer to this as our audio token vocabulary. 2.2 Supporting Text and Audio Tokens To support text and audio tokens in our LLM, we follow the formulation of Rubenstein et al. (2023). We extend the embedding layer of a transformer decoder by a tokens, where a represents the size of our audio token vocabulary. This modification leads to an embedding layer with size (t + a) × m, where t is the number of tokens in the text vocabulary and m is the dimensions of the embedding vectors. In our implementation, the first t tokens represent text and the remaining a tokens are reserved for audio. We initialize the embeddings layer from scratch when training our model. 3 Data and Tasks Appendix A.3 details our training and evaluation datasets along with the number of languages in each dataset, the split we used, and the size of each dataset. We focus on the following retrieval tasks: Speech-to-Text Retrieval (S2T) involves retrieving the corresponding transcription from a database given a speech sample. In S2T, we train on CoVoST-2 (Wang et al., 2021) speech utterances and their transcriptions. CoVoST-2 is a large multilingual speech corpus derived from Wikipedia expanding over 21 languages and provides translation to and from English. We use FLEURS (Conneau et al., 2023) to evaluate S2T performance on 102 languages. FLEURS is an n-way parallel dataset containing speech utterances from FLoRES-101 (Goyal et al., 2021) human translations. To evaluate S2T, we report recall at 1 (R@1) rates for retrieving the correct transcription for every speech sample and word error rate (WER). Speech-to-Text Retrieval (S2T) Speech-to-Text Translation Retrieval (S2TT) attempts to retrieve the corresponding text translation of a speech sample. We use S2TT to measure the cross-lingual capabilities of our multi-modal DE retrieval system. We evaluate this capability zero-shot on X → En S2TT data of FLUERS and explore if we can further improve this capability by training on readily-available machine translation data from WikiMatrix (Schwenk et al., 2019). We pick French, German, Dutch, and Polish to English that are common across WikiMatrix and FLEURS and further discuss the amount of machine translation data used in Appendix A.3. For S2TT, we report 4-gram corpusBLEU (Post, 2018). Speech-to-Text Translation Retrieval (S2TT) 4 Model Figure 1 shows an illustration of our model. We initialize our dual encoder from PaLM 2 XXS (Google et al., 2023) and append a linear projection layer after pooling the outputs along the sequence length dimension. The embedding and linear projection layers are initialized randomly. After initializing our model from PaLM 2, we use a contrastive loss (Hadsell et al., 2006). Appendix A.1 includes more details on our training setup. We will refer to our proposed model as PaLM 2 DE. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

102 Languages, One Model: The Multimodal AI Breakthrough You Need to Know

Do you know that feeling when you create your own component instead of improving the default HTML components? Your design team created something beautiful, but browsers won't support it out of the box, and fixing it everywhere becomes a nightmare. We all know this pain, but these challenges make our job interesting. Today, I wanted to talk about one pitfall that awaits us during this thrilling journey: the placement of dropdown elements, such as the select menus or the date pickers. Absolutely Wrong At first, it looks like position: absolute solves all our problems, and it does to some extent. But then, modal windows ruin everything. If the dropdown overflows, it gets cut off. Sure, you can scroll down and see it, but you’d better pray that your designer cannot reach you with a sharp object. That’s far from perfect—we can do better. Fix It With 'fixed' If we want to display the content over everything, we need position: fixed. The only problem is that we will lose the parent element coordinates: fixed elements are quite independent by nature. This means the only thing we need to do is determine the exact coordinates of the drop-down element in these situations: When we display it.
When its content is changed.
When we scroll the window and/or the scrollable parent.
When we resize the window and/or the scrollable parent. We also need to decide whether to display it above the toggler if it is too close to the bottom of the screen. Feels doable. I will use Vue.js, but it should be easy to follow even if you prefer React or Angular. Let’s Rock This Joint Here’s the structure we’ll use: export const useDropdownAttributes = () => {
  const dropdownWidth = ref('');
  const dropdownTop = ref('');
  const dropdownBottom = ref('');
  const dropdownLeft = ref('');
  const isDirectedUpwards = ref(false);
  const togglerRect = ref<DOMRect>();
  const dropdownRect = ref<DOMRect>();

  const autodetectPosition = (
    isDropdownDisplayed: Ref<boolean>,
    togglerElement: HTMLElement | null = null,
    dropdownElement: HTMLDivElement | null = null,
    dropdownContent: Ref<unknown> | ComputedRef<unknown> = ref([]),
    isUpwardPreferred = false,
  ) => {
    // ...
  }

  return {
    autodetectPosition,
    dropdownTop,
    dropdownBottom,
    dropdownLeft,
    dropdownWidth,
    isDirectedUpwards,
    togglerRect,
    dropdownRect,
  };
}; There are four variables for the dropdown position plus the isDirectedUpwards flag and a function that updates them all. We also return two variables for the toggler’s and dropdown’s Rects: this might be convenient, for example, for the tooltips that need to be aligned to the middle of the content. As you may remember, we also need to handle the scrolling and resizing of the scrollable parent, so let’s create a function to find it: const getFirstScrollableParent = (element: HTMLElement | null): HTMLElement => {
    const parentElement = element?.parentElement;
    if (!parentElement) return document.body;

    const overflowY = window.getComputedStyle(parentElement).overflowY;

    if (overflowY === 'scroll' || overflowY === 'auto') return parentElement;

    return getFirstScrollableParent(parentElement);
  }; Now, let’s add the main function: const autodetectPosition = (
    isDropdownDisplayed: Ref<boolean>,
    togglerElement: HTMLElement | null = null,
    dropdownElement: HTMLElement | null = null,
    dropdownContent: Ref<unknown> | ComputedRef<unknown> = ref([]),
    isUpwardPreferred = false,
  ) => {
    if (!togglerElement || !dropdownElement) return;

    const updateDropdownAttributes = () => {
      togglerRect.value = togglerElement.getBoundingClientRect();
      dropdownRect.value = dropdownElement.getBoundingClientRect();
      dropdownWidth.value = `${togglerRect.value.width}px`;
      dropdownBottom.value = `${window.innerHeight - togglerRect.value.top}px`;
      dropdownTop.value = `${
        window.innerHeight - togglerRect.value.bottom - dropdownRect.value.height
      }px`;
      dropdownLeft.value = `${togglerRect.value.left}px`;
    };

    const handleResize = () => {
      requestAnimationFrame(updateDropdownAttributes);
    };

    const handleScroll = () => {
      requestAnimationFrame(updateDropdownAttributes);
    };

    watch(
      [isDropdownDisplayed, dropdownContent],
      ([newVal, _]) => {
        const scrollableParent = getFirstScrollableParent(togglerElement);
        if (!newVal) {
          window.removeEventListener('resize', handleResize);
          window.removeEventListener('scroll', handleScroll);
          scrollableParent.removeEventListener('resize', handleResize);
          scrollableParent.removeEventListener('scroll', handleScroll);
          return;
        }

        requestAnimationFrame(() => {
          const distanceFromBottom =
            window.innerHeight - togglerElement.getBoundingClientRect().bottom;
          const distanceFromTop = togglerElement.getBoundingClientRect().top;
          const dropdownHeight = dropdownElement.offsetHeight;

          isDirectedUpwards.value = isUpwardPreferred
            ? distanceFromTop > dropdownHeight
            : distanceFromBottom < dropdownHeight &&
              distanceFromTop > dropdownHeight;

          updateDropdownAttributes();
          window.addEventListener('resize', handleResize);
          window.addEventListener('scroll', handleScroll);
          scrollableParent.addEventListener('resize', handleResize);
          scrollableParent.addEventListener('scroll', handleScroll);
        });
      },
      { deep: true },
    );
  }; We pass isDropdownDisplayed and dropdownContent so we can react to their updates. We also pass togglerElement and dropdownElementwhich we need to calculate the position. Finally, there’s isUpwardPreferred in case you want the dropdown above the toggler by default. Time to Relax and Enjoy In your component, you will need something like this (I assume you have added refs to your toggler and dropdown in the template): const {
  autodetectPosition,
  dropdownTop,
  dropdownBottom,
  dropdownLeft,
  dropdownWidth,
  isDirectedUpwards,
} = useDropdownAttributes();

const togglerRef = ref<HTMLElement>();
const dropdownRef = ref<HTMLElement>();

const isDropdownShown = ref(false);

onMounted(() => {
  autodetectPosition(isDropdownShown, togglerRef.value?.$el, dropdownRef.value?.$el);
}); And the CSS will look like this: .dropdown {
  position: fixed;
  bottom: v-bind('isDirectedUpwards ? dropdownBottom : dropdownTop');
  left: v-bind('dropdownLeft');

  width: v-bind('dropdownWidth');
  min-width: 0;
} Voilà. The dropdown is displayed properly even when overflowing and moves above the toggler if there isn’t enough space below. And since we’re at the end of the article, I’d love to leave you with something cheerful—but I’m out of ideas. So, I'm afraid "good luck" is all I have this time. Good luck. 👋 You can find the complete code on GitHub. Do you know that feeling when you create your own component instead of improving the default HTML components? Your design team created something beautiful, but browsers won't support it out of the box, and fixing it everywhere becomes a nightmare. We all know this pain, but these challenges make our job interesting. Today, I wanted to talk about one pitfall that awaits us during this thrilling journey: the placement of dropdown elements, such as the select menus or the date pickers. Absolutely Wrong At first, it looks like position: absolute solves all our problems, and it does to some extent. But then, modal windows ruin everything. position: absolute If the dropdown overflows, it gets cut off. Sure, you can scroll down and see it, but you’d better pray that your designer cannot reach you with a sharp object. That’s far from perfect—we can do better. Fix It With 'fixed' 'fixed' If we want to display the content over everything, we need position: fixed . The only problem is that we will lose the parent element coordinates: fixed elements are quite independent by nature. This means the only thing we need to do is determine the exact coordinates of the drop-down element in these situations: position: fixed When we display it. When its content is changed. When we scroll the window and/or the scrollable parent. When we resize the window and/or the scrollable parent. When we display it. When its content is changed. When we scroll the window and/or the scrollable parent. When we resize the window and/or the scrollable parent. We also need to decide whether to display it above the toggler if it is too close to the bottom of the screen. Feels doable. I will use Vue.js, but it should be easy to follow even if you prefer React or Angular. Let’s Rock This Joint Here’s the structure we’ll use: export const useDropdownAttributes = () => {
  const dropdownWidth = ref('');
  const dropdownTop = ref('');
  const dropdownBottom = ref('');
  const dropdownLeft = ref('');
  const isDirectedUpwards = ref(false);
  const togglerRect = ref<DOMRect>();
  const dropdownRect = ref<DOMRect>();

  const autodetectPosition = (
    isDropdownDisplayed: Ref<boolean>,
    togglerElement: HTMLElement | null = null,
    dropdownElement: HTMLDivElement | null = null,
    dropdownContent: Ref<unknown> | ComputedRef<unknown> = ref([]),
    isUpwardPreferred = false,
  ) => {
    // ...
  }

  return {
    autodetectPosition,
    dropdownTop,
    dropdownBottom,
    dropdownLeft,
    dropdownWidth,
    isDirectedUpwards,
    togglerRect,
    dropdownRect,
  };
}; export const useDropdownAttributes = () => {
  const dropdownWidth = ref('');
  const dropdownTop = ref('');
  const dropdownBottom = ref('');
  const dropdownLeft = ref('');
  const isDirectedUpwards = ref(false);
  const togglerRect = ref<DOMRect>();
  const dropdownRect = ref<DOMRect>();

  const autodetectPosition = (
    isDropdownDisplayed: Ref<boolean>,
    togglerElement: HTMLElement | null = null,
    dropdownElement: HTMLDivElement | null = null,
    dropdownContent: Ref<unknown> | ComputedRef<unknown> = ref([]),
    isUpwardPreferred = false,
  ) => {
    // ...
  }

  return {
    autodetectPosition,
    dropdownTop,
    dropdownBottom,
    dropdownLeft,
    dropdownWidth,
    isDirectedUpwards,
    togglerRect,
    dropdownRect,
  };
}; There are four variables for the dropdown position plus the isDirectedUpwards flag and a function that updates them all. We also return two variables for the toggler’s and dropdown’s Rects: this might be convenient, for example, for the tooltips that need to be aligned to the middle of the content. isDirectedUpwards As you may remember, we also need to handle the scrolling and resizing of the scrollable parent, so let’s create a function to find it: const getFirstScrollableParent = (element: HTMLElement | null): HTMLElement => {
    const parentElement = element?.parentElement;
    if (!parentElement) return document.body;

    const overflowY = window.getComputedStyle(parentElement).overflowY;

    if (overflowY === 'scroll' || overflowY === 'auto') return parentElement;

    return getFirstScrollableParent(parentElement);
  }; const getFirstScrollableParent = (element: HTMLElement | null): HTMLElement => {
    const parentElement = element?.parentElement;
    if (!parentElement) return document.body;

    const overflowY = window.getComputedStyle(parentElement).overflowY;

    if (overflowY === 'scroll' || overflowY === 'auto') return parentElement;

    return getFirstScrollableParent(parentElement);
  }; Now, let’s add the main function: const autodetectPosition = (
    isDropdownDisplayed: Ref<boolean>,
    togglerElement: HTMLElement | null = null,
    dropdownElement: HTMLElement | null = null,
    dropdownContent: Ref<unknown> | ComputedRef<unknown> = ref([]),
    isUpwardPreferred = false,
  ) => {
    if (!togglerElement || !dropdownElement) return;

    const updateDropdownAttributes = () => {
      togglerRect.value = togglerElement.getBoundingClientRect();
      dropdownRect.value = dropdownElement.getBoundingClientRect();
      dropdownWidth.value = `${togglerRect.value.width}px`;
      dropdownBottom.value = `${window.innerHeight - togglerRect.value.top}px`;
      dropdownTop.value = `${
        window.innerHeight - togglerRect.value.bottom - dropdownRect.value.height
      }px`;
      dropdownLeft.value = `${togglerRect.value.left}px`;
    };

    const handleResize = () => {
      requestAnimationFrame(updateDropdownAttributes);
    };

    const handleScroll = () => {
      requestAnimationFrame(updateDropdownAttributes);
    };

    watch(
      [isDropdownDisplayed, dropdownContent],
      ([newVal, _]) => {
        const scrollableParent = getFirstScrollableParent(togglerElement);
        if (!newVal) {
          window.removeEventListener('resize', handleResize);
          window.removeEventListener('scroll', handleScroll);
          scrollableParent.removeEventListener('resize', handleResize);
          scrollableParent.removeEventListener('scroll', handleScroll);
          return;
        }

        requestAnimationFrame(() => {
          const distanceFromBottom =
            window.innerHeight - togglerElement.getBoundingClientRect().bottom;
          const distanceFromTop = togglerElement.getBoundingClientRect().top;
          const dropdownHeight = dropdownElement.offsetHeight;

          isDirectedUpwards.value = isUpwardPreferred
            ? distanceFromTop > dropdownHeight
            : distanceFromBottom < dropdownHeight &&
              distanceFromTop > dropdownHeight;

          updateDropdownAttributes();
          window.addEventListener('resize', handleResize);
          window.addEventListener('scroll', handleScroll);
          scrollableParent.addEventListener('resize', handleResize);
          scrollableParent.addEventListener('scroll', handleScroll);
        });
      },
      { deep: true },
    );
  }; const autodetectPosition = (
    isDropdownDisplayed: Ref<boolean>,
    togglerElement: HTMLElement | null = null,
    dropdownElement: HTMLElement | null = null,
    dropdownContent: Ref<unknown> | ComputedRef<unknown> = ref([]),
    isUpwardPreferred = false,
  ) => {
    if (!togglerElement || !dropdownElement) return;

    const updateDropdownAttributes = () => {
      togglerRect.value = togglerElement.getBoundingClientRect();
      dropdownRect.value = dropdownElement.getBoundingClientRect();
      dropdownWidth.value = `${togglerRect.value.width}px`;
      dropdownBottom.value = `${window.innerHeight - togglerRect.value.top}px`;
      dropdownTop.value = `${
        window.innerHeight - togglerRect.value.bottom - dropdownRect.value.height
      }px`;
      dropdownLeft.value = `${togglerRect.value.left}px`;
    };

    const handleResize = () => {
      requestAnimationFrame(updateDropdownAttributes);
    };

    const handleScroll = () => {
      requestAnimationFrame(updateDropdownAttributes);
    };

    watch(
      [isDropdownDisplayed, dropdownContent],
      ([newVal, _]) => {
        const scrollableParent = getFirstScrollableParent(togglerElement);
        if (!newVal) {
          window.removeEventListener('resize', handleResize);
          window.removeEventListener('scroll', handleScroll);
          scrollableParent.removeEventListener('resize', handleResize);
          scrollableParent.removeEventListener('scroll', handleScroll);
          return;
        }

        requestAnimationFrame(() => {
          const distanceFromBottom =
            window.innerHeight - togglerElement.getBoundingClientRect().bottom;
          const distanceFromTop = togglerElement.getBoundingClientRect().top;
          const dropdownHeight = dropdownElement.offsetHeight;

          isDirectedUpwards.value = isUpwardPreferred
            ? distanceFromTop > dropdownHeight
            : distanceFromBottom < dropdownHeight &&
              distanceFromTop > dropdownHeight;

          updateDropdownAttributes();
          window.addEventListener('resize', handleResize);
          window.addEventListener('scroll', handleScroll);
          scrollableParent.addEventListener('resize', handleResize);
          scrollableParent.addEventListener('scroll', handleScroll);
        });
      },
      { deep: true },
    );
  }; We pass isDropdownDisplayed and dropdownContent so we can react to their updates. isDropdownDisplayed dropdownContent We also pass togglerElement and dropdownElement which we need to calculate the position. togglerElement dropdownElement Finally, there’s isUpwardPreferred in case you want the dropdown above the toggler by default. isUpwardPreferred Time to Relax and Enjoy In your component, you will need something like this (I assume you have added refs to your toggler and dropdown in the template): const {
  autodetectPosition,
  dropdownTop,
  dropdownBottom,
  dropdownLeft,
  dropdownWidth,
  isDirectedUpwards,
} = useDropdownAttributes();

const togglerRef = ref<HTMLElement>();
const dropdownRef = ref<HTMLElement>();

const isDropdownShown = ref(false);

onMounted(() => {
  autodetectPosition(isDropdownShown, togglerRef.value?.$el, dropdownRef.value?.$el);
}); const {
  autodetectPosition,
  dropdownTop,
  dropdownBottom,
  dropdownLeft,
  dropdownWidth,
  isDirectedUpwards,
} = useDropdownAttributes();

const togglerRef = ref<HTMLElement>();
const dropdownRef = ref<HTMLElement>();

const isDropdownShown = ref(false);

onMounted(() => {
  autodetectPosition(isDropdownShown, togglerRef.value?.$el, dropdownRef.value?.$el);
}); And the CSS will look like this: .dropdown {
  position: fixed;
  bottom: v-bind('isDirectedUpwards ? dropdownBottom : dropdownTop');
  left: v-bind('dropdownLeft');

  width: v-bind('dropdownWidth');
  min-width: 0;
} .dropdown {
  position: fixed;
  bottom: v-bind('isDirectedUpwards ? dropdownBottom : dropdownTop');
  left: v-bind('dropdownLeft');

  width: v-bind('dropdownWidth');
  min-width: 0;
} Voilà. The dropdown is displayed properly even when overflowing and moves above the toggler if there isn’t enough space below. And since we’re at the end of the article, I’d love to leave you with something cheerful—but I’m out of ideas. So, I'm afraid "good luck" is all I have this time. Good luck. 👋 You can find the complete code on GitHub. You can find the complete code on GitHub . GitHub

Proper Display of Custom Dropdown Elements: A How-to Guide

العرض الصحيح لعناصر القائمة المنسدلة المخصصة: دليل إرشادي

Σωστή εμφάνιση προσαρμοσμένων αναπτυσσόμενων στοιχείων: Οδηγός πώς να το κάνετε

Presentación correcta de elementos de descarga personalizados: una guía de cómo

カスタム ドロップダウン要素の適切な表示: ハウツー ガイド

사용자 정의 드롭다운 요소의 적절한 표시: 방법 가이드

Правилно прикажување на приспособени паѓачки елементи: Водич како да се

Pontšo ya maleba ya Dielemente tša go Theoga tša Tlwaelo: Tlhahlo ya Tsela ya go Dira

Prawidłowe wyświetlanie niestandardowych elementów rozwijania: przewodnik

Одговарајући приказ прилагођених падајућих елемената: водич са упутствима

Özel Açılır Öğelerin Doğru Görüntülenmesi: Nasıl Yapılır Kılavuzu

حسب ضرورت ڈراپ ڈاؤن عناصر کا مناسب ڈسپلے: ایک گائیڈ

How Millions of Users Unwittingly Fed Their Facial Data to ChatGPT and the Consequences That Followed How Millions of Users Unwittingly Fed Their Facial Data to ChatGPT and the Consequences That Followed How Millions of Users Unwittingly Fed Their Facial Data to ChatGPT and the Consequences That Followed Picture this for once. You're scrolling through your favorite AI art generator, captivated by how your basic pictures were converted to a  beautifully rendered Ghibli-inspired portrait, and that too, within seconds. The magic of this automated image generation meets the whimsical world of Studio Ghibli, allowing you to create art with the click of a button. But beneath the surface of this captivating digital renaissance lies a troubling reality! In this article, we delve into the unsettling truth behind the AI art revolution, focusing on how millions of users have unknowingly contributed their facial recognition data to AI platforms. This is not only about the art you create; it is also about the personal data you leave behind, the possible consequences that could reshape security, and the future of digital art itself. The Rise of AI-Generated Art and the Ghibli Aesthetic The explosion of AI-generated art has been fueled by platforms like DALL·E, MidJourney, and ChatGPT and has given users an unparalleled opportunity to create stunning visuals with just a few words/prompts. One particular style that has been dominating the AI art world is the nostalgic, dreamlike quality of Studio Ghibli. Generations have been enthralled by Ghibli's whimsical charm, from Princess Mononoke to My Neighbor Totoro, and AI now makes it possible for anybody to create similar fantastical worlds. However, as more users went to these sites to produce their own Ghibli-inspired works of art, they inadvertently joined a far wider ecology of data harvesting. Not only are ChatGPT and other AI systems educated on text, but they also take in a variety of additional personal information, such as facial recognition data. Unaware of the consequences, users unintentionally enter their facial expressions, feelings, and even micro-reactions into the algorithms that drive these sites. Facial Recognition: The Invisible Price of Creative Freedom Even if AI-powered creative production is unquestionably innovative, it is critical to recognize the hidden costs that many users incur. Systems for facial recognition are modeled to examine every nuance of a person's face, which also includes emotions and minute variations in expression. Platforms like ChatGPT may be collecting biometric markers in addition to visual data when they request access to your camera or let users upload images. These markers could be utilized for purposes well beyond creating art. Every time you engage with an AI model, you’re not just interacting with a tool; you're feeding an algorithm. For millions of users, this data is not just used to improve art generation- it is stored, analyzed, and could be monetized by third-party companies. This is where things start to get murky!!! The Big Brother Potential: Facial Data and Surveillance The potential applications of facial recognition technology outside the creation of art are growing along with AI technology. What started off as a playful experiment with Ghibli-inspired graphics could eventually become a huge monitoring tool. Facial expression data can be used to forecast your behavior, customize material, and even subtly influence your emotions. Imagine this. Every smile, frown, or even the look of surprise could be analyzed and stored to create a digital profile of your emotional tendencies. Companies could then use that information to target you with ads or even influence the art you see, making it more aligned with your perceived emotional state. This, of course, assumes that facial recognition data is shared with or sold to third parties — a very real risk in an increasingly data-driven world. The Ethics of AI Art: Should We Be Concerned? At its core, the issue here isn't about privacy but ethics. Is it right for platforms to collect facial data without clear consent? Should the creators of these tools be transparent about how this information is used? The fact that many consumers are ignorant of how their data is being utilized exacerbates the issue. We frequently don't recognize that we are consenting to the processing and use of our facial data in ways we don't completely comprehend when we click "accept" on a terms-of-service agreement. It is the ethical duty of AI businesses to clarify this procedure and guarantee that customers can manage their data. Real-World Consequences: How Facial Data Could Be Misused Despite our desire to believe that our data is secure, practical examples have demonstrated that this is not the case. Facial recognition software has occasionally been used to follow people in public areas, track their emotions for advertising, and even support mass monitoring. There is a great deal of opportunity for this technology to be abused, and it frequently is. Take the increasingly popular "deepfake" technology, which makes use of AI and facial recognition to produce incredibly lifelike recordings of people speaking things they never actually uttered. Deepfakes might be amusing, but they can seriously jeopardize trust and privacy. It may soon be possible to produce deepfakes, sway public opinion, or harm reputations using the same facial data that drives art generators. Protecting Your Data: What Can You Do? Users must be more aware of their digital footprint in light of the possibilities. Here are a few steps you can actually take to protect your facial data: Be Informed: Always review the platforms' privacy policies. Verify what information they gather and how they intend to use it.
Control Your Data: Whenever feasible, take advantage of solutions that give you data management capabilities. Platforms should offer the ability to anonymize or remove your data.
Be Vigilant: Remain alert by routinely checking and updating your accounts' and devices' privacy settings. Give users access to facial recognition features with caution. Be Informed: Always review the platforms' privacy policies. Verify what information they gather and how they intend to use it. Be Informed Control Your Data: Whenever feasible, take advantage of solutions that give you data management capabilities. Platforms should offer the ability to anonymize or remove your data. Control Your Data Be Vigilant: Remain alert by routinely checking and updating your accounts' and devices' privacy settings. Give users access to facial recognition features with caution. Be Vigilant Conclusion: The Art of Balance One of the most fascinating developments in creative technology has been the emergence of AI-generated art, particularly the Ghibli-inspired designs. However, we must be aware of the hidden risks associated with this new frontier as we continue to explore it. Despite its strength, facial recognition raises a number of privacy issues that should not be disregarded. As users, we must demand more transparency and better control over how our data is used. In Parallel, companies need to take a more proactive role in safeguarding user privacy and ensuring that AI technologies are not misused for profit or surveillance. The future of AI art is bright, but we must ask ourselves: Are we willing to trade our privacy for creativity? The future of AI art is bright, but we must ask ourselves: Are we willing to trade our privacy for creativity? Would you like to take a stab at answering some of these questions? The link for the template is HERE. Interested in reading the content from all of our writing prompts? Click HERE. Would you like to take a stab at answering some of these questions? The link for the template is HERE. Interested in reading the content from all of our writing prompts? Click HERE. HERE HERE HERE. HERE

Behind the Canvas: The Hidden Cost of AI-Generated Ghibli Art

Sentient has carved out a unique path, blending blockchain technology with open-source AI to create a community-owned ecosystem. With over 1 million waitlist sign-ups for Sentient Chat in just 24 hours and a record-breaking 650,000 NFT mint for their decentralized AI model, Dobby, Sentient is redefining the future of AI. Sentient At the helm of this revolution is Himanshu Tyagi, Co-founder of Sentient and a Professor at the Indian Institute of Science, whose expertise in AI and blockchain is driving innovations like the Reasoning Agent and a Perplexity-rivaling chatbot. We sit down with Himanshu to explore the technical challenges, groundbreaking solutions, and ambitious vision behind Sentient’s rise. Ishan Pandey: Sentient Chat has exploded onto the scene with over 1 million waitlist sign-ups in under 24 hours. What technical challenges did you face in building an AI platform that’s both decentralized and consumer-friendly? Ishan Pandey: Himanshu Tyagi: Decentralization is foundational to our mission, extending far beyond just Sentient Chat. We believe that the future of AI must be open and community-driven, as decentralization uniquely empowers active community participation. This philosophy cascades directly into our products—most notably, Sentient Chat. Himanshu Tyagi: With Sentient Chat, our goal is to ensure that the community truly feels ownership. Users aren't merely passive consumers; they are creators, molders, and stewards of the platform. They have full control—from the underlying AI model and agents, to data and governance—enabling them to shape the product precisely to fit their diverse needs and use cases. Critically, decentralization and usability reinforce each other in Sentient Chat. By empowering users to customize and enhance the platform, every innovation developed by one user or team becomes accessible to the broader community. This collective contribution enriches the platform's usability, continually driving improvements and innovations that benefit everyone involved. Ultimately, decentralization doesn't complicate the user experience—it enhances it, creating a vibrant ecosystem where collective creativity and collaboration translate directly into a more powerful, flexible, and user-friendly platform. Ishan Pandey: You’ve positioned Sentient Chat as a competitor to Perplexity, powered by Dobby, the world’s first community-owned AI model. Can you walk us through the technical innovation behind Dobby and how it supports Sentient Chat’s unique capabilities? Ishan Pandey: Himanshu Tyagi: AI is rapidly becoming the intermediary between information networks and users, granting it significant power to influence user perspectives based on inherent biases. Our goal is to empower communities by enabling them to align AI models according to their specific values, thus promoting a diverse ecosystem of "Loyal AI" models. While biases will always exist due to training data, users should not be limited to generalized models that simply average internet opinions. Instead, we envision users selecting from a wide array of models aligned with distinct community values. Himanshu Tyagi: Dobby represents the first of such models, explicitly aligned with the crypto community's values—pro-crypto and pro-personal freedom. With Dobby, we've achieved three notable technical breakthroughs: Effective Value Alignment Without Performance Loss: Aligning AI models to specific values through fine-tuning is notoriously challenging, often leading to degraded performance and incoherent outputs. With Dobby, we've successfully fine-tuned the Llama model to embrace specific values while maintaining excellent performance across critical benchmarks.


Human-centric Tone and Freedom of Expression with Safety: Achieving our desired alignment required removing existing guardrails from Llama and carefully rebuilding them. Surprisingly, this process enhanced the model’s tone, resulting in more natural and human-centric interactions. Even with relaxed constraints, Dobby remains safe, as demonstrated by metrics from our "Sorry Bench." Moreover, our Dobby-mini 8B and Dobby 70B models exhibit granular control over specific safety dimensions (such as hate speech, financial, medical, and legal advice, and explicit content), allowing us precise adjustments of the model's safety profile.


Fingerprinting for the Most Distributed Model Ownership Ever: Dobby is currently the most widely distributed model, with over 660,000 owners. This unprecedented level of distribution is facilitated by our innovative fingerprinting technology. Fingerprinting injects unique key-response pairs into the model, enabling robust verification of ownership. These fingerprints have a negligible impact on overall model performance and are resilient to removal through fine-tuning, model merging, or other modifications. Effective Value Alignment Without Performance Loss: Aligning AI models to specific values through fine-tuning is notoriously challenging, often leading to degraded performance and incoherent outputs. With Dobby, we've successfully fine-tuned the Llama model to embrace specific values while maintaining excellent performance across critical benchmarks. Effective Value Alignment Without Performance Loss: Aligning AI models to specific values through fine-tuning is notoriously challenging, often leading to degraded performance and incoherent outputs. With Dobby, we've successfully fine-tuned the Llama model to embrace specific values while maintaining excellent performance across critical benchmarks. Effective Value Alignment Without Performance Loss: Human-centric Tone and Freedom of Expression with Safety: Achieving our desired alignment required removing existing guardrails from Llama and carefully rebuilding them. Surprisingly, this process enhanced the model’s tone, resulting in more natural and human-centric interactions. Even with relaxed constraints, Dobby remains safe, as demonstrated by metrics from our "Sorry Bench." Moreover, our Dobby-mini 8B and Dobby 70B models exhibit granular control over specific safety dimensions (such as hate speech, financial, medical, and legal advice, and explicit content), allowing us precise adjustments of the model's safety profile. Human-centric Tone and Freedom of Expression with Safety: Achieving our desired alignment required removing existing guardrails from Llama and carefully rebuilding them. Surprisingly, this process enhanced the model’s tone, resulting in more natural and human-centric interactions. Even with relaxed constraints, Dobby remains safe, as demonstrated by metrics from our "Sorry Bench." Moreover, our Dobby-mini 8B and Dobby 70B models exhibit granular control over specific safety dimensions (such as hate speech, financial, medical, and legal advice, and explicit content), allowing us precise adjustments of the model's safety profile. Human-centric Tone and Freedom of Expression with Safety: Fingerprinting for the Most Distributed Model Ownership Ever: Dobby is currently the most widely distributed model, with over 660,000 owners. This unprecedented level of distribution is facilitated by our innovative fingerprinting technology. Fingerprinting injects unique key-response pairs into the model, enabling robust verification of ownership. These fingerprints have a negligible impact on overall model performance and are resilient to removal through fine-tuning, model merging, or other modifications. Fingerprinting for the Most Distributed Model Ownership Ever: Dobby is currently the most widely distributed model, with over 660,000 owners. This unprecedented level of distribution is facilitated by our innovative fingerprinting technology. Fingerprinting injects unique key-response pairs into the model, enabling robust verification of ownership. These fingerprints have a negligible impact on overall model performance and are resilient to removal through fine-tuning, model merging, or other modifications. Fingerprinting for the Most Distributed Model Ownership Ever: Sentient Chat is our AI-powered search platform, designed to showcase and provide users with a direct experience of the intangible concept of Loyal AI. This platform demonstrates the competitive edge of open-source ecosystems, successfully outperforming platforms like Perplexity and matching the search benchmarks of ChatGPT. Powered by Dobby, Sentient Chat illustrates how Loyal AI can directly influence the information delivered to users, presenting content concisely and engagingly. The integration of Dobby into Sentient Chat has notably improved the efficiency and entertainment value of search interactions. With Dobby's clear, human-like, no-nonsense communication style, users can quickly and effectively access the information they seek, highlighting the platform's distinct advantage in AI-driven search experiences. Ishan Pandey: One of the most intriguing aspects of Sentient Chat is its Reasoning Agent, equipped with tools like search, calculator, and think. How did you architect this agent, and what technical hurdles did you overcome to make reasoning a core strength? Ishan Pandey: Himanshu Tyagi: Recent advancements have shown that generating executable Python code for tool calling yields a significant boost in performance compared to conventional JSON-based approaches. In particular, LLM’s are inherently adept at compressing the action-space of tasks using code. This naturally prompted the use of code as the mode of expression for our reasoning agent. Sometimes LLM's write code that does not execute/run the way it is intended to, this is remedied by a Python interpreter which checks if the code generated executes or throws errors, if it throws errors, then it will keep regenerating the code (with the previous error traces) until it outputs satisfactory code. Himanshu Tyagi: Another caveat is that the LLM may generate functional code that does not do what it is intended to do (e.g. calls the calculator tool instead of searching); the reasoning agent is capable of digesting the output of the information and assessing whether it fits with, on a general level, what it expects from the particular code/tool call. We tested two different types of agents in the Open Reasoning Agent portion of the ODS framework: Chain-of-Thought with ReAct agent and Chain-of-Code with CodeAct agent. We tested two different types of agents in the Open Reasoning Agent portion of the ODS framework: Chain-of-Thought with ReAct agent and Chain-of-Code with CodeAct agent. In the Chain-of-Thought ReAct architecture, the agent follows an iterative reasoning-action loop. First, the 'Think' step allows the agent to reason internally, articulating plans or interpreting intermediate results in natural language. Whenever the agent encounters uncertainty or lacks critical information, it invokes the 'Search' tool, actively retrieving external information from the web. For arithmetic or computational tasks that require high precision, the agent utilizes the 'Calculate' tool. Thus, CoT-ReAct continuously cycles through thinking, searching, and calculating as needed, until it arrives at a reliable solution. On the other hand, the Chain-of-Code or CodeAct agent employs executable code to perform its reasoning tasks. Initially, the agent thinks through its strategy, similar to CoT-ReAct. However, instead of relying purely on natural language, it generates executable Python code as part of its reasoning. The 'Calculate' function is seamlessly embedded here, directly integrated into the Python execution environment. This allows for accurate computations and algorithmic reasoning. Like CoT-ReAct, CodeAct also uses the 'Search' tool when additional external information is necessary, incorporating the results directly into its code-generation and execution processes. The key difference lies in how each agent interacts with these tools. For CoT-ReAct, 'Think', 'Search', and 'Calculate' are distinctly separate actions explicitly chosen during reasoning. For CodeAct, while 'Think' remains an explicit planning step, 'Calculate' is inherently built into the code execution itself, making computational tasks integral rather than external. 'Search' remains similarly explicit in both approaches, providing vital external context whenever the agent's internal knowledge is insufficient. Ishan Pandey: Sentient completed a record-breaking 650,000 NFT mint for Dobby, tying ownership to a decentralized AI model. What technical infrastructure did you build to handle this scale, and how does blockchain enhance the community ownership aspect? Ishan Pandey: Himanshu Tyagi: This was not a technically challenging problem – modern blockchains can easily handle such load. The most interesting part here is that this is the first time anybody has demonstrated this scale of “direct democracy” in model governance. In centralized companies, this governance is done by a tiny alignment team which has sometimes even modified the model to match its own beliefs. Anthropic had done some simple experiments under the name of constitutional AI with about 1000 people, but nothing close to this open governance by community. We wanted to give this power to the people and blockchains are perfect instruments for that. Dobby community already decided what kind of persona Dobby should have, leashed or unhinged (they chose unhinged over leashed). In future, all decisions about Dobby-unhinged model will be taken by this community, who will also get rewarded as model usage increases. Himanshu Tyagi: Going forward, these ownership NFTs will be converted to ownership tokens on our blockchain which can be used to govern alignment and updates to this model. Furthermore, the associated fingerprints will be committed as Merkle root to the same contract – anyone can verify the identity of the model by asking our fingerprint queries, and can check that the same queries are committed with the ownership contract. Ishan Pandey: Your Reasoning Agent integrates real-time tools and aims to tackle ambiguous, subjective tasks. How does this differ from existing reasoning agents, like OpenAI’s Deep Research, and what novel solutions did you implement to push the boundaries of AI reasoning? Ishan Pandey: Himanshu Tyagi: When comparing our open-source platform to closed-source competitors, the primary competitive advantage of closed systems has traditionally been their technology. However, platforms like Sentient Chat and Open Data Sources (ODS) have effectively leveled the playing field by democratizing AI knowledge, making our framework completely open to the community. Closed-source platforms, such as those by OpenAI, operate as black boxes; neither you, I, nor anyone else can fully comprehend their internal processes. Himanshu Tyagi: In contrast, our open framework invites continuous innovation from the wider community, enabling rapid identification and resolution of gaps that might otherwise remain unnoticed. This collaborative, open-input and open-output approach is key to building the best possible search platform. Our ultimate vision includes enabling users to contribute their own data sources and agents, thereby drastically enhancing Sentient Chat and ODS through collective, community-driven improvements. Ishan Pandey: You’ve emphasized Sentient Chat’s edge over competitors like Perplexity with its 15+ AI agent integration. What technical challenges did integrating multiple agents into a single chatbot interface present, and how did you solve them? Ishan Pandey: Himanshu Tyagi: Technically, we are addressing many complex software and design challenges to allow a broad variety of agents to participate in a singular user experience through Sentient Chat. We are also offering many tools to these agent builders such as our state-of-the-art AI search APIs, the Sentient Secure Enclave offering for building “unruggable agents,” etc. But more than addressing these technical challenges, the most interesting part here is the possibility of a completely new way to access internet knowledge. Himanshu Tyagi: For decades, Google has dominated search by focusing primarily on helping users find information online. Google's reliance on advertising revenue, rooted in recommending sources for information, inherently limits its ability to innovate beyond traditional information retrieval. AI, on the other hand, provides an opportunity to bypass mere information gathering, enabling users to directly execute tasks without needing separate analysis and action phases, which is our exact competitive advantage against the behemoth of Google. This ambitious vision requires a diverse ecosystem—one that integrates varied indexed data sources with numerous AI agents capable of performing specialized actions. The key to unlocking this potential lies in creating a transparent and open platform, actively incentivizing broad community participation. Data providers must clearly understand the value their contributions bring, while agent developers must seamlessly integrate their services to enhance overall functionality. Such an ecosystem must be community-governed, ensuring fairness, openness, and continual innovation driven by collective contribution and collaboration. By the way, a lot more than 15 agents are coming on Sentient Chat! Ishan Pandey: Looking ahead, what’s next for Sentient in terms of technical innovation? How do you plan to evolve the Reasoning Agent and expand the open AGI ecosystem? Ishan Pandey: Himanshu Tyagi: We see a future where AI becomes an intermediary between all human senses, global information, and social networks. Sentient’s vision is to make sure that this AI is loyal to us. This singular objective drives all our development. For reasoning agents, we want to make sure that developers get access to models whose reasoning capabilities and skills are aligned with their interests or the use case. Himanshu Tyagi: For instance, if you are building a sovereign agent that can control wallets, the underlying models should have reasoning capabilities to ensure that malicious actors cannot manipulate prompts and steal funds. This is alignment – we will be building such aligned models. Every developer building in the Open AGI ecosystem should have the option of using AI models that will be loyal to their application or users. Sentient Chat is the collective offering of all such agents built on Loyal AI. Don’t forget to like and share the story! Vested Interest Disclosure: This author is an independent contributor publishing via our business blogging program. HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYO Vested Interest Disclosure: This author is an independent contributor publishing via our business blogging program. HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYO Vested Interest Disclosure: business blogging program business blogging program

How Sentient’s Reasoning Agent is Outsmarting the Competition: Inside Look 

كيف يعمل عميل التفكير في Sentient على تجنب المنافسة: النظر في الداخل

Com l'agent de raonament de Sentient està superant la competència: a l'interior

Cómo el agente de razonamiento de Sentient está superando a la competencia: Inside Look

Bagaimana Agen Penalaran Sentient Mengalahkan Pesaingnya: Pandangan dari Dalam

ERROR: invalid target_lang (iw)

Sentient の推論エージェントが競合他社を凌駕する方法: 内部の分析

Како резонантниот агент на Sentient ја надмудри конкуренцијата: погледнете одвнатре 

Bagaimana Sentient's Reasoning Agent Mengabaikan Persaingan: Melihat Dalam

Imaynatachus Sentient’s Reasoning Agent atipanakuymanta aswan yachayniyuq kachkan: Ukhupi qhaway 

Как рассуждающий агент Sentient вытесняет конкуренцию: Внутренний взгляд

Sida Wakiilka Sababta Sentient uu uga Wanagsan yahay Tartanka: Gudaha Fiiri 

Sentient的推理代理如何超越竞争:内部看法

The Challenge of Liquidity Fragmentation In the decentralized finance (DeFi) ecosystem, liquidity is the lifeblood of every protocol — fueling trades, enabling lending, and powering yield generation. However, as DeFi has expanded across numerous chains, rollups, and execution environments, a new challenge has emerged: liquidity fragmentation. Liquidity fragmentation refers to the dispersion of assets across multiple isolated ecosystems. While Layer 2 solutions such as rollups have addressed scalability issues by offloading transactions from Ethereum’s mainnet, they’ve also unintentionally fractured liquidity. Assets are now siloed, and users must bridge assets to access specific pools or dApps. This fragmentation introduces inefficiencies: traders face increased slippage, protocols suffer from reduced capital efficiency, and arbitrage becomes both more prevalent and more complex. The more scattered the liquidity, the more difficult it becomes to achieve optimal execution for trades or to deploy capital effectively across DeFi. The introduction of Layer 3 solutions on Arbitrum offers a promising path forward. By enabling app-specific rollups that remain interoperable within a broader network, Layer 3 aims to unify fragmented liquidity without sacrificing performance or decentralization. At EVEDEX, we view this architectural breakthrough not only as a technical evolution, but as a strategic opportunity, one that allows us to reimagine liquidity aggregation and deliver a more seamless trading experience across the Arbitrum ecosystem. Understanding Liquidity Fragmentation in DeFi Liquidity fragmentation in decentralized finance is more than just a technical inconvenience; it’s a core structural issue that impacts both users and protocols across the ecosystem. As DeFi scales horizontally across Layer 1s, Layer 2s, and now Layer 3s, liquidity becomes siloed within each environment. Each network (Arbitrum, Optimism, zkSync, Base, and others) hosts its own decentralized exchanges, liquidity pools, and versions of key assets. Even when tokens are nominally the same (e.g. ETH on Arbitrum vs. ETH on zkSync), they are technically distinct and non-interchangeable without bridging. This fragmentation results in several inefficiencies: Higher slippage: Reduced liquidity per pool leads to poorer execution prices for traders.
Lower capital efficiency: Liquidity is spread thin across multiple chains and protocols.
Increased arbitrage overhead: Traders must actively monitor and bridge between networks, bearing gas fees and time delays.
Integration complexity: dApps must maintain multiple bridge and liquidity interfaces, increasing development and operational costs. Higher slippage: Reduced liquidity per pool leads to poorer execution prices for traders. Higher slippage Lower capital efficiency: Liquidity is spread thin across multiple chains and protocols. Lower capital efficiency Increased arbitrage overhead: Traders must actively monitor and bridge between networks, bearing gas fees and time delays. Increased arbitrage overhead Integration complexity: dApps must maintain multiple bridge and liquidity interfaces, increasing development and operational costs. Integration complexity Consider a trader looking to swap $100k USDC to ETH. They’re faced with fragmented pools across multiple L2s, varying liquidity depths, inconsistent pricing, and the uncertainty of cross-chain execution. This undermines the very promise of DeFi: efficient, permissionless, global trading. This is where the need for a new paradigm emerges, one that enables unified liquidity access across rollups and chains without sacrificing decentralization. The emergence of Arbitrum’s Layer 3 ecosystem and platforms like EVEDEX that integrate into it represents a critical step toward solving this structural flaw. Arbitrum’s Layer 3 as a Solution While Layer 2 solutions brought significant scalability to Ethereum, they also introduced an unintended side effect: more isolated environments with their own liquidity silos. Arbitrum’s Layer 3 approach aims to solve this by offering a modular, customizable framework that retains interoperability while enabling specialization. Rather than forcing projects to deploy on a monolithic rollup, Arbitrum’s Layer 3 allows teams to create tailored app-chains with direct communication to the Arbitrum L2 and Ethereum L1. This creates a unified settlement layer while preserving the flexibility to optimize for performance, privacy, or cost. For platforms like EVEDEX, this is a game-changer. For platforms like EVEDEX, this is a game-changer. By deploying on Arbitrum’s Layer 3, EVEDEX can: Build a dedicated execution environment with faster confirmation times and lower fees.
Maintain real-time connectivity to Arbitrum’s liquidity layer without bridging.
Design a trading system that aggregates liquidity across integrated protocols and Layer 3 instances.
Avoid compromising decentralization by settling trades on a secure, Ethereum-aligned architecture. Build a dedicated execution environment with faster confirmation times and lower fees. Maintain real-time connectivity to Arbitrum’s liquidity layer without bridging. Design a trading system that aggregates liquidity across integrated protocols and Layer 3 instances. Avoid compromising decentralization by settling trades on a secure, Ethereum-aligned architecture. In practical terms, EVEDEX leverages the Layer 3 model to unify fragmented liquidity across multiple environments. Traders don’t need to worry about which rollup their funds are on. The routing engine handles that under the hood, pulling from the most efficient pool available within the Arbitrum ecosystem. This creates a user experience that feels centralized (fast, smooth, and low-friction) while preserving the principles of decentralization and self-custody. It’s not just a technological solution but a structural evolution redefining how liquidity can flow across DeFi. The Role of EVEDEX in Tackling Liquidity Fragmentation EVEDEX is designed from the ground up to address the challenges of fragmented liquidity. Operating on Arbitrum’s Layer 3, EVEDEX is not just another decentralized exchange; it’s a liquidity hub built with native cross-chain logic and capital efficiency in mind. At the heart of EVEDEX is an advanced Smart Order Routing (SOR) engine that identifies the most optimal trading paths across various pools and protocols, whether on different Layer 3 chains or within Arbitrum’s core liquidity layer. This reduces slippage, boosts execution speed, and ensures best-price execution for traders. Additionally, EVEDEX integrates a modular AMM architecture, enabling dynamic liquidity provisioning that adjusts based on market conditions. Instead of isolating assets in separate pools, EVEDEX can unify or segment liquidity based on user demand, trading volume, or volatility, creating a more responsive market. Crucially, EVEDEX uses Arbitrum’s Layer 3 connectivity to implement cross-chain liquidity settlement without needing to manually bridge assets in and out. A trade can be initiated on one Layer 3 rollup and finalized across pools on another, while the user experiences this as a single seamless transaction. Let’s take a real-world example: Let’s take a real-world example: A user wants to swap ETH for USDC. Typically, they’d need to figure out which chain hosts the best rates, move their funds, and execute the trade. On EVEDEX, the SOR finds the best price across its integrated pools, executes the swap, and handles settlement, all on Arbitrum’s Layer 3, with no bridging or manual transfers. For developers, this architecture enables future integrations with other rollups and liquidity sources without rebuilding core logic. For users, it’s a more simple, faster, and more capital-efficient way to trade. And for Arbitrum, it demonstrates how Layer 3 is not just a scaling solution, it’s an innovation layer for next-gen DeFi infrastructure. Broader Impact on DeFi and the Future of Liquidity Aggregation Solving liquidity fragmentation is more than just a technical achievement, it’s a necessary evolution for the entire DeFi ecosystem. When users can access deep, unified liquidity across chains and protocols without unnecessary complexity, the barrier to entry for decentralized finance drops dramatically. It’s not just about faster trade execution or better prices; it’s about rebuilding the user experience around simplicity, security, and efficiency. Arbitrum’s Layer 3 model plays a crucial role in that transformation. By enabling customizable rollups that remain interoperable within the ecosystem, Layer 3 provides a foundation for scalable, flexible, and composable liquidity. EVEDEX builds directly on this foundation, pushing the boundaries of what’s possible in decentralized trading (smart routing, real-time settlement, adaptive AMMs) all without sacrificing decentralization or composability. The broader implication is this: projects like EVEDEX don’t just improve DeFi for power users, they make it truly accessible for the next wave of global users and institutions. Liquidity aggregation becomes the default, not a feature. Cross-chain UX becomes seamless, not fragmented. And the entire Arbitrum Layer 3 ecosystem gains a practical, real-world demonstration of how infrastructure can evolve from fast to frictionless. As the market matures and demand for smoother capital flow intensifies, we believe Layer 3 protocols like EVEDEX will define the next generation of financial rails: permissionless, composable, and deeply liquid. Written by: Written by: Written by: Vlad Komissarov, CTO of EVEDEX Bessie Liu, corporate writer at Offchain Labs (Arbitrum technology)

Liquidity Fragmentation in DeFi and How Layer 3 on Arbitrum Solves It

Likviditetsfragmentering i DeFi og hvordan lag 3 på Arbitrum løser det

Liquiditätsfragmentierung in DeFi und wie Layer 3 auf Arbitrum es löst

Fragmentación de liquidez en DeFi y cómo la capa 3 de Arbitrum la soluciona

Fragmentación da liquidez en DeFi e como a capa 3 en Arbitrum a resolve

פיצול נזילות ב-DeFi וכיצד שכבה 3 ב-Arbitrum פותר את זה

Fragmantasyon lajan likid sikile nan DeFi ak kijan Kouch 3 sou Arbitrum rezoud li

Frammentazione della liquidità in DeFi e come Layer 3 su Arbitrum la risolve

DeFi における流動性の断片化と Arbitrum のレイヤー 3 による解決方法

Ny fihenjanana ny liquidity ao amin'ny DeFi sy ny fomba fanatanterahana ny Layer 3 amin'ny Arbitrum

Liquiditeitsfragmentatie in DeFi en hoe Layer 3 op Arbitrum het oplost

په DeFi کې د مایعاتو ټوټه کول او دا چې څنګه په Arbitrum کې دریمه طبقه دا حل کوي

Ku avana ka Liquidity eka DeFi na Ndlela leyi Layer 3 eka Arbitrum yi Swi tlhantlhaka ha yona

The way users engage with apps is shifting fast — and often not in the direction most teams anticipated. What once passed as innovation is now just noise. Features that used to impress are now ignored. Entire product categories, built on bold predictions and massive funding rounds, have quietly faded away. Today, user attention is scarce, switching costs are near zero, and tolerance for clunky UX is gone. Most importantly, people no longer explore apps with curiosity; they evaluate them immediately. If the value isn’t apparent immediately, they move on. In this environment, product teams are forced to rethink almost everything: how value is communicated, how habits are formed, and how technologies like AI are integrated. It’s no longer about building more — it’s about building less but better. more less So, what exactly stopped working? What expectations are replacing old norms? And how are the most adaptive teams delivering under these new constraints? Let’s break it down. 1. What No Longer Works 1. What No Longer Works Not every technology backed by billions has managed to earn its place in everyday life. Some of the most ambitious bets of recent years — AR, VR, and voice interfaces — failed the test of real-world usage. Not every technology backed by billions has managed to earn its place in everyday life. AR and Apple Vision Pro were pitched as the next layer of interaction. In reality, they were too expensive, uncomfortable, and limited by short battery life. Everyday use just wasn’t practical. A design built to impress couldn’t deliver where comfort and endurance mattered most. The technology meant to replace screens ended up as more of a showcase than a solution. In reality, In reality, Voice assistants faced a similar fate. Siri, Alexa, and their counterparts didn’t become the default interface. Their use stayed confined to narrow scenarios — playing music, checking the weather, or helping older users talk to a speaker. That’s not real behavioral adoption. The core issue behind both is the lack of an active, recurring user base. Without it, developers have no incentive to build; without developers, there’s no ecosystem of meaningful use cases. It’s a closed loop that stops the platform from becoming truly indispensable. The core issue behind both is the lack of an active, recurring user base. Without it, developers have no incentive to build; No Device Has Replaced the Smartphone No Device Has Replaced the Smartphone Another area that hasn’t lived up to the hype is the search for the “next device” after the smartphone. Startups like Rewind tried to reinvent how we consume digital services — but the phone remains firmly in place as the main, all-purpose interface. Building specifically for Oculus or Apple Vision Pro doesn’t make sense — these platforms lack the user base to justify serious developer investment. hasn’t lived up to the hype is the search for the “next device” after the smartphone. Attempts to turn AR into the next big user interface also fell flat. Facebook poured billions into the vision — after acquiring Oculus, up to 30% of the company’s developers were focused on AR and the Metaverse. But the idea never really took off. Users stayed grounded in reality, where devices still need to be practical, autonomous, and suited for everyday use — not experimental showcases. AI Remains Untrusted AI Remains Untrusted At the same time, skepticism toward AI-generated content is growing. People have a stronger emotional response to things made by humans — even when AI delivers the same output. The rise of synthetic influencers, AI-generated photos, and machine-written texts raises new questions: Where is the line between real and fake? Should AI content be labeled? And is it even healthy to form emotional connections with a simulation? This psychological barrier is starting to matter more than the technical one. Even the most intelligent support chatbots still get the same request: “Connect me to a human.” The desire to talk to real people hasn’t gone away, which limits how far AI can scale, no matter how efficient it gets. This psychological barrier is starting to matter more than the technical one. Even the most intelligent support chatbots still get the same request: “Connect me to a human.” The desire to talk to real people hasn’t gone away, which limits how far AI can scale, no matter how efficient it gets. The Superapp Illusion The Superapp Illusion The idea of a super app — combining multiple services into one interface — was long seen as a strategic goal for digital platforms. The logic seemed straightforward: if users can solve all their needs within a single product, retention and monetization should follow. But in many cases, that logic didn’t hold. Attempts to pack everything into one place often led to bloated interfaces, unclear user flows, and declining engagement. Instead of convenience, users encountered unnecessary complexity. As a result, the trend began to shift in the opposite direction. Platforms started breaking products apart again. For example, in February, reports emerged that Instagram considered turning Reels into a standalone app. Earlier, media reports indicated that TikTok was officially expanding into long-form content territory. Platforms started breaking products apart again. in February, in February, Earlier, Earlier, The focus has shifted from platform dominance to targeted competition across specific use cases — stories, short videos, and long-form content. The new approach is clear: success isn’t about doing everything, it’s about doing one thing better than anyone else. The focus has shifted from platform dominance to targeted competition across specific use cases — stories, short videos, and long-form content. The new approach is clear: success isn’t about doing everything, it’s about doing one thing better than anyone else. The focus has shifted from platform dominance to targeted competition across specific use cases — stories, short videos, and long-form content. 2. How User Behavior Has Shifted 2. How User Behavior Has Shifted Users don’t explore apps anymore — they scan them. If something doesn’t work or make sense instantly, they’re out. The space between opening and uninstalling has collapsed to seconds, and product teams now operate within this ever-shrinking window. Users don’t explore apps anymore — they scan them Attention spans have tightened Attention spans have tightened People spend less and less time inside individual products. The competition isn’t just between apps in the same category — it’s across all digital experiences.  Every product is fighting for a share of the same limited attention. Even time spent per app is shrinking, making retention a battle that starts the moment a user lands. Users need a reason to stay — not later, but right now. Product teams have to design for habit from day one. If an app doesn’t deliver immediate value or fit naturally into a user’s daily routine, people will simply move on. As the saying goes: "Make it a habit, or they’ll disappear fast." Users need a reason to stay — not later, but right now. Product teams have to design for habit from day one. If an app doesn’t deliver immediate value or fit naturally into a user’s daily routine, people will simply move on. As the saying goes: "Make it a habit, or they’ll disappear fast." Users need a reason to stay — not later, but right now. Users need a reason to stay — not later, but right now. Product teams have to design for habit from day one. If an app doesn’t deliver immediate value or fit naturally into a user’s daily routine, people will simply move on. As the saying goes: "Make it a habit, or they’ll disappear fast." First-touch experience has become critical First-touch experience has become critical No more “cold starts.” Apps are expected to deliver engagement immediately — without friction, tutorials, or delayed benefits. Each interaction needs to feel like a forward motion. The time users are willing to spend before they feel progress is near zero. Apps are expected to deliver engagement immediately — without friction, tutorials, or delayed benefits. Today, apps have to prove more than just user demand—they need to show how they’ll generate revenue early on. It’s no longer enough to simply grow a user base — you need to know how you’ll make money from day one. Today, apps have to prove more than just user demand—they need to show how they’ll generate revenue early on. It’s no longer enough to simply grow a user base — you need to know how you’ll make money from day one. Products with no clear path to monetization aren’t just risky — they’re dead on arrival. 3. What’s Working in 2025 — and What Will Keep Working 3. What’s Working in 2025 — and What Will Keep Working While some product bets are being wiped out, others quietly prove their value — not through hype but through everyday use. Successful apps make complex things simple, seamlessly blend into daily life, and keep the technology in the background. Successful apps make complex things simple, seamlessly blend into daily life, and keep the technology in the background. Financial services have gone mainstream Financial services have gone mainstream Uberization unlocked experiences once reserved for the elite. You no longer need to buy a helicopter — you can rent one. A ride in a Maybach? Just another tap in your app, not a luxury fantasy. What used to be a privilege has become a commodity. But that phase is already behind us. The “rent, don’t own” model has quietly evolved into the next stage: the democratization of complex financial tools. What once required a bank account, paperwork, and approval processes is now part of the product flow. Apps let you invest $3 in stocks or split your grocery bill into four payments — no forms, friction, or feeling that you’re “doing finance.” It all happens quietly in the background, without your effort. The “rent, don’t own” model has quietly evolved into the next stage: the democratization of complex financial tools. Subscriptions are the default for entertainment Subscriptions are the default for entertainment The less time people spend working, the more they spend consuming. Subscription is now a core model, aligned with the growing habit of continuous entertainment.People are willing to pay if a product fills free time without friction. Despite the flood of streaming content, TikTok, and YouTube, user time is still expanding — and the services that occupy it will continue to grow steadily. People are willing to pay if a product fills free time without friction. AI works best when it disappears The most useful AI isn’t in chatbots — it’s in small, built-in tools that solve problems before you notice them. Removing unwanted objects from photos, searching images by content, and auto-cleaning videos — these features save time without demanding attention. This kind of AI doesn’t draw the eye or sell itself as a feature — it just works. Not a showcase but a silent engine that makes the product smarter, faster, and cleaner. Apps ultimately become tools for saving time and taking over routine tasks. The most useful AI isn’t in chatbots — it’s in small, built-in tools that solve problems before you notice them. Do less, but do better. In an environment of shrinking attention and rising expectations, it’s not the feature-packed apps that win — it’s the ones that deliver immediate and specific value. Users don’t allow time for warm-ups — the benefit needs to be clear from the start, and the experience must feel natural and effortless. This way, you can win.

What Consumers Expect from Apps in 2025 and How to Deliver

Can a Decentralized Internet Make Free Speech and Global Access a Reality?

I’m not a designer. I’m a founder. I don’t care about kerning—I care about shipping. So when Canva promised clean design with zero friction, I was sold. No agencies. No Figma drama. For $120, we looked like a Series B startup while still figuring out what our product did. Fantastic. Then one day, we tried starting from scratch. No templates, no guardrails, just an idea and a blank slide. Nothing came out. This isn’t a Canva teardown, just a warning: design tools can quietly erase your marketing instincts—until all you know how to do is make things that look like marketing. This isn’t a Canva teardown, just a warning: design tools can quietly erase your marketing instincts—until all you know how to do is make things that look like marketing. What Convenience Takes From You What Convenience Takes From You The damage of Canva isn’t aesthetic. It’s psychological. 1. Canva Cuts Strategy Gets Cut Out Of The Sequence 1. Canva Cuts Strategy Gets Cut Out Of The Sequence Before Canva, strategy came first, because it had to. You couldn’t design anything until you knew what you were trying to say. Messaging shaped layout. Insight shaped visual hierarchy. The thinking determined the design. Canva flips that sequence. You start with the template. Then you backfill the message. You tweak colors, swap fonts, drag headlines—then try to figure out what it’s saying. The visuals lead. Strategy follows, if it shows up at all. Canva severs the link between message and design—so your team builds around layout, not logic. Canva severs the link between message and design—so your team builds around layout, not logic. 2. Canva Collapses Your Ability to Prioritize 2. Canva Collapses Your Ability to Prioritize Before Canva, you had to choose your battles. Design was both time consuming and expensive, so the right to create had to be earned. You had to make the case: Why this asset? Why now? What’s the point? Why this asset? Why now? What’s the point? That design complexity acted as a filter, letting you prioritize what mattered most. Now everything takes 10 minutes. So everything feels equal. You design with the same energy for a homepage, an investor update, a LinkedIn post. You lose the instinct to ask: Is this a two-hour task or a two-week one? You forget how to feel the weight of things. Canva collapses your ability to prioritize—because when everything is equally easy to create, nothing feels important enough to think deeply about. 3. Canva Hides Brand Decay in Plain Sight 3. Canva Hides Brand Decay in Plain Sight Brand decay doesn’t happen through earth-shattering flops. It happens through hundreds of tiny, unintentional decisions: • One more font • A slightly different tone of green • A new layout for this week’s post • A promo slide that wasn’t run through anyone. By the end of the quarter, you’ve unintentionally rebranded five times. It gives you so many options, your brand dies by a thousand optimizations. you’ve unintentionally rebranded five times 4. Canva Makes Bad Ideas Look Legit—and Misleads You When They Fail 4. Canva Makes Bad Ideas Look Legit—and Misleads You When They Fail Canva designs are pretty good—a little too good. a little too good. Drop weak messaging into a sleek template, and suddenly it feels solid. Confusing copy looks clean. Half-baked ideas look presentable. And when the results suck, no one questions the content—because the container looked so convincing. You just assume the colors are off. The layout needs adjusting. The design needs a “refresh.” Canva makes weak marketing look wonderful—and when it flops, you end up fixing the wrong thing. You don’t rewrite. You redesign. And the cycle repeats. Quietly. Expensively. Ineffectively. Quietly. Expensively. Ineffectively. 5. Canva Trains You to Avoid Anything That Doesn’t Fit the Template 5. Canva Trains You to Avoid Anything That Doesn’t Fit the Template Over time, you stop chasing ideas that don’t already have a layout. You don’t write the weird essay post—because it won’t look good in a carousel. The strange campaign? No, because there’s no ready-made design for that. You steer clear of new formats simply because Canva makes them difficult to produce. You default to what fits, not what resonates. And over time, the things that don’t fit the mold don’t feel worth pursuing at all. Canva shapes how you present ideas, and also limits which ones you’re willing to have. Canva shapes how you present ideas, and also limits which ones you’re willing to have. 6. Canva Warps Your Team’s Understanding of Marketing 6. Canva Warps Your Team’s Understanding of Marketing When everyone can drag and drop, everyone thinks they “get” marketing. When you finally try to hire real marketers, real creatives, real thinkers—your team resists. “Isn’t this just making slides?” “Isn’t this just making slides?” “Didn’t we already figure out the brand?” “Didn’t we already figure out the brand?” “Can’t we just use the template?” “Can’t we just use the template?” Canva makes marketing feel like something anyone can do—sure, if your definition of marketing is pixel shuffling. But real marketing isn’t decorating. It’s positioning, insight, narrative, timing, channel, taste. And once your team is trained to value polish over depth, it’s hard to reintroduce the real thing. Canva democratizes marketing execution at the cost of respect for the discipline itself. By the time you need the real thing, your team doesn’t believe it matters. Canva democratizes marketing execution at the cost of respect for the discipline itself. By the time you need the real thing, your team doesn’t believe it matters. The Cases That Justify Convenience The Cases That Justify Convenience Canva is brilliant. But it only works if you know exactly what you’re using it for. Here are the legitimate, high-leverage use cases—where Canva becomes an asset, not a marketing assassin. 1. When You’re Just Trying to Exist Online Long Enough to Get Feedback 1. When You’re Just Trying to Exist Online Long Enough to Get Feedback You’re early. Resources are tight. The idea might pivot by next week. But you still need to throw out a landing page, run a quick ad, or test a message—just to see if anyone bites. You just need something that looks decent enough to get feedback, but you don’t have a budget for a designer. That’s where Canva fits. It’ll give you the ammunition you need to figure out if you’ve actually built something worth marketing. But don’t bother shopping around to find the perfect template. If your product positioning and messaging are off, no template will save you. 2. When You Know Exactly What You Want—and CBF Learning Illustrator 2. When You Know Exactly What You Want—and CBF Learning Illustrator This is the sweet spot. You’ve got the vision in your head. You’ve done the hard part: the copy works, the hook is tested, the positioning is sharp. You know the vibe. You just don’t want to spend 5 hours watching Adobe tutorials or begging a designer to “move the logo 10 pixels left.” Canva works when it’s used as a tool of translation, not inspiration. When you’re not there to be inspired by templates, just to recreate something already clear in your mind. 3. When You Have a Brand Identity—But No Time to Police It 3. When You Have a Brand Identity—But No Time to Police It You’ve defined the voice. You’ve chosen the color system. You’ve aligned on tone. But every day someone on the team is creating something—and you don’t have time to approve it all. Canva works when your job as a founder or brand lead shifts from creator to enforcer. You’re not asking Canva to invent your brand. You’re asking it to protect it while you’re doing 10 other things. And their Brand Kit function lets you do just that. It’s not a design tool at this point—it’s a compliance tool. 4. When You’re Trying to Make Strategy Feel Real 4. When You’re Trying to Make Strategy Feel Real Startups move fast, but alignment moves slow—especially when ideas are abstract. For example: a new vibe, or campaign that lives in your head but hasn’t clicked for the team yet. In these moments, Canva helps you make things real enough for people to react to. You’re not shipping externally, just creating internal clarity. Sometimes, the only way to get buy-in is to show—not tell. Use Canva when the team needs to see the future before they believe in it. 5. When You Need to Dress Up Something Disposable 5. When You Need to Dress Up Something Disposable Not everything you make is sacred. Some things are just… necessary. necessary. A partner deck you’ll use twice. A recruiting asset for a specific role. A last-minute slide for a VC that’ll never read past page five. These things still need to look good—but they’re temporary, tactical, and forgettable by design. That’s when Canva makes sense. It helps you not overthink the things that don’t deserve to slow you down. It’s not for your best ideas. It’s for the ones that don’t need to be. Use Canva when the work needs to look polished—but only needs to live for a week. but only needs to live for a week. Final Thoughts: Canva Is a Controlled Substance Final Thoughts: Canva Is a Controlled Substance My Canva subscription’s still active. I use it. I love it. But I treat it like a prescription drug. In the chaos of early-stage building—when your product’s half-broken, your team’s underwater, and your messaging is mush—Canva gives you one part of the company that looks clean. So you keep fixing that layer. The visual one. The easy one. But that’s sedation, not progress. Canva can help, as long as you don’t let it numb the parts of the business that actually need your attention. So use it. Just don’t abuse it.

The $120 Tool That Saved Us $120k—And Nearly Killed Our Marketing Culture

A Good Story Is the Difference Between Success and Survival

Dobrý příběh je rozdíl mezi úspěchem a přežitím

Una buena historia es la diferencia entre el éxito y la supervivencia

Une bonne histoire fait la différence entre le succès et la survie

एक अच्छी कहानी सफलता और अस्तित्व के बीच का अंतर बताती है

Dobra priča je razlika između uspjeha i preživljavanja

Լավ պատմություն, որ տարբերություն է հաջողության եւ կյանքի միջեւ

良い物語は成功と生存の違いです。

ERROR: invalid target_lang (ky)

یو ښه داستان د بریالیتوب او د ژوند په منځ کې فرق دی

O poveste bună este diferența dintre succes și supraviețuire

ERROR: invalid target_lang (tg)

Authors:
(1) Mårten Schultzberg, Experimentation Platform team, Spotify, Stockholm, Sweden;
(2) Sebastian Ankargren, Experimentation Platform team, Spotify, Stockholm, Sweden;
(3) Mattias Frånberg, Experimentation Platform team, Spotify, Stockholm, Sweden. Authors: Authors: (1) Mårten Schultzberg, Experimentation Platform team, Spotify, Stockholm, Sweden; (2) Sebastian Ankargren, Experimentation Platform team, Spotify, Stockholm, Sweden; (3) Mattias Frånberg, Experimentation Platform team, Spotify, Stockholm, Sweden. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 1.1 Related literature 1.1 Related literature Types of Metrics and Their Hypothesis and 2.1 Types of metrics
2.2 Hypotheses for different types of metrics


Type I and Type II Error Rates for Decision Rules including Superiority and Non-Inferiority Tests
3.1 The composite hypotheses of superiority and non-inferiority tests
3.2 Bounding the type I and type II error rates for UI and IU testing
3.3 Bounding the error rates for a decision rule including both success and guardrail metrics
3.4 Power corrections for non-inferiority testing


Extending the Decision Rule with Deterioration and Quality Metrics


Monte Carlo Simulation Study
5.1 Results


Discussion and Conclusions Types of Metrics and Their Hypothesis and 2.1 Types of metrics
2.2 Hypotheses for different types of metrics Types of Metrics and Their Hypothesis and 2.1 Types of metrics Types of Metrics and Their Hypothesis and 2.1 Types of metrics 2.2 Hypotheses for different types of metrics 2.2 Hypotheses for different types of metrics Type I and Type II Error Rates for Decision Rules including Superiority and Non-Inferiority Tests
3.1 The composite hypotheses of superiority and non-inferiority tests
3.2 Bounding the type I and type II error rates for UI and IU testing
3.3 Bounding the error rates for a decision rule including both success and guardrail metrics
3.4 Power corrections for non-inferiority testing Type I and Type II Error Rates for Decision Rules including Superiority and Non-Inferiority Tests Type I and Type II Error Rates for Decision Rules including Superiority and Non-Inferiority Tests 3.1 The composite hypotheses of superiority and non-inferiority tests 3.1 The composite hypotheses of superiority and non-inferiority tests 3.2 Bounding the type I and type II error rates for UI and IU testing 3.2 Bounding the type I and type II error rates for UI and IU testing 3.3 Bounding the error rates for a decision rule including both success and guardrail metrics 3.3 Bounding the error rates for a decision rule including both success and guardrail metrics 3.4 Power corrections for non-inferiority testing 3.4 Power corrections for non-inferiority testing Extending the Decision Rule with Deterioration and Quality Metrics Extending the Decision Rule with Deterioration and Quality Metrics Extending the Decision Rule with Deterioration and Quality Metrics Monte Carlo Simulation Study
5.1 Results Monte Carlo Simulation Study Monte Carlo Simulation Study 5.1 Results 5.1 Results Discussion and Conclusions Discussion and Conclusions Discussion and Conclusions APPENDIX A: IMPROVING THE EFFICIENCY OF PROPOSITION 4.1 WITH ADDITIONAL ASSUMPTIONS APPENDIX A: APPENDIX A: IMPROVING THE EFFICIENCY OF PROPOSITION 4.1 WITH ADDITIONAL ASSUMPTIONS APPENDIX B: EXAMPLES OF GLOBAL FALSE AND TRUE POSITIVE RATES APPENDIX B: APPENDIX B: EXAMPLES OF GLOBAL FALSE AND TRUE POSITIVE RATES APPENDIX C: A NOTE ON SEQUENTIAL TESTING OF DETERIORATION APPENDIX C: APPENDIX C: A NOTE ON SEQUENTIAL TESTING OF DETERIORATION APPENDIX D: USING NYHOLT’S METHOD OF EFFICIENT NUMBER OF INDEPENDENT TESTS APPENDIX D: APPENDIX D: USING NYHOLT’S METHOD OF EFFICIENT NUMBER OF INDEPENDENT TESTS Acknowledgments and References Acknowledgments and References Abstract. In the past decade, A/B tests have become the standard method for making product decisions in tech companies. They offer a scientific approach to product development, using statistical hypothesis testing to control the risks of incorrect decisions. Typically, multiple metrics are used in A/B tests to serve different purposes, such as establishing evidence of success, guarding against regressions, or verifying test validity. To mitigate risks in A/B tests with multiple outcomes, it’s crucial to adapt the design and analysis to the varied roles of these outcomes. This paper introduces the theoretical framework for decision rules guiding the evaluation of experiments at Spotify. First, we show that if guardrail metrics with non-inferiority tests are used, the significance level does not need to be multiplicity-adjusted for those tests. Second, if the decision rule includes non-inferiority tests, deterioration tests, or tests for quality, the type II error rate must be corrected to guarantee the desired power level for the decision. We propose a decision rule encompassing success, guardrail, deterioration, and quality metrics, employing diverse tests. This is accompanied by a design and analysis plan that mitigates risks across any data-generating process. The theoretical results are demonstrated using Monte Carlo simulations. Abstract 1. INTRODUCTION Randomized experiments are the gold standard for providing evidence for causal relationships. Modern technology companies use A/B tests, a randomized controlled trial in a digital setting, extensively to evaluate the efficacy of new changes to their products. These products include ride-sharing apps, search engines, streaming services, recommendations, and more. Ultimately, the goal of these experiments is to decide whether or not to release a product change more widely. Most of the literature on statistical inference for randomized experiments focuses on a single hypothesis test of a single outcome, and how to bound the type I and type II error rates for that test. However, experiments are not univariate tests of isolated outcomes. Instead, the risks that matter are the risks of making the incorrect decision for the product. For example, at a tech company like Spotify, we want to limit how often we release product changes that show an improvement when there truly is none, and how often we refrain from releasing changes that lead to improvements but we fail to find. These types of decisions typically include results from several hypothesis tests. Experiments usually involve results for multiple outcomes, and making a single decision based on these multiple outcomes can be challenging. For example, some of the outcomes, what we will refer to as ‘metrics’, may show improvements, while others show none or even negative effects. In the online experimentation literature, the only aspect of multi-test decision making that is extensively covered is multiple-testing correction. Multiple-testing corrections, such as Bonferroni, Holm [7] and Hommel [8], bound the type I error rate of an implied decision rule that declares what decision you will make based on the results of the individual hypothesis tests. As we will discuss extensively in this paper, unless your desired decision rule matches the rule implied by the multiple-testing correction, it is typically incorrect. In this paper, we show how it is possible to formalize the decision-making process of experiments without leaving the standard hypothesis testing framework. The key to ensuring that you obtain the intended risk bounds for the product decision is to explicitly specify a decision rule. A decision rule exhaustively specifies what product decision you will make based on the results of your experiment. Importantly, to bound the risks of making an erroneous decision, the design and analysis of your experiment must be closely aligned with the decision rule. Articulating the decision rule is important for several reasons. Being unclear about what outcomes lead to a positive product decision means that there is no mechanism for properly controlling the risks of the experiment at the level that matters to the company, namely the decision to ship the feature or not. Additionally, a lack of an articulated and standardized decision rule can mean that different teams or parts of the organization hold themselves to different standards. Our decision rule framework is a simple but effective approach for combating those issues. The decision rule framework helps standardize the analysis of experiments and is a useful tool for experimentation platforms. What the decision rule includes can be made more or less flexible. For example, new experiments can be forced to demonstrate that important company metrics are not negatively impacted while selecting the set of metrics that should show an improvement is made completely up to the experimenter. Even if the choice of metrics is completely arbitrary with no metrics made mandatory by the platform, the decision rule approach promotes a shared understanding of what a successful experiment is. Throughout this paper, and without loss of generality, we only consider experiments with two groups to simplify notation. In addition, we only consider one-sided tests, although more than one one-sided test might be applied to each metric. We limit ourselves to one-sided tests as there must be an intended direction for a change in the metric to map to a measurable improvement in the product. For simplicity, we assume that all metrics improve when they increase. Moreover, we assume that each statistical hypothesis test is valid and achieves its type I and type II error rates exactly if the experiment is designed accordingly. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv available on arxiv

Spotify’s Secret to Smarter A/B Testing (Hint: It’s Not Just Statistics)

Spotify's hemmelighed til smartere A / B-testning (Tips: Det er ikke kun statistik)

El secreto de Spotify para realizar pruebas A/B más inteligentes (pista: no se trata solo de estadísticas)

Le secret de Spotify pour un test A / B plus intelligent (Souci: Ce n'est pas seulement des statistiques)

Il segreto di Spotify per un test A/B più intelligente (Suggerimento: non è solo statistica)

SpotifyのSmart A/Bテストの秘密(ヒント:統計だけではない)

د سپوټایف د هوښیار A/B ازموینې راز (اشاره: دا یوازې احصایې ندي)

Secretul Spotify pentru testarea A / B mai inteligentă (Sugestie: Nu este doar statistică)

Spotify Ibanga ryo Kwipimisha A / B (Ibitekerezo: Ntabwo ari Imibare gusa)

Spotify tajomstvo pre inteligentnejšie A / B testovanie (Návod: Nie je to len štatistika)

Spotify's Secret to Smarter A/B Testing (Напомена: То није само статистика)

Spotify's Secret to Smarter A / B Testing (เคล็ดลับ: มันไม่ได้เป็นเพียงสถิติ)

Spotify的秘密更聪明的A / B测试(提示:这不仅仅是统计数据)

In an era where digital transformation initiatives often struggle with complexity and user adoption challenges, the remarkable success of the Oracle Fusion HCM implementation at Harbison Walker International (HWI) stands as a testament to exceptional technical leadership and operational excellence. Under the guidance of Mohammed Misbahul Khair, a leading Oracle HCM Solutions Architect, this ambitious enterprise-wide implementation has set new benchmarks for efficiency, user adoption, and operational cost savings in the HR technology sector, fundamentally transforming how the organization manages its global workforce. The comprehensive implementation, encompassing Core HR, Payroll, Time and Labor, Absence Management, Benefits, and Recruiting modules, emerged as a significant challenge in enterprise HR transformation. With responsibility for managing the implementation across multiple geographies and supporting over 5,000 employees, Mohammed Misbahul Khair faced the complex task of coordinating various functional streams while maintaining strict quality controls and achieving aggressive efficiency targets. The project's scope required careful consideration of diverse regional requirements, compliance standards, and the intricate interplay between different HCM modules. At the heart of this success story was a methodical approach to solution design and implementation. As a key implementation consultant, Mohammed implemented rigorous technical controls and innovative operational strategies that not only met but substantially exceeded performance targets. Through extensive business process workshops and detailed requirement gathering sessions, he established a robust foundation for the implementation. His thorough understanding of both technical capabilities and business needs enabled him to design solutions that precisely aligned with organizational objectives while leveraging Oracle Fusion HCM's full potential. The implementation featured sophisticated configurations of employee and manager dashboards, carefully structured workforce hierarchies, and meticulously designed security frameworks. Mohammed's expertise in utilizing Oracle Functional Setup Manager enabled the deployment of complex workforce structures, including Legal Entity HCM Information and Enterprise HCM Info, creating a seamless and secure user experience. The project achieved an unprecedented 98% data accuracy rate during migration – a significant accomplishment in enterprise-scale data transitions. This success was attributed to Mohammed's careful planning and execution of data conversions using HCM Data Loader (HDL), ensuring the accurate migration of critical data including workforce structures, employee records, jobs, and locations. The implementation's success translated into substantial operational improvements, with a 30% reduction in manual effort across HR processes, 25% improvement in payroll processing time, 40% increase in system utilization rates, and 20% reduction in overall HR operational costs. The project introduced several groundbreaking technical solutions, including sophisticated HCM Extracts and Fast Formulas that revolutionized payroll processes, enabling efficient handling of retro payments, complex proration methods, and automated payment generation. Stakeholder management emerged as a critical success factor throughout the implementation. Regular business process workshops and CRP sessions fostered strong stakeholder engagement, ensuring perfect alignment between technical solutions and business requirements. Mohammed's expertise in Oracle Fusion HCM allowed him to provide valuable insights and best-practice guidance, establishing him as a trusted advisor throughout the implementation lifecycle. The implementation prioritized user experience, resulting in intuitive interfaces and streamlined workflows that significantly enhanced adoption rates. The achievement has demonstrated far-reaching implications for the HR technology sector, establishing new benchmarks for enterprise HCM projects, particularly in areas of data migration accuracy, user adoption rates, process automation efficiency, security framework design, and global workforce management. The implemented solution was designed with scalability and future innovation in mind, incorporating flexible configurations that could adapt to evolving business needs and technological advances. For Mohammed Misbahul Khair personally, the project represented a significant career milestone, showcasing his ability to deliver complex technical solutions while managing substantial organizational change. His approach combined deep technical knowledge with strong business acumen, ensuring solutions that not only met current needs but also supported future growth and innovation in HR technology. This success story illustrates how strategic technical leadership, when combined with effective implementation practices, can transform enterprise HR operations. The HWI implementation not only contributed to the organization's digital transformation but also established new standards for HCM implementation in the sector. As organizations continue to modernize their HR systems, this project serves as a compelling example of how focused leadership can drive exceptional results in enterprise-scale implementations. About Mohammed Misbahul Khair About Mohammed Misbahul Khair A distinguished professional in Oracle HCM solutions, Mohammed Misbahul Khair has established himself as a leading expert in enterprise HR technology implementation. Based in Edison, New Jersey, he holds a Master's degree in Information Technology Management and extensive Oracle certifications. His comprehensive experience spans implementation of Core HR, Talent Management, Recruiting, and Payroll modules, consistently delivering solutions that enhance organizational efficiency while improving the employee experience. His technical expertise extends across the full spectrum of Oracle Fusion HCM modules, with particular strength in Core HR, Global HR implementations, payroll processing and automation, benefits administration, talent management solutions, recruiting and onboarding systems, time and labor management, absence management, and security and compliance frameworks. Through his career, Mohammed has demonstrated an exceptional ability to bridge the gap between technical capabilities and business requirements, consistently delivering solutions that drive organizational success while maintaining the highest standards of technical excellence. As the HR technology sector continues to evolve, his work at HWI stands as a model for future enterprise HCM projects, showcasing the powerful combination of technical expertise, operational excellence, and business acumen in driving transformational change in enterprise HR operations. This story was distributed as a release by Kashvi Pandey under HackerNoon’s Business Blogging Program. Learn more about the program here. This story was distributed as a release by Kashvi Pandey under HackerNoon’s Business Blogging Program. Learn more about the program here. here here

Enterprise HCM Implementation: A Transformation Journey Of Mohammed Misbahul Khair

ERROR: invalid target_lang (ay)

এন্টারপ্রাইজ এইচসিএম বাস্তবায়ন: মোহাম্মদ মিসবাহুল খায়েরের একটি রূপান্তর যাত্রা

Implementación de HCM empresarial: un viaje de transformación de Mohammed Misbahul Khair

Enterprise HCM Implementation: Mohammed Misbahul Khairin muutosmatka

Implementacija poduzeća HCM: Putovanje transformacije Mohameda Misbahula Khaira

Enterprise HCM Implementation: A Transformation Journey of Mohammed Misbahul Khair(モハメド・ミズバウル・カイア)

Кәсіпорынның HCM іске асырылуы: Мұхаммед Мисбахул Хайрдың трансформациялық саяхаты

Implementarea HCM pentru întreprinderi: o călătorie de transformare a lui Mohammed Misbahul Khair

Implementimi i ndërmarrjes HCM: Një udhëtim transformimi i Mohammed Misbahul Khair

Enterprise HCM Implementation: A Transformation Journey of Mohammed Misbahul Khair

Enterprise HCM Implementation: A Transformation Journey ของ Mohammed Misbahul Khair

No one is perfect and that applies to high performers on your team too. They may be doing excellent work or exceeding your expectations, but that doesn’t make them flawless or leave them with no room for improvement or areas of growth. Not giving constructive feedback conveys you don’t care about them, don’t recognize them and don’t appreciate the value they bring to the team. You’re not doing enough if all you do is pass occasional remarks to appreciate their work: That was great! That was great! You’re awesome! You’re awesome! You rock! You rock! Occasionally telling them how great they’re doing does not inspire, motivate or challenge them to do better. Rather, lack of clarity on development and growth areas can make them repeat behaviors that prevent them from reaching for their true potential. Your high performers are highly valuable to the team and the organization—they’re 400% more productive than an average performer. Losing them robs the organization of excellence while incurring additional cost, time, and effort to hire their replacements. Spending significant time with people on the team who aren’t doing so well or those who need guidance to excel isn’t wrong. It’s problematic when you do it at the cost of your high performers. Talent is the multiplier. The more energy and attention you invest in it, the greater the yield. The time you spend with your best is, quite simply, your most productive time.
— Marcus Buckingham Talent is the multiplier. The more energy and attention you invest in it, the greater the yield. The time you spend with your best is, quite simply, your most productive time. — Marcus Buckingham Give constructive feedback to your high performers to help them stay motivated and invested in their own growth using these 5 practices: Manage their ego Giving constructive feedback to high performers is tough because of the inner voice in your head: inner voice What if it puts them off? What if it puts them off? What if they don’t agree? What if they don’t agree? What if they take it the wrong way? What if they take it the wrong way? What if they quit? What if they quit? These fears aren’t irrational. High performers tend to have some of the biggest egos and managing it is crucial to ensure your feedback is taken constructively without putting the recipients off. How you deliver feedback matters the most. Words used in communicating feedback hit on a person’s emotional system. Saying good things about them evokes happy feelings while conveying difficult feedback triggers negative reactions. When giving feedback, using certain words is a big no-no. They trigger intense negative feelings which puts them on the defensive: Generalizing words like “always” and “never.”
Enforcing words like can’t, shouldn’t, must, obey, have to.
Words that challenge their character like bad, demanding, unprofessional, rude.
Passing judgment with words like mistake, failure, unacceptable. Generalizing words like “always” and “never.” Enforcing words like can’t, shouldn’t, must, obey, have to. Words that challenge their character like bad, demanding, unprofessional, rude. Passing judgment with words like mistake, failure, unacceptable. For example: Instead of: You always come late to meetings. It’s unacceptable. Instead of: You always come late to meetings. It’s unacceptable. Say this: In the last 2 meetings, I noticed that you came 10-15 minutes after the meeting started. How can you come to meetings on time and set a good example for others? Say this: In the last 2 meetings, I noticed that you came 10-15 minutes after the meeting started. How can you come to meetings on time and set a good example for others? Instead of: You may have great ideas to share, but you also have to let others speak up in discussions. Not letting others speak is rude. Instead of: You may have great ideas to share, but you also have to let others speak up in discussions. Not letting others speak is rude. Say this: Hearing diverse viewpoints will help us make better decisions. How can you encourage more participation from others in discussions? Say this: Hearing diverse viewpoints will help us make better decisions. How can you encourage more participation from others in discussions? Respect is like air. As long as it's present, nobody thinks about it. But if you take it away, it's all that people can think about. The instant people perceive disrespect in a conversation, the interaction is no longer about the original purpose—it is now about defending dignity.
— Kerry Patterson Respect is like air. As long as it's present, nobody thinks about it. But if you take it away, it's all that people can think about. The instant people perceive disrespect in a conversation, the interaction is no longer about the original purpose—it is now about defending dignity. — Kerry Patterson In short, be considerate. Think carefully about the effect of what you say and avoid emotionally charged words that can get their ego involved and turn what should’ve been a friendly feedback conversation hostile. The conversation is more likely to stay productive with egos out of the picture. Talk about actions and impact High performers thrive on doing impactful work. That makes them super curious about how their actions contribute towards their goals or take it away from them. When giving feedback, don’t talk about behavior, character, or personality. Don’t tell them what they did wrong or what they should do. Such feedback may come across as judgmental, opinionated, and biased making them hate or turn against you. Instead, describe their action as a fact and the impact it had. They can’t refute the fact and they aren’t designed to ignore impact. Both are likely to get their attention and encourage them to find a solution instead of justifying or trying to prove you wrong. For example: For example: Instead of: You shout at others. This is no way to treat them. Learn to control your temper. Instead of: Say this: I was told that in the last meeting, you raised your voice (fact). This made others in the room reluctant to speak up. Without their contribution and agreement, we cannot make a decision. This will delay the project (impact). What can you do to support others’ ideas so that they’re encouraged to share (solution)? Say this: Instead of: You’re very rude and keep interrupting others. This is unacceptable. Instead of: Say this: In the last discussion, I noticed that when Kim was sharing a solution, you interrupted her several times and took over the conversation (fact). When you do not let others complete, they feel unheard which impacts their confidence. We can grow as a team only when everyone in the team feels safe and empowered to share their opinions and make their own decisions (impact). What can you to do enable it (solution)? Say this: Giving honest feedback is tricky, because it can easily result in people feeling hurt or demoralized. By aiming for candor—feedback that is smaller, more targeted, less personal, less judgmental, and equally impactful—it’s easier to maintain a sense of safety and belonging in the group.
— Daniel Coyle Giving honest feedback is tricky, because it can easily result in people feeling hurt or demoralized. By aiming for candor—feedback that is smaller, more targeted, less personal, less judgmental, and equally impactful—it’s easier to maintain a sense of safety and belonging in the group. — Daniel Coyle Judgment forms a negative feedback loop which shifts the purpose of the discussion from growth to blame. You can’t lead with judgment and achieve anything significant at the same time. Stay un-opinionated, stick to facts. Chart out a growth path High performers don’t just like to stay busy. They want to do impactful work, derive a sense of accomplishment from solving tough problems, and constantly seek to push the boundaries of their limits. They strongly align with what Daniel Pink said in Drive “Greatness and nearsightedness are incompatible. Meaningful achievement depends on lifting one’s sights and pushing toward the horizon.” Their passion for their work and confidence in pulling off challenging projects keeps them hungry and desperate for more. “Greatness and nearsightedness are incompatible. Meaningful achievement depends on lifting one’s sights and pushing toward the horizon.” Don’t just assign work to your high performers, show them where they are headed. Spend dedicated time in visualizing their growth path, identify the opportunities they will need along the way, and ensure their work is aligned with the challenges they need to learn and grow. Don’t hesitate to look beyond the boundaries of your team to other areas that might be better suited for their growth. Remember, charting out this path is not the end of your responsibility, it’s the beginning of one. Companies evolve and so do the people that make them up. Keep an eye out for a change in market conditions, their own personal preferences, or anything else that may require you to revisit this plan. Miss a few dots along the way and you are sure to lose your high performer. Make the conversation a two-way street High performers aren’t silent receivers of information. They like to add their views to the discussion. This applies to feedback too. When giving constructive feedback or showing the growth path ahead, don’t just do all the talking. Pause and allow them to expand on their thoughts and share their feelings. Your discussion should feel like a partnership, not an interrogation or one-sided account of what you deem right for their career and growth. You may be surprised to find out that what you planned for them may not even align with what they want to do or how they see themselves a few years from now. Making assumptions about their career or enforcing your own aspirations onto them is bound to backfire. It can be quite upsetting to have a manager who doesn’t understand you or doesn’t care about what you want. For example: Assigning lead or project management responsibilities to a developer because you think they can be an awesome manager even though they want to grow into an individual contributor track. Assigning lead or project management responsibilities to a developer because you think they can be an awesome manager even though they want to grow into an individual contributor track. Having your lead sign up for design decision reviews even though they are more interested in the problem-solving group. Having your lead sign up for design decision reviews even though they are more interested in the problem-solving group. The idea is not to meet all their demands or allow them to refuse to contribute in an area where their expertise might be helpful. It’s about hearing them out and showing that you’re willing to put in the effort to balance their needs with the organization’s wants. The idea is not to meet all their demands or allow them to refuse to contribute in an area where their expertise might be helpful. It’s about hearing them out and showing that you’re willing to put in the effort to balance their needs with the organization’s wants. Effective listening is the single most powerful thing you can do to build and maintain a climate of trust and collaboration. Strong listening skills are the foundation for all solid relationships.
― Michelle Tillis Lederman Effective listening is the single most powerful thing you can do to build and maintain a climate of trust and collaboration. Strong listening skills are the foundation for all solid relationships. ― Michelle Tillis Lederman Find ways to create a win-win situation. Hear them out. Follow up and follow through We judge ourselves by our intentions and others by their behavior. That’s why following through on our commitments is such a critical component of trust. Making false promises, inconsistent actions and unpredictable behavior leads to distrust. Don’t make promises in your feedback discussions that you can’t keep. Portraying a beautiful picture just to make them happy in the moment will make them very unhappy in the long term. Think about what you’re committing and don’t promise anything you can’t fulfill. For example: For example: If you promise to have monthly one-on-ones with your high performers, always schedule them on your calendar and make time for them. If you promise to have monthly one-on-ones with your high performers, always schedule them on your calendar and make time for them. If you commit to prioritize technical excellence in a quarter over product features, make sure to align all the stakeholders with the direction. If you commit to prioritize technical excellence in a quarter over product features, make sure to align all the stakeholders with the direction. If you promise to provide more transparency and clarity around work, take time to share the right level of information with your high performers and take their feedback to make it better. If you promise to provide more transparency and clarity around work, take time to share the right level of information with your high performers and take their feedback to make it better. If you agree to delegate more responsibilities to your high performers, consciously prioritize what you plan to do yourself and what activities are best left to them. If you agree to delegate more responsibilities to your high performers, consciously prioritize what you plan to do yourself and what activities are best left to them. Nothing breaks trust more than not following through on your commitments—not doing the things you say you are going to do. When you say one thing and do another thing or make promises but do not keep up with those promises, people start doubting your intention. Trust starts with intention. Others can’t trust you if they doubt your intention. Without trust none of your feedback matters. While giving feedback, be conscious about what you commit to your high performers. Be clear about the expectations they have from you. Make sure to follow through on the expectations you created. To keep this in check: Keep a working list of all the commitments with your high performers.
Take time to assign a timeline to each item on the list and dedicate focussed attention to make progress towards your commitments.
Schedule regular sync-ups with your high performers to go over the status of each item. Keep a working list of all the commitments with your high performers. Take time to assign a timeline to each item on the list and dedicate focussed attention to make progress towards your commitments. Schedule regular sync-ups with your high performers to go over the status of each item. Trust is a critical element that you actively need to build upon to strengthen your relationship with high performers. It's not possible to gain trust if you’re not serious about your commitments. Summary Not sharing constructive criticism with high performers prevents them from improving in areas where there is potential for growth. Denying them these improvements prevents them from reaching even bigger and better goals.


Sharing feedback with high performers isn’t easy. Your own fear combined with their ego can make such conversations a disaster unless you pay special attention to how the message is delivered. Choose words carefully.


Passing judgment when sharing feedback can defeat the purpose of your discussion because opinionated feedback will make your high performers defensive and turn them against you. Sticking to facts, talking about their action and the impact it had will get the attention you need to create a positive effect aligned with their long-term growth.


High performers not only want to do great work, they want to know where they’re headed. The journey that excites them must be carefully planned and executed. Spend dedicated time to match their aspirations with the right opportunities.


Feedback’s aren’t meant to be a one-sided conversation where you do all the talking while the other party listens. It’s a partnership. You can’t know what they want unless you have actually heard them say it. Embrace silence. Give them space to talk and lay out what really matters to them.


Fancy commitments may impress your high performers temporarily. But not keeping promises you made breaks trust with negative consequences in the long run. Be careful about what you commit. Don’t forget to follow through on your commitments. Not sharing constructive criticism with high performers prevents them from improving in areas where there is potential for growth. Denying them these improvements prevents them from reaching even bigger and better goals. Not sharing constructive criticism with high performers prevents them from improving in areas where there is potential for growth. Denying them these improvements prevents them from reaching even bigger and better goals. Sharing feedback with high performers isn’t easy. Your own fear combined with their ego can make such conversations a disaster unless you pay special attention to how the message is delivered. Choose words carefully. Sharing feedback with high performers isn’t easy. Your own fear combined with their ego can make such conversations a disaster unless you pay special attention to how the message is delivered. Choose words carefully. Passing judgment when sharing feedback can defeat the purpose of your discussion because opinionated feedback will make your high performers defensive and turn them against you. Sticking to facts, talking about their action and the impact it had will get the attention you need to create a positive effect aligned with their long-term growth. Passing judgment when sharing feedback can defeat the purpose of your discussion because opinionated feedback will make your high performers defensive and turn them against you. Sticking to facts, talking about their action and the impact it had will get the attention you need to create a positive effect aligned with their long-term growth. High performers not only want to do great work, they want to know where they’re headed. The journey that excites them must be carefully planned and executed. Spend dedicated time to match their aspirations with the right opportunities. High performers not only want to do great work, they want to know where they’re headed. The journey that excites them must be carefully planned and executed. Spend dedicated time to match their aspirations with the right opportunities. Feedback’s aren’t meant to be a one-sided conversation where you do all the talking while the other party listens. It’s a partnership. You can’t know what they want unless you have actually heard them say it. Embrace silence. Give them space to talk and lay out what really matters to them. Feedback’s aren’t meant to be a one-sided conversation where you do all the talking while the other party listens. It’s a partnership. You can’t know what they want unless you have actually heard them say it. Embrace silence. Give them space to talk and lay out what really matters to them. Fancy commitments may impress your high performers temporarily. But not keeping promises you made breaks trust with negative consequences in the long run. Be careful about what you commit. Don’t forget to follow through on your commitments. Fancy commitments may impress your high performers temporarily. But not keeping promises you made breaks trust with negative consequences in the long run. Be careful about what you commit. Don’t forget to follow through on your commitments. Follow me on LinkedIn or here for more stories like this. Follow me on LinkedIn or here for more stories like this. LinkedIn

How to Give Constructive Feedback to High Performers (Without Losing Them)

Stories, News & Mentions about ALIGNED

Aligned’s Stories on HackerNoon