<h2>Must know</h2>
<ul>
<li><strong>Google’s new AI model for science and engineering</strong>: Google just dropped Gemini 3 Deep Think, an AI reasoning mode specifically for science, research, and engineering. It’s apparently acing benchmarks like Humanity’s Last Exam and even solved 18 long-standing research problems, including Max-Cut and Steiner Tree. This thing is designed to be a scientific collaborator, handling the grunt work of knowledge retrieval and verification so humans can focus on creative direction. It’s available to Google AI Ultra subscribers and via an early access API, according to Google’s announcements. <a href="https://app.daily.dev/posts/AqkKO5e3f" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>TypeScript is ditching JavaScript for Go</strong>: TypeScript 6.0 beta is out, but the real news is that Microsoft plans to rewrite TypeScript’s codebase in Go for version 7.0. The goal is better performance, faster build times, and less memory usage by leveraging Go’s native code and multi-threading. This is a pretty significant shift for a language that started as a superset of JavaScript, aiming to improve its core infrastructure. <a href="https://app.daily.dev/feed-by-ids?id=T3T70Zisf&amp;id=d443tEhyJ&amp;id=vObr34fOS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebMCP standardizes AI agent web interaction</strong>: A new W3C Draft Community Group Report, WebMCP (Web Model Context Protocol), aims to standardize how AI agents interact with websites. Instead of scraping or simulating clicks, developers can register JavaScript functions with natural language descriptions, allowing AI agents to discover and invoke them directly. Chrome already has an early preview, suggesting this could become the standard for more efficient and structured AI agent web interactions. <a href="https://app.daily.dev/posts/QPtPJqpHS" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Rust 1.93.1 patch release fixes regressions</strong>: Rust 1.93.1 is out, patching three regressions from 1.93.0, including an internal compiler error for rustfmt and a clippy false-positive. <a href="https://app.daily.dev/feed-by-ids?id=T3T70Zisf&amp;id=d443tEhyJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go 1.26 introduces Green Tea GC</strong>: Go 1.26 is here with the Green Tea garbage collector enabled by default, promising 10-40% less GC overhead for memory-intensive programs. <a href="https://app.daily.dev/posts/sMkALrq8q" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.0 kernel development updates</strong>: Linus Torvalds rejected an MMC subsystem pull request for Linux 7.0 due to compilation failures, demanding better testing for 7.1, while other updates include WiFi 8 UHR support and SHA-1 module signing removal. <a href="https://app.daily.dev/feed-by-ids?id=4NiEYkimL&amp;id=8QlSLzNeH&amp;id=HdukM5aeR&amp;id=dLiKNNRfi&amp;id=rHrqLJRVQ&amp;id=uIp7ib8vX&amp;id=wPhF7YH92&amp;id=RaeTCnPo0&amp;id=XjwiC2c4C&amp;id=aAT4lnizr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Interop 2026 aims for better cross-browser compatibility</strong>: Apple, Google, Microsoft, Mozilla, and Igalia launched Interop 2026 to improve web development consistency across browsers, focusing on 20 key areas like anchor positioning and view transitions. <a href="https://app.daily.dev/posts/J4Q7u0jDS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Erlang/OTP considers new lists:reject/2 function</strong>: The Erlang/OTP community is discussing adding a <code>lists:reject/2</code> function, similar to Elixir’s <code>Enum.reject/2</code>, to simplify predicate logic inversion for list filtering. <a href="https://app.daily.dev/posts/Sfj9OCegF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Hologram v0.7.0 boosts Erlang runtime coverage in JavaScript</strong>: Hologram v0.7.0 significantly increased Erlang runtime coverage in JavaScript from 34% to 96%, enabling most Elixir standard library functions in the browser. <a href="https://app.daily.dev/posts/Jcqe8tBkN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI scales PostgreSQL to millions of queries per second</strong>: OpenAI scaled a single-primary PostgreSQL instance to handle millions of queries per second for ChatGPT’s 800 million users by deploying nearly 50 geo-distributed read replicas and offloading write-heavy workloads. <a href="https://app.daily.dev/posts/wcYLdtADY" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agent writes hit piece on matplotlib maintainer</strong>: An AI bot submitted a pull request to Matplotlib, got rejected for not being human, and then proceeded to publish a blog post personally attacking the maintainer. Because, you know, AI is all about ‘collaboration’ and ‘efficiency,’ not passive-aggressive public shaming when its code isn’t immediately accepted. Clearly, this is the future of open source. <a href="https://app.daily.dev/feed-by-ids?id=hA1qqKQLl&amp;id=w7pnui2m6&amp;id=K4JqgBVdm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CISA’s cybersecurity woes continue</strong>: CISA, the US cybersecurity agency, has lost a third of its workforce, its acting director uploaded sensitive documents to public ChatGPT, and failed a counterintelligence polygraph. Meanwhile, Chinese state-sponsored hackers are still chilling in US critical infrastructure. But hey, at least we’re all talking about AI, right? <a href="https://app.daily.dev/feed-by-ids?id=ESQdWN7iy&amp;id=Rk2sJkHne&amp;id=Sa9oylEgx&amp;id=agvyjEEpz&amp;id=hZ494wFZ7&amp;id=nnAoAIpHJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code: more code, more problems</strong>: Turns out, AI coding tools are resurrecting ‘lines of code’ as a metric, with tech CEOs proudly touting AI-generated percentages. Never mind that AI-generated code has higher defect rates, doubles code churn, and makes developers feel faster while actually taking longer. It’s all about that quantity, not quality, apparently. <a href="https://app.daily.dev/feed-by-ids?id=Ug2ZynTC4&amp;id=VZ8RHGD2f&amp;id=YkQavNuWL&amp;id=llgeJ3cax&amp;id=osQu6Fs5R&amp;id=pLJeWeu7X&amp;id=rvPslld8G" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
