This depends on how transformative the act of encoding the data in an LLM is. If you have overfitting out the ass and the model can recite its training material verbatim then it’s an illegal copy of the training material. If the model can only output content that would be considered transformative if a human with knowledge of the training data created it, then so is the model.
In the long term, alternative platforms need to be built on something different than outrage and “not being the bad company”. In the end, the vast majority of people cares very little about the underlying technology, they just want their content and people to interact with. Mastodon is in decline already, the fediverse shouldn’t be a place where people come to say “wow, [company] sure does suck”, and then go back to that company if they actually need a piece of information or reach a person that does not know or care what an API or federated protocol is, aka 99% of the population.
Not sure what to think of this honestly. Like imagine a small email provider decided to block Gmail, that’s a death sentence. It’s impossible to get people to switch apps when they have to leave behind all of the content and people they used an app to interact with. And let’s be honest, threads is going to run at a loss for a long time to grow their userbase before they start pulling weird shit. We need to have a migration path when that happens, and if threads is blocked everywhere, people will lose their content and contacts upon switching, so they won’t do it.
Wefwef. It’s incredibly polished and fast, especially for a webapp. It runs circles around new reddit performance wise, even with the servers being a raspi trampled by a horde of reddit refugees
Luckily the arm macs don’t need fans most the time. They do have a finite amount of ram though, and boi, is Linux virtualization inefficient on mac
Wait, ipv6 doesn’t require port forwarding to expose something to the internet?