These coding challenges used to be a very good measure of ability, but now applicants are just memorizing the most common problems for each company. We're not at the point where these tests are useless but we will get there in the near future. The problem is that I haven't seen any good alternatives.
I'm skeptical. The stereotypical interview question involves reciting algorithms from memory, where the thing you actually want to know of a software engineer is do they appreciate how things fit together to meet the constraints of the project.
> The problem is that I haven't seen any good alternatives.
Something we've done before (and had good success with in hiring good candidates) has been to pose a real problem we're dealing with now or have recently solved (so the solution and problems are fresh in the reviewers' mind) and ask the candidate for their solution at a high level. Not whiteboard coding, though. Something like if I tell you we're trying to filter out strings that contain creatively obfuscated obscenities, what should we do in order to maximize true positives while minimizing false negatives (and as the candidate offers solutions, provide sample strings that would fail their solution and see how they adapt)?
I wish more companies would do this. The hardest part for me at least is staring at a blank document.
Huge companies like google are for the most part hiring people they can slot into almost any team in the org so it makes sense for them to place less emphasis on tooling expertise. They have plenty of time and money to upskill employees.
The same can't be said for the vast majority of consumer start ups that really just need people who can rapidly prototype.
Of course those groups are not mutually exclusive, I just think it's sad that everyone has adopted the approach that works for a select few companies
SAT is basically an IQ test, which is of course a reasonably strong predictor of white collar job performance (ceterus paribus).
That's it. Not about being smart or educated.