Recent content by Vanadium 50

I agree, but most of "AI" is not what people think of as AI. You say LLMs are like autofill on steroids. I agree. But a lot of AI inferences are statistics on steroids. Some of the most interesting work is on things we don't even consider as "intelligence" such as motion and balance.

AI can write plausible sounding theory paper titles and abstracts already: see the "SnArXiv:,

Let me describe a building I worked it (somewhat simplified) so the "better locks" idea can be more thoroughly explored. There is a nice glass lobby, leading to two wings. Anyone with a brick and motivation can get into the lobby. The North Wing has lecture halls and seminar rooms and you can...

What could go wrong? Not using the YubiKey is like leaving a door (one of several in series) open. Is that better or worse than having your locksmith keep a copy of your key to that one door?

That is so 60's! If you make the doors too hard to get into, first responders can't get into them either. And remember, during an incident, you may not have power. Further, a door is only as strong as its frame: a vault-type door (yes, that's a thing) doesn't work so well if its set in a...

Sure - if someone says to me "Here's $25 for a share of your profits - but you don't have to actually give me any of your profits" what do you expect me to say? Besides "Are you sure you only want one?" If someone wants to bet money that they can buy something worthless and sell it for more to...

For some odd reason, writers want their wormhole network (to use @BWV s term) to map on to stars we can see. Why is this even necessary? In most stories, it would work out just fine if you popped out of the wormhole at the scene of your adventure with no earthly idea (literally) as to where...

I saw that, but my reading of Andy's post is that he was talking about something wider than a few protestors led by a single individual at one university. If the point is "you just have to wait it out and they'll lose interest," well, maybe that will work. It will certainly work until it...

Please tell me. (Boy, I'm glad I am not a university administrator. One group's slogan is "Don't kill the pigs" and the other is the opposite....)

Is that what's going on? It's pretty clear what GME's thinking is: it's easier to sell stock than to sell games.

If two points are "congruent" they are at the same position. Your drive doesn't have to work very hard. But it's also of limited utility.

Newton or Leibniz? Must be Taylor Swift.

That's actually not in the paper itself. What the paper says is that IF there is an a(1810) AND it is the isopartner of the f(1710), that supports the notion that it is the f(1500) as opposed to the f(1710) that is primarily glueball. I have no objection to that. Note that in neither case is...

Their problem is to create a ML training dataset of particle trajectories, and the hope is that they can simulate them without using a full physics-based simulation, which is slow. They are looking at special cases, which makes doing it the right way even slower. Their code uses backtracking...

Third question - what have you done to answer this question? Where have you looked, what did you find, and where should we start?