Episode Details
Back to EpisodesHacker Newsroom for 26 July: Android ADB Limits, Hannah Fry Prize, Claude Opus 5, Postgres Notifications
Description
Hacker Newsroom for 26 July recaps major Hacker News stories, moving through android adb limits, hannah fry prize, claude opus 5, postgres notifications.
1. Android ADB Limits
The next story is a Kitsumed blog post warning that Android may soon restrict on-device ADB after a Google ADB maintainer suggested binding the daemon to the Wi-Fi interface instead of localhost, a change the post says would break Shizuku, libadb-based tools, rootless call recording, and phone-only developer workflows. The blog post argues this is not an official announcement and that loopback ADB could be disabled by default if Google wants more safety, but should remain available behind a persistent opt-in because malicious apps still need multiple manual user actions before they can abuse it.
2. Hannah Fry Prize
The next story is about Cambridge mathematician and broadcaster Hannah Fry winning the 2026 Leelavati Prize, an award from the International Congress of Mathematicians for increasing public awareness of mathematics. The Cambridge article presents it as recognition for Fry’s outreach career, from books and documentaries to her current role as Professor of the Public Understanding of Mathematics, treating math communication as work that matters in its own right.
3. Claude Opus 5
The next story is about Artificial Analysis putting Claude Opus 5 at the top of its AI intelligence leaderboard, a comparison page that ranks hundreds of models across benchmark performance, price, speed, latency, and context window. The post matters because the gap at the top looks narrow, and the data also suggests that cheaper or smaller variants can come surprisingly close once you factor in cost per task, token efficiency, and different reasoning settings.
4. Postgres Notifications
The next story is about a DBOS post arguing that Postgres LISTEN/NOTIFY can scale much better than its reputation suggests, if you stop firing a notification on every write and instead buffer and batch those signals, which let the team move from about 2. 9 thousand to 60 thousand stream writes per second with millisecond latency on one server.
5. Open Weight AI
The next story argues that open-weight AI is hitting a Kubernetes-style moment, where good enough base models plus open tooling can turn a model family into a platform, and it warns that a U. S.
6. Interview Ghosting
The next story is Did They Ghost You, a project built around the blunt question in its subtitle: Ghosted After a Job Interview? Report the Company.
That's it for today, I hope this is going to help you build some cool things.