Fortunately at least a few people got it: the point of these polls was to explore the tradeoff between "voting as preference satisfaction" and "voting as information gathering".

Thread:

If the purpose of a vote is to choose between proposals that satisfy differing end goals of different people, then even if the majority group has very similar opinions, going with the majority still serves the goal of making more people happy.

On the other hand, if the goal of a vote is to aggregate the information in different perspectives, then two perspectives agreeing on something should count more the more different they are. And two people copied 30 minutes ago are almost the same as one person.

There's gradations here: what if people were copied 5 years ago and lived separately since then? What if people are only partially correlated, because they come from the same backgrounds but do live separate lives? From the info-gathering perspective, individuality is a spectrum.

Finally, if one person will be copied *30 minutes in the future*, paradoxically enough from a preference satisfaction standpoint that should be almost the same as them being copied 30 minutes in the past. From an info-gathering standpoint too.

One poll I did not have space to make is: what if the 10% supporters are elves who will live 10000 years, the 90% are humans who will on average live 40 years, and the decision will have consequences for 10000 years?

I'd go with the elves, but the question is a very noisy one, because you could argue the elves are wiser because many must already be thousands of years old etc.

To end it's worth noting that the current version of quadratic funding that @gitcoin uses, https://ethresear.ch/t/pairwise-coordination-subsidies-a-new-quadratic-funding-design/5553, already has a mechanism that counts two participants' contributions for more if those participants are otherwise less correlated.

(read the post)