Sorry about that. That was dumb of me. I realized it later but was past the edit limit.
I think that would require the >50% attacker to be >50% forever. But let me concede it may have a problem. I want to turn it around and say “a proper FTL enforces a PTL on every block”. A miner who refuses to mine on top of a timestamp 7200 into the future of his own timestamp is just enforcing the FTL. If all nodes not only enforce the FTL but require every miner to have also enforced it, it would indirectly implement a PTL. Not enforcing it like this means nodes are knowingly allowing a dishonest block.
I called the sweeping consensus changes “better” only in a theoretical sense and said they were the most dangerous (due to being sweeping changes) and thereby hardest to implement. I wanted to discuss the tangents because that’s what’s interesting to me. Thinking about it is why I suspected there was a hack. It’s similar to wanting a PTL on every block. I wasn’t able give a reason to push it except it’s “prettier”. It was annoying that I couldn’t find a justification. Seeing a “properly-enforced FTL” = PTL is like pulling out a thorn. I’ve always felt there was something unsatisfyingly “open-ended” about the way miners enforce the FTL. “His timestamp is wrong” isn’t as satisfying as “His timestamp won’t let me be honest.”
All future nodes would know someone had been dishonest if any ancestor timestamps were > 2 hr in the future of any descendants, which would be a stricter and more logical form of the FTL that could remove the need for the MTP. But I’m not saying that as a recommendation.