This would be better than treating transactions differently, in my opinion. However, it has to be implemented similarly to how XMR does it. Raising the block size would result in a reduction in the block subsidy; otherwise, whatever the max cap is will always be activated by greedy miners. This would lead to larger pools absorbing all available well-paying transactions, leaving smaller transactions with almost nothing.

If we have three pools—A (45%), B (45%), and C (10%)—block C, with the current protocol, is almost always guaranteed to mine 10% of the transactions. So, if there are 100 transactions paying 0.1BTC each, and each block can only take 10 of them, it would, on average, make 1 BTC out of that. However, if pools A and B can increase the limit from 10 to 15, pool C (which only finds 1 in 10 blocks) will be left with nothing but dust. The current way relies solely on luck; a small miner may hit a 5 BTC fee block, whereas large miners who find twice as many blocks may not get 5 BTC combined.

In other words, as of now, potential rewards from fees are based more on "luck and market conditions" than your hashrate. The other model will shift that, making it more about your hashrate and leaving little room for luck.

However, looking at Suzanne5223's comment above this one, if you read what user Kiba wrote right below Satoshi's post, he said:

Reading this 13 years later, it certainly makes a whole lot of sense; it's pretty difficult to do now.

@philipma1957

I agree that leaving a small amount of BTC for emergency cases like this in somebody else's custody may not be a bad idea. My only issue is with the KYC part; I know many people are okay with KYC-ing themselves, but I try to avoid that as much as I possibly can.