Posted by pfdietz 9 hours ago
They are thinking about what it is they do and what their intent was in doing it. Their adaptation is not a blind resistance to change but to preserve their intent.
I especially that they explicitly state that this is what they are trying and that not dropping a decree of new rules that are set in stone.
More platforms should take this approach.
But I thought the whole idea of the list was that it catalogued precisely the problems Erdos posed that he did not solve. The broader set of "all problems Erdos found interesting, including those he resolved" would be both larger and I suspect qualitatively different.
What I think you might be missing, though, is that Erdos was inherently playful with these problems. He'd sometimes just drop in the middle of a paper "hey, this variant of this lemma is neat, I bet it's true, but I'll give cash to anyone who disproves it" and move on. Or challenge a colleague while visiting and collaborating with them, and the problem would get spread by word of mouth. The problems often weren't things he thought long and hard about individually, but they appealed to his sense of aesthetics and curiosity and play. I think that's maybe close to what Bloom is talking about there. It's handy to have a catalog, but fundamentally, creating a set of unsolved math problems was not what Erdos was doing.
From the link:
"The main way that people publicly interact with the site now is to advertise their AI-generated proofs, often without any attempt to explain them, but as a way to record a (increasingly meaningless) priority claim.
This is very different to what I imagined, and I don't want to manage a website which does this."
Just as one does not open a restaurant in an abattoir, it is important that there be a separation between such repositories and a site which aims to promote the actual questions, place them in an appropriate context, and give a useful overview of the current state of human understanding."
I think this is a very reasonable blog post overall. A site for agents to read has many different needs and concerns than one for humans. Really anthropic or openai should be whipping this up for theirs.
I especially like that he’s allowing AI if it generates a better learning proof. That’s so considerate.
I do not particularly care to debate a theory of why it is so, nor press a charge of malice or intent; mere observation suffices to convince me that this shit is corrosive to the better nature of humanity.
Do you think they also are mostly deficient of scruples?
I had been expecting this since updates became irregular.
Perhaps this will force those solving these problems to produce something worthy of arxiv submission. Or maybe someone will produce a knock off site that continues to track these problems.
Of all sad words of tongue or pen,
the saddest are these:
a tragedy of the commons has happened, again.