Totally off topic: Yeah so, about that 90% number... You you really think that is the best they can do?
I mean, 90% is pretty good for, like, dog food ads and car commercials, but with lives, a 90% true-positive rate is just garbage. Like, with all that terrific amount of data, are they still just doing T-tests, ANOVA, MW-U-tests? Like, what is their p-value, still 0.05? I know this is super stats-wonky for this thread, but I mean, come on, they have to have some super secret stats and mathy stuff that they are doing, right? Like, formulas and theories that are just really good. It's been, like, 15 years they have had this scale of data, and it's only growing, right? If so, nothing at all has been sent out to the academic community, which, for math theorems, is kinda hard to believe. I now signal-to-noise is super important for NatSec, but it's also super important for DrugDev.
But yeah, a camera that is meant to watch me dress and then order shit for me, that is a super no-no. It, like, actually gives me goosebumps.
"Totally off topic: Yeah so, about that 90% number... You you really think that is the best they can do?"
For the political parties specifically, yes, by the standard they care about. Remember how we're always talking about how the centrists generally end up with the deciding vote, and how the polls are oscillating around by ~10% in the several weeks leading up to the Presidential election? Those people themselves don't really know who they're voting for or whether they support the "correct" person, for any given definition of support, so it's a bit much to expect anyone else to accurately guess.
In the event of a true police state which gives you the choice of vigorous fealty or the Gulag I would expect they can go much higher and that the initial competent execution would be almost inescapable. (Over time it would decay due to various forces, but I can't put a very solid time frame on it... between 10 to 50 years to develop very serious holes in it, probably, but even 10 years is an awful long time.)
I mean, 90% is pretty good for, like, dog food ads and car commercials, but with lives, a 90% true-positive rate is just garbage. Like, with all that terrific amount of data, are they still just doing T-tests, ANOVA, MW-U-tests? Like, what is their p-value, still 0.05? I know this is super stats-wonky for this thread, but I mean, come on, they have to have some super secret stats and mathy stuff that they are doing, right? Like, formulas and theories that are just really good. It's been, like, 15 years they have had this scale of data, and it's only growing, right? If so, nothing at all has been sent out to the academic community, which, for math theorems, is kinda hard to believe. I now signal-to-noise is super important for NatSec, but it's also super important for DrugDev.
But yeah, a camera that is meant to watch me dress and then order shit for me, that is a super no-no. It, like, actually gives me goosebumps.