Although personally i'd rather avoid dynamic block size altogether since there's concern it could be manipulated by miner who decide which and how much transaction included on their mined block.
Not only that. There is an attack, that could lead to fast verification for the attackers, and extremely slow verification for everyone else.
1. You pick some random, and secure seed.
2. You generate a lot of addresses (deterministically, from your HD wallet).
3. You generate deterministic transactions (yes, it is possible).
4. You include a lot of self-transfer, only to increase block size, according to consensus rules.
And then, you know, that all of your self-transfers are valid. More than that: you can recreate them on the fly, just by knowing the master public key. Which means, you, and your network, can easily validate a lot of your own self-transfers, and mark them automatically as valid. While the rest of the network has to check that, or risk mining an invalid block.
Then, your strategy is simple: always fill the whole block with your dummy transactions, no matter what.
1. It doesn't affect miners, but only competing mining pools (and their block verification time) or other full nodes.
2. It doesn't affect your network, because by knowing master public key, your nodes can skip those transactions as valid.
3. It forces high fees all the time, because even if blocks will be full, then your self-transfer will always fill the remaining space, and make some fee market.
4. It centralizes the network, because if blocks are bigger and bigger, then only your nodes can know upfront, which transactions are fake, and can be skipped as valid. Everyone else waste time on your dummy transaction verification.