500 HTTP Error

How are you confident in the code, coding style and practices simply because the LLM says so. How do you know it is not hallucinating since you don't understand the codebase?

Now instead of having accountants audit transactions you will have accountants audit LLM output for possible hallucinations. Seems counter productive.

> Someday, people will have to realize we live in a society. What will it take?

Anarchism, socialism and communism can work perfect in a small village where everyone knows and trusts each other. But if you scale it up it does not work well because people can be corrupt. If you want to scale up to a Geo Global level that is trust-less the best way we know is to use Capitalism, but Capitalism ends up becoming more and more centralized.

Because Capitalism is inherently competitive there will always be winners and losers and these are not just businesses it's everyone in the system because capital is required to partake in the system. This competitiveness is also what leads to the lack of "morality".

What will it take?

I think you cannot have the benefits of capitalism without these side-effects.

I think we now have the technology to make decision-making and resource allocation systems that do not need to centralize power. If we can do that, then it wouldn't matter that people can be corrupt, because there would be no positions of power that people can abuse.

I also believe that technology is the solution. But all the key technology is centralized Chips, AI, Batteries, Cryptography, Email, Internet access, Radio Waves

Most people aren't looking to eliminate capitalism. They just want constraints to be put on it. Higher taxes on wealth, stricter antitrust enforcement, investing in social infrastructure, or passing laws that protect consumers don't prevent capitalism from working.

Australia has social healthcare and massive mining companies. They coexist just fine. There really is a lot of wiggle room between fully embracing socialism and going full anarcho-capitalist, and maybe the tradeoffs of shifting towards the socialism side of things are worth considering.

Although, George seems to just want to flip the table out of the belief that real reform that would impact most people positively will never get passed in a democracy. It would require too much change.

> Most people aren't looking to eliminate capitalism. They want sensible constraints to be put on it. Things like higher taxes on wealth, stricter antitrust enforcement, or investing in social infrastructure don't prevent capitalism from working.

In capitalism the capitalists end up being the government. They can choose who gets elected, the laws, they even start political parties.

That's an oversimplification. Yes, wealthy individuals inevitably have more influence. But there are numerous countries whose governments regularly act against corporate interests. For example, as much as I dislike GDPR, it is a strong example of governments implementing a policy that is explicitly against corporate interests. Another example is the OECD global minimum corporate tax.

So, there are governments that oversee capitalist countries that are willing to implement policies that hurt corporate interests with the goal of helping consumers. I'd say the problem is that often these policies made with good intentions, like GDPR, end up being poorly implemented and therefore harming consumers as well as hurting corporations... but that's an entirely different problem.

What are the benefits of capitalism?

With it, you don't have to die from hunger. Not much, but compared to the alternatives, it is one of the strongest arguments in favor of capitalism.

> It can generate data that mimics anything humans have produced...

No, it can generate data that mimics anything humans have put on the WWW

The frontier model developers have licensed access to a huge volume of training data which isn't available on the public WWW.

Shor's algorithm, originally designed for integer factorization, can also be adapted to solve the discrete logarithm problem in polynomial time on a quantum computer. There is also the less efficient Grover's algorithm can also be used for unstructured search problems on a quantum computer.

I was thinking more along the lines of solving in polynomial time on a conventional computer.

Bitcoin is designed with clever incentives to prevent this kind of thing. If you can afford to bruteforce wallets the incentive would be to just mine bictoin which is more probable and it also help secure the network. If you can bruteforce wallets bitcoin is effectively worthless. Or you could even use all of that compute to mine something else for example Monero.

People always forget about the great Wei Dai, who like Nakamoto is already sort of pseudonymous, he also created Bmoney which is allot like Bitcoin. He is also the creator of the Crypto++ cryptography library for C++ (bitcoin is written in C++) From all the OG Cypherpunks he ticks allot of the boxes.

True. Nothing gets built in isolation. Bitcoin was the result of decades of work by cryptographers, economists, investors, and engineers

> I audit everything myself before making PRs and test rigorously

How do you audit code from an untrusted source that quickly, LLMs do not have the whole project in their heads and are proned to hallucinate.

On average how long are your prompts and does the LLM also write the unit tests?

The auditing is not quick. I prefer cursor to claude code because I can review its changes while it’s going more easily and stop and redirect it if it starts to veer off course (which is often, but the cost of doing business). Over time I still gain an understanding of the codebase that I can use to inform my prompts or redirection, so it’s not like I’m blindly asking it to do things. Yes, I do ask it to write unit tests a lot of the time. But I don’t have it spin off and just iterate until the unit tests pass — that’s a recipe for it to do what it needs to do to pass them and is counterproductive. I plan what I want the set of tests to look like and have them write functions in isolation without mentioning tests, and if tests fail I go through a process of auditing the failing code and then the tests themselves to make sure nothing was missed. It’s exactly how I would treat a coworkers code that I review. My prompts range from a few sentences to a few paragraphs, and nowadays I construct a large .md file with a checklist that we iterate on for larger refactors and projects to manage context

I think it will push opensource/ free software hackers to close source their code because it is being used to feed LLMs. Similar to how allot of hardcore free software proponents don't use Github. Is closed source the future?

> Is closed source the future?

No. I don't believe that. I personally want my code to outlast me and help people in the future, but I don't want allow anyone to just scrape it, strip its license and use for whatever. I use (A)GPLv3+, because I believe in "Freedom for the user", not "Freedom for the developer" which permissive licenses provide.

My code is not free labor for anyone. It has conditions attached.

This is the problem that AI solves, though: rather than steal our code directly, now the thieves will just ask their favorite AI to generate a new project that does exactly what our (A)GPLv3+ projects did, which it will be able to do only because it read our code. And, even if the result is eerily similar to what we publish -- we might, after all, be one of the few good examples in the training set for this problem -- it will be difficult to demonstrate, as the AI is more effective at the process of laundering licenses than a human (and no one seems to want to admit that, the same way that a human can be tainted by reading the source code of a project they want to reimplement -- making them have to walk a tightrope if they later want to develop anything similar -- an AI might be similarly tainted). In this shitty new world, our code, is, in fact, free labor for people who are using Cursor to rip it off.

Ouch!

I believe in OSS. But damn. I had not really considered this move.

I had a stray thought and that is most SI content I have looked at has watermarks of a sort. Perhaps this could be used?

I dunno, even after considering that move, I'll continue to publish FOSS like before.

I always did it without any expectation of gains from it, and with the intention for people to use it for whatever they want. That calculation hasn't changed, even considering machines will slurp it up now.

I do agree that it sucks for people who do care about what the code is used for, and I hope these people migrate to other licenses that support their ideas about control and ownership.

We already did migrate to that license: (A)GPLv3+. You can use my code if-and-only-if you won't then hoard your own changes from the world and lock users of your derivative software away from having the same empowerment you did. It isn't about "expectation of gains", and that's a ridiculous way of portraying the situation: it is about a social contract that happens to be enforced by copyright.

And, as such, when your favorite AI generates code similar to my code after having read my code, that's infringement, the same as if a human had done the same thing... only, the AI doesn't bother to consider that angle, and, even if you know to care, you have no way to know what is going on, in the way a human at least usually can know when it is cribbing off of what it knows (though even a human can do this accidentally).

I will do the same. I am aligned with ESR basically, as expressed in "The Clue Train Manifesto."

Use value of OSS remains high. Because of that, when I can add to the body of OSS, I do. People will do what they do.

All I control is me. They do them.

We all benefit from the high use value.

I do wish those who have made fortunes would contribute more and keep their roots, and the labor of many high quality humans just a bit more firmly in mind.

>I am aligned with ESR basically, as expressed in "The Clue Train Manifesto."

You mean "The Cathedral and the Bazaar". Both were published about the same time, long ago, but Eric had nothing to do with "Clue Train".

Yes, my confusion! Great catch!

This will not stop the AI companies from using it in their training data.

From an open source software perspective, I don't understand the feelings around LLM ingestion.

The models isn't generally recreating your software, but might be spreading your way of thinking in pieces.

I get it from the artists and to a lesser degree, writers. I just don't understand it from software projects.

I guess if you think of it as something to replace you, but since you are already a creator, it is also a way to unlock much greater capacity for turning your ideas into solutions.