Nodes don’t reject blocks with transactions that have a massive amount of arbitrary data. They are still technically valid. Nodes simply filter these transactions out of their mempools and prevent them from being relayed to other nodes and miners. Spammers can circumvent these filters by submitting their spam directly to miners and paying them a high enough fee. Check Marathon’s Slipstream product.
If nodes really wanted to assert their authority, they would have to start rejecting blocks with too much arbitrary data. This would be extremely difficult as there would be disagreements about what counts as arbitrary data.
