tl;dr: Om Malik, the influential tech journalist, founder of GigaOM, and partner at True Ventures, has died. The page is filled with tributes from prominent figures across Silicon Valley journalism, startups, and venture capital, who remember him as a pioneering Web 2.0-era writer, mentor, photographer, and uncommonly kind presence whose insightful coverage of telecom, mobile, and cloud helped shape the industry.

tl;dr: Researchers have virtually unwrapped and read PHerc. 1667 in full — the first Herculaneum scroll read end-to-end without physical unrolling — using high-resolution phase-contrast X-ray microtomography at the ESRF combined with machine learning to detect faint ink traces. The recovered text appears to be a 2nd-century BC Stoic philosophical treatise on ethics, naming Aristocreon (disciple of Chrysippus). Two other scrolls yielded results too: independent confirmation of the 2023 Grand Prize reading via direct 3D ink detection, and identification of PHerc. 139 as Philodemus' On Gods, Book 8. All data and code are openly released.

tl;dr: Australia's under-16 social media ban, now being emulated by the UK, EU, and various US states/federal proposals like KOSA, forces platforms to verify users' ages via government IDs, biometrics, or third-party services—creating massive privacy risks, as evidenced by a recent Discord breach exposing 70,000 Australians' ID data. Beyond data breach exposure, these mandates effectively end online anonymity, chill speech on sensitive topics, and may extend to cracking down on VPNs. The author argues this "papers, please" infrastructure of surveillance, once built, will be nearly impossible to dismantle—and isn't even effectively keeping kids off social media.

tl;dr: A browser-based game (also available as a desktop app via Python) that puts you in the role of an OS, requiring you to juggle process scheduling, memory allocation, and I/O events before impatient users reboot you. Built in Python with a web version and sandbox/automation modes for development, it's open source under GPLv3. Playable at https://plbrault.github.io/youre-the-os or on itch.io.

tl;dr: Summary not available

tl;dr: IBM unveiled a 0.7 nm ("7 angstrom") chip using a new "nanostack" 3D architecture that vertically stacks and staggers nanosheet transistors, fitting ~100 billion transistors on a fingernail-sized die. The company claims up to 50% better performance or 70% better energy efficiency over its 2 nm node, plus 40% SRAM scaling, with production possibly within five years. Work continues at IBM's Albany facility alongside ASML High-NA EUV and partners like Lam, TEL, and SCREEN.

tl;dr: OpenKnowledge is a local-first, GPL-3.0 markdown editor and LLM wiki positioned as an open-source alternative to Obsidian and Notion, with WYSIWYG editing and built-in integrations for Claude, Codex, and Cursor via MCP/CLI. It ships as a macOS app or a CLI-launched web app for Linux/Windows/Intel Mac, and uses git/GitHub under the hood for team sharing and sync.

tl;dr: Summary not available

tl;dr: Summary not available

tl;dr: OS9Map is a native OpenStreetMap browser for Mac OS 9 on PowerPC, supporting smooth pan/zoom, Nominatim-based address search, and bookmarks. It requires 16MB RAM and an Open Transport TCP/IP connection. Version 1.0.0 marks the first public release.

tl;dr: Summary not available

tl;dr: Zig's `@bitCast` has been redefined with new semantics based on a type's "logical bit layout" rather than reinterpreting memory bytes, making behavior endian-agnostic and enabling operations like converting `[2]u3` to `@Vector(3, u2)`. The change was driven by LLVM backend improvements that now extend arbitrary bit-width integers (u4, u13, etc.) to ABI-sized types in memory—matching Clang's `_BitInt(N)` lowering—which restored missed optimizations and yielded ~5% performance gains in the Zig compiler itself. The new semantics are implemented across all backends and comptime execution, landing in 0.17.0.

tl;dr: Summary not available

tl;dr: A developer building a virtual running app (In the Long Run) built a pipeline using GeoNames, Wikipedia, DuckDB, and Parquet to surface points of interest along routes, with Claude Haiku providing subjective relevance ratings. Hallucinations forced him to abandon LLM-generated summaries in favor of Wikipedia text, relegating AI to a scoring role alongside traditional signals like Wikipedia language counts. The hardest part was evaluation: there's no ground truth or unit test for "taste," and each route needed custom tuning to balance natural, historical, and populated landmarks.

tl;dr: Hacker Trends is a Google Trends–style tool that charts how often topics, tools, or people appear across 18 years of Hacker News posts and comments, built on a date-histogram over 45M items using Upstash Redis Search. Users can overlay multiple terms to compare traction over time and drill down into the underlying stories and comments. The site includes curated comparisons illustrating shifts like Webpack→Vite, MySQL→Postgres, TensorFlow→PyTorch→JAX, and Cursor→Claude Code→Codex.

tl;dr: Summary not available

tl;dr: Summary not available

tl;dr: LastPass is notifying users of a breach involving third-party market research firm Klue, which exposed customer contact info and support case data (but not password vaults). The attackers accessed data via Klue's integrations with Salesforce and Gong; LastPass has revoked access, rotated API tokens, and warned users to watch for phishing. This marks the latest in a string of security incidents for LastPass, following major breaches in 2015 and 2022.

tl;dr: OpenAI announced its first custom inference chip, codenamed Jalapeño, developed with Broadcom and reportedly designed with help from OpenAI's own AI models. The chip targets inference workloads (not training) and claims significantly better performance-per-watt than current alternatives, aimed at reducing reliance on Nvidia GPUs and cutting operating costs for running models like Codex. It follows similar custom-silicon moves by Google and Amazon, and is still in testing.

tl;dr: Cloudflare has opened self-managed OAuth to all customers, letting developers create their own OAuth applications for delegated API access instead of relying on API tokens. To support this, they upgraded their underlying Hydra OAuth engine through a staged 1.X then 2.X migration, using a blue-green strategy with a Cloudflare Queues-based revocation replay system to avoid downtime. The upgrade yielded notable performance gains, including a 45% drop in P95 API latency and 37% lower CPU usage.