Thanks for the link. But that only reduces the variance, it doesn't increase the expected gains from forging. Pools "work" (well, used to work) for Bitcoin and friends because expected gains (used to be) > costs even for small-timers, but the variance was extremely high. Nxt has low expected gains from forging. Poor shibes like me will likely have have expected gains of < 0 (after electricity and hardware costs), the only profitable ones may be ones that can do it in scale and have market advantages in the different parts of the cost equation, to keep costs down.

I share the concerns of centralization, but seems to me the way to deal with that is for leasers to be able to have some kind of influence on the forger's behavour (i.e. what txes they reject, scripts they don't run and how they blacklist/weigh other forgers). Maybe moving leasing power to someone else is sufficient, or maybe there needs to be some way for leaser's to inflict a penalty (but maybe with their own reputations on the line).