gmaxwell - do you think that many pools are reliant on patched bitcoind nodes to maximise their profits (since most a pps it is their profit not the miners that is impacted by orphans/empty block trade-off).

Those 20 pools are arguably the most critical users of bitcoin core. And should it not include every possible optimisation (even if at expense of SPV) to ensure they run vanilla code. As that then ensures that future soft/hard forks go smoothly as those 20 users just have to do an upgrade from version 2x ->2y - rather than having to try to merge in some old patch and try to test it works as expected.

Or is issue that those patches are proprietary and what gives them their competitive advantage.

I guess since the pools are operating off 2.5% margin on pps - it magnifies by a factor of 40 P&L impact of orphans.