Since a good bit before 2014, even. They used to recruit very heavily from MIT (while Palantir was overrepresented at Stanford and Quora at Harvard--all of these reflecting the almae matres of the founders). Note that this was before the popularity of Leetcode and the whole cottage industry around trying to game algorithmic-type interviews. I'm not sure if similar companies founded today would push these algorithm-heavy interviews as hard, since they've probably lost some signal now & prevailing attitudes have changed a bit.
At any rate, it doesn't surprise me at all that Dropbox engineers do better than FANG engineers on these technical metrics. The average Dropbox engineer is almost certainly a bit smarter and a bit better at algorithms than the average FANG engineer. Of course those attributes don't automatically translate into being a better engineer, though, nor do they automatically translate into company success or anything.
At any rate, it doesn't surprise me at all that Dropbox engineers do better than FANG engineers on these technical metrics. The average Dropbox engineer is almost certainly a bit smarter and a bit better at algorithms than the average FANG engineer. Of course those attributes don't automatically translate into being a better engineer, though, nor do they automatically translate into company success or anything.