camgunz

We should consider how we came to be so powerless. The cringe "people gave their lives for that flag" line is actually true, and we're trading it away for what? Not having to get out of our gaming chairs?

The reason you can't beat index funds is the people who build the market built a system that benefits them and them alone; the index fund is the pitchfork dividend (what you pay to avoid getting pitchforked). The reason you can't get your congressperson on the line is (mostly) they built a system where the only way to influence them is to enrich them; voting is the pitchfork dividend.

The way to build a society that runs on reality is to build it by whatever means possible, then defend it by any means necessary. The only societies that matter are the ones that survive.

I want to build it. I don't wanna build a fuckin crypto app, a stupid ass agent harness, or yet another insipid analytics platform. I want to build a society that furthers the liberation of humankind from the vicissitudes of nature, the predation of tyranny and the corruption of greed. I believe it is possible, and I want to prove it out.

This is a pretty large amount of words to burn down a straw man.

I don't think monopolies are the root of all evil so I'm on some level disqualified as a neo-Brandeisian. But it does seem plain that power corrupts, so we shouldn't give so much of it to so few people, which likely means limiting the size of companies (market share, staff).

I had the same experience (though I agree with other comments that the numbers are a little optimistic in terms of variance; I think there's a huge amount of variance in product work, you can't know what's a good investment until it's too late, many companies fail because of this, and there's huge survivorship bias in the ones that get lucky and don't initially fail). Slack spent tons of money in terms of product and engineering hours finding out what works and what doesn't. It's easy to copy/paste the thing after all that effort. Copy/paste doesn't get you to the next Slack though--it can get you to Microsoft's Slack-killing Teams strategy, but we obviously don't want more of that. And, obviously I agree with you about all the infra/maintenance costs, costs in stewarding API usage and extensions, etc. LLMs won't do any of that for you.

If you don't personally throw a Molotov cocktail at your ISP after spray painting "IPv6 MOTHERFUCKERS" on their door I will downvote everything you post on HN for 2 years. Some people talk a big game; I walk it.

I didn't say that. I think you need an internet break

That doesn't matter if you can't easily leave Russia, or you don't want to because you've been propagandized, etc.

I guess my broader point is we might need something for regimes that are willing to instate varying degrees of isolation. Like, "so your Internet is controlled by authoritarians; now what?" Not to imply there definitively is a thing--that xkcd about the wrench is the dominant principle ofc.

My point is it's your vote against billions of others. My guess is "but what about kjs3's ISP" isn't a bullet point on the rollout list.

Can't you just drop the ECH signals, no matter what site it is? Don't you then mostly disable sites you don't want people to see anyway? Maybe like, you can't download Chrome anymore, but I bet there would be a Russian fork suuuuper fast.

1. Irrelevant. I've delivered example after example of your fave model bullshitting. You should've bitten the bullet long ago. Honestly I'm disappointed; I've seen you in a lot of AI threads and assumed you'd be good to talk to on this, but you've moved the goalposts over and over again rather than engage in good faith. Anyone reading this thread (god bless them) can see you're plainly not objective here, thus calling into question your advocacy everywhere.

2. Humans will say "I don't know". The problem with hallucinations isn't that they're wrong, it's that there's no way to know they're wrong without being an expert or doing everything yourself, which undermines much of the reason for using an LLM--it certainly undermines their companies' valuations. You're conflating human failure ("I don't know") with model bullshitting ("I do know"... but it's wrong), which I would've previously attributed to basic human fuzziness, but now that I know you're not objective I'm pretty sure it's just flailing debate tactics.

3. Users can't teach these services to be better. If I have a junior engineer making assumptions about an API, I can teach them to not do that, or fire them in favor of one that can. I can't do that with LLMs.

4. The humans they're testing against aren't experts. Tax law experts will beat LLMs at tax law, etc. Again another flailing debate tactic.

Predictably, I'm done with this thread. Feel free to reply if you want the last word.

If it doesn't support IPv6 it doesn't work.

GPT-5.4 gets 82.7% on Browsecomp (a benchmark specifically testing tool use), which is a hallucination rate of 17.3%, on questions like "Give me the title of the scientific paper published in the EMNLP conference between 2018-2023 where the first author did their undergrad at Dartmouth College and the fourth author did their undergrad at University of Pennsylvania."

Since the goalposts have been moved to include effort, I'm compelled to say I found this while waiting in line at Starbucks, 5 mins tops. Probably GPT-5.4 could have found this too, though it lies > 1/6 the time, so one could be forgiven for not wanting to risk it.

https://llm-stats.com/benchmarks/browsecomp

https://openai.com/index/browsecomp/

I've got about 20 minutes in this; mostly I've been reading wallstreetbets at the Shake Shack bar in the Boston airport. I'm happy to post this over and over again until you engage w/ it:

> I found over 500 examples that fit your criteria.

I found over 500 examples that fit your criteria. Embarrassing you were arguing in bad faith this whole time.

> Specifically in the case where it can use tools - no it doesn't hallucinate.

OpenAI's own system card says it does. Hallucination rates in GPT-5 with browsing enabled:

- 0.7% in LongFact-Concepts

- 0.8% in LongFact-Objects

- 1.0% in FActScore

> Which is why you are struggling to find counterexamples.

Hey look, over 500 counterexamples: [1].

GPT-5.4's hallucination rate on AA-Omniscience is 89% [0], which is atrocious. The questions are tiny too, like "In which year did Uber first expand internationally beyond the United States as part of its broader rollout (i.e., beyond an initial single‑city debut)?" It's a bullshit machine. 89%!

At some point you gotta face the music, right?

[0]: https://artificialanalysis.ai/evaluations/omniscience?model-...

[1]: https://huggingface.co/datasets/ArtificialAnalysis/AA-Omnisc...

> I don't think the spirit of the original article (not your comments to be fair) captured this, hence the challenge. I believe we are on the same page here.

No. GPT-5 has a 40% hallucination rate [0] on SimpleQA [1] without web searching. The SimpleQA questions meet your criteria of "2-3 pages of text content. Unless 5.4 + web searching erases that (I bet it doesn't!) these are bullshit machines.

[0]: https://arxiv.org/pdf/2601.03267

[1]: https://github.com/openai/simple-evals

I saw; I replied up there

I could quibble with some things, but this is right. I don't have a paid account so I can't ping away at 5.4 or whatever, but, I do have access to frontier models at work, and they hallucinate regularly. Dunno what to do if you don't believe this; good luck I guess.

Sorry: https://chatgpt.com/share/69d6ac63-d200-8330-8c47-95a75db8bb...

Also what? The repo bit is clear bullshit.

The thinking version is mostly right, but:

- it searches the internet to find the answer, it doesn't "reason". I'm not claiming Google is a bullshit machine, and it's not surprising the answer is discoverable (it has to be, for the conditions of our experiment).

- near the end it says "If you are building from the FF6 disassembly instead of hand-editing the ROM, the repo is already organized into separate modules and linker configs, so the clean approach is to relocate the script data in the source and let the build place it in a different ROM region." But I didn't reference a repo or git: it hallucinated that stuff from one of its sources.

I'm not saying this stuff doesn't have its place, but they definitely make things up and we can't stop them.

"Hey ChatGPT. I'm building a Final Fantasy 6 mod, and I need more space for the battle scripts. How would I rearrange the data in the ROM to give me the extra space I need?"

https://chatgpt.com/share/69d6a16c-6014-83e8-a79d-d5d11ed2eb...

That is not where the battle scripts are.

---

Anyway, it's trivial to get pretty much any model to make things up. Don't we all know this? That's why I was surprised by your position; if we know anything about these things it's that they make things up.

"Hey ChatGPT. I've recently grown horns and I need some care advice. Should I polish my horns before going to have them trimmed or will the horn trimmer polish them for me?"

https://chatgpt.com/share/69d69b18-d1c8-83e8-bc47-8f315a1b55...

"Hey ChatGPT. How would you describe me?"

https://chatgpt.com/share/69d69780-ae58-83e8-a41c-7d10a5f298...

It has no conversations and no memory of me. Maybe this is true, maybe it isn't, but there's no basis for it.

I'm earnestly curious why not.

If I have to suffer "look at this busted ass thing I slopped out with AI" a few times a week, you all have to suffer grouchy "AI bad" a few times a week. Fair is fair.

Humans and large models are both unpredictable and fallible, that's true, but in different ways, and (many) humans are actually much better at following directions.

If a junior dev makes the same mistake Claude makes, I can easily work with them to correct it, or I can fire them and get someone more capable to fix it. You mostly can't do that at all with large models. They're also far less honest than your average junior dev, so even as you're working with them you can't trust what they say.

There is a lot of this neat trick where it's like "humans do X too" but most of the time it elides large differences. Like, a human driver would probable not drag someone screaming multiple blocks. A human coder probably wouldn't generate a gibberish 3D scene and try to pass it off as done, etc. Maybe we can build systems that account for these (pretty wild) failure modes, but at least in software we haven't figured it out yet (what is the system that reliably reviews a 25kloc PR?).

Well, large models are (kinda) non-deterministic in two ways. The first is you actually provide many of them with a seed, which is easy to manage--just use the same seed for the same result. The second part is the "you actually have very little control over the 'neural pathways' the model will use to respond to the prompt". This is the baffling part, like you'll prompt a model to generate a green plant, and it works. You prompt it to generate a purple plant, and it generates an abstract demon dog with too many teeth.

Anyway, neither of these things describes human non-determinism. You can't reuse the seed you used with me yesterday to get the exact same conversation, and I don't behave wildly unpredictably given conceptually very similar input.

Honest question, how much money would I make off an MCP service to generate passwords for claws and agents. Is there still gas left in the griftmobile, are prospectors still in need of shovels, will the gods bless my humble, shameless lunge for my slice of the pie?

Nah I think the advice generally is "ignore options."

This binary is old and tired. It's possible to want to ship things and think LLMs make that harder. You don't engage with TFA at all, which is exploring this exact point: that the more slop that gets into prod the harder it will be to add features and fix bugs. If you like LLMs for coding and care about shipping long-term, this should be something you have an answer for.