The usage and access of data in your data store when building a realtime application. If you were building a chat-app these days you are given a number of very easy to adopt (minimal complexity/edge cases) NoSQL solutions - Firebase, Parse, RethinkDB.
Whereas with SQL you have the option of using cache layers (which you could say is NoSQL anyway), message queues, naive timely polling and log monitoring.
With NoSQL the solutions are just simpler. For example Firebase has a tutorial for creating a real-time chat [1] - it takes 5 minutes.
I get your point of view. My hypothesis for this behaviour, or part of the reason for the behaviour, is that I think the education system has traditionally been geared to teaching about SQL systems over their NoSQL counter-parts. This creates developers with strong opinions of one, without necessarily having a sound understanding of the other(s).
The people who cause pain in your organisation are almost definitely committing a logical fallacy - see [https://yourlogicalfallacyis.com/]
Outperform on what? Can you give specific examples? There are a large number of Combinatorial Optimisation Problems. It is a very sweeping statement to say 'gradient descent techniques far outperform evolutionary ones'.