Sure, individual situations vary but you can average across small stakeholders and conclude that they lose out as a group. Unless of course the group is self-selected to include only those with the necessary resources, but this is still going to be wealth-concentrating in general.
Another factor which others were trying to convey is the risk-adjusted return per stake is lower for smaller stake even if pure expected value per stake is the same.
resource requirements are pretty low for most PoS systems
I think this basically assumes the system is a failure, no (or at least, if successful only in a very niche way)? For something with "high" usage (say as high as current Ethereum or potentially much higher), the resource requirements will not be low.
Yes, these are fair points.
I think I'm trying to explain that these centralization considerations for PoS, are a bit different than the considerations for elected positions (and the potential gamification of those positions) for DPoS.
In PoS, you could at least expend system-external resources to protect your stake, but in a monopolized DPoS this could become impossible.
Yes I agree, although 51%-type attack scenarios still apply to PoS where enough stake (either with one owner or a cartel) could censor minority stakeholders or somewhat less (?) maliciously charge them to not censor, which amounts to very similar form of enforced monopolization where even external resources can't help.
In such "attack scenarios", those participants would need to collude to create what is effectively a byzantine fault.
In this DPoS case, the participants need not collude or "attack" at all, they can each adhere honestly to the rules of the system and yet still collectively monopolize it. Of course, if they did collude, they could perform "attacks" like that as well.
Just trying to illustrate the differences here.
Sure, I agree with that too. But the end result may be the same, and it is a very mild sort of 'fault'. From the point of view of users who aren't trying to stake, everything would be fine (in fact possibly better since the dominant stakeholders may have better infrastructure and not have the entire network held back for the benefit of the smallest).