This doesn’t ring true to me. Everybody makes mistakes and it seems to me that popular big engineering companies are more likely to try to understand root (technical) causes of outages and mitigate them rather than blaming people for being around when a fragile system fails (which is what it sounds like you are suggesting with the emphasis on not making errors.)
The point I was trying to make was that even if 99% of the job doesn't involve any algorithmic knowledge and is boring crud work, the one day a year (or even less) where one of those engineers has to recognize how to do optimal graph traversal or memoization makes it all worth it.
When seconds of latency even for hours can be hundreds of thousands of dollars or more, hiring to make sure someone is less likely to screw up one day a year is worth it.
