If model collapse is such an issue for LLMs, then why are humans resistant to it? We are largely trained on output created by other humans.

If a small group of people get to outcompete the rest of society thanks to having exclusive access to more powerful LLMs, I’m sure (🤞) that would lead to a lot of unrest. Not only from us plebs, but also from the rest of business. I can see it leading to some lawsuits concluding that AI advancement must be shared (if at a fee) with the public.

I wish they also had an LLM. I need a paid backup for when ChatGPT enshittifies. That won’t eventually extort me like VC backed corpos will

How does his massive penis fit inside that tiny car??

Imagine having been the last person alive to know the location of this lost city

Wouldnt they have been better off attaching tyres horizontally

*Insert Have I Been A Good Tundra meme here*

Funny that you mention Nebula, you might be interested in this other idea I had. Perhaps I should try contacting the Nebula team

Hmm good point, it’s true that a human would probably be better at getting the most for FOSS out of your money than a dumb algorithm. Just like is the case for stock brokers

Yeaaaaah, this would need WM integration which woukd be a pain to write

You can use the switch in the first picture to exclude apps from the donation

Haha this is KDE, just styled to look like good old W7 :)

Woaaaaaaah ok that survey idea is super cool. Although I’d argue that paradoxically it’s the rubbish apps that the bulk of the money should be directed to, to give their authors the resources to make them as good as the other apps. I guess it could almost be a Kickstarter-style system where you have to publish a barely-working MVP but if ppl vote for it then it’ll get a larger share of people’s donations. Wait I think I’ve just re-invented the wheel here.

FOSS projects may not have completely obvious donation schemes, let alone ubiquitous and automatable schemes, for starters.

Yes this is something I’ve already run up against. I’ve written a pretty functional scraper for donation pages. There seem to be ~6 donation platforms with the biggest one (PayPal eeugh) having 30% adoption. And yes not many have an API for one-to-many payments, although LiberaPay which comes a close second does. Hopefully if enough people use this app it’ll be lucrative enough for more projects to open LiberaPay accounts.

In addition, the dependencies suggestion is a pretty important and massive chestnut. It will certainly turn out that many core dependencies are used a huge portion of the time. Some disproportionately so. It will also certainly be very difficult to find and include any dependencies that are compiled in to things, and may not have obvious signatures without analyzing the source.

Yup. I wish there was an API for this. My plan right now is to scrape the debian repos as packages contain dependency information.

Note to self: in the settings dialog, let the user create custom donation entries with a regex to look for in ‘ps’ and a donation url.

Yes this is definitely something thing I want to add. It could also donate to Wikipedia and nonprofit news sites. I plan to host JSON files woth payment info in a separate git repo so that people can add their sites in and crowdsource the data this way.

I like the idea, but i would want it to be more granular so that i also donate to all dependencies i use.

Writing a scraper that would follow dependencies is definitely on the todo list. But currently I need an MVP so it just tracks flatpacks (donate links are part of their metadata).

I personally also do not care about the shares each projects get so i would prefer if nothing runs in the background it only collects once a month all my installed packages and donates to them evenly.

Interesting. I might add an option for this then.

https://lobste.rs/c/2aacvs