Hard disagree. This is a problem every web service has had to deal with since the beginning of the web: what happens when a host (either the machine or the person) stops working? How do you keep the service up?
Centralized services solve that problem with internally funded, transparent redundancy. Federation solves the problem with externally funded, highly-visible redundancy. They’re still the same solution, just a different way of going about it.
You could argue that user identity is lost due to the discontinuity between instances, but that’s probably something the Lemmy devs could fix without too much hassle.