> The simplest case is a single tainted block, after which other miners could roll back by consensus. Invalid blocks are rejected trivially by any node as soon

> The simplest case is a single tainted block, after which other miners could roll back by consensus. Invalid blocks are rejected trivially by any node as soon as it sees them. You can't "force" an invalid block to be accepted by throwing hashrate at it and miners don't need to do anything to make an invalid block be recognized as such. The issue is on malicious actors trying to rewrite/replace recent blocks (to double spend, usually) which was the core innovation, but that also includes trying to censor txs or generally DoS the network by refusing to allow txs to be accepted.