RE: Should validators be able to reject transactions?

That, is interesting. Could you make my research on that easier by explaining a bit on why that happens?

Let's say 99 out of 100 block producers (miners) are actively censoring your transactions with a softfork on a POW network. That means 1 out of every 100 blocks is going to include the censored transaction and it will be included on the blockchain even though 99% of the network was against allowing it. That means that even if 99% of the network opposes what you are doing on Bitcoin you're still going to make it through after around 17 hours of waiting.

Put this context in the situation of a hacker who just stole millions of dollars.
Having to wait 17 hours to move the money somewhere else or syphon it through a privacy protocol like CoinJoin is a trivial affair. It's not going to matter to the hacker one bit that 99% of the network opposes their action.

On a DPOS or POS chain getting 99% consensus is the same as 100% consensus and they can rollback the chain and prevent the block from ever being minted. On Hive I think you need 16/20 witnesses to agree for the block to be finalized. 16/20 is only 80% which is how these networks are able to censor things. POW can't do this even with 99%.