There's a difference between $2.5B/year hard, and $500M/year hard, and between those and $50M/hard.
The front end (tight "amount of data and latency requirements") are probably doable on $50M/year, as WhatsApp was doing something comparable on much less than that.
I know where some of the money went - tens of projects like Bootstrap, which have benefited the community at large, but whose value to twitter is probably not on par with the costs. But that still does not explain even a small fraction of where the money goes.
to be fair, an issue with whatsapp is fundamentally inward facing. lose a couple of old messages? you're on the free tier, what do you want?
twitter on the other hand - a publicly liked and massively supported tweet/twitter account disappears or starts having malfunctions and the whole service is maligned. "free speech is being threatened" etc etc
twitter has been used for mass protests/revolutions because of it's reach and stability. do you think whatsapp would have done the same? they were still transmitting plain text between users and it was trivially easy to view someone else's texts for quite some time
I disagree. There are whatsapp groups of hundreds of members, some used for political action. If they indeed lose messages, people notice.
I'm not sure why you bring up plaintext (it is irrelevant, regardless of Twitter still being plaintext itself).
Engineering wise, the user-visible side of Twitter is more or less as complex as the user visible side of whatsapp. Both can be done on very modest hardware with modest operations if they are properly done. Whatsapp was lean (and probably still is). Twitter never was.
apologies i brought up plaintext as an example of a thing messaging clients should fundamentally not do but whatsapp never got around to fixing until it became a real issue for them, hell p2p encryption was only done this year
The front end (tight "amount of data and latency requirements") are probably doable on $50M/year, as WhatsApp was doing something comparable on much less than that.
I know where some of the money went - tens of projects like Bootstrap, which have benefited the community at large, but whose value to twitter is probably not on par with the costs. But that still does not explain even a small fraction of where the money goes.