I think stronger oversight and less trust is sort of implicit in a pseudo-anonymous model. An attack like I described doesn’t require something as obvious as “c

I think stronger oversight and less trust is sort of implicit in a pseudo-anonymous model. An attack like I described doesn’t require something as obvious as “code that steals money” either, it could be something much more innocuous that even afterwards appears to have been a mistake. Almost any type of bug combined with a well timed leveraged short position could result in big profits.