[gaelow]

reliably <> accurately

Wow, some psychology professor who read Ender's Game (Probably watched it since, come on, psychology is science at its laziest and the timeframes match) decided to make a quick buck by publishing an online paywalled paper on the "discovery" "for the first time" that people give up signs of their intelligence when observed performing an activity that requires them to use it (which matches the description of an infinite number of activities...).

Hey, I can reliably measure your intelligence by reading your comments on Hacker News. May I have your cash-money bucks, please? $_$ I have been working hard with hundreds of my students to prove it! For the first time ever! $_$

The fact is that you need specifically designed games for an accurate test. Most video games just won't make the cut. Even a combination of them: take the top 10 from steam. I dare anyone to accurately measure the IQ of their players just by looking at them them play in whatever lab environment they want. An those would be pretty boring. They would be in fact, tests.

[stuxnet79]

Publish or perish mentality ... or ... good science hindered by a hyperbolic, misleading title. The former is cancerous but the latter can be easily rectified. I myself can't call it because I didn't click the link hoping I would get a TLDR in the comments.