If there is a way to identify irrational spam txts (like this guy do: https://mobile.twitter.com/sbetamc/status/943433491097714688 ) there should be a way to make them illegal without censhorship the blockchain...or not?
Systematically prohibiting/blacklisting transactions is censorship, and that would be particularly problematic at the protocol level. Miners are free to reject transactions that appear to be spam. Priority calculations were historically used to identify spam, and now the fee market deters it.
The problem is that an attacker with significant hash power can do the following: significantly drop the hash rate after difficulty adjusts upwards, spam the blockchain at different fee levels, and collect the block rewards (including his own spam). Part of the genius of Bitmain setting up several different pools is that it's very difficult to definitively prove this activity is occurring.