Even if L2 txns incur no special costs to miners, no added complexity etc., still the miners are inevitably the entity to receive users’ txn fees and to enjoy surge in demand for L2 txns, b/c they are the (only) gatekeepers to the entire system, so the user must incentivize them – bribe, if you may – in order to be sequenced sooner rather than later. The more pressure and demand for L2 txns – the more impatient users are willing to pay to enter the sequencing in a timely manner – and if this willingness to pay does not transfer to the miners themselves in the form of greater revenue from these impatient txns then the miner would ignore these high priority txns, which destroys the QoS and practically leads to a DoS for L2.
Whether you need another L2-dictated form of payment or mechanism to satisfy the economics or sustainability of provers – I did not form a strong opinion on that, there are several options that come to mind. One could argue, as you
mentioned, that provers’ load is const and therefore they should require no extra compensation, others can argue that a healthy macro dynamic is one where service providers receive higher compensation in case of higher interest/demand in their services. L2’s can decide their own mechanism here, as long as they comprehend that fees for prioritization under demand must inherently flow to the miners.