Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. AI-generated posters don’t have to be horrible(john.hartnup.uk ↗)
    43comments
  2. Human brain is two separate organs, Stanford Medicine-led research finds(stanford.edu ↗)
    104comments
  3. If math is more than proof, we need to better celebrate the rest of it(terrytao.wordpress.com ↗)
    73comments
  4. GPT-6 Astra Solves a WWI German Radio Cipher(prinzai.com ↗)
    60comments
  5. Android 17 is the first since 3.x to add new APIs without releasing to the AOSP(grapheneos.social ↗)
    421comments
  6. San Francisco Onion Futures Company(onionfutures.com ↗)
    70comments
  7. Apple M6 Pro Achieves the Highest Single-Core CPU Score in Geekbench 7(geekbench.com ↗)
    41comments
  8. "The Secret Life of Circuits" is here(coredump.cx ↗)
    14comments
  9. Cloudflare Quick Tunnels(cloudflare.com ↗)
    285comments
  10. SDCC – Small Device C Compiler(sourceforge.net ↗)
    18comments
  11. Science Is Open Software(jepedersen.dk ↗)
    38comments
  12. How to Write with an LLM(sockpuppet.org ↗)
    335comments
  13. Saving another 100TB of RAM(cloudflare.com ↗)
    79comments
  14. Why building a Rust LSP is hard(rust-glancer.github.io ↗)
    32comments
  15. You can run Git on object storage if you re-make packfiles(tigrisdata.com ↗)
    15comments
  16. How OpenAI Used Its Own LLMs to Design Its Jalapeño Chip(ieee.org ↗)
    88comments
  17. Ctenophores: Wonders of Biology(quantamagazine.org ↗)
    6comments
  18. Communication by means of modulated Johnson noise(pnas.org ↗)
    discuss
  19. NASA-IBM Lunar Foundation open-Source Geospatial AI Model(usra.edu ↗)
    2comments
  20. The first new cat species discovered in 100 years(nationalgeographic.com ↗)
    105comments
  21. OpenJev(openjev.com ↗)
    267comments
  22. Goroutine Leak Profiles(go.dev ↗)
    4comments
  23. Show HN: Cactus Needle 3: 8-29MB automation models can match DeepSeek V4 Flash(cactuscompute.com ↗)
    89comments
  24. Photon-Emission-Guided Laser Fault Injection Enables RP2350 Secure Debug(ledger.com ↗)
    71comments
  25. Typesafe-computer-use drives a Mac toward a goal for 1/50th of a cent per step(github.com/awlevin ↗)
    52comments
  26. Cache-to-Cache: Direct Semantic Communication Between LLMs (2025)(arxiv.org ↗)
    13comments
  27. Warez: The Infrastructure and Aesthetics of Piracy (2021)(archive.org ↗)
    71comments
  28. Roman Timekeeping(wikipedia.org ↗)
    discuss
  29. Inside ZCode: Silently uploading your Git history to the cloud(ferstar.org ↗)
    100comments
  30. Cyclomatic Complexity in C#(ndepend.com ↗)
    23comments

Shirky: A Group Is Its Own Worst Enemy

57 pointsby 15y agoshirky.com
15 comments
15y agoHN ↗

Thanks for that, the full text is a much better read than Jeff's cut down version.

15y agoHN ↗

Great, informative read. Have to complain about this bit tho:

But learning from experience is the worst possible way to learn something. Learning from experience is one up from remembering. That's not great. The best way to learn something is when someone else figures it out and tells you: "Don't go in that swamp. There are alligators in there."

Works great for avoiding alligators. Fails miserably for riding bikes.

15y agoHN ↗

I dunno about that. I think it's probably easier to learn to ride a bike with someone experienced helping you.

If someone sets up the seat for you and explains how to shift and brake and rides with you a little bit, I'm sure learning would go much faster than if you were just left in a driveway with a bike and told to figure it out.

15y agoHN ↗

Not to detract from your point necessarily, but my anecdotal experience contrasts with this. I had no luck being told how to ride (I guess someone skilled in teaching rather than bike riding may have helped). In the end I was literally left in the driveway with a bike and not allowed back inside until I could ride. I picked it up in an afternoon.

15y agoHN ↗

When learning to ride a bike brakes and shifters are really irrelevant. You're going to crash anyway.

As for how I learned to ride, my dad took me to a hill, sat me on the bike and pushed me down. I learned how to ride pretty quick...well until I hit a small stump hidden by the grass.

After that I was mostly fine and learned to ride pretty quick.

15y agoHN ↗

Indeed.

If you want a kid to learn to ride a bike, remove the fear of failure. Do it on a gently sloping grassy hill that levels out, for example.

When I first learned to ice skate I learned that falling on ice fucking hurts. I didn't learn to ice skate. I learned to not fall over. Then my friend dressed me up in his team's goalie's pads - I pretty much bounced back up again if I fell over. By the end of an hour I could do the sprint exercises with the team and keep up (over short distances anyway!).

15y agoHN ↗

Yeah, I flinched at that part too, but for a different reason: it assumes that everyone else is honest and has the same goals you do.

Hiking destinations. I have a longtime friend and hiking buddy who really likes tough, exposed hikes up to the top of something, with a perverse derision towards anything resembling a freshwater source anywhere along the hike. But he also can't stand mosquitoes. At all. So, he might say, "Don't go in that meadow, there are mosquitoes in there." For me, they're annoying, but I like little hidden meadows, so maybe I want to go there.

And, also, there's a chance that you'll figure out something that nobody else has. You go in that swamp, even though someone warned you about the alligators, and you manage to avoid the alligators but you get a cut while you're in there and you discover that the swamp muck does something interesting with cuts, it actually disinfects them and helps them heal better.

My dad used to say that a smart man learns from his mistakes and a wise man learns from others'. The older I get, the more skeptical I become about that.

15y agoHN ↗

More likely, some exotic bacteria gets in the cut and you develop necrotizing fascitis.

Learning from your own mistakes instead of others' is a high-variance strategy. Our culture is better off because of the people that do it, but if you're selfishly looking to maximize your own expected value, you'll put great weight on the majoritarian view.

15y agoHN ↗

I am often grateful that humans are not perfect game theory machines.

15y agoHN ↗

It would certainly be depressing to live in a world of John Nashes. On the other hand, if everyone were perfect bounded rationalists, we'd merely have to replace Causal Decision Theory and Evidential Decision Theory with something that can solve coordination problems--probably a variant of Timeless Decision Theory, which could already solve this particular problem in many ways.

15y agoHN ↗

LambdaMOO is a fascinating virtual world. Users are allowed to expand the virtual world by building areas, and users with the Programmer Bit are also allowed to program new objects and areas.

15y agoHN ↗

Is there any MUD/MUSH/MOO where this doesn't happen? SMAUG allows you to do that as well.

15y agoHN ↗

A recklessness on the part of the author within an otherwise thought-provoking set of ideas: his analogy concerning Calvinism and the "Book of Revelations" limps badly for a strawman. Evidently, not much time was spent reading John Calvin's "Brève Instruction chrétienne" (Brief Outline of the Christian Faith) nor "Revelation" (singular).

15y agoHN ↗

Yes, it's a good read, but it's eight years old (not that old is bad, just that it's not "news") and this essay has appeared in many places over the years, including the book "Best of Software Writing I", edited by Joel Spolsky.