https://news.ycombinator.com/item?id=48647272

Third time's the charm

1- Tradeoff individual account risk, for systemic risk. You may argue password managers are safe, but few would argue that the risk model reduces the risk of individual password leaks more than the risk of all your passwords leaking. It's a tradeoff.

2- Cat and mouse security: There's a class of security decisions that work because they are new and different. First the weakness was that passwords were short, then you make passwords long but unmemorable, so people rely on some other mechanisms to authenticate, like a file on their computer, a drive, a fingerprint, facial recog, which may in turn be protected by a second factor password.

At first the new security model will not be stressed, but as more users migrate from one security model to the next one, that's when you are able to compare the security of both technologies, it starts being a juicy enough target that it becomes attacked.

So we are at the point where password managers are used enough that they start becoming worthwhile targets of attack (to overcome the difficulty of vulnerating them).

Also worth noting that these attacks are more winner-takes-all. In the sense that rather than seeing one account hacked every couple of hours, you will see them all hacked at once, because you introduced a vendor in the password supply chain AND because the vendor centralizes all of the passwords. So target that one vendor and from a single attack you get all the spoils. So when comparing the security of the olden method and the new, just 1 incident is enough to undo all of the reputational gains it has made over the years.