Think about all the times in llm gets it wrong, the fact that would have helped to get it right is something that was lost. I suppose this isn't proof it's lossy just maybe we don't know how to get the data out.
Or look at it another way LLMs or just text prediction machines, whatever information doesn't help them predict the next token or conflicts with the likelihood of the next token is something that gets dropped.
Or look at it another way these things are often trained on the many terabytes of the internet yet even a 200 billion parameter network is 100 or 200 GB in size. So something is missing, and that is a way better compression ratio then the best known algorithms for lossless compression.
Or we can look at it another way, these things were never built to be lossless compression systems. We can know by looking at how these things are implemented that they don't retain everything they're trained on, they extract a bunch of statistics.
Just thinking that an unexpected event can't happen on the road shows a huge lapse in judgment and that's just the first sentence. Thinking they're special and different other people is another big red flag.
Yeah appealing to that book is literally conspiracy nonsense when you're talking about a group of engineers trying to do good engineering. That appeal fundamentally means that there are people trying to actively sabotage a thing, that by itself is a conspiracy theory particularly when you consider that much of the standards committee has been doing standardization work for ages, and all of them are experienced engineers focused on engineering.
Prima facie claiming that people who are joining an optional group that puts out optional rules that companies can opt into implementing for the sake of sabotaging something in such an esoteric way is complete conspiracy nonsense.
Edit - LOL, step one of my conspiracy and get a PhD and work at a company for 10 years using a programming language so I can get someone kicked off an optional committee! Brilliant plan, no notes!!!
Why does it so often seem that the people complaining about censorship are the ones punching down?
Why is it so often someone's right to complain and make problems for others but never concern about people's right to be tolerated when they're being decent humans?
Either people need to be banned who insult others and use slurs and those who maliciously push right up against the rules, or they will bully people out. Look at modern X/Twitter allowing hate speech has pushed out advertisers and something like half of the users?
This is basic Paradox of Tolerance stuff, decent people aren't Banning anyone for pointing out actual arguments like discussing if "question" is okay, asking for extra context if this guy did something else or if this is council overreach. But people complaining about wokeness, DEI, diversity hires, or other technically allowable but obviously hostile nonsense are clearly just trying to attack other people and often in ways that are racist dog whistles. If people insist on being hostile up to the amount allowable by the rules instead of just trying to get along then the rules need to keep changing and adjusting and of course the people who are willfully choosing to be assholes will scream "censorship". Before teaming up with someone complaining about censorship be sure they're actually at risk of censorship and not just trying to use Free Speech as a shield to hurt others.
The article clearly has gaps simply by not being an hour long read on the Atlantic, one obvious gap is that it doesn't seem to have a statement from the person removed after they were removed.
The last few times I've seen technical groups kick someone out over racism, including anti-semitism, they picked one point where the person kept doubling down after a very long history of borderline rule following behavior that was clearly malicious. I would be very surprised to see if this was different.
While I agree in principle that we can't allow the word question to be destroyed by hate speech, there are always assholes who ride up to some line to be dicks to someone using whatever the boundaries of the rules are. I want to know what happened here and if that was the case.
I don't think it matters if the judge agrees with me.
Cases will be brought before judges based on money. Big pirates that are somehow making money doing it will be brought in front of a judge and some love will impact them.
Anybody hypothetically pirating one steam game to play for themselves because they felt alienated by Steam isn't going to be dragged in a judge.
Yeah, totally, working out what is for dinner is great way to learn about people you have dinner with. What foods do they like? Why do they like, dislike, or simply cannot partake in some foods? And less direct but deeper notions like: What makes arguments happen with this person? What happens when this person lashes out on a small scale? Can I react sensibly to this person disagreeing with me?
This and a million other little nuances are super important for successful relationships.
What is the dividiing line between single organism and clones with connecting structure, like shared roots?
It seems like a fuzzy gradient to me. Maybe some biologist can share what makes the distinction clear, but I can imagine a gradient ranging from fully distinct autonomous disconnected clones all the way to clearly a single organism that only grows outward into a large sphere.
Clearly Pando is somewhere in the middle of this gradient. What is Pando's position on that gradient and why is "bunch of individual clones" somewhere else? How is another tree sharing a roots not a single individual too?
On a team you are emotionally (or maybe even just financially) invested in it feels bad, but when I was at EA they almost worked hard to make it hard to become emotionally invested.
At a company the size of EA almost certainly this will be used to play politics and even if it hurts the company as a whole people will use it to have larger control over the now smaller company.
Isn't this taking something that should be relationship building and outsourcing to an app? How long can a relationship last when people refuse to settle the littlest of disagreements like rasonable adults?
Oh yeah Cobol absolutely worked as a programming language, but it never lived up to the promise of being a plain English replacement for code which was a common promise at the time.
There's no it way won't suck, so it's good to github built in a way to iterate and called it experimental. We need much better llms, or some other technique, before we can have anything like this.
People have been trying this building software in plain English bullshit for at least 65 years, look into the history of cobol. It didn't work last time, or the time before, or the 50 times before that. It absolutely will work someday, but there's no compelling reason to believe that day is coming in the next month or two.
It was mandated by right to repair laws, it provides the absolute minimum, and they've attempted the price out people wanting to do repairs. The only way it could be more hostile to users is by literally being illegal.
They could go out of their way to make things actually easy to work on and service, but that has never been the Apple Way. Compare to framework or building your own PC, or even repairing a laptop from another OEM.
Or look at it another way LLMs or just text prediction machines, whatever information doesn't help them predict the next token or conflicts with the likelihood of the next token is something that gets dropped.
Or look at it another way these things are often trained on the many terabytes of the internet yet even a 200 billion parameter network is 100 or 200 GB in size. So something is missing, and that is a way better compression ratio then the best known algorithms for lossless compression.
Or we can look at it another way, these things were never built to be lossless compression systems. We can know by looking at how these things are implemented that they don't retain everything they're trained on, they extract a bunch of statistics.