Oof, these highlighted parts from only one video are already enough for me. This looks very stressful, I don't think I could finish a whole ride with one of these.

This is a very unpopular opinion in online discourse, where discussion can only ever be "won". I think this is an inherent property of the medium and cannot be changed. But one can dream...

I'm using Cerakey caps on my Glorious and I absolutely love them. Can't speak for the board itself, though.

Those muppets should just have read Camus and dealt with it. Instead, they created their own stupid religion and are now ruining everything for everyone. Good job, idiots.

Fedora bluefin is a much bigger project and a much larger paradigm shift in how Linux distros can be understood than what you make out to be. Tweaking system files might be a good choice for users who need to go beyond what comes with the standard, but it's not something a wide majority of users will or should need.

This is a clown show and the judge who let this travesty happen is an idiot.

That official White House X-account is really something. The utter shamelessness combined with the harrowing inhumanity would make Goebbels blush.

At the same time it is so honest and clear about the Trump administration's perspective on things it's refreshing. They don't hold back, there is no secret agenda, they just put it all out in the open for the Americans and the whole world to see. They really feel invincible.

Trust me, the analogy works if you understand what you're actually getting back from an LLM. Most people think of it as an answer to a question, but in reality you get a series of tokens in response to your query of tokens. In many cases this will very, very closely resemble something useful, like human language or source code, but that's really just our human interpretation of the result. To the LLM it's just a probable series of numbers without any meaning whatsoever. And, given the same inputs and settings, the responses are entirely predictable, there is no intelligence at work here. It's really a very complex way to query a very complex and large dataset which incidentally spits out results that have an uncanny resemblance to human language. Somehow this is enough to fool the majority of people into thinking this system is intelligent, and you can see why. But it's not. And the companies involved do their very best to keep this illusion alive for as long as they can.

As usual I wonder if this exchange is real. But it probably is, so... ow!

That "article", though. I thought I had just read the intro, but it was the whole thing. It's this what passes for online journalism now?

It would appear that sub's not ready yet.

What federation protects from is the singular owner of the platform sweeping in and setting/enforcing new rules for some or all communities. This could still happen on one instance, but new instances can mitigate the effects. Single communities can still turn bad, but it will be up to the users to decide whether to stick around or move to other communities.