Yeah the exponential diminishing of the block reward means that FPPS becomes an unviable reward sharing scheme (even within 8-12 years people are thinking). FPPS is great when block rewards are large, but if volatile tx fees become a sizeable portion of miner/pool revenue then PPLNS becomes the best pool payout system.

FPPS works because the pool can anticipate future block rewards and average out the reward per share, but when future rewards become unpredictable then FPPS becomes unpredictable.

PPLNS is great because it maximally reduces variance while maintaining a no-deficit policy for the pool: miners only get rewarded when the pool finds a block. Pools don’t need to anticipate rewards, they just share the reward with the miner after they receive the reward.

FPPS is the major liquidity driver of mining pool centralization as all miners and pools want to reduce the variance of their hashrate with consistent guaranteed payouts. When the payout per share becomes volatile, then all miners and pools gain exposure to variance risk, and smaller pools can become more tenable for larger miners. 

Until then, everyone will be drinking the FPPS kool aid until the pools can no longer maintain those payouts.