Hacker News

Article URL: https://medium.com/towards-artificial-intelligence/what-if-ai-models-learned-like-humans-do-c69c19f29d0c

Comments URL: https://news.ycombinator.com/item?id=47135762

Points: 1

# Comments: 0

Article URL: https://github.com/RiccardoGrin/BankViz

Comments URL: https://news.ycombinator.com/item?id=47135761

Points: 1

# Comments: 0

I prefer using Markdown files when taking notes or writing. Even more so these days when working with AI. So I thought I'd build a Notion clone on top of GitHub.

You can try it out at pullnotes.com, or install it yourself: https://github.com/hunvreus/pullnotes

It's not perfect, but good enough for me to use it.

This is on my todo:

- Auto-save - File merge - Media upload - Drag and drop for pages

Comments URL: https://news.ycombinator.com/item?id=47135757

Points: 1

# Comments: 0

The IETF has specs for IP-over-HTTP/3 (MASQUE CONNECT-IP, RFC 9484) and Multipath QUIC, but no open-source implementation combines both. I implemented MASQUE CONNECT-IP on XQUIC (which already had Multipath QUIC), and wrote a new multipath scheduler designed for QUIC Datagrams, then built a VPN layer on that.

This scheduler (WLB) distributes TCP flows across paths proportional to capacity — with asymmetric paths, it reaches 319 Mbps (84% of theoretical max), +21% over the default MinRTT scheduler at 16 parallel flows. Failover is zero downtime.

Benchmarks and graphs in docs/benchmarks_netns.md.

Comments URL: https://news.ycombinator.com/item?id=47135745

Points: 1

# Comments: 0

Article URL: https://www.nature.com/articles/s44271-026-00428-5

Comments URL: https://news.ycombinator.com/item?id=47135736

Points: 1

# Comments: 0

Article URL: https://github.com/aiqualitylab/ai-natural-language-tests

Comments URL: https://news.ycombinator.com/item?id=47135715

Points: 1

# Comments: 0

Article URL: https://www.appsoftware.com/blog/openclaw-running-a-secure-capable-lowcost-claw-hetzner-tailscale-discord-zapier-mcp

Comments URL: https://news.ycombinator.com/item?id=47135694

Points: 1

# Comments: 0

We took a vague 2-sentence client request for a "Team Productivity Dashboard" and ran it through two different discovery processes: a traditional human analyst approach vs an AI-driven interrogation workflow.

The results were uncomfortable. The human produced a polite paragraph summarizing the "happy path." The AI produced a 127-point technical specification that highlighted every edge case, security flaw, and missing feature we usually forget until Week 8.

Here is the breakdown of the experiment and why I think "scope creep" is mostly just discovery failure.

The Problem: The "Assumption Blind Spot"

We’ve all lived through the "Week 8 Crisis." You’re 75% through a 12-week build, and suddenly the client asks, "Where is the admin panel to manage users?" The dev team assumed it was out of scope; the client assumed it was implied because "all apps have logins."

Humans have high context. When we hear "dashboard," we assume standard auth, standard errors, and standard scale. We don't write it down because it feels pedantic.

AI has zero context. It doesn't know that "auth" is implied. It doesn't know that we don't care about rate limiting for a prototype. So it asks.

The Experiment

We fed the same input to a senior human analyst and an LLM workflow acting as a technical interrogator.

Input: "We need a dashboard to track team productivity. It should pull data from Jira and GitHub and show us who is blocking who."

Path A: Human Analyst Output: ~5 bullet points. Focused on the UI and the "business value." Assumed: Standard Jira/GitHub APIs, single tenant, standard security. Result: A clean, readable, but technically hollow summary.

Path B: AI Interrogator Output: 127 distinct technical requirements. Focused on: Failure states, data governance, and edge cases. Result: A massive, boring, but exhaustive document.

The Results

The volume difference (5 vs 127) is striking, but the content difference is what matters. The AI explicitly defined requirements that the human completely "blind spotted":

- Granular RBAC: "What happens if a junior dev tries to delete a repo link?" - API Rate Limits: "How do we handle 429 errors from GitHub during a sync?" - Data Retention: "Do we store the Jira tickets indefinitely? Is there a purge policy?" - Empty States: "What does the dashboard look like for a new user with 0 tickets?"

The human spec implied these were "implementation details." The AI treated them as requirements. In my experience, treating RBAC as an implementation detail is exactly why projects go over budget.

Trade-offs and Limitations

To be fair, reading a 127-point spec is miserable. There is a serious signal-to-noise problem here.

- Bloat: The AI can be overly rigid. It suggested microservices architecture for what should be a monolith. It hallucinated complexity where none existed. - Paralysis: Handing a developer a 127-point list for a prototype is a great way to kill morale. - Filtering: You still need a human to look at the list and say, "We don't need multi-tenancy yet, delete points 45-60."

However, I'd rather delete 20 unnecessary points at the start of a project than discover 20 missing requirements two weeks before launch.

Discussion

This experiment made me realize that our hatred of writing specs—and our reliance on "implied" context—is a major source of technical debt. The AI is useful not because it's smart, but because it's pedantic enough to ask the questions we think are too obvious to ask.

I’m curious how others handle this "implied requirements" problem:

1. Do you have a checklist for things like RBAC/Auth/Rate Limits that you reuse? 2. Is a 100+ point spec actually helpful, or does it just front-load the arguments? 3. How do you filter the "AI noise" from the critical missing specs?

If anyone wants to see the specific prompts we used to trigger this "interrogator" mode, happy to share in the comments.

Comments URL: https://news.ycombinator.com/item?id=47135683

Points: 1

# Comments: 0

Article URL: https://www.kicad.org/blog/2026/02/Three-New-Importers-in-KiCad-10-Allegro-PADS-and-gEDA/

Comments URL: https://news.ycombinator.com/item?id=47135676

Points: 1

# Comments: 0

Article URL: https://mnt.io/articles/about-memory-pressure-lock-contention-and-data-oriented-design/

Comments URL: https://news.ycombinator.com/item?id=47135617

Points: 1

# Comments: 1

Built this because annotation teams working on serious AV/robotics datasets often can't send data to a third-party cloud, and existing self-hosted options have no real multi-user workflow.

Runs on Docker + PostgreSQL. WebGL renderer handles 1M+ points at 60fps in the browser. Full annotator/reviewer/admin roles with a review pipeline, issue tracking, and audit trails. Supports local filesystem or S3-compatible storage.

Happy to answer questions about the rendering approach or the multi-user architecture.

Comments URL: https://news.ycombinator.com/item?id=47135616

Points: 1

# Comments: 0

Article URL: https://github.com/theotrama/pico-tls

Comments URL: https://news.ycombinator.com/item?id=47135612

Points: 1

# Comments: 0

Article URL: https://hidden-phenomena.com/articles/rh

Comments URL: https://news.ycombinator.com/item?id=47135611

Points: 1

# Comments: 0

Article URL: https://www.wired.com/story/the-righteous-ev-owners-who-wont-let-their-broken-cars-die/

Comments URL: https://news.ycombinator.com/item?id=47135607

Points: 1

# Comments: 0

Article URL: https://github.com/cylonmolting-creator/agora-oracle

Comments URL: https://news.ycombinator.com/item?id=47135597

Points: 1

# Comments: 0

Article URL: https://github.com/remenoscodes/git-native-issue

Comments URL: https://news.ycombinator.com/item?id=47135581

Points: 1

# Comments: 1

Article URL: https://www.apple.com/newsroom/2026/02/apple-accelerates-us-manufacturing-with-mac-mini-production/

Comments URL: https://news.ycombinator.com/item?id=47135576

Points: 2

# Comments: 0

Perusing out of pure curiosity the May 1980 edition Practical Computing. I came upon a rather intriguing short story "And there was light" by a certain John Abbatt, which I was happily absorbed in until, without warning, pages 73 and 74 were nowhere to be found.

I understand that finding an excerpt nearly half a century old, from a little-known magazine and a little-known author, is bound to be challenging.

Comments URL: https://news.ycombinator.com/item?id=47135571

Points: 1

# Comments: 0

Article URL: https://github.com/GregoryLi360/Agentic-Browser-Automation

Comments URL: https://news.ycombinator.com/item?id=47135308

Points: 1

# Comments: 1

Article URL: https://developer.apple.com/news/?id=ueeok6yw

Comments URL: https://news.ycombinator.com/item?id=47135306

Points: 1

# Comments: 0