the slow operational bitrot of twitter remains absolutely fascinating to witness. i get the feeling some people will point to it as a case-study against microservices, when realistically it's a case-study against firing the overwhelming majority of your employees and leaving a skeleton crew struggling to keep a major social media platform online while forcing them to implement the dumbest possible features.
i strongly recommend against doing this.
I'm sort of in awe of the past engineers there even if the new ones are mostly trapped and rapidly burning out.
because this is like watching an entire season of improbable Star Trek engineering feats under impossible odds, with tooling and automated safeguards completely alien to most of them, made by an ancient, now lost, technically advanced culture.
sure, the space whale is rotting but you aren't getting home in that other ship.
the current team's capacity seems lacking at times (though at how big Twitter is, who knows. half of them haven't slept since November), but think about who came before them: they designed the most clown resistant worldscale machine/system we've seen a case study for.
the last time this happened was Sun/Oracle and that was just sad. (and VERY not clown proof)
