This kind of thing has been around the block for a bit -- the reality in my opinion is that "programming ability" is simply not something we've defined and possibly not a single thing. The many decades of interest in predicting programming ability have sometimes succeeded at pushing against our stereotypes that it is math associated (as this work), but "math ability" is ALSO a fraught measure. It's important to bring a lot of context to the prediction of ability...
How do we think about aptitude, ability, performance and potential -- these are massive and complex arguments even in the areas of psychology where we have done the most work and have the strongest evidence to draw on. I've been reading a lot about "predicting programming aptitude" and this work here is better than a lot of what I've been reading, but in the entire area I see a lot of failure to integrate with modern education research.
Gotta read this whole thing to have an informed opinion but "EEG to large claim about extremely complex real world implication about aptitude" 🧐 not usually a fan (and I'm speaking as someone whose PhD was in an EEG lab)
We recruit any huge number of capacities that we have available to us in order to succeed at solving real world problems and do our complex knowledge work. Looking for the "real strongest predictor" can be a very dangerous game even when it might line up with something we want to believe. Language feels more inclusive vs math ability! But is it? Take it too far and what are we saying about the enormous % of people in the world with language disabilities?
So I have no doubt there are interesting things to surface using the many technologies available to us in measurement. But folks in tech are generally not aware, and should be more aware, of the enormous critiques leveled against biological reductionism in how we talk about aptitude and the many ways modern neuroscience and cognitive science both give us strong signals against this. I've done a few threads on how cog sci "basics" like working memory are questioned or measure differently
Individual differences research cannot be divorced from its very long and very ugly history of exclusionary, discriminatory arguments. We need to understand that history when we seek to extend research on "human ability" into areas that have such enormous economic and social implications for people. Against this isn't really against this one paper but a caution to not overindex on this kind of approach. Like I said..."programming ability" is not necessarily one thing in the real world.
I'm not quite in the middle of watching Sapolsky's lectures on biology of human behaviour, and a topic that repeats there is how reasons for interpersonal variances were gotten very obviously wrong (or less obviously, but still badly wrong), and how to recognize such problems.
(The lectures are altogether pretty long, available iiuc as audio only, and done at a "podcast pace" -- you can listen to them with nonfull attention and don't need to backtrack.)
