No, I actually meant decreasing over time. The idea is to have decreasing rewards over time such that nodes will not delay including the forced transactions just to increase the rewards. With a decreasing scheme, the forced transaction may be included faster. The forced transactions have a deadline so one can define a decreasing reward function for this fix period of time
This makes sense if we think stakers have some incentive of not including a transaction. I’m a bit worried about pushing them to add these txs faster as validating them means seeing they are included in L1 which means having a node that is as synced as the proposer’s L1 node which is risky.
Maybe a clean solution (proposed by Adi from Informal Systems) for avoiding these timeouts would be to have a proof market
There’s nothing in the design that prevents a proof market, but I don’t want to enforce it. In particular a market means pricing. What if a proposer sees the proof they need is too expensive? if they end up proving themselves, then will the market be just for round > 0 proposers? What happens if they see a high price. We need to have rules that allow to progress regardless. But again, if such a market will exist, it can be used.
Another solution which avoid a proof market, is to incentivize proposers for rounds strictly greater than 0 to compute as a backup the necessary proof by giving them some coupon that they can use only once to increase their chance to become proposers for next heights
This sounds like a complication for something that should rarely happen (say 1% of blocks). Now the scheduling needs to handle coupon balances, refresh them (per epoch i guess), etc. And, this still doesn’t mean someone will create proof hoping round 0 will fail. It is a lot of effort that usually will be wasted.