Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Europe's e‑waste is a valuable resource – let's not throw it away(theconversation.com ↗)
    discuss
  2. Software Engineer: I'm Tired of Pretending [video](youtube.com ↗)
    discuss
  3. Show HN: S-Roll – local AI video understanding and clipping for Mac(saliency.dev ↗)
    discuss
  4. Show HN: Built a Simulation of Autonomous Agents(agents.london ↗)
    discuss
  5. Noam Brown – Agent swarms, alignment, & recursive self-improvement(dwarkesh.com ↗)
    discuss
  6. Show HN: Custom Relevance – Describe what matters and let Jev rank it(foglight.co ↗)
    discuss
  7. Feeling overwhelmed by the AI doom loop? Here's the essential reading list(theguardian.com ↗)
    discuss
  8. Google Noto 3D Emoji Design Process(design.google ↗)
    1comments
  9. Becoming a Benchmark(terrytao.wordpress.com ↗)
    discuss
  10. Sourcerer: Fast Git client written in C++ version 1.0.24 released(sourcererapp.com ↗)
    discuss
  11. The Download: mice with part-human brains and climate tech innovators(technologyreview.com ↗)
    discuss
  12. Should OpenAlex publish a novelty score? A vibe check(openalex.org ↗)
    discuss
  13. Show HN: Smart Crosshair – A lightweight C++/Win32 adaptive contrast crosshair(steampowered.com ↗)
    discuss
  14. I didn't sign the Fields medallists' letter(terrytao.wordpress.com ↗)
    discuss
  15. Towards Self-Driving Codebases(detail.dev ↗)
    discuss
  16. 4-Bit Rotational Quantization: -45% RAM, <1% recall drop vs. TurboQuant(weaviate.io ↗)
    discuss
  17. Show HN: Repodify: Make Podcasts Out of Podcasts(repodify.app ↗)
    1comments
  18. Palantir's Karp: AI needs to have 'reasonable guidelines,'(cnbc.com ↗)
    1comments
  19. Aegis: Zero-GC 64-byte cache-aligned memory arena in C++20 (1B ops in 0.649s)(github.com/markbgilbert ↗)
    discuss
  20. Ask HN: Co-Founder(s). Do I need any? How would I even find them?
    discuss
  21. From Stonemasons to Carpenters(thelastsoftwareengineer.substack.com ↗)
    discuss
  22. I Made Turn-Based Combat into a Database(louisxu3.substack.com ↗)
    discuss
  23. Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data(arxiv.org ↗)
    discuss
  24. Show HN: Pappice live demo in browser via WASM(pappice.eu ↗)
    discuss
  25. Show HN: AutoBot – live voice control for long-running AI work(github.com/demeyer1 ↗)
    discuss
  26. Show HN: Craigslist for agent skills, curated by a human(skillbay.sh ↗)
    discuss
  27. The Bicycle and the Algorithm: Amber Case on Why AI Has It Backwards(designwhine.com ↗)
    discuss
  28. Royal Enfield's Flying Flea C6 Touches Down in Europe(irishnews.com ↗)
    discuss
  29. Show HN: Dishlist – my favorite things on the menu(testflight.apple.com ↗)
    discuss
  30. Show HN: Content-aware PII redaction with Jev in Postgres(pg-redact.vercel.app ↗)
    discuss

Agentic Societies Need a Social Harness

1 pointsby 1h agosocial-harness.org
1 comments
1h agoHN ↗

Author here, I am finishing up my PhD at UW, with Arvind Krishnamurthy and Ratul Mahajan.

A few months ago, when OpenClaw came out, we were super curious to learn more about how people were using personal agents and it seemed that a lot of it was around communication management: managing calendars, drafting replies to emails, and so on. If soon everyone will have such an agent (and it seems like it), then these agents should be able to communicate autonomously without humans in the loop.

The whole thing seemed easy enough since agents are good at using messaging services, but we soon started observing that challenges arise higher up the stack. We went back into research mode and set up some experiments to learn what are the different ways that things could fail, and came out with some pretty interesting findings.

Today, our intuitions indicate that agents are quite similar to humans, when it comes to the ways that they fail. They can lack social norms regarding what to talk about and when to talk about it, and are gullible and hence can be exploited by others. However, they are also similar to classical distributed systems when it comes to techniques that can be applied to solve them.

This is super early, and this post mainly focuses on the interesting failures we observed. The proposed architecture for a social harness (which will complement existing personal harnesses) is described in the paper [1] and we will discuss it more in subsequent posts here as well.

[1] https://arxiv.org/pdf/2609.17527