kestreloats

Why Taleb?

I hate that piece because most of what he's referring to has been dealt with for decades. It's like "narcissist who knows nothing about a field doesn't bother to learn anything about it and as a result takes down a strawman to make himself look good."

So I agree with you about SAT <-> IQ, and IQ <-> SES. However...

IQ isn't the only correlate of success. I think if you look at it all, there's a large dose of conscientiousness (which standardized testing companies are now going after), and attractiveness/charisma...

... but that's on the individual side of things. There's also a whole host of societal and random stuff that is outside the control of the individual, or maybe is significant to a person, but that studies tend to treat as irrelevant.

Also, saying that the SAT is useful as a selection device doesn't mean it's the only useful selection device, or that as a selection device it's very good. Offhand I don't remember the numbers, but standardized test score probably correlates .45 or so with first-year college grades? Think about that for a second. First, that's a ton of noise. That's not very predictive at all. Second, that's first-year college grades. Change the criterion but it's still the same: your best tool really is a pretty fuzzy predictor.

So now take this very fuzzy predictor, add some other fuzzy predictors that at best get things up to like maybe .6 correlation? Still fuzzy. Now you're going to be really selective on these things? What you're going to end up with is a lot of people who would have done as well but for whom the stars didn't align right at a particular period in their life. But now we as a society have this crazy income inequality, rent-seeking monopolies of all forms, and a general winner-takes all climate, so these small meaningless differences get amplified tenfold.

The conversations about this too have this kind of all-or-nothing quality, like you're forced to choose between "standardized tests are meaningless" or "standardized tests are valid predictors for a large portion of people so we need to treat them as infallible predictors for everyone." The truth is really much greyer than these positions: yes they predict, but they predict pretty weakly, all things considered, and generally for people who fit into a certain box. This all might be fine, except now we've structured our society in part around these oversimplified assumptions, pretending it works when it doesn't really. It might be ok when there's lots of second chances, lots of opportunities for people to get back on their feet from the vagaries of life, and good opportunities in general for everyone, but when resources gets hoarded by fewer and fewer, there's more noise, that's compounded by people gaming the system, etc. etc. etc.

Just as a thought exercise: what do most human traits look like in terms of distribution? They're pretty normal, pretty Gaussian. What does income look like? Not that, not at all. The discrepancy between them should be shocking to everyone.

Somehow the idea that this is all caused by people applying to more places doesn't cut it to me. It might account for some of it, but not all of it. A lot of this is probably fairly chaotic and complex. For example, the GI bill probably contributed directly somewhat also, but also probably led to cultural shifts because of the veteran population, which then in turn changed career and educational expectations, but the GI bill itself reflects WWII, which also reflects cataclysmic societal changes (that is, WWII caused many significant things, but was also caused by many significant things).

Speaking as someone who has done research on selection and standard testing, and whose first-hand experiences support what I see from the theory and research, this is kind of horrifying, because it inevitably leads to increases in bullshit.

The accuracy of any of the selection methods we use in education is very poor. It just is. Standardized tests, "holistic review", all of it. It's not useless, but it is poor. It's the sort of thing that works well for a certain level of selectivity, but not past it. Once you go past it, you're selecting on well-impressioned noise, and incentivizing bullshit.

Somehow this all seems related in my mind to the rise of concerns about replicability in science, college admissions scams, and the age of fraud in general (https://www.theatlantic.com/entertainment/archive/2019/04/th...). Something is broken. Probably many things.

So... I've been programming in R for over 20 years. I've been ready for an alternative for performance reasons for about half that time. Julia seems like a promising alternative, that I wish I could use all the time instead of R but it's just not there; I'd prefer Nim most of all but that's even less well-resourced in terms of libraries. Maybe a zero-cost abstracted offshoot of Rust will eventually come to have a role? Who knows.

R, like Python, has far outgrown its initial scope. I don't think it was initially envisioned to be used the way it is today. But both have been kept in use as costumes for C/C++.

One of the things I've noticed the most in the last 20 years, to your point, is that the language used to be a lot more straightforward and simpler, more predictable. Over the years a lot has been added in a sort of haphazard way, and as a result today you have this kind of Frankenstein language that isn't what it started with.

As for data structures, though, I don't really see R as being that different from other languages. Many of them are the same as in other languages, but just have different names (and I do wish they used similar terminology). Others have been taken up in other languages as people have come to appreciate their utility.

Being a wrapper for C/C++ can only go so far. Eventually you have to write in R (or Python) and the speed shows, if you have enough data to deal with.

I think what a healthy diet is fairly individual-specific. I think there are probably general guidelines that are pretty good for most people, but the details depend on how different people metabolize different things, their taste preferences etc.

I think I feel the best eating vegetarian, whole grains, lots of fruits and vegetables, nuts and legumes, no restrictions on unsaturated fats but little or no saturated fats. I basically try to avoid processed foods, saturated fats, artificial colors, artificial flavorings, and preservatives, and meats (including fish). I try to avoid added sugars too but am not too strict about it, and don't have any restrictions on complex carbs. I've kind of done intermittent fasting before it was a "thing" and I do think that's been helpful/feel better with that at the right times, as has not eating after a certain time at night, and keeping breakfasts light or waiting a bit later.

It's worked for me but I think everyone's optimal diet is probably a bit different. I think in general you can't go wrong with whole grains and nuts, whole vegetables, and whole fruits. The rest of it is kind of tweaking for an individual.

I think the poster was just pointing to the parallels. It's salient to me as a professor.

FWIW, I think the issues are definitely related though. There's lots of STEM hype in academics as well as in commerce which creates a labor oversupply in both.

The problem is that there are genuine advancements in STEM that are interesting and rewarding, from the academic as well as commercial side, but they're sort of buried in a sea of bullshit, and being able to connect with the worthwhile stuff (financially or otherwise) can often be due to random or sketchy reasons. So people see the great stuff going on, and then either promote it (in the case of politicians and public figures) or pursue it (as someone seeking a career) not realizing that throwing more people at something will necessarily produce more of what we like.

So you end up with the Gallipoli that is modern academics.