It's a matter of ROI. The initial skim is just the sorting phase. And sure, I may miss a great candidate, but that's the price one pays. It's much better to get a false negative than a false positive.

A false negative costs me very little. A false positive is a massive burden. I'm optimizing to avoid false positives.

Also to be clear, I will read the full resume of a candidate I'm interviewing. And their blog and GitHub and whatever else they tell me about. And I'll take notes so I can ask them about interesting things I find.

The skimming is just for the initial "should we consider this person for an interview" screen.