There must be some clever way to compress or summarize and validate the utxo set or like I said Moore’s law can continue and prior to it being a problem large amounts of RAM today will be trivially cheap, fast, and reliable? Or some combination of solutions?
https://zerosync.org/
Is this anywhere close to being production ready? How would a merchant use this?
I dunno. Haven’t looked into it much. @benthecarman do you know anything about the status of ZeroSync?
It can do header validation but not yet able to run it as a full node. Probably pretty far away from that
I think assumeutxo may be the closest thing I have seen to a mitigation, though I haven't seen much movement trying to implement it in a user-friendly way, and of course you still need to be able to do the sync eventually even if your node is usable ~immediately, so you can help bootstrap other nodes.
I wonder if any work has been done with utxo set for selectively deciding which to hold in memory vs disk lookups based on age.
Yes, I have wondered this as well. It would surprise me a lot if *no* work had been done.