MarceliusK

4 ms·

1.
by MarceliusK 11d ago
That's an interesting wrinkle. If anything, externally funded international students seem like exactly the case where you'd still want the interview, since the funding answers "who pays for this?" but not "does this
2.
by MarceliusK 11d ago
I think this is true of defenses as they exist today, but that may be partly because the thesis currently does most of the signaling.
3.
by MarceliusK 11d ago
The tricky part is that live review is much more expensive than async review. But maybe that just means we were getting away with a cheap proxy because it happened to work reasonably well.
4.
by MarceliusK 11d ago
The key shift seems to be from "can you produce mathematics?" to "do you understand mathematics?"... Those used to be correlated strongly enough that a thesis or paper could serve as evidence of both. If AI breaks that c
5.
by MarceliusK 18d ago
A phone is a pretty bad proxy for what someone is actually doing with their attention
6.
by MarceliusK 18d ago
The first rule reduces platform power. The second potentially requires platforms to build even larger classification and moderation systems. I'm very much in favor of the first, but I don't think they belong under the same concept
7.
by MarceliusK 25d ago
And the incentives are arguably worse than QA: you get little credit if the result replicates, and potentially a lot of grief if it doesn't. Novel results build careers. Checking them mostly doesn't
8.
by MarceliusK 25d ago
I'd be pretty wary of stretching wire fraud law to cover this. Prosecutors deciding whether scientific results are fraudulent sounds like it could create its own set of ugly incentives
9.
by MarceliusK 25d ago
At this point the surprising part isn't that another Ariely result has problems, it's that some of these datasets apparently went decades without this kind of basic scrutiny
10.
by MarceliusK 1mo ago
I'm really sorry for your loss. There's not much to say to something like that except that I hope this treatment came too late for fewer people than it did for your sister
11.
by MarceliusK 1mo ago
If molecular glues keep working like this, I wonder how many other "undruggable" targets are really just targets we've been trying to attack in the wrong way
12.
by MarceliusK 1mo ago
Pancreatic cancer is one of those diseases where I've gotten used to reading depressing numbers, so seeing 13.2 months vs 6.7 months actually made me stop for a second
13.
by MarceliusK 1mo ago
This is the classic difference between testing whether the software works and testing whether the system works
14.
by MarceliusK 1mo ago
That does make the 10 million figure more meaningful but I still think it is doing a lot of work rhetorically
15.
by MarceliusK 1mo ago
Without the total number of attempts, number of failures and ideally how those failures are distributed, it tells us almost nothing
16.
by MarceliusK 1mo ago
I mostly agree but I think the bar for the software still has to be unusually high when the failure mode is "you are not allowed to enter the country you legally live in"... The human fallback is essential, but it shouldn't b
17.
by MarceliusK 1mo ago
"Known issue" and "you may not board the flight home" really shouldn't appear in the same system. If a database error can effectively invalidate your legal status at the border, there needs to be a fallback that doe
18.
by MarceliusK 1mo ago
IKEA Poland missed an opportunity here
19.
by MarceliusK 1mo ago
This is exactly why the names work
20.
by MarceliusK 1mo ago
I always assumed the names were mostly random Swedish words picked for the vibe. The fact that there's a taxonomy behind them makes it much funnier somehow
21.
by MarceliusK 2mo ago
So it is less "England has a program and the others don't" and more that each NHS is pursuing the same WHO goal independently, with different reporting and timelines
22.
by MarceliusK 2mo ago
My guess is that devolution makes these programs and statistics look more separate than they really are
23.
by MarceliusK 2mo ago
The downside of that model is that "be proactive" assumes people know what they should be proactive about
24.
by MarceliusK 2mo ago
A disease that can stay silent for years, but is also highly curable once detected, seems like exactly the thing you want to screen for more aggressively
25.
by MarceliusK 2mo ago
There is a pretty striking contrast here with the infected blood scandal. The NHS went from being responsible for one of the worst hepatitis C transmission disasters in the country's history to potentially becoming one of the first hea
26.
by MarceliusK 2mo ago
The problem is that one essential app being unavailable can outweigh the other 95% of the phone working perfectly
27.
by MarceliusK 2mo ago
The frustrating part is that security increasingly means proving you are using an approved device and software stack, rather than proving that you are actually the account owner
28.
by MarceliusK 2mo ago
The distinction still matters legally and administratively
29.
by MarceliusK 2mo ago
If it lets the US use that identity to query European databases and retrieve associated information, that is a much larger change
30.
by MarceliusK 2mo ago
It changes the nature of the scheme quite a bit

More ›