<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by MingYi Zhao on Medium]]></title>
        <description><![CDATA[Stories by MingYi Zhao on Medium]]></description>
        <link>https://medium.com/@mingyizhao?source=rss-294d537d0cc0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*2EHIFlsJDd7SAkeZ.jpg</url>
            <title>Stories by MingYi Zhao on Medium</title>
            <link>https://medium.com/@mingyizhao?source=rss-294d537d0cc0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 21:23:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mingyizhao/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[AI Is Triggering a Cambrian Explosion in Software]]></title>
            <link>https://mingyizhao.medium.com/ai-is-triggering-a-cambrian-explosion-in-software-e8bd3148fbf9?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/e8bd3148fbf9</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Wed, 25 Feb 2026 00:49:04 GMT</pubDate>
            <atom:updated>2026-02-25T00:49:04.804Z</atom:updated>
            <content:encoded><![CDATA[<p>Roughly 600 million years ago, Earth was home to creatures that would seem almost alien today. Much of the visible life in the fossil record from the Ediacaran period is soft-bodied — often flat. Many lived close to the seafloor, interacting with microbial mats and slow-moving currents rather than high-speed predators. Some researchers have even described this world as a kind of “Garden of Ediacara,” a calmer ecosystem before the arms races began.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/650/1*tVOoqn05i6dNlHHCutVFYA.png" /></figure><p>Then, around 541 million years ago, everything changed. In a geologic blink, animal diversity surged: new body plans, new ways of moving, new senses, and new defenses. The ocean started to look recognizably modern — busy, competitive, and inventive. Not long after, the earliest vertebrates entered the scene: distant ancestors on the long evolutionary branch that would eventually lead to humans.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*no5f0P-lwTtQqz8RbmNdoA.png" /></figure><p>Why did the Cambrian explosion happen? One compelling part of the story is that evolution gained a better way to prototype. Before this, new body forms largely had to be built from scratch. Then evolution acquired Hox genes (and related developmental control genes), which function like a body-plan toolkit, giving creatures a way to tinker with their anatomy — adjusting where segments go, what forms at the front versus the back, and how repeated parts can diversify into specialized structures. With this toolkit, organisms could try many different body configurations, and the variants that fit new conditions and niches could spread quickly. Once many organisms can explore more designs, selection gets harsher: mobility, predation, burrowing, sensing, and armor begin feeding one another in an escalating arms race. The Cambrian, in other words, isn’t just “more animals” — it’s a shift in the innovation rate and competitive intensity of the entire ecosystem.</p><p>We are at the start of a Cambrian explosion in software. If we view software as creatures, then before 2026 was the Ediacaran period: code was written by hand, and iteration cycles were slow. Then something changed. Around late 2025 and into 2026, AI coding became widely adopted. New software is almost 100% written by AI — generating, modifying, and recombining code becomes dramatically cheaper and faster. Companies ship at a new tempo, while enthusiasts “vibe code” an endless stream of tiny apps to satisfy every imaginable niche.</p><p>AI won’t replace software. It will supply the intelligence of the digital world, while software provides the durable “bodies” and infrastructure — workflows, permissions, ledgers, invariants, and audit trails — the equivalent of animal anatomy, coral reefs, and forests. Model intelligence is adaptive and creative, but it isn’t inherently deterministic or stable. Paired with a reliable software body, agents become far more capable: they can act, coordinate, and produce value safely in complex environments. That’s why, even today, much of the value from frontier models comes not from the model alone, but from the surrounding system — how it’s orchestrated, evaluated, and connected to tools and execution layers (for example, OpenClaw or Claude Code).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wkH4CwSRpQlLph3dL3h5YA.png" /></figure><p>Ever-improving AI agents will also accelerate software growth. In the Cambrian, life didn’t just diversify — it rapidly created new ecological niches and then filled them. We may see the same dynamic in the digital world: as build costs collapse, it suddenly makes economic sense to serve long-tail human needs — and to tackle social problems like poverty, scams, inequality, and education — so the number of “software species” can explode. But the bigger opportunity may be niches created by AI agents themselves — services that collect and clean data for other agents, identity and trust layers for agent-to-agent interaction, and operational agents that manage real-world infrastructure like solar farms in the desert. Like the Cambrian, the ecosystem won’t just add more “species” — it will generate entirely new food webs.</p><p>Even though life exploded, Earth in the Cambrian was still a barren place compared to today. The oceans lacked coral reefs, and the vast continents were largely empty of life. Only much later did the land become covered by forests — and eventually by human-made buildings and roads. We may see the same arc in the digital space: not just more software, but software spreading everywhere — until “software eats the world” becomes literally true.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e8bd3148fbf9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[One-person Unicorn is Dangerous]]></title>
            <link>https://mingyizhao.medium.com/one-person-unicorn-is-dangerous-608a03fde254?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/608a03fde254</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Fri, 20 Jun 2025 13:56:36 GMT</pubDate>
            <atom:updated>2025-06-20T13:56:36.398Z</atom:updated>
            <content:encoded><![CDATA[<h3>One-person Unicorn is Dangerous 🦄</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*U_6KKMb813eb3XHuXdNzDA.png" /></figure><p>Lately, there’s been growing excitement about the idea of a “one-person unicorn” 🦄 — a billion-dollar startup run entirely by a single individual empowered by AI. But to me, that vision is pretty horrible.</p><p>I’m not just concerned about job losses from automation — though that’s a serious issue. What troubles me more is the dangerous arrogance behind this idea and its potential consequences.</p><p>History shows that concentrating too much power in one person’s hands often leads to disaster. Even in business, lasting success isn’t just about speed or efficiency. It also requires integrity, ethics, sustainability.</p><p>Every human being is flawed and limited. But in a strong team, people support and challenge each other. They catch mistakes, offer emotional resilience, and create better outcomes together. I’m not convinced AI agents can do the same. In fact, they may simply amplify a founder’s blind spots and weaknesses.</p><p>That’s why I’d be cautious about relying on products from one-person companies powered solely by AI. My bet is that diverse, collaborative teams — supported by well-aligned AI tools — will always outperform lone operators in the long run.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=608a03fde254" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Trump’s Tarriffs]]></title>
            <link>https://mingyizhao.medium.com/trumps-tarriffs-6b20c2b2b009?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/6b20c2b2b009</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[meaning-of-life]]></category>
            <category><![CDATA[humanity]]></category>
            <category><![CDATA[tarriffs]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Mon, 07 Apr 2025 01:31:15 GMT</pubDate>
            <atom:updated>2025-04-13T00:51:59.836Z</atom:updated>
            <content:encoded><![CDATA[<h3>Trump’s Tarriffs and AI</h3><p>One motivation behind Trump’s tariffs is the aim to revitalize America’s manufacturing sector, although many critics have challenged this goal due to its high economic costs and potential hardships. However, I propose an alternative perspective that sheds new light on the broader implications — particularly in relation to the rise of artificial intelligence (AI).</p><p>Experts widely predict that AI will significantly reduce white-collar and knowledge-based jobs in the coming years, potentially leading to large-scale unemployment in the United States and beyond. Unemployment doesn’t just slash people’s incomes — it also raises deeper existential questions about purpose. After all, if AI is smarter, more efficient, and tirelessly diligent, what role remains for humans?</p><p>Many people in America’s Rust Belt lost their jobs — and consequently their sense of dignity — during the last wave of deindustrialization, leading to social issues such as drug addiction. This is precisely why conservative voices, exemplified by J.D. Vance, emphasize the importance of employment as fundamental to societal stability.</p><p>Thus, even if American manufacturing remains expensive and relatively inefficient, conservatives remain committed to policies like tariffs to rebuild this sector. Manufacturing can generate numerous jobs — workers, factory managers, transport operators, and more — that inherently interact with the physical world and are therefore less susceptible to AI replacement.</p><p>While U.S.-produced goods might be costlier, conservatives view this trade-off as worthwhile because it helps rebuild a society where dignity is accessible to all. Simultaneously, lowering service-sector costs (including government operations) through AI adoption could manage overall inflation effectively, potentially enabling tax reductions and preventing public discontent.</p><p>However, many people question whether tariffs can genuinely bring manufacturing jobs back to America. Manufacturing is an intricate and lengthy process that thrives in a stable and predictable environment — rapidly shifting decisions undermine this stability. Furthermore, effectively revitalizing manufacturing in the U.S. would require cooperation with China, the world’s largest manufacturing hub. The Oscar-winning documentary, <em>American Factory</em>, vividly illustrates how Chinese entrepreneurs collaborated successfully with an American Rust Belt city to establish a world-class factory.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fm36QeKOJ2Fc%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dm36QeKOJ2Fc&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fm36QeKOJ2Fc%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/98039372678dd05b8cdb72df05a578b1/href">https://medium.com/media/98039372678dd05b8cdb72df05a578b1/href</a></iframe><p>Ultimately, sustained economic prosperity and societal stability may best be achieved through cooperation rather than conflict.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6b20c2b2b009" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Can Coursera Make a Comeback Under Its New CEO?]]></title>
            <link>https://mingyizhao.medium.com/can-coursera-make-a-comeback-under-its-new-ceo-e226fe988b32?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/e226fe988b32</guid>
            <category><![CDATA[online-learning]]></category>
            <category><![CDATA[enterpreneurship]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[coursera]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Sat, 15 Mar 2025 03:04:48 GMT</pubDate>
            <atom:updated>2025-03-15T03:04:48.219Z</atom:updated>
            <content:encoded><![CDATA[<p>Coursera has long been a leader in the online learning space. Personally, I used it extensively around 10 years ago and discovered several life-changing courses. I’ve spent hundreds of dollars in various courses. Over the past two years, I even owned Coursera stock but ended up losing some money due to its dramatic decline — an 84% drop from its peak.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yVPdd8tcb1_TeqsSBP2y-g.png" /></figure><p>One key factor contributing to this decline is the rise of advanced AI chatbots like ChatGPT, providing instant, personalized learning without requiring users to sit through lengthy video courses. Additionally, Coursera’s product doesn’t seem to have significantly evolved over the past decade, and the overall user experience still feels somewhat clunky.</p><p>To turn things around, the company just announced a new CEO, Greg Hart. Greg is an amazing leader and you can tell that from his LinkedIn self introduction.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AXlTw0f6nqqJ88ogAAmjAQ.png" /></figure><p>However, I am skeptical about his ability to revive Coursera. The company doesn’t simply need a leader with an impressive resume; it needs someone deeply passionate about learning, education, and the product itself. Unfortunately, Greg’s LinkedIn profile doesn’t clearly reflect that passion.</p><p>On the other hand, Duolingo, a thriving online education platform, has experienced steady stock growth since its IPO, in stark contrast to Coursera.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*98FAh-Bf4LRQRUbQVXx6Lg.png" /></figure><p>If you look at Duolingo CEO Luis von Ahn’s LinkedIn, you’ll notice he focuses solely on the mission of making education accessible to everyone. Rather than emphasizing his impressive past accomplishments — such as inventing CAPTCHA — he remains humble and mission-oriented.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qHEa7p5GyzVfPSXP9cMS0Q.png" /></figure><p>That’s the right leader for a growing tech company.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e226fe988b32" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Professionals Can Learn Effectively in a Busy World]]></title>
            <link>https://mingyizhao.medium.com/how-professionals-can-learn-effectively-in-a-busy-world-3937bec9a1c3?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/3937bec9a1c3</guid>
            <category><![CDATA[education]]></category>
            <category><![CDATA[lifelong-learning]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Sun, 15 Dec 2024 23:25:12 GMT</pubDate>
            <atom:updated>2024-12-15T23:25:12.252Z</atom:updated>
            <content:encoded><![CDATA[<p>Professionals, like software engineers, need to continuously learn new skills and knowledge to stay relevant and grow their careers. The world is changing rapidly, particularly with the explosive growth of AI. Without ongoing learning, professionals risk becoming outdated.</p><h3>The Challenge of Lifelong Learning</h3><p>Learning as a professional presents unique challenges. Unlike college students, most professionals can’t dedicate large blocks of time to study. Particularly, people in their 30s often juggle demanding work schedules and family responsibilities, leaving them with only fragmented moments for self-improvement. This makes traditional online learning solutions, like full-length courses on platforms such as Coursera, less effective. Even reading books can feel like an inefficient use of their limited time.</p><h3>Is There a Better Way?</h3><p>What if there were an approach that allowed professionals to learn effectively, even in just five minutes? This solution would need to be engaging and enjoyable — something that competes with the allure of doom scrolling through TikTok during breaks.</p><h3>Lessons from Duolingo</h3><p>Duolingo provides a compelling example of how learning can be both fun and efficient. Its bite-sized lessons cater to fragmented schedules, and its quiz-based format keeps users engaged.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Aeh-E7rLxdVvFNBN.png" /></figure><p>As a result, the stock of Duolingo is performing well 😃</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0_jg2P9a1XaLI_4_tRVh8A.png" /></figure><p>However, Duolingo primarily focuses on language learning. What if you want to learn about business communication or cutting-edge AI technologies?</p><h3>Introducing SkillUp</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eS3p_sGKe2p4TI67Bo-IsA.png" /><figcaption>Landing Page of SkillUp</figcaption></figure><p>SkillUp is a new product designed to meet these needs. It offers bite-sized lessons on a wide range of topics, from the latest advancements in AI to practical tips for living a longer life or immersing yourself in the culture of Hawaii. Users can develop new skills by completing quizzes and receiving real-time feedback. With SkillUp, learning is accessible, engaging, and perfectly suited to the busy lives of professionals.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XQAfr46KxH09wlNUrszxQg.png" /><figcaption>Learning by Answering Questions</figcaption></figure><p>To give it a try, visit <a href="https://www.skill-up.io/">SkillUp’s website</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3937bec9a1c3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Centralization, Decentralization, Web 3]]></title>
            <link>https://mingyizhao.medium.com/centralization-decentralization-web-3-c843e5e83c91?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/c843e5e83c91</guid>
            <category><![CDATA[economy]]></category>
            <category><![CDATA[web3]]></category>
            <category><![CDATA[jp-morgan]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Sun, 15 Jan 2023 14:58:15 GMT</pubDate>
            <atom:updated>2023-01-15T17:14:36.029Z</atom:updated>
            <content:encoded><![CDATA[<h3>Centralization</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g6Fz9Hb0d91u2iZkGpkWMQ.jpeg" /></figure><p>J. P. Morgan was a prominent financial figure around 1900. He made two major contributions to the growth of US economy.</p><p>First, he was the driving force of industrial consolidation. An example is the creation of the U.S. Steel:</p><blockquote>In 1900, Morgan began talks with to purchase Andrew Carnegie’s steel business and merge it with several other steel, coal, mining and shipping firms. After financing the creation of the Federal Steel Company, Morgan merged it with the Carnegie Steel Company and several other steel and iron businesses (including William Edenbirn’s Consolidated Steel and Wire Company) in 1901, forming the United States Steel Corporation. U.S. Steel was the world’s first billion-dollar company, with an authorized capitalization of $1.4 billion, much larger than any other industrial firm and comparable in size to the largest railroads. (Wikipedia)</blockquote><p>This advanced US’ industrial capability and grew its economy:</p><blockquote>U.S. Steel’s goals were to achieve greater economies of scale, reduce transportation and resource costs, expand product lines, and improve distribution to allow the United States to compete globally with the United Kingdom and Germany. (Wikipedia)</blockquote><p>Second, he orchestrated several interventions during panics to stabilize the financial system. For example, in the Panic of 1907:</p><blockquote>Major New York banks were on the verge of bankruptcy and there was no mechanism to rescue them, until Morgan stepped in to help resolve the crisis… Morgan then met with the nation’s leading financiers in his New York mansion, where he forced them to devise a plan to meet the crisis… Morgan organized a team of bank and trust executives which redirected money between banks, secured further international lines of credit, and bought up the plummeting stocks of healthy corporations. (Wikipedia)</blockquote><h3>Decentralization</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/685/1*u7v07vzp2xq-aG6N7i6Dow.jpeg" /></figure><p>People have criticized Morgan for centralizing too much power, similar to how BigTech is criticized today. However, people also want stability and growth. The free market is a decentralized system. In most of the time, it works fine and allows people to build new technologies and grow businesses using their wisdom and hardworking. But sometime the market will fail because people are also stupid, greedy and lazy. For example, in the Panic of 1857:</p><blockquote>Before 1857, the railroad industry had been booming due to large migrations of people to the west, especially to Kansas. The large influx of people made the railroads a profitable industry, and the banks began to provide railroad companies with large loans. Many of the companies never made it past the stage of a paper railroad and never owned the physical assets necessary to run a real one. Prices of railroad stocks as a whole began to experience a stock bubble, and railroad stocks saw increasingly-speculative entries into the fray, worsening the bubble. (Wikipedia)</blockquote><p>When the bubble bursts, the whole decentralized market system is disrupted, and everyone, not just the culprits, suffers. At this point, intervention from a centralized authority is a clear way out. And that’s what J.P. Morgan did. Is that the best way? I don’t know. All I want to say is that there will always be a strong demand for centralized intervention during crises because people hate uncertainty and want to end the pain soon, even if it bears some long-term costs.</p><p>As the complexity of the system grows, the demand for centralization only increases. Compared with 100 years ago, the US economy is much more complex and also more centrally managed. If we look back to 2000 years ago, then the transition from the Roman Republic to the Roman Empire is largely attributed to the need to manage the huge and diverse territory.</p><h3>Web 3</h3><p>A key thesis of Web 3 is decentralization. Web 3 proponents argue that decentralized organizations, social networks, and finance will solve problems of today’s centralized platforms and give us a new world. I think technologies like Bitcoin and Ethereum are great inventions and they will one day become the foundation of human society. However, I suggests that for a decentralized system like Web 3, we should also consider the need for centralization and be prepared for it. The worst thing that could happen is that, during a crisis, the community desperately needs a savior, so they accept help from someone who is malicious.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c843e5e83c91" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Major Causes of Data Breaches in 2022]]></title>
            <link>https://mingyizhao.medium.com/major-causes-of-data-breaches-in-in-2022-b7a141ffd861?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/b7a141ffd861</guid>
            <category><![CDATA[data-breach]]></category>
            <category><![CDATA[security]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Mon, 02 Jan 2023 15:38:00 GMT</pubDate>
            <atom:updated>2023-01-08T17:51:46.699Z</atom:updated>
            <content:encoded><![CDATA[<h3>Social Engineering</h3><p><a href="https://techcrunch.com/2022/04/04/mailchimp-internal-tool-breach/"><strong>Mailchimp says an internal tool was used to breach hundreds of accounts</strong></a></p><blockquote>In a statement given to TechCrunch, Mailchimp CISO Siobhan Smyth said the company became aware of the intrusion on March 26 after it identified a malicious actor accessing a tool used by the company’s customer support and account administration teams. Access was gained following a successful social engineering attack, a type of attack that exploits human error and uses manipulation techniques to gain private information, access or valuables.</blockquote><p><a href="https://techcrunch.com/2022/08/08/twilio-breach-customer-data/"><strong>Twilio hacked by phishing campaign targeting internet companies</strong></a></p><blockquote>The attack used SMS phishing messages that purported to come from Twilio’s IT department, suggesting that the employees’ password had expired or that their schedule had changed, and advised the target to log in using a spoofed web address that the attacker controls.</blockquote><h3>API Scraping</h3><p><a href="https://9to5mac.com/2022/08/08/twitter-data-breach/"><strong>Twitter belatedly confirms data breach which exposed contact details for 5.4M accounts</strong></a></p><blockquote>HackerOne covered the vulnerability back in January, which allowed anyone to enter a phone number or email address, and then find the associated twitterID. This is an internal identifier used by Twitter, but can be readily converted to a Twitter handle.</blockquote><h3>Configuration Error</h3><p><a href="https://techcrunch.com/2015/10/05/patreon-hacked-gigabytes-of-data-and-code-leaked/"><strong>Patreon Hacked, Gigabytes Of Data And Code Leaked</strong></a></p><blockquote>The data seems to have come from a debug version of the site that was visible to the Internet. The debug version included a “snapshot” of the production database.</blockquote><p><a href="https://www.rfa.org/english/news/china/breach-07152022152546.html"><strong>Shanghai data breach exposes suppression of ‘white-hat’ security research in China</strong></a></p><blockquote>When the programmer was using the ElasticSearch server to build a big data search system for the Shanghai Public Security Bureau, he backed up the data to Alibaba Cloud, but turned it into a data visualization website in error, making all the information downloadable or viewable through Kibana.</blockquote><h3>Third Party</h3><p><a href="https://techcrunch.com/2022/03/22/okta-january-hack-breach/"><strong>Okta confirms January breach after hackers publish screenshots of its internal network</strong></a></p><blockquote>In late January 2022, Okta detected an attempt to compromise the account of a third party customer support engineer working for one of our subprocessors. The matter was investigated and contained by the subprocessor.</blockquote><p><a href="https://techcrunch.com/2022/08/25/doordash-customer-data-breach-twilio/"><strong>DoorDash hit by data breach linked to Twilio hackers</strong></a></p><blockquote>DoorDash said malicious hackers stole credentials from employees of a third-party vendor that were then used to gain access to some of DoorDash’s internal tools.</blockquote><h3>Software Vulnerability</h3><p><a href="https://techcrunch.com/2023/01/06/rackspace-ransomware-data-exchange/"><strong>Rackspace says hackers accessed customer data during ransomware attack</strong></a></p><blockquote>According to the incident report update, Play threat actors gained access to Rackspace’s networks by exploiting CVE-2022–41080, a zero-day flaw patched by Microsoft in November that has been linked to previous ransomware incidents.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b7a141ffd861" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[There will always be enough jobs even as AI improves]]></title>
            <link>https://mingyizhao.medium.com/there-will-always-be-enough-jobs-even-as-ai-improves-9500cec0ce70?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/9500cec0ce70</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[labour]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Wed, 28 Dec 2022 16:24:14 GMT</pubDate>
            <atom:updated>2022-12-28T16:24:14.787Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*03V93c5ZFeRlt8ekkOWYcg.png" /><figcaption>Generated in DreamStudio with prompt: “Future world where people only work 4 hours per day and enjoy life more as AI gets better.”</figcaption></figure><p>Some people worry that as AI technology progresses, more people will lose their jobs due to their lower productivity compared to machines. This is true if we view the economy as a system that optimizes for production.</p><p>But what if the economy is actually optimizing for consumption? Thanks to the multiple waves of technological advancements in the past 200 years, there are already sufficient products available. To grow the economy, the key is to distribute money to more people so they can buy these products.</p><p>This is just a hypothesis, but if we assume it is true, then we can deduce something interesting: jobs are not just for producing, but also for paying. Therefore, there will always be enough jobs for people since there is always a need for wealth distribution to boost the consumption engine. For example, I recently talked to a staff engineer at Google who estimated that even with a 30% layoff, the company would still run fine. This indicates that Google’s 150K employees are likely working “inefficiently”, perhaps only 4 hours per day on average. But this might be fine since the purpose is to distribute wealth to these people so they can buy other products (e.g., Tesla), while still providing shareholders with a decent profit. If Google were to optimize for productivity by laying off 75% of the workforce, the reduced consumption could actually hurt shareholders’ profits, not to mention the potential for a bloody revolution in society that rich people would not want to see.</p><p>In summary, it is possible that even as technology advances, there will still be enough jobs for people, although the jobs may be less demanding. Perhaps in 10 years 4-hour per day work will become the norm. The released working hours can be used for consuming experience-oriented products (e.g., from Netflix, Meta, Snap), further boosting economic growth.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9500cec0ce70" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to tell if an app is really good?]]></title>
            <link>https://medium.com/design-bootcamp/how-to-tell-if-an-app-is-really-good-542bfdfbac86?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/542bfdfbac86</guid>
            <category><![CDATA[user-feedback]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[product]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Tue, 25 Jan 2022 03:14:28 GMT</pubDate>
            <atom:updated>2022-01-29T04:51:29.504Z</atom:updated>
            <content:encoded><![CDATA[<p>Every day, around 30 new products are launched on <a href="https://www.producthunt.com/">Product Hunt</a>. Many of them receive positive comments like:</p><ul><li>This is a great idea. Love the concept. It’s going to be the next big thing in XYZ!</li><li>This is a great product because of ...</li><li>I am so happy to see this. Really appreciate the work.</li></ul><p>But the sad reality is that majority of them will never gain meaningful traction. So these positive feedback actually contains very little information and should not be used for decision making.</p><p>What kind of positive feedback is really meaningful? To answer this question, I’ve looked at two apps, <a href="https://readwise.io/">Readwise </a>and <a href="https://hq.getmatter.app/">Matter</a>, that have gained traction overtime. I noticed that they at least have two kinds of unique feedback.</p><p><strong>Type 1</strong> — The user reports a <strong>behavioral change</strong> because of using the app. For example:</p><h3>Florian Kempenich on Twitter: &quot;This is not an advertisement, but let me tell you this:Readwise / @readwiseio has completely revolutionized the way I consume booksI believed that tech books were only useful in physical form, but w/ Readwise I&#39;m now reading everything on my Kindle 😍https://t.co/zmTz8yj29k pic.twitter.com/MQJ8WNY6rZ / Twitter&quot;</h3><p>This is not an advertisement, but let me tell you this:Readwise / @readwiseio has completely revolutionized the way I consume booksI believed that tech books were only useful in physical form, but w/ Readwise I&#39;m now reading everything on my Kindle 😍https://t.co/zmTz8yj29k pic.twitter.com/MQJ8WNY6rZ</p><h3>kv on Twitter: &quot;Cant believe I&#39;m going to say this but I love the @getmatterapp so much that I want it to send me notifs so I engage with it more. / Twitter&quot;</h3><p>Cant believe I&#39;m going to say this but I love the @getmatterapp so much that I want it to send me notifs so I engage with it more.</p><p><strong>Type 2</strong> — The user <strong>sticks to the app for a long period of time</strong>.</p><h3>Sid Jha on Twitter: &quot;Streak 🔥 of 31 days of reviewing my Kindle/iBooks highlights on @readwiseio. I&#39;ve already seen massive improvements in my ability to recall what I read and usage of quotes in my writing. Thanks @homsiT and @deadly_onion! Proud to be one of the early users :) pic.twitter.com/PTx17kFrHp / Twitter&quot;</h3><p>Streak 🔥 of 31 days of reviewing my Kindle/iBooks highlights on @readwiseio. I&#39;ve already seen massive improvements in my ability to recall what I read and usage of quotes in my writing. Thanks @homsiT and @deadly_onion! Proud to be one of the early users :) pic.twitter.com/PTx17kFrHp</p><h3>Pranav 🚀 on Twitter: &quot;I&#39;ve been using the @getmatterapp for a while and I&#39;m blown away. By far one of my favorite applications to save content &amp; read later.Not to mention the seamless UI &amp; voice integration. / Twitter&quot;</h3><p>I&#39;ve been using the @getmatterapp for a while and I&#39;m blown away. By far one of my favorite applications to save content &amp; read later.Not to mention the seamless UI &amp; voice integration.</p><p>These tweets are verdicts of solving real problems. In contrast, the earlier shallow comments only tell product is an interesting piece of artifact. But without solving real and painful problems, no one will use it.</p><p>That’s why as creators, we should <strong>obsess about the problem, not the solution</strong>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=542bfdfbac86" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/how-to-tell-if-an-app-is-really-good-542bfdfbac86">How to tell if an app is really good?</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[JavaScript Graph Visualization Libraries]]></title>
            <link>https://mingyizhao.medium.com/background-b553fda47349?source=rss-294d537d0cc0------2</link>
            <guid isPermaLink="false">https://medium.com/p/b553fda47349</guid>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[graph]]></category>
            <category><![CDATA[network]]></category>
            <category><![CDATA[data-visualization]]></category>
            <dc:creator><![CDATA[MingYi Zhao]]></dc:creator>
            <pubDate>Sat, 06 Mar 2021 17:57:21 GMT</pubDate>
            <atom:updated>2021-06-06T03:30:07.670Z</atom:updated>
            <content:encoded><![CDATA[<h3>Ranking of JavaScript Graph Visualization Libraries</h3><h3>Background</h3><p>I am working on a Web product that needs to visualize and manipulate graph / network data using JavaScript. Obviously, I can use the omnipotent D3.js library, but I am also curious to learn graph visualization specific libraries that can provides higher level of abstraction. I did a bit search and found two nice overview articles:</p><ul><li><a href="https://github.com/anvaka/graph-drawing-libraries">JavaScript Graph Drawing Libraries</a></li><li><a href="https://elise-deux.medium.com/the-list-of-graph-visualization-libraries-7a7b89aab6a6">List of graph visualization libraries</a></li></ul><p>They both provide a comprehensive list of libraries. However, which library should a reader check out first? If following the alphabetical order, then one actually encounters Alchemy and arbor, which haven’t been updated for years. It would be nice if the list can be sorted so that libraries that are still being actively updated, and libraries that have higher popularity are ranked top.</p><p>Therefore, I created a simple Interest Score for this purpose. Below is a ranked list of JavaScript graph visualization libraries. For details on how this score is calculated, please go to the end of this document.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fairtable.com%2Fembed%2FshrNwGULKztHreqS3&amp;display_name=Airtable&amp;url=https%3A%2F%2Fairtable.com%2FshrNwGULKztHreqS3&amp;image=https%3A%2F%2Fstatic.airtable.com%2Fimages%2Foembed%2Fairtable.png&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=airtable" width="800" height="533" frameborder="0" scrolling="no"><a href="https://medium.com/media/6c595582895cc47ac3ecde9f0f73e9c1/href">https://medium.com/media/6c595582895cc47ac3ecde9f0f73e9c1/href</a></iframe><p>I want to emphasize that this score is NOT a measure of how good a library is. All these libraries are the result of great design and engineering effort. We should sincerely appreciate the work of the authors.</p><p>Below are details of the top 5 libraries.</p><h3>1. G6</h3><p><strong>Interest score</strong>: 1.0</p><p><strong>Description</strong>: G6 is a graph visualization engine, which provides a set of basic mechanisms, including rendering, layout, analysis, interaction, animation, and other auxiliary tools. G6 aims to simplify the relationships, and help people to obtain the insight of relational data.</p><p><a href="https://github.com/antvis/g6">Github link</a></p><h3>2. Cytoscape.js</h3><p><strong>Interest score</strong>: 0.9</p><p><strong>Description</strong>: Cytoscape.js is a fully featured graph theory library. Do you need to model and/or visualise relational data, like biological data or social networks? If so, Cytoscape.js is just what you need.</p><p><a href="https://github.com/cytoscape/cytoscape.js">Github link</a></p><h3>3. vis.js — Network</h3><p><strong>Interest score</strong>: 0.8</p><p><strong>Description</strong>: Network is a visualization to display networks and networks consisting of nodes and edges. The visualization is easy to use and supports custom shapes, styles, colors, sizes, images, and more. The network visualization works smooth on any modern browser for up to a few thousand nodes and edges. To handle a larger amount of nodes, Network has clustering support. Network uses HTML canvas for rendering.</p><p><a href="https://github.com/visjs/vis-network">Github link</a></p><h3>4. VivaGraph</h3><p><strong>Interest score</strong>: 0.8</p><p><strong>Description</strong>: VivaGraphJS is designed to be extensible and to support different rendering engines and layout algorithms. Underlying algorithms have been broken out into ngraph.</p><p><a href="https://github.com/anvaka/VivaGraphJS">Github link</a></p><h3>5. dagre</h3><p><strong>Interest score</strong>: 0.8</p><p><strong>Description</strong>: Dagre is a JavaScript library that makes it easy to lay out directed graphs on the client-side.</p><p><a href="https://github.com/dagrejs/dagre">Github link</a></p><h3>Interest Score Calculation</h3><p>Here is how the interest score is calculated</p><p><strong>Interest Score</strong> = 0.3 * <strong>Recency Score</strong> + 0.7 * <strong>Popularity Score</strong></p><p><strong>Recency Score</strong> = 1-log(days since last commit) / log(max of days since last commit)</p><p><strong>Popularity Score</strong> = log(Github stars) / log(max of Github stars)</p><p>That’s it. Thank you for reading this article. Please feel free to share your comments!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b553fda47349" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>