7,330 karma · joined May 12, 2008
This new bound is sensitive to energies above roughly 10^13 electron-volts — more than an order of magnitude beyond what the LHC can currently test.
Wow.
Regime change is hard. Creating a functioning, not corrupt state in the aftermath is harder.
The barrier to Russians forming a coup is state repression. The FSB is good at routing out any coup before it can get far. Unless the upper ranks of the military fear for their lives under Putin, their calculus is that they're better off with the status quo than a coup.
But for OP, the point isn't having humans understand. It's having the computer check the proof. You still use traditional journal papers for communicating the proof to humans. You're just also able to say "The proof has been verified and really is true."
The idea would be to use the expected phrasings and idioms, format, etc. Of course, all content would have to be checked, as LLMs are prone to BS. But if the problems were mostly that it didn't follow the conventions of that academic subgenre, this seems exactly what an LLM could do. If the problem was more about e.g. not citing previous work, the LLM probably wouldn't do a good job there.
Human nature, and publish-or-perish rewards, mean the researcher will say "well, what if I try this other model? Or tweak this parameter, or ..."
In other words, it's unrealistic to say you have one shot to confirm that have, indeed, solved it.
Programming is essentially solving puzzles, so you can find the work interesting even if the topic area isn't interesting. No seven year old says "when I grow up, I want to write a memory allocator!" but doing it well, with as few CPU cycles as possible, without fragmenting memory too much, is an interesting challenge.
I wouldn't worry about what people have in their job descriptions. There's a quote from John Carmack about an ad for a Oculus dev that he technically wouldn't qualify for, and DHH who invented Ruby on Rails, was called by a recruiter asking whether he had 10 years of Rails experience when Rails was only 9 years old. A "Senior C++" engineer might just mean more than 2 years of experience with C++, who knows.
At least that's my guess, I'm not an expert.
That's certainly true in marketing material of fund managers. But it's not true in honest academic papers, like this one. There's lots of data out that about total returns of the stock market, that takes into account dividends and survivorship bias.
Maybe write it now and revise it as your understanding grows? Also, posting it now might encourage others to evaluate it for you, doing the work you may never get to.
Just a suggestion. Thanks for your deep dive into this and writing it up in the HN comment! Very insightful.
"Last updated: October 11, 2022
Building access protocols remain largely unchanged from the spring 2022 semester, with several additional spaces now open to the public including in buildings 32, 76, and E25. MIT ID holders continue to have wide-ranging access to campus buildings; MIT campus outdoor spaces remain open to all. MIT continues to welcome visitors to open events and welcomes other guests who are registered using Tim Tickets or escorted by MIT ID holders."
1. Explanation of the subject. Could be videos or textbook.
2. Problems to work on. Much like learning guitar, you have to learn math and physics by doing.
3. Feedback on my solutions. Did I solve it right, or did I have some fundamental misunderstanding? Also, when I'm stuck, it can be useful for someone to look at it and say "here's where you went wrong."
4. Community and deadlines. It's easy to put off the next lesson/chapter without those. We talk about students with "sufficient motivation," but motivation isn't a fixed thing, and having a community and deadlines can really push up motivation.
A web page or video can explain a subject. A textbook will have problems to work on. But feedback, community and deadlines are something you get from a University class or a MOOC. And there are precious few MOOCs for even 3rd or 4th year undergraduate topics.
Impressive.
"In late September Micron, an Idaho-based maker of memory chips, reported a 20% year-on-year fall in quarterly sales. A week later amd, a Californian chip designer, slashed its sales estimate for the third quarter by 16%. Within days Bloomberg reported that Intel plans to lay off thousands of staff, following a string of poor results"
But most of the article is about how they're building new fabs in America, right when sales to China are being restricted. So its about a future surge in supply and reduction in demand.
That's the only time I've ever used them though.
[I've spent 450 hours in Factorio, but stopped soon after my partner nick named it Divorcio.]