Earlier quoted context omitted.
Or former insider. I spent several years pointing out to my last employer that every former employee could have walked off with secrets that allowed them access to our backends. The were already slowly working on hardening write access but read access was still being worked on a couple months before I left, when I got to write about half of the last mile code for the user facing bits. This is not a unique experience…
Are these business-owned exchanges and managed wallets not fundamentally incompatible with making guarantees of security? Is anyone doing it the "right" way and what does the right way even look like?
But one mistake we make over and over is that we write code that just does its best to answer questions as quickly as possible. And when those questions show up 10x as quickly as they have any other time in our company history, they either just plug right along or maybe throw an error.
Someone shouldn't be able to empty a billion dollars out of an exchange in 10 minutes, unless they do $250B in daily traffic. And I suspect most of them can be, and in even less time than that.