I think the refusal of service is less problematic than the attack where they keep mining empty blocks without including any transactions. The refusal of service will only extend the block time which will be resolved in the next difficulty re-adjustment. It also requires significant percentage of the miners to agree to co-operate.
With the empty blocks attack, they prevent difficulty re-adjustment and also get rewarded with new btc unlike the refusal of service where they'll just be wasting their electricity without any rewards and the small miners will be able to produce blocks albeit in a much longer time than ~10 minutes.
> The refusal of service will only extend the block time which will be resolved in the next difficulty re-adjustment. It also requires significant percentage of the miners to agree to co-operate.
That's the premise: If you want to do something that strongly goes against the interest of all miners, that cooperation will form naturally. If 99% of miners agree on something, the block time would be several hours. Difficulty adjustment would have to be patched in.
In the meantime, the miners on the "rogue chain" are mining blocks and clearing transactions. Who says that this chain is not Bitcoin? Why should all the stakeholders consider a broken chain with 1% of the hash power as the "one true Bitcoin", as opposed to a failed fork?
With the empty blocks attack, they prevent difficulty re-adjustment and also get rewarded with new btc unlike the refusal of service where they'll just be wasting their electricity without any rewards and the small miners will be able to produce blocks albeit in a much longer time than ~10 minutes.