This is a divisive issue, as many things are. There is a problem that needs to be resolved. Pick-to-win from the reward pool is not ok or good for platform. Pay-to-win, well that's actually someone putting in something of their own for the potential to win. There is a real risk-reward to mitigate the involvement based on what people are willing to lose. With no risk, of course people want to jump in on it, no loss, you end up just getting more and more of it. It's motivated behavior. So make it real betting, yes.
Thanks for the feedback. I will be posting more on this, as it is a real issue that needs to be resolved. Isn't that what you want? For people to come to a consensus for what to do? So this needs to be dealt with. People need to be aware so the problem can be resolved. I disagree that there is no objective answer. I will try to make my case in forthcoming posts.
I had "vote-buying" in my title originally, but that can be argued against since not everyone wins... but it's still vote-buying even if you're not guaranteed to win, you're guaranteed not to lose, which is itself the motivator to participate in a free lunch. All they see is short term gains, and upvote to win, which is buying people's votes. Even if they don't realize it, that is the underpinning psychological motivational factor.
Any given post isn't vote buying, but if outcome is perfectly random then over multiple rounds it is as if everyone was vote buying. Even without being perfectly random, a simple strategy of "voting with the herd" or creating two accounts with reputation and "voting both sides" will turn it into vote buying.
I think the real innovation would be to have a different algorithm for allocating the "community rewards" to these games. I would base it on the total amount of private funds placed into which ever side has the least bet.
That's something I have thought of and have some things written, on allocation of rewards for certain content types. For example, art, photos, fiction, non-fiction, etc.? Since you're considering it for "pick games" for the reward pool, then it has a measure to apply and a rule to code to apply a measurement, meaning it would have an objective valuation. The same could be done for others content types, where they each had their own reward pool? This would make "over fishing", shall we say, less of an issue, but problematic to evaluate how to divide the rewards per content type. Thanks.