Here is another way to realize how ridiculous @Peter R's (Bitcoin Unlimited's) thesis is.

Let's assume every miner had the same hashrate and the same propagation delay, i.e. perfect equality, and so the equilibrium block size was established at rate at which the waste costs of orphans for the entire system was balanced with what the market was willing to pay for transaction fees. But every miner would have an incentive to have a higher share of the hashrate and a faster than average propagation delay, because those who did would win a disproportionate share of the rewards (if you don't understand why then you need to study the original selfish mining paper and the followup on optimal mining strategies).

So the natural market outcome is that miners would sign their blocks with a public key representing their reputation to not send invalid blocks. So then miners (pools) would being to trust each other and send only the signed block headers so they can begin mining on the new blocks as fast as possible with only constant factors of propagation (independent of block size). Miners who didn't participate would be less profitable and lose share of hashrate over time. So then you end up with no free market limit on block size, i.e. no fee market.

Q.E.D.