No strong opinions but:

Denial of service attacks can become 5x cheaper.

If you don't thoroughly test https://github.com/bitcoin/bitcoin/issues/16499 these changes you can end up with bugs that can cause issues on p2p network.

On Fri, Aug 21, 2020 at 4:00 AM Dan Bryant via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:
It's been 5 years since minrealytxfee was lowered.  At the time
bitcoin was trading for $255 and it was agreed that the fee of 5000
sat/vkB was too high.  It was lowered to 1000 sat/vkB.  In regards to
how much anti-DoS protection that provided, it comes out to $0.00255 /
vkB in USD terms.  To have parity with the last reduction, we would
need to reduce minrealytxfee to 22 sat/vKB, though an even more
conservative reduction to 100 or 50 sat/vKB would be welcome.

With the growing adoption of LN, there is a need for ultra-low-fee
on-chain TXNs.  Having these queue and confirm overnight, or even
waiting until the Sunday lull would still probably be welcome to many
users.  The fact that the mempool is going empty at least every week
indicates that miners have not reached the floor of what they are
willing to mine.

About 2 years ago there was a PR (#13922) to try to make a reduction
from 1000 to 200 sat/vkB.  It was widely accepted but the submitter
eventually closed it in favor of PR #13990.

If minrelaytxfee is already parameterized and configurable in
bitcoin.conf, how could it be detrimental to operation of a node to
change the default?

References:

* https://github.com/bitcoin/bitcoin/commit/9e93640be6c49fa1505ba5c5df8c89210da5a6e4
* https://github.com/bitcoin/bitcoin/pull/13922
* https://github.com/bitcoin/bitcoin/pull/13990
_______________________________________________
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev