It makes no difference, except your node will make poor fee estimates because it has no idea what’s going on in other mempools.

Forwarding strings run by an overwhelming majority of the propagating network have a substantial difference on what gets into blocks and add a reasonable marginal cost. This is why like 99.9% of OP_RETURNs over the past decade have adhered to the network forwarding settings… because they work.

None of this suggests you can’t get around them. You can get around a fence, you can run a hash cash stamp and still send spam emails, you can get around all sorts of “filters” that are used at this layer of operation in any network or environment we could use as an example. Almost by their nature barriers are usually just a matter of cost.

But a small barrier that lets honest transactions through, and that would catch the majority of spam, allowing normal activity to deal with X in costs, while spam has to deal with X + 20% (or whatever) in time, alternative software, fees, whatever, then that’s a proper and meaningful balance to disincentivize the activity. And we have a very long period of bitcoin history to show this very thing, and that simply continuing to make small

Adjustments to ongoing spam evolution will continue to diminish its impact.