shredlefiddle
No bio...
User ID: 1727
shredlefiddle
1yr ago
Isn't it possible that some people in the deep state think Biden is getting old, and want to see him replaced by a younger Democrat? That's what voters in general want, if I recall correctly, and it seems reasonable to take voter sentiment as a prior for government employee sentiment. (Actually I think govt employees lean Dem heavily but that pretty much reinforces my point)
shredlefiddle
1yr ago
Interesting, I think this theory implies that sports rivalries are helping to make America less racist?
Unfortunately, it appears that education is increasing IQ but not increasing general cognitive ability (Lasker & Kirkegaard, 2022; Hu, 2022; Kirkegaard, 2022). This is reflected in the fact that not all g-loaded test items see improvement. It is as if you purchased my cheat sheet and became good at the test but noticeably saw no improvement on certain items, namely the ones not on the cheat sheet.
Seems like a bad argument. You claim that education represents an effort to "game the test". But if so, why would it increase IQ? Very few schools give lessons in how to pass an IQ test!
"Does generalize" vs "doesn't generalize" seems like a false dichotomy. It sounds like education generalizes to some degree, but not to the point of increasing "general cognitive ability". Call it an increase in "general scholarly ability" or something like that. Increasing "general scholarly ability" could still be a huge deal. Lots of important intellectual tasks probably depend on "general scholarly ability" in addition to "general cognitive ability".
BTW, I made the argument above due to my knowledge of causal graphs, which is itself something I learned in an educational context. Does my knowledge of causal graphs cause me to score higher on an IQ test? Probably not. But it does make me a better thinker and scholar.
Furthermore, your post misses an important point: Average IQ might not matter much for national prosperity. It may be the case that what matters is the IQ of the top 5% of the population. The top 5% will be over-represented in key administrative roles and in innovation clusters that drive economic growth. See https://www.institutostrom.org/en/2018/09/09/hive-mind-how-your-nations-iq-matters-so-much-more-than-your-own-interview-with-garett-jones/
So instead of discussing the impact of education on the population as a whole, perhaps we should be focused on the impact of education on the so-called "cognitive elite". I think there are a number of reasons to believe this impact is positive. In the absence of education, it seems likely to me that much of the "cognitive elite" would fail to acquire the belief that scholarship is important, get nerd sniped by computer games, and fail to develop self-discipline. Our education system teaches the "cognitive elite" to be snooty nerds who think that ability to solve tricky calculus problems is what's important in life, because they're surrounded by peers who can't solve those problems, and their ability to solve those problems makes them feel special. Without an education system, those same nerds would just be an unusually talented bicycle mechanic in an African peasant village, letting their potential go to waste.
I think the education of the cognitive elite matters a lot, because scholarship is a force multiplier on general cognitive ability. (By "force multiplier", I mean if your general cognitive ability is low, scholarship won't help much, but if it is high, scholarship can help a lot.) A few thousand years ago, humans had similar biological potential and general cognitive ability, but ancient civilizations weren't able to do cool stuff like modern civilizations -- essentially, because their wise men were focused on divining the will of the gods from sheep entrails instead of arguing about causal graphs.
Suppose we did a study in Ancient Greece which found that being tutored by Aristotle had no effect on the lives of 99% of Athenians. It seems like whatever Aristotle has to teach us cannot help Athenians with everyday tasks like farming and shopkeeping. We conclude that Aristotle is a fraud and learning from him is a waste of time. Then Alexander the Great gets tutored by Aristotle and conquers a huge fraction of the known world.
So overall my argument is something like: A country whose bureaucrats are familiar with e.g. causal graphs will make better policy than a country whose bureaucrats don't understand causal graphs. And education is a way to increase the fraction of bureaucrats who understand causal graphs. If we abandoned our commitment to education, none of our bureaucrats will understand causal graphs, and that will cause them to make bad policy, which will have bad downstream effects. I don't think this argument is refuted by your analysis.
Genghis Khan was also known for his focus on recruiting experts from every region that he conquered. He might not have been a scholar himself, but he valued scholarship.
In any case, you are correct that small-n tells us little. I mention these examples as intuition pumps.
shredlefiddle
1yr ago
Chris Rufo has been championing legal challenges of this sort:
https://twitter.com/search?q=from%3Arealchrisrufo%20%22civil%20rights%20law%22
shredlefiddle
1yr ago
it's all just clockwork and springs
Sure, and humans are just neurons firing...
Microsoft has displayed it cares more about the potential profits of a search engine than fulfilling a commitment to unplug any AI that is acting erratically. If we cannot trust them to turn off a model that is making NO profit and cannot act on its threats, how can we trust them to turn off a model drawing billions in revenue and with the ability to retaliate?
shredlefiddle
1yr ago
Should this really be reassuring though? Suppose you could order a science kit in the mail that allowed you to grow a brain in a vat. Imagine someone was worried about crime in their neighborhood. You respond by reassuring them: "Criminal brains are just human brains, made of neurons. Order the brain-in-a-vat kit and play with one yourself. Once you do, you'll recognize the various ways that brains can veer off track and get weird."
There's nothing inherent about token prediction which prevents Bing from doing scary stuff like talking to a mentally ill person, convincing the mentally ill person they have a deep relationship, and hallucinating instructions for a terrorist attack.
shredlefiddle
1yr ago
You're talking about a guy that reacted to Internet Feminism by contemplating suicide or chemical castration.
Side note, I suspect unless you yourself have been dogpiled on a national scale the way Aaronson was, it's difficult to understand the emotions it produces.
shredlefiddle
1yr ago
It seems like this pretty much explains the replication crisis, no? There just isn't much expertise accumulating on how to do actual science, in particular data analysis. By the time you get to be a senior grad student, you might have accumulated some knowledge on what a solid statistical analysis looks like, but then you're either up (to an administer role where you don't get your hands dirty with data) or out of academia into industry.
shredlefiddle
1yr ago
unscrupulous younger students could wait until an older student had worked out nearly all the kinks in an experiment and then swoop in to take credit for the results.
Did you swap older/younger here?
shredlefiddle
1yr ago
Why did they suck? Convince us.
A few weeks ago I linked to a discussion in the NYT about affirmative action. The most popular NYT comments were at least weakly supportive of the conservative Supreme Court's coming affirmative action ban.
Here's an NYT story from a few days ago about black New Yorkers being priced out of the city. I'm bolding sentences of interest.
2nd most recommended comment (427 Recommend)
4th most recommended comment (338 Recommend)
Others chime in with similar stories:
What does it take to achieve "friendly interactions between blacks and whites as the norm rather than exception"? What are the success stories of positive race relations (including in a non-American context) that we can learn from? I'm interested in scientific data, commenter anecdote, and everything in between. Let's identify and replicate successes like these.
More options
Context Copy link