> 1. But this is making assumptions on why other people run nodes, right?

Yes. The specific assumption is: some people run a node with a mempool to help relay some other transactions on the network, but do not want to extend that help to spam transactions.

> I don't think you can honestly argue that 1MB of OP_RETURNS is worse for the performance of your node than 1MB (or more) of data in any other part of a block?

If the data was identical, and if I was excluded from using the argument about base space being more valuable than witness space (to be considered in the next paragraph), then I would not object to the statement that 1MB of OP_RETURNS do no further harm to a node's performance than the same data in another part of a block.

> Do you think spam will persistently outbid monetary transactions for block space?

I do not think so because I do not think I am a reliable predictor of the future. But if I thought I could reliably predict that nothing will change, then I would also predict that spam will often outbid monetary transactions in the future, because I think they often do so in the present.