I agree that we need to be very very cautious about changing anything with regard to the incentive structure.

That being said, it does feel like current incentives are not as aligned as they could be. I have not yet read the specific post that you linked on bitcoin-dev, but here is one of my favorite (soft-fork) ideas which might better align incentives:

example soft-fork w/ burning: require all transactions over a certain size (including witness) to include an op_return output which burns a certain number of sats/byte

Burning sats is not a very popular idea. If need be, we could probably replace burning with a suitably long timelocked anyone_can_spend output, something like 10 or 15 years would be my preference. Then it is almost like burning, but the sats still are available to be (re)mined in the future.

This mechanism still has some vulnerabilities (for example, people would probably just chop their larger transactions into smaller transactions to get under the limit, so more research is certainly necessary), but the idea of burning sats is at least interesting to me because it fairly (proportionally and instantly) rewards all remaining hodlers -- and the node operators we care most about and want to protect are the hodlers. These are the people who are in bitcoin for the long run.

At least then when your node is verifying all these large transactions, every time you get to the op_return output, you get a little bit richer and, by definition, the creator of that offending transaction gets a little bit poorer.