Hacker News Reader

Comment by r053bud | Hacker News Reader

None of your arguments hold any water if big-corpo AI restricts open-weight models - which they very clearly want to do. AI is NOT “the best thing that ever happened to the Free Software world” if our only option is to pay a select few companies to use it.

View on HN

Those arguments are independent of the state of OSS models. There appears to be no moat whatsoever on creating state of the art models commercially - the US threw the kitchen sink at China in terms of sanctions and barriers ... yet it has been unable to keep Chinese companies away from the frontier (arguably ahead in small local models with Qwen). Even if all the models were proprietary we'd expect the costs and capabilities to be widely available. The Europeans might even be able to field their own advanced models eventually.

OSS models are strategically important, but that is a different matter to getting value out of them for OSS software development.

IMO, AI will have the same problems no matter which area or country uses it to dominate over others.

Consumers are way too fickle to pursue open models en masse. Even if they are 90% as good and fairly easy to use (works acceptably on most machines, easy installation, whatever), they will still go for the ones controlled by big corportions/big tech.

Look at how consumers deal with free, open source software, alternate Operating Systems, alternate social networks, alternate search engines, alternate chat protocols... even when the barriers are low they just don't give a damn. I honestly think most people would choose to join the Borg out of sheer convenience. It's futile to think this won't happen with AI.

Big corporations can't do shit about open-weight models that already exist, because those are... just bags of floats. Once released, there is no unreleasing them. The Internet never forgets.

Download the files as they come out if you're worried, but there's no way to roll LLMs back beyond what we have in open-weight models right now. It's a ratchet.

Big entertainment has been trying to restrict the distribution of movies online since that became possible, how's that going for them?

Pretty well, honestly. Somebody with both technical knowledge and familiarity with a particular subculture can get access to entertainment videos with some work. But they've managed to keep piracy economically negligible. Basically everybody uses streaming services, 90% of US households pay for at least 1, and the average household pays for 4. [1]

Brand was wrong. Information doesn't want to be free, just cheap.

[1] https://variety.com/2026/tv/news/how-much-us-households-spen...

This is only the case because streaming services today are so cheap and convenient. Jack up the price, or decrease the convenience, and see people sail the high seas en masse again..

Indeed, the origin stories of these streaming services are deeply intertwined with combating piracy if you go back 20+ years. And I agree, if they outlaw open weight models, then only outlaws will have open weight models. But good luck finding and prosecuting the outlaws beyond a few whales.

You can watch a lot of TV series online on some random pirate sites by simply googling "show x free online". It doesn't even require the relatively simple technical knowledge of using a BitTorrent client.

Quite well. Lots of people in jail and it's quite obscure how to download movies as a casual.

The long term trend seems to be in their favor as governments want to include more traceability and ability to shut down sites into the Internet for their own reasons which will help the entertainment industry as a byproduct. Plus they can use LLMs just as much as anyone else in their quest to build better roadblocks.

We have open-source models, and they're getting better and better.

But, rather that a wide Edenic expanse of communication, we seem headed toward walled gardens with gatekeepers and a need to pay money for all the usual reasons.

This seems both inevitable, and an occasion to tout the https://www.fsf.org

Eh? Where exactly is that a reality though? I just spent like 20$ last week on various Chinese frontier models using openrouter and ppq building like 10 different projects that would've cost way more than double or quadruple per token at the big 3 if I didn't have those alternative options. Hell, for small tasks I even fell back to a local 27b Qwen model.

for now.

They will literally never get worse. It's a one-way street.

I agree. The question is if they will remain legal.

The law doesn't prevent someone from doing or using something. It punishes them for doing or using something (if caught), but never prevents.

If laws had the power to prevent, Chicago would be some sort of utopia.

Chicago is pretty great, bub. This year I opened a pinball museum here, so I can testify that it's a fine place to run a business. Some friends just came from the Bay Area for a week and had a lovely time. If only for your mental health, you should step out of your media bubble now and again

... if you can afford the ever increasing cost of hardware, a rise directly caused by the same AI companies.

I don't think it will be ever-increasing. It might last a long time, but sooner or later supply will increase to meet the demand (or the demand will drop, or both) and hardware will be cheap again.

It also seems pretty reasonable to worry that Oracle isn't banning LLM code contributions to OpenJDK https://www.theregister.com/ai-and-ml/2026/08/03/as-larry-el... ( HN discussion at https://news.ycombinator.com/item?id=49213754 ) purely on quality grounds. If things carry on as at present, and it turns out that indeed it really was legally safe to copy and paste almost any code that came out of an LLM, then that's one thing; but if instead there's a rash of code copyright (and indeed patent) lawsuits a few years from now then the big and middling companies will simply pay each other settlements, while FOSS projects and their small-fry users will likely be in a much worse situation.

That’s like saying private libraries are bad for freedom to read, or Blockbuster was bad for freedom to view movies

Reply on news.ycombinator.com