The "cost to rollback blocks" is a cost in computational power (to run a 51% attack), _not_ the cost of the block rewards

In theory a higher block reward will incentivize and allow for greater computational power to be thrown at it, thereby increasing the cost to attack it.

The only way to measure that is by looking at expense over longer timescales.

https://www.f2pool.com/coins shows Bitcoin on top in daily dollar issuance. Ethereum was on top a few months ago, but not for over a year.

That appears to not include fees.

https://bitinfocharts.com/

Ethereum ($38,451,803.7)

Bitcoin ($34,875,489.84)

Before EIP 1559 fees were much more important. Ethereum has higher rewards since about Q3 2020

edit: even worse, f2pool just multiplies the base reward (2 eth) - ignoring uncle rewards and fees, by the wrong number of blocks (based on 15 sec block time - it's actually 13.2s). Trash calculation.

Thanks for the correction. Is there a site providing historic graphs of daily miner revenue?