frgtpsswrdlame

>I think we should revisit the policy of an actual forced savings account - rather than the bad proxies for it: social security, home equity, 401k.

I think social security is a far better setup than forced savings account. The fact that it's much harder to raid, that it's not really 'your money' locked up somewhere is a big reason why it has lasted as long as it has.

So this guy had a few twitter exchanges with Luigi and he thinks the US healthcare system isn't that bad? Doesn't really seem like an article for HN. But as long as we're here I think, sort of like Luigi, I'll let other people argue about whether the US healthcare system functions well or efficiently. But as for this:

>While it’s true that UnitedHealthcare has the highest denial rate for medical claims, the CEO doesn’t set the approval rate of a health insurance company’s payouts — that’s done by the actuaries, who themselves are constrained by various considerations, such as the need to keep costs low, including for policyholders.

By this logic, what can Brian Thompson be said to be responsible for? It's very strange to me to assign more responsibility for a companies denial rates to its actuaries than the actuaries' bosses bosses boss. So why exactly does UHC have higher denial rates than other companies? It just happens to be that it's actuaries are more frugal? This explanation doesn't hold water and I think it's a strange response to the shooting of a CEO to say, well actually CEOs aren't responsible for the things their corporations do. Of course they are, now they aren't solely responsible, but they probably have more responsibility than any other single individual. It can be completely true that killing Brian Thompson was wrong AND that he was no saint, being responsible for (and getting rich off of) large amounts of human misery inflicted on UHC policy holders by the organization he headed.

Regarding this part:

Clarification of specialty role

  - Less strict on the direct link between degree/job responsibilities

  - Recognizes that AI may require multiple academic background

You really won't need to clarify whether the role is a specialty one or not if you just increase the minimum wage for H1Bs. I really don't know why we don't have some rule that pins H1B wages to like the 90th percentile wage.

10 people died from the Boars Head contaminated meat. I guess I don't really mind if a few lawyers chase those ambulances and extract a lot of money out of that company?

I thought the issue was deregulation? Something like Trump rolled back some rules about oversight and then Biden never put them back into place? Does the USDA even have the power to fix this?

Anyways, having a pregnant wife right now, it's worrying that listeria is on the rise. We're already avoiding stuff like deli meat but when there's listeria in the frozen waffles, mushrooms, and vegetables you have to wonder what's really going on.

If you're the only one that can come close. Kaggle competition prizes are about focusing smart people on the same problem. But it's very rare for one team to blow all the others out of the water. So if you wanted to make a business out of the problem kaggle will (probably) show the best you could do and still have no moat.

>characteristic of these use cases is that there is no one right answer

I think what you mean is that they work best in cases where it's very hard to measure how well they are working.

Yeah but what made Twitter twitter wasn't really the usercount, it was the mix of established voices and complete randos mixing. If the journalists and economists and politicians go to Bluesky, it will win. I definitely don't think that's a given but it seems much more plausible now than it did a few months ago.

I mean what if it is a toss up? Were we talking about a literal, actual coin flipping, calling it as 50-50 is a better insight than calling it as either heads or tails.

I like the dashboarding more than the analysis lol. I think FL is just a red state now and positing it as anything else smells a little fishy to me. Also, I think there's a thing now where all sorts of analysts know they can get a name out of making one big correct prediction that everyone else got wrong. That's lead to a lot of predictions being made with the intention of being contrarian.

Judging by my twitter feed, some scientists are starting to push back on the use of UMAP (and t-SNE.)

I still don't get the impetus or desire to make NNs work better for tabular data. Regression works pretty well and is easy to interpret/diagnose/work with. GBMs work really well (given a few considerations) and is trickier to work with but nothing crazy. When I see all the fancy hijinks people get up to when applying NNs to audio/text/pictures I think it's really cool but also not something I'd want to have to do if I didn't absolutely need to when working with data out of a relational db. And anyways, how much of a benefit could it actually bring? GBMs are already capable of fitting and dramatically overfitting most datasets.

I actually think Nate is wrong on this, especially on how he states the wrong-ness of the two parts of the story. It seems very likely to me that initial death counts were overstated. But the probability that it was an Israeli bomb that went off in the parking lot carries a lot more uncertainty and I think there's still a decent chance that it's true. There's still a lot of investigation to be done but, if when all is said and done, it turns out the bomb likely was from Israel, will Nate stick to his own standards of retraction?