Trace – Freddie Enrichment and IQ Debate: Why Trace is Wrong

Trace has gotten into a spat with Freddie deBoer over educational enrichment and IQ. Freddie soundly rejects the blank slate, whereas Trace, finding some agreement with Freddie, still sees value of enrichment. I am going to give this one to Freddie, and I think Trace is off the mark here.

It started with an article by Freddie deBoer, who asserts that enrichment like chess or piano lessons does not make children smarter. Freddie’s point is that parents who hope that enrichment and extracurriculars will make their otherwise dull or average kids stand out, will be disappointed to learn that enrichment only reveals disparities of innate ability.

In a string of Tweets, Trace rebutted Freddie’s article:

But what predicts level of expertise? It’s not like expertise is equally distributed. Trace’s major blind spot in this exchange is failing to see that domain-specific expertise for g-loaded activities is highly correlated with IQ or other innate factors, like talent, which ultimately determines one’s ranking or placing when many people are competing at said activity. I don’t think too many people would dispute that playing chess, learning the piano, or creative writing have some value, but if one actually seeks remunerative returns or accolades instead of merely being a participant, that means having to compete against other people who also want to be good at that very thing and also reap the benefits. If everyone is practicing and putting in their ‘10,000 hours’, who lands the lucrative career or contract? More competition means innate factors become more important.

Again, what predicts relative differences? IQ. And it also happens that smarter kids do better at this too. Smarter kids learn at faster rates, retain more, make better inferences–things we would expect from people who are smarter. Having top 1% pattern recognition or working memory helps for a multitude of domains, not just taking IQ tests.

Again, Trace gets the causality wrong. What would you guess Faustino’s IQ is? I would hazard to guess it’s way above average.

What separates Magnus Carlsen from other chess players who have lost to him even though they also practice a lot? It helps he has a very high IQ, or at least it can be reasonably assumed to be:

Carlsen showed an aptitude for intellectual challenges at a young age. At two years, he could solve 500-piece jigsaw puzzles; at four, he enjoyed assembling Lego sets with instructions intended for children aged 10–14.[11]

His father, a keen amateur chess player,[12] taught him to play at age five, although he initially showed little interest in it.[13] He has three sisters, and in 2010 stated that one thing that first motivated him to take up chess seriously was the desire to beat his elder sister at the game.[14]

He was not even initially interested in chess but quickly became the best.

And a photographic memory:

When he was 5 years old, Magnus memorized the populations, capitals, areas and flags of all the 200 countries in the world, much to the surprise of his family, who had no idea he was even attempting such a thing. He had even taken it further, though, by memorizing the same numbers for all the 400+ municipalities of Norway. At memory games he was always unbeatable.

This again suggests innate qualities that made him so unbeatable at chess. Working memory is a major part of IQ.

Then how does it spring up? Studying and learning are great, but Freddie’s point still stands, which is that IQ predicts how much millage one gets out of studying. Same for rank-ordered success for g-loaded activities, in which everyone is practicing and studying but only one can be the best. Consider a powerlifting tournament: everyone practices, but in the end, only the strongest lifter wins, not the lifter who practiced the longest.

“Playing piano or learning chess is great for the sake of learning how to play piano or chess.” I don’t think anyone is arguing against that. But to Freddie’s point, society tends to reward exceptional individual talent, not merely practicing or participating. That is why the maligned participation trophy has always been the butt of jokes.

This is a misinterpretation of his beliefs. Freddie doesn’t say it’s possible, practical, or even desirable to eliminate economic or other societal disparities. Rather, he believes in a strong social safety net for those who may not be smart enough or otherwise fortunate enough to contribute and fall behind.

Trace expounded on his beliefs in a much longer blog-tweet in response to a private reply by Freddie, published with the latter’s permission [0]:

Trace writes:

Schools are generally very, very bad at teaching for expertise. They are simply not in the learning maximization business, caught as they are between a thousand different priorities. Education policy is a frustrating morass, and translating policy to improvements for any given kid is a struggle in the best of circumstances. But make no mistake: there are better and worse ways to teach, better and worse ways to learn, and concrete ways education in narrow fields can make a difference for people. We understand much of this, but it clashes with our instincts, and we are slow to implement the best tools.

Imagine a hypothetical school in which teaching is 100% optimized. Thus, innate differences must matter, because environment has been optimized. This is the point Freddie has always made, and Trace has not refuted this and attempts to sidestep this by going off on digressions about how learning is good for the sake of learning, or how schools do a poor job of teaching. Learning is good; no one disputes this. But those who make money and achieve success in material terms or social status, are not just good; crucially, they are good relative to others, which is what matters the most, and is Freddie’s point.

The overall impression I get is that Trace is arguing against a strawman. No one is saying that expertise and learning are not worthwhile, but rather that IQ is predictive of how much expertise one attains or how much progress one makes relative to others for competitive, g-loaded endeavors. For example, many high school students compete in math competitions, especially in high-SES areas, but generally it’s the smartest competitors who tend to place the highest and win the medals, holding environment (e.g. studying, effort, cramming etc.) constant.

Given the high stakes of these competitions for elite college admissions–when GPAs and other metrics have become inflated to the point of uselessness and acceptance rates keep falling (plus record-high salaries and status in FAMNG, banking, consulting and other sectors for elite school graduates)–top placings in competitions is one of the few, objective ways to stand out from the masses. And you can be sure competitors are availing themselves of aforementioned options, such as studying and coaching, leaving IQ as the deciding factor.

Continuing:

IQ was designed to measure how far someone’s aptitude deviates from average. It follows a normal distribution, such that an equal amount are above and below. Many cognitive tests also aim to measure more domain-specific aptitudes. It measures something real, if hard to pin down.

But skills don’t follow a normal distribution. They follow something closer to a power law: the vast majority of people are indescribably worse at chess than grandmasters. If you measure everyone’s chess skill, almost everyone clusters at the base. Every once in a while, some hotshot will claim to be a “super learner” and will try to ascend the skill ladder very, very quickly—memorably, Max Deutsch, who humiliated himself against Magnus Carlsen. They invariably fail as he did. No matter your aptitude, the process to build chess expertise takes a very, very long time.

Yes, the distributions are different, yet obey the same exponential-decline pattern or behavior at the left-side/tail, so they are effectively the same. If one was to truncate the normal distribution or use the log-normal, one would obtain a distribution that resembles a power law. So thus, IQs above 150 are very rare, as are grandmaster chess ratings. I am not sure how this disproves Freddie, or who he is even arguing against, or what point he is making.

Because expertise follows a power-law distribution and not a normal one, everyone who is reading this is capable at becoming better than most people in specific, narrow domains. You may not be able to reach the absolute peak. You will hit slowdowns and, perhaps, hard limits. But you can learn. You can learn languages, you can learn math, you can learn chess, you can learn the law. You can specialize, you can drill into specific topics, and you can outperform the great majority of people within them. Your kids, too, can do this, much earlier and much more than the default path will give them the chance to demonstrate.

Yes, anyone can make marginal gains at chess or whatever by practicing, and with directed/focused practice, one can get even better. Maybe even win a local tournament. But others are doing that too. So when you got lots of people practicing and studying to attain recognition for some rank-ordered, g-loaded endeavor, what variable, in the end, determines the superlative attainment of ranking and the glory and other benefits that come with it? Likely it would be something innate, like IQ, aptitude, or talent, given that we have controlled for the other stuff, like practice, coaching, etc.

Overall, Trace is trying some intellectual sleight of hand or goalpost moving where he argues that because people can get better with practice, or that schools do a poor job of teaching mastery, that this invalidates the role of IQ or other innate factors, overlooking or ignoring that environmental factors can also be reproduced, leaving innate variables as the deciding factor. I have seen this first-hand where people try and try, but because they lack that certain creative spark, or what some would call talent, never make the leap from good to great. He also ignores that success or mastery at g-loaded activities are highly correlated with IQ.

It’s like we want to be reassured that IQ does not matter. Sure, maybe for piano IQ does not matter that much, but it’s generally good-paying, competitive, high-status things in which talent or other innate factors are necessary (even if insufficient). And as long as status, by definition, must be scarce and unequally distributed, I don’t see any reason for this to change.

[0] As an addendum, the use of Twitter as a de-facto blogging platform agrees with my post from earlier in 2023 in which I argued that Twitter was overtaking Substack for blogging, as Twitter is superior for generating virality compared to Substack. Many of the Substack bloggers I follow are using Twitter a lot more, and blogging less.