I tested a filtered Bitcoin node with mempool against the public mempool.space instance
Both fee estimates were within tolerances and either both confirmed or both didn’t on the next block
Discussion
So non-issue for force closures?
Yes. It must also be understood that mempool consistency is not guaranteed and nodes must be able to tolerate drifts in fee estimates
There’s also the fact that LND has its own proprietary fee estimator, and therefore had a lot of force closures during fee volatility periods with CLN/Eclair nodes
So this would have to be fixed even with uniform mempools