Gonna evict whatever N months of mining it took to recognize a problem?  If that's possible might as well give up on Bitcoin.


That would reduce reduce the attackers ability to embed data by one bit per 32 bytes.

You cannot non-interactively combine other people's ECC public keys.  For sorting, that could reduce the capacity by log2(n) bits per output (e.g. the cost that is require to encode a hidden counter in the encoded data.


It's also important to keep in mind what problem you're solving.   Lets imagine that it were possible to keep the data out of witnesses (it's not but lets imagine)--  the embedder now has to pay more.  But did they particularly care how much they were paying?  No, or they wouldn't use Bitcoin at all.  Most of the time the users are upset about the embedding they're upset because of the impact they have on fees-- but excluded from the witness they have 4x the impact on fees! Maybe their traffic levels go down a bit but they have to go down 4x just to be back where they were in terms of fee impact.  I think because people fall into a combative mindset of viewing them as an enemy they start thinking in terms of how much they hurt their opponent, but hurting the opponent isn't the same as helping everyone else.   Worse, evicted from the witnesses the cost difference for the spammer from using prunable op_return vs some unprunable output embedding is small, at the latter has less further blocking potential.  So you go from spam but at least it's prunable, to spam with 4x the fees impact and maybe it's not even prunable.  Maybe the total bytes sent are somewhat less, but was that ever anyone's issue?

It's like one could also "solve" spam by reducing the block capacity a lot-- if it's reduced to zero no spam at all.  But even if it is just reduced a lot then obviously the amount of spam is reduced too.   But that just seems vindictive to me.  The thing that got most users up in arms is that spam increases fees and cutting capacity is like a giving the spammer that reduced space for free, forever, in terms of increased fees.

So it would reduce the amount of illegal content they could put in, but I think mostly only a few developers have ever really been concerned there, and that attack doesn't need large amounts in total in any case.  So it's not like bitcoin with 150vkb blocks as luke-jr argues for would be immune to a very wealthy attacker that wants to put some unlawful stuff in the chain to make node operators worry they'll get in trouble.  But in any case as I say I don't think that concern is what has animated people.

And then we get to the argument that at least some of the embedders are potentially just trying to disrupt bitcoin, that they're actually malicious.  Well then countermeasures would be even less effective for them, if the want to spent lots of money to drive up fees they can also do so with txn that no one would say are not legitimate finical transactions.   And to their extent that their goal is stuff like creating drama or getting the public to abuse long standing developers that they can't buy off-- well they're winning on that respect right now, aren't they?