Recruiters hardly ever seem to dig into things... But if you get far enough in the process, the people doing the interviews sometimes will dig a bit. Sometimes. Sometimes they print off your resume 10 minutes before, ask one or two things about it, and throw you a couple whiteboard problems to take up the rest of the time.
A college degree in CS or related used to be a kind of 'community certification test', but then interviewers started getting bad candidates even though they had BS or MS degrees, and when they started testing for ability to program a FizzBuzz, found that some MS students couldn't do it! So its value as a test has gone down significantly. This is why I like the idea of an IQ test better, because it can't be cheated and corrupted the same way.
Classification of company profiles is kind of what Glassdoor et al. purport to be, but sometimes I can get the feeling I won't fit in just from the company's website and how it represents itself (especially if they have a dev blog). e.g. One company had a picture of a scrum team doing pushups together at standup. I'm not opposed to exercise, or the idea of getting together with some coworkers to go work out at a gym, but I am opposed to such group dynamics and conformity pressures at work. If I worked on a team that already had the established culture of doing pushups during the daily standup, I would decline, and that probably indicates there would be other cultural problems down the line.
I'm not familiar with Gould's discussion of IQ but I generally don't trust Gould's work at all and don't bother reading it. https://www.lesswrong.com/posts/BahoNzY2pzSeM2Dtk/beware-of-... convinced me of that, what he did to evolutionary biology isn't really forgivable. From the first sentence of your wiki link I would worry he did the same to IQ research by representing everyone in the field as whatever he characterizes are "biological determinists". IQ is merely the best single predictor we have for a large variety of "success" metrics. Its validity predicting job performance, as one of those metrics, depends on the job and can range from 20% to 80% (i.e. knowing nothing about a person but their IQ, for some jobs, you can predict their performance bracket accurately say 75% of the time -- they will be better or worse than predicted the other 25%). I suspect software job performance would be towards the upper end of predictive power. Throwing a second data point of Conscientiousness in, your predictive accuracy rises even more, but Conscientiousness alone isn't as useful as IQ alone.