I would propose dropping reward in half for each consecutive PoW blocks that are mined on top of each other instead of making such complex adjustment to protocol.
The discontinuity implied by any approach to count the number of consecutive blocks will most likely result in various forms of gaming the system. Generally, the block reward for pow is greater than pos. Linking reward to the sequence of blocks results in miners who have stake attempting to manipulate the sequence of blocks so that they maximize their reward, which is not what we want.
While such discontinuities are easy to imagine, their safe implementation leaves a lot to be desired. Linking in a check against the previous block header into the block reward is inherently more complicated than simply doing some additional algebra using the variables provided.
The simpler alternative I laid out in the RFC is to simply change the slope of the equation. This would be far simpler for faster target regulation than trying to ban or penalize consecutive blocks.
This has been entirely rewritten to act on nInterval instead of nAugmentSpacing.