First it should be understood that Multi-PPS aspires to solve only one aspect of mining centralization - the centralization of many small miners in a few large pools. It doesn't solve the problem of huge mining entities that enjoy economies of scale - though, I don't think that problem is too great to begin with, as I claimed in here.
Under the assumption that the cause of mining pool centralization is the built-in performance advantage (in terms of variance and fees) of large pools (as opposed to psychological factors and features), any solution must be based on having each miner target multiple pools simultaneously. This way, there can be many small pools while the miners enjoy the performance of large pools.
Multi-PPS specifically is best suited for a future scenario when Bitcoin mining is dominated by transaction fees. However, that scenario has some inherent challenges that will need to be solved in the context of any reward framework, Multi-PPS included.
Multi-PPS is an extension of classical PPS, in the same way that mining in multiple pools is an extension of classical PPLNS. Multi-PPS allows PPS with low fees, and mining in multiple pools allows PPLNS with low variance.
For either of these to work, mining pools need to become more of a commodity, so that the mining client can automatically connect to many pools to optimize the performance.