achierius

> We should just naturally have the natural things that we naturally want, supplied by pixies.

Is this how you see roads? Are we entitled for wanting those to be paid for by the state? What about the police? Should we have to pay whenever a police officer stops a mugging -- or is the wage of that officer, too, supplied by pixies?

> Having their output employ randomness by default is a choice, not a requirement

This is not really meaningfully true. E.g. batching, heterogeneous inference HW, and even differences in model versions can make a difference in what result you get, and these are hard to solve.

I didn't say anything like that! Like I said I just don't think that this opinion is somehow associated with "vibe coders"; if anything I'd expect the opposite.

> Centaur outperformed all methods including CMA-ES alone by using the LLM on only 30% of trials. The LLM receives CMA-ES's full internal state (mean vector, step-size, covariance matrix), the top-5 configurations, and the last 20 trials. A 0.8B LLM already suffices to outperform all classical and pure LLM methods. Scaling from 0.8B (0.9766) to 27B (0.9763) to Gemini Pro (0.9767) yields no improvement, suggesting a capability plateau [which Claude sliightly beats]

> We ablate the LLM ratio: higher ratios degrade performance, confirming that CMA-ES should retain majority control.

One of those rare papers where the code speaks for itself. They do a bunch of comparisons but the most salient is comparing Karpathy's autoresearch (verbatim, best as I can tell) vs. some HPO algorithms, and as of yet the Tree-structured Parzen estimator still wins out -- but just barely!

More interesting though is that the best results come from 'centaur' approaches, where an LLM is hooked up with a standard HPO. Somewhere around 1:3 LLM:HPO control seems to work best, with more LLM control degrading performance. But either way this method far outperforms either the naive autoresearch loop or the bare HPO approach.

I think a lot of non-vibe-coding types also hold similar opinions -- in fact they might dislike Anthropic products even more, given that they (however few they might be) choose not to use them.

Not if the big labs have anything to say about it! They're working to fix the 'problem', and with Mythos we no longer have any guarantees that the frontier will even be available to distill.

That's fair. I'll adjust and say that I think there's a mix: some people certainly are bashing without understanding, but there are also a lot of engineers here whose day to day work is held to a higher standard than I think we see coming out of Anthropic, at least w.r.t. the product side of things (obviously the models are great).

Do you not think people here work at big companies with big products? I do, and we have a much higher bar for shipping.

What people dislike is the boom-bust cycle inherent to all levels of a market economy. During some years, these companies suck people up like a vacuum -- that can be bad if you're on the inside and all of a sudden the culture goes out the window, or if you're expected to onboard 3-4 people at the same time, or you end up with a reorg every quarter. Then, on the other end of the spectrum, companies shut down (non-backfill) hiring entirely and layoff huge percentages of the company, with no guarantee that you'll be safe just because you're doing a good job.

Human lives do not work like this. If you're getting married, if you have an unexpected hospital expense, if you want to buy a house -- these are not things that "market cycles" will plan around, but you have to.

Being quick to hire or fire is not the problem. Massive overhiring and massive layoffs are.

Maybe not 1/10, but definitely on-the-order-of 1/4th or 1/6th as many.

We aren't building dozens of new datacenters to host more webapps.

> Assembly to Python creates a lot of Intent & Cognitive debt by his definition, because you didn't think through how to manipulate the bits on the hardware, you just allowed the interpereter to do it

I agree! You often see this realized when projects slowly migrate to using more and more ctypes code to try and back out of that pit.

In a previous job, a project was spun up using Python because it was easier and the performance requirements weren't understood at that time. A year or two later it had become a bottleneck for tapeout, and when it was rewritten most of the abstract architecture was thrown out with it, since it was all Pythonic in a way that required a different approach in C++

GP meant moving the driver into userspace, which is much less painful due to the stable userspace APIs.

> And the fact that having outline calls to methods of value objects is so expensive

Is this tied to unions? Or otherwise, when does this happen? I don't see the connection w/ invisicaps or &c

Clearly they don't. They don't tend to occupy other countries, not outside of immediate territorial claims like Tibet (if you think that constitutes an "other" country)

> whether ... the concept of "bad actors coming in and ruining it for everybody by taking without giving back" even makes sense.

This is pretty clearly answered by the GPL: yes, it does, and this concept has been around since the very beginning.

> The information is still there

True

> as is the community that you've built

Untrue. At this point it's well understood that AI is substitutionary for many of the services that would have once afforded people a way to monetize their production for the community. Without the ability to make a living by doing so, even a small one, people will be limited to doing only what they can in the little free time they get outside of work.

That's the whole problem -- that AI, as it exists today, is taking away from the public, and hurting it at the same time. That's closer to robbery than it is to "sharing in the community".

> The last time a property class was removed was _slaves_.

Easy counterexample: titles of nobility. Also perpetual bonds, delegated taxation rights, the ability to mint currency. The list goes on.

If you're going to use history to support your AI bull agenda, you should at least pre-fly it with the AI first -- it would have pointed this out.

> Arguing that copyright is good because a subset of big tech doesn't want it around is as stupid as arguing that slavery is good because the robber barons don't like it.

Sorry, who's saying it's good? You are, actually, insofar as you're willing to support the right of AI companies to take people's information and use it to create copyrighted model weights. Why do you care less about the intellectual property of billionaires than that of the common man? Do you really think they're on your side?

Are you serious? In what world did we agree "someone may train incredibly important systems on our every utterance, without any compensation, and we will do what we can to not impede them"?

Can you not see how there's a difference?

I can tell you that it's not that he's setting aside speed -- the fact that it's as fast as it is is an achievement. But there is a degree of unavoidable overhead -- IIRC his goal is to get it down to 20-30% for most workloads, but beyond that you're running into the realities of runtime bounds checks, materializing the flight ptrs, etc.

Are they? I'm sure they're vulnerable to certain jailbreaks, but many common ones were demonstrably fixed.

> Steve Jobs was a middle-class guy trying to find his place in the world; the kind to travel India to to meet Neem Karoli Baba, shaved head, barefoot, wearing kurtas

I'm gonna be real: I've seen guys with shaved heads and kurtas at Apple Park. Admittedly they did have shoes but Indian people aren't "exotic", they're just people.

Essentially all Apple engineers use macOS ~exclusively (at least in the SWE org). That these bugs persist is in spite of that fact.

Frankly there's just a lot going on, and between all the kinds of bugs that have higher priority -- crashers, panics, memory regressions (your app gets Jetsam'd), power bugs (your phone runs out of battery mid-day), perf regressions (could violate EU regulations against planned obsolescence), security bugs (people can lose their money, journalists can get killed) -- bugs like this often fall by the wayside for quite some time until they're fixed.

And yet feudalism was abolished, and the map of Europe remade.

Do you expect people to wait by while billionaires pour trillions of dollars into replacing them? Evidence takes years to mount; present events are moving far faster than that. Your argument is the exact same as that of COVID denialists in 2019 -- that we don't know how bad it'll be yet, that there's so little evidence, that we shouldn't jump to hasty action before getting results in. Empiricism can only go so far.

If I knew someone was spending hundreds of thousands of dollars building a big laser pointed at my house, I would not wait for "quantified evidence" of its effect to take some sort of action. The only real debate is what kind of action.

> also, if the worst case scenario does happen and most of the population finds itself without money. there are other ways to live with very little money.

If you have so little attachment to your money, why hold on to it at all? Do not be upset that other people are operating on a slightly larger time horizon than you are, and are interested in their livelihood not just today, but three or five years from now.

Well for one, nobody was killed here. But second of all, sure -- because Hacker News are not the class of people involved in drive-by gang shootings; to most of us they are essentially abstractions, barely more real than the trolley problem. If you went around asking people who knew a guy that was shot, you'd eventually find someone who said he had it coming -- he got involved with the wrong guys, he shot at one of them first, he did something he shouldn't have (a common thread: the livelihood of the people involved). This is obviously atrocious: nobody should go around shooting people on the streets. But we can recognize that both are playing with fire, and understand the violence in that context -- such that the solution to gang violence is not, "moralize at the gang members until they stop shooting eachother", but rather "improve socio-economic conditions until they stop wanting to". So yes, there are elements of HN's population that will cheer these events on. But this should not be surprising -- the ruling class is playing with fire.

While I don't agree with your conclusion, I like the phrase -- "raped society" does quite well capture the feeling of violation I think many feel at having their own publications turned into machines meant to impoverish them.

[flagged]

Exactly -- that means that any analysis based on the current (as of 1926) 'reserves' or 'production capacity' for rubber/fertilizer/coal/wood would have been invalidated as soon as we switched to using oil instead. Imagine if instead of harvesting helium directly we find an economic way to split nitrogen (somehow, who knows). At that point, what you'd have to have forecasted would be the 'reserves' of nitrogen, which are functionally infinite.

Cities predated agriculture. Look up Göbekli Tepe -- this is a common misconception and worth correcting yourself on.