Hacker News Digest
Tuesday, May 19, 2026
In This Issue
- Hacker News
- Elon Musk has lost his lawsuit against Sam Altman and OpenAI
- Show HN: Files.md – open-source alternative to Obsidian
- Garry Tan, the CEO of venture YC, accused me of unethical reporting
- The last six months in LLMs in five minutes
- We stopped AI bot spam in our GitHub repo using Git's –author flag
- Anthropic Acquires Stainless
- Codex-maxxing
- What political censorship looks like inside an LLM's weights (Qwen 3.5)
- Open-Source Agentic QA Harness with Memory
Zipper Data Brief
May 19, 2026
Your daily digest of the best from Hacker News
Top 6 Trending
#1
985 points
· nycdatasci
· comments
Discussion Summary
Musk lost his lawsuit against OpenAI on a statute of limitations technicality rather than on the merits, with many commenters speculating his real goal was generating bad publicity and discovery rather than winning, while others debate whether the nonprofit-to-for-profit conversion deserves scrutiny regardless of the lawsuit outcome.
Musk lost his lawsuit against OpenAI on a statute of limitations technicality rather than on the merits, with many commenters speculating his real goal was generating bad publicity and discovery rather than winning, while others debate whether the nonprofit-to-for-profit conversion deserves scrutiny regardless of the lawsuit outcome.
#2
645 points
· zakirullin
· comments
Discussion Summary
The discussion reveals strong interest in Obsidian alternatives, with commenters praising Files.md's approach to automating metadata but debating whether open-source, native, lightweight tools or existing solutions like Joplin and Logseq better serve markdown-based note-taking. Key themes include frustration with Electron apps, the desire for true data ownership, and skepticism about whether any single tool can replace Obsidian's plugin ecosystem.
The discussion reveals strong interest in Obsidian alternatives, with commenters praising Files.md's approach to automating metadata but debating whether open-source, native, lightweight tools or existing solutions like Joplin and Logseq better serve markdown-based note-taking. Key themes include frustration with Electron apps, the desire for true data ownership, and skepticism about whether any single tool can replace Obsidian's plugin ecosystem.
#3
525 points
· gok
· comments
Discussion Summary
Commenters largely criticize Garry Tan for attacking journalist Radley Balko's reporting on criminal justice as unethical, arguing Tan misunderstands free speech and is using his wealth to suppress legitimate journalism that contradicts his preferred narrative. The consensus is that Balko's work represents quality reporting despite his acknowledged progressive lean, while Tan's response demonstrates a troubling Silicon Valley pattern of dismissing criticism as corruption.
Commenters largely criticize Garry Tan for attacking journalist Radley Balko's reporting on criminal justice as unethical, arguing Tan misunderstands free speech and is using his wealth to suppress legitimate journalism that contradicts his preferred narrative. The consensus is that Balko's work represents quality reporting despite his acknowledged progressive lean, while Tan's response demonstrates a troubling Silicon Valley pattern of dismissing criticism as corruption.
#4
503 points
· yakkomajuri
· comments
Discussion Summary
The discussion reflects sharp disagreement about LLM progress: while some see genuine breakthroughs in coding agents and agentic AI enabling dramatic productivity gains, others argue the improvements are marginal and overhyped, with concerns about job displacement, security vulnerabilities, and the hype cycle around benchmarks like the "pelican on a bicycle" test.
The discussion reflects sharp disagreement about LLM progress: while some see genuine breakthroughs in coding agents and agentic AI enabling dramatic productivity gains, others argue the improvements are marginal and overhyped, with concerns about job displacement, security vulnerabilities, and the hype cycle around benchmarks like the "pelican on a bicycle" test.
#5
472 points
· ildari
· comments
Discussion Summary
The discussion reveals that while the featured repo used a clever git workaround to filter bot spam, the root problem persists—GitHub lacks basic safeguards like CAPTCHAs or deposit systems, and the broader issue stems from hype around AI code generation incentivizing low-quality automated submissions.
The discussion reveals that while the featured repo used a clever git workaround to filter bot spam, the root problem persists—GitHub lacks basic safeguards like CAPTCHAs or deposit systems, and the broader issue stems from hype around AI code generation incentivizing low-quality automated submissions.
#6
468 points
· tomeraberbach
· comments
Discussion Summary
Anthropic acquired Stainless (an SDK generation tool) in an acquihire, immediately shutting down the product—a move seen as talent acquisition and competitive consolidation that frustrates existing customers and raises concerns about AI companies using venture capital to eliminate developer tools rather than compete fairly.
Anthropic acquired Stainless (an SDK generation tool) in an acquihire, immediately shutting down the product—a move seen as talent acquisition and competitive consolidation that frustrates existing customers and raises concerns about AI companies using venture capital to eliminate developer tools rather than compete fairly.
AI / Machine Learning
88 points
· dnw
· comments
Discussion Summary
The discussion is largely critical of "Codex-maxxing"—obsessively using LLMs for coding and work—with commenters questioning whether it produces real outcomes or just creates an illusion of productivity, while a few defend specific use cases like accessibility work or managing multiple AI agents on concrete projects.
The discussion is largely critical of "Codex-maxxing"—obsessively using LLMs for coding and work—with commenters questioning whether it produces real outcomes or just creates an illusion of productivity, while a few defend specific use cases like accessibility work or managing multiple AI agents on concrete projects.
79 points
· s314
· comments
Discussion Summary
The discussion centers on whether political censorship in LLMs like Qwen 3.5 can be detected and manipulated through analysis of model weights, with commenters raising concerns about the article's credibility (possibly LLM-generated), discussing techniques to identify censorship circuits, and questioning whether obfuscation will eventually hide such mechanisms.
The discussion centers on whether political censorship in LLMs like Qwen 3.5 can be detected and manipulated through analysis of model weights, with commenters raising concerns about the article's credibility (possibly LLM-generated), discussing techniques to identify censorship circuits, and questioning whether obfuscation will eventually hide such mechanisms.
50 points
· pranshuchittora
· comments
Discussion Summary
The creator introduced agent-qa, an AI-powered testing tool that writes tests in plain English using Playwright/Appium while learning from each run, addressing the testing bottleneck created by AI-assisted coding. Commenters questioned its necessity compared to existing solutions and asked about Claude integration compatibility.
The creator introduced agent-qa, an AI-powered testing tool that writes tests in plain English using Playwright/Appium while learning from each run, addressing the testing bottleneck created by AI-assisted coding. Commenters questioned its necessity compared to existing solutions and asked about Claude integration compatibility.
233 points
· theanonymousone
· comments
Discussion Summary
Users praise Qwen's rapid iteration and strong performance of the 3.6 models for local inference, while expressing concerns about fragmented benchmarking standards and hoping the company maintains its open-source commitment as models approach proprietary performance levels.
Users praise Qwen's rapid iteration and strong performance of the 3.6 models for local inference, while expressing concerns about fragmented benchmarking standards and hoping the company maintains its open-source commitment as models approach proprietary performance levels.
152 points
· Bender
· comments
Discussion Summary
The discussion praises the article's catchy headline while debating whether AI hype is driven by manufactured FOMO, with commenters noting that companies are increasingly skeptical as AI projects fail to deliver ROI and the real competitive threat lies in how AI empowers smaller teams to disrupt established businesses. There's broad agreement that moving slowly with concrete use cases is wiser than rushing into AI adoption driven by fear of missing out.
The discussion praises the article's catchy headline while debating whether AI hype is driven by manufactured FOMO, with commenters noting that companies are increasingly skeptical as AI projects fail to deliver ROI and the real competitive threat lies in how AI empowers smaller teams to disrupt established businesses. There's broad agreement that moving slowly with concrete use cases is wiser than rushing into AI adoption driven by fear of missing out.
Startups / Business
18 points
· jbredeche
· comments
Discussion Summary
A jury ruled in favor of OpenAI and Sam Altman in Elon Musk's lawsuit, rejecting claims that the company breached its founding agreement by pursuing for-profit ventures. The decision suggests courts found insufficient evidence that OpenAI violated its original non-profit mission commitment to Musk.
A jury ruled in favor of OpenAI and Sam Altman in Elon Musk's lawsuit, rejecting claims that the company breached its founding agreement by pursuing for-profit ventures. The decision suggests courts found insufficient evidence that OpenAI violated its original non-profit mission commitment to Musk.
46 points
· mrcoldbrew
· comments
Discussion Summary
InsForge, an open-source Heroku-like platform for coding agents, has launched with positive reception for its smooth onboarding experience and potential to consolidate multiple services. The team is prioritizing safety features including dynamic permissions for agents and reversible operations with Git-based snapshots.
InsForge, an open-source Heroku-like platform for coding agents, has launched with positive reception for its smooth onboarding experience and potential to consolidate multiple services. The team is prioritizing safety features including dynamic permissions for agents and reversible operations with Git-based snapshots.
6 points
· simonmysun
· comments
Discussion Summary
No comments available.
No comments available.
More Stories (36)
313 points
· mitchbob
· comments
240 points
· LucidLynx
· comments
148 points
· geox
· comments
106 points
· thm
· comments
140 points
· redbell
· comments
114 points
· ck2
· comments
88 points
· jgrahamc
· comments
68 points
· saikatsg
· comments
55 points
· giuliomagnifico
· comments
322 points
· stephenlthorn
· comments
48 points
· Gaishan
· comments
52 points
· KnuthIsGod
· comments
318 points
· srameshc
· comments
228 points
· ankitg12
· comments
36 points
· momentmaker
· comments
35 points
· Kapura
· comments
309 points
· tekkertje
· comments
42 points
· Teever
· comments
36 points
· aworks
· comments
37 points
· JumpCrisscross
· comments
33 points
· JumpCrisscross
· comments
23 points
· Crunsher
· comments
107 points
· JaakkoP
· comments
22 points
· Tomte
· comments
44 points
· Reaktornano
· comments
27 points
· ent101
· comments
147 points
· speckx
· comments
19 points
· fodmap
· comments
281 points
· lukaspetersson
· comments
65 points
· dogscatstrees
· comments
207 points
· jonbaer
· comments
16 points
· reaperducer
· comments
16 points
· zzzeek
· comments
65 points
· surganov
· comments
23 points
· mhb
· comments
Created by Zipper Data Co.
· 2026-05-19 12:01 UTC
· Unsubscribe
Get digests like this delivered to your inbox every morning.
Subscribe Free