I'm against a dynamic block size because it fails to take into account the possibility of the miners intentionally colluding and manipulating the fee market, by either introducing scarcity or otherwise intentionally gaming the block size. Problem with dynamic block size is that it is difficult to accurately provision higher block size for the correct period, due to time lag mainly as the sample size is way too huge.
I'd like to remind you that any miner with a lot of hashrate can inflate the fees by choosing to not pick the transactions that pay a low fee. It is actually easier for a miner to do that with an inelasic supply, if the supply was elastic then inflating the fees would also increase the block size limit which would ultimately reduce the fees (it would then be pointless to inflate the fees in the first place).
I would rather just proposing a predictable block size increase, than to have a dynamic block size because it is honestly quite redundant. I don't expect Bitcoin to survive solely on on-chain transactions, that isn't feasible.
I see but what factor would you take into account to calculate block size limit? I believe scaling can only be done by increasing the network's throughput, I don't think the LN is a viable solution but that's another topic...