that's not the pt.
the full nodes should and will care only about what tx volume they have the capability or desire to validate. for me, despite not getting paid, i will want to max out those validation capabilities to whatever i'm willing to pay (donate) to the network, which is much higher than it is being used today. i want Bitcoin to grow in size and price so i'm going to donate to my hearts content trying to make that happen. the Cripplecoiners will say that is no way to run a full node but i'll bet there are thousands of guys like me who are willing to do this for the prospect of taking the price to the next 10x level. there is nothing wrong with wanting a higher price b/c a much higher price is essential to Bitcoin's overall success and ability to move large scale tx's as we've argued about before. it has to be able to allow large tx's in the $millions to fund purchases of real estate, yachts, planes or bigger items so as not to perturb the overall exchange price, unlike what we have today.
as more users come onboard, merchants will have the incentive, not to mention fiduciary responsibility to run full nodes. that's a good thing.
but if someone designs a full node fee mkt, i won't be complaining either.
the optimal solution is looking more and more like 101 to me.
I think the sybil attack is a good reason keep nodes from forking to the wrong block size. I'm not sure a sybil attack is the way to look at it, with this new proposal. What is there to sybil attack? Every node operator independently selects the max block size that
he is willing to validate. The
effective block size limit is equal to the largest block that has successfully been included in the longest proof-of-work chain. Any miner can attempt to increase the limit, simply by publishing a slightly larger block than has ever been published before. They can assess the probability that their block will be mined upon (rather than orphaned) by any number of methods. If they try to publish too large a block, then it will likely be orphaned.
As sickpig suggested, it would be a recognition that the block size limit is not part of the consensus layer, but rather part of the transport layer.