One thing I don't understand: Would it even be stealing or just declare coins which anyway are dust "dead"? Stampchain outputs for example seem to try to match the dust limit as close as possible. And thus moving them would cost almost the same money than they're worth.

So if the owner of these dust outputs doesn't speculate on a future where the transaction fees are 10 times or more lower than now so he can move the coins retaining most of their value, the "BIP" achieves almost nothing regarding incentives mitigating spam. And that with massive overhead (regularly having to run buggy external (!) tools to see what we can confiscate, etc.).

There was the 1 sat/vByte to 0.1/vByte move of the minimum standard relay fee recently, so for old Stampchain stuff it could have made a difference if these "old dust" outputs were confiscated. But everybody would move them probably before the softfork gets activated.

In other words, the proposal fails to its own goal - can we just bury it?  

(what can be done instead is to implement tools to "clean" the UTXO set from unspendable outputs, like I wrote here.)

If somebody here knows: Has Core planned something like this? Or are there already tools to "clean" unspendable UTXOs? (Here I don't refer to Utreexo but to tools for the "good old standard UTXO set database", I know there are Utreexo-style tools and clients but for Core it's still in the early BIP stage.)

Anyway I feel all these proposals are completely missing the main reason for "spam": relatively small token/memecoin transactions with BRC-20, SRC-20 or Runes protocol. Which are standard even by pre-Core-30 policy. The JPEGs and audios everybody's talking always were only a tiny fraction of the Ordinals, Stampchain and OP_RETURN transactions and never took really significant block space.

This dust incentive would be even less effective for these tokens probably, because as long as you don't have a lot of tokens in your wallet, consolidation techniques would not save you lots of fees.