Rubber hoses still exist. I think GP was trying to correctly point out that having a single point of failure is a weakness in any system, cryptographic or otherwise.
But Moxie's point in the blog post that is referenced is that with federated systems you have a lot of failure points. You can't keep everyone on the latest update in a federated system. You need all those actors to function perfectly, most of whom aren't going to be experts. This makes federated systems not great for anyone but experts.
> federated systems [..] have a lot of failure points
They sure do, but you also have access to all the parts needed to maintain your own instance and so does everyone else. Failures, when they happen, are limited to instance who aren't "functioning perfectly" instead of extending to every user of a centralized system.
Just to keep everyone reading this thread on the same page: The point of failure being discussed here is the human factor. See comment ancestors for context.
> Just to keep everyone reading this thread on the same page: The point of failure being discussed here is the human factor.
Thanks for reiterating this, because I think it is a key thing people who are scanning this thread are missing.
I think this is actually the argument for a centralized system though. At least for something that's a replacement to text messaging. I think the issue with a federated system is that most people don't know how to run a server, even pretty technical people. So this makes sense as a slack replacement, but not text messaging. At least to me.