I agree with you about some of this, but I just want to point out that the filters are what's causing bandwidth strain because nodes and other miners that aren't aware of certain transactions and have to download them to validate a recently mined block. Bandwidth usage is already limited by block size and blocks are almost always full anyway, so it's not like spam would actually increase the amount of data that needs to be shared every time a block is mined. Having everyone aware of all the transactions available to be mined would reduce latency and keep the network running smoother. As for the rest, I agree that spam is marginally detrimental to the monetary use case, but I also think it will be priced out pretty easily at higher fee rates. My current opinion is that I would like to see fees reach a higher baseline to preemptively price out spam before the filters are dropped, but I do think they should be dropped eventually.

Replies (2)

I've read that post and I know about the block reconstruction, but that doesn't address my argument about bandwidth. If everyone is running filters then nodes and miners will never see certain transactions to cache them.
There is no strain on bandwidth from missed transactions (even though that also isn’t happening anyway). The transaction is downloaded either way, either when broadcast, or when added to a block. But default compact block filtering actually keeps all transactions that are valid, even those being filtered, in expectation of them being in a block. Essentially this argument being paraded around has no example of it in reality. It’s just some theory being tossed around that kinda sounds good, imo. Transactions are downloaded by every node no matter what the relay policy is. The only question is when.