Another question is whether the average Joe thinks he's an average Joe -- given how self-serving biases work, I doubt it -- and if not, why would he vote for more power for someone else?
The average Joe knows that he is not a millionaire, he is not a movie star, and he did not get Nobel price. His biases will probably make him believe that he is between 60th and 70th percentile. He can still vote for less power for the top 10%, or more popularly 1%.
The rest of your questions... I don't really know. Right at this moment it occurred to me that perhaps the feeling of power is more important that the power itself; most people don't notice the difference. Saying "senator Sam will reduce crime and give you free healthcare if you vote for him, Joe" makes Joe feel powerful. Saying "milionaire Mark made his money legally, you can't take his money away and use it as you want, Joe" does not make Joe feel powerful. Even if in reality senator Sam has more money than millionaire Mark, and senator Sam makes some rules that reduce Joe's freedom, while Mark only provides cheap shiny toys for everyone. Generally, if Joe feels that he can influence the state (even if that influence is mostly illusory), a more powerful state will make Joe feel more powerful. But I don't know how to measure this feeling precisely.
The tale of Senator Sam and Millionaire Mark is written as if the average Joe trusts more the former and less the latter (please correct me if it wasn't your intention to make it sound that way). This is still somewhat contrary to my experience, in which people generally distrust both millionaires and senators, hard to tell whom of them more. (This is more of a side note than an objection.)
Anyway, the hypothesis that people support democracy because voting makes them feel powerful even if it doesn't make any difference is plausible (althought it is certain...
I've seen several people on Less Wrong recommend Mencius Moldbug's writings, and I've been curious about how he became so popular here. He's certainly an interesting thinker, but he's rather obscure and doesn't have any obvious connection to Less Wrong, so I'm wondering where this overlap in readership came from.
[EDIT by E.Y.: The answer is that he's not popular here. The 2012 LW annual survey showed 2.5% (30 of 1195 responses) identified as 'reactionary' or 'Moldbuggian'. To the extent this is greater than population average, it seems sufficiently explained by Moldbug having commented on the early Overcoming Bias econblog before LW forked from it, bringing with some of his own pre-existing audience. I cannot remember running across anyone talking about Moldbug on LW, at all, besides this post, in the last year or so. Since this page has now risen to the first page of Google results for Mencius Moldbug due to LW's high pagerank, and on at least one occasion sloppy / agenda-promoting journalists such as Klint Finley have found it convenient to pretend to an alternate reality (where Moldbug is popular on LW and Hacker News due to speaking out for angry entitled Silicon Valley elites, or something), a correction in the post seems deserved. See also the Anti-Reactionary FAQ by Scott Alexander (aka Yvain, LW's second-highest-karma user). --EY]