Which is why this outage is so weird: the entire point of paying MarkMonitor is ensuring that absolutely nothing goes wrong with a very fraught process, and they seem to have just taken down one of the biggest brands they support.

Precisely. You pay a company like this the nosebleed-inducing fees they charge so that this exact event never happens. That assurance, and not the mechanics of domain registration or canned web searches or whatever else, is their product.

It's like, as I'm sure I'm paraphrasing from something I read God alone knows how many years ago, if your publicist lets you walk into a press event with a giant blob of snot hanging out of your nose. There surely is a reason why that error occurred, and it probably is at least a pretty good reason. But no one is very surprised to see the intro invite from your new publicist.

It isn't a relationship you blow up on a whim, but Zoom that can't route call traffic is Zoom that's not generating revenue, and while the reputational impact is negligible if it happens once, it had really better happen only once. Zoom is the incumbent; no one remembers they were revolutionary once, now everyone only notices the parts they don't like. (Being a skilled but politically naïve sysadmin is much the same.)

Basically, this is why Ma Bell - which had about the only stronger possible "uptime" expectation, in that no one uses Zoom for 911 - was so uptight you couldn't even plug in a modem until about five minutes before divestiture, and specified everything down to the number of turns in the splices their technicians made. There was a fad among programmers, when I was a child, to consider such practices stodgy.