Under no circumstances do I think we should *increase* the dust limit. That would have a mildly confiscatory effect on current Lightning Channel operators, among others.

Generally, the UTXO set will grow. We should work to accommodate the worst case scenario under current consensus rules. I think this points to using things like Utreexo or similar rather than meddling in the user's business.

I am skeptical that 0 value outputs are a real spam problem given the cost to create. Generally one creates an output when one either believes it would make sense to redeem it in the future. So surely this is a market problem, if people want them they can pay what it is worth for them to have it. Again, it's not my business.

Matt proposes that people might use a nominal amount of bitcoin on a zero value input so that it doesn't look like dust. What Matt is asking for is that in any protocol you pay for your space not via fees, but instead via an assurance bond that you will eventually redeem it and clean the state up. In my opinion, this is worse than just allowing a zero value input since then you might accrue the need for an additional change output to which the bond's collateral be returned.

With respect to the check in the mail analogy, cutting down trees for paper is bad for everyone and shipping things using fossil fuels contributes to climate change. Therefore it's a cost borne by society in some respects. Still, if someone else decides it's worth sending a remittance of whichever value, it is still not my business.

With respect to CT and using the range proofs to exclude dust, I'm aware that can be done (hence compromising allowed transfers). Again, I don't think it's quite our business what people do, but on a technical level, this would have the impact of shrinking the anonymity set so is also suspect to me.

---------------

If we really want to create incentives for state clean up, I think it's a decent design space to consider.

e.g., we could set up a bottle deposit program whereby miners contribute an amount of funds from fee revenue from creating N outputs to a "rolling utxo" (e.g., a coinbase utxo that gets spent each block op_true to op_true under some miner rules) and the rolling utxo can either disperse funds to the miner reward or soak up funds from the fees in order to encourage blocks which have a better ratio of inputs to outputs than the mean. Miners can then apply this rule in the mempool to prioritize transactions that help their block's ratio. This is all without directly interfering with the user's intent to create whatever outputs they want, it just provides a way of paying miners to clean up the public common.

Gas Token by Daian et al comes to mind, from Eth, w.r.t. many pitfalls arbing these state space freeing return curves, but it's worth thinking through nonetheless.