Way back when, it was a pretty common screwup to accidentally saturate the nodes you were packeting from. So then your C&C couldn't get them to respond, either. Oops.
I don't think that's actually true, or at least is certainly cannot be taken for granted. Instead, it appears ransom has followed more of the path of Silicon Valley VCs:
.It sounds perverse but the incentives require it: if payment didn't bring resolution, no one would pay. As a result, all of the big gangs avoid scamming.
What you're describing is the expected Game Theory outcome over long periods in an iterated game. This works as long as the payment amount is towards the <salary> side of the potential payment spectrum, where each payment may well be decent money for the work the ransomers put in but not so much that they don't need new ransoms. The problem comes if/when the absolute amount of payment moves from "salary" to the "Exit"/"Retirement" side of the spectrum, ie, heads into what VC would call "Unicorn" status. At some level of money it reaches the point where the ransomers need never work again in their lives, it's enough money to get out of the risky business and live off of it indefinitely. It's now no longer an iterated game but a single game, and in single games defection can be rewarded. It no longer matters if reputation is burned, on the contrary it might be the moment to cash all accumulated rep in.
I think in general, both on the bright and dark sides, this sort of "phase change" in a given market space is worth trying to keep an eye out for because it can result in significantly changed behavior "out of nowhere" that can head in ugly directions very fast.