I've created 14 markets asking, "What is Manifold's favorite X?" that resolve based on a survey where participants can rank each work as "Excellent", "Good", "Okay", "Bad", or "Haven't seen/read/played/heard". The way I will choose Manifold's favorite is just by taking 2*(no. of "Excellent" rankings) + (no. of "Good" rankings) - (no. of "Bad" rankings). This means that a work that more people are familiar with has an advantage, since it's more likely to get more Excellent or Good rankings.

Although I think this is the most faithful interpretation of "favorite", it's also interesting to see which work Manifold judges as the best, that is, "What do the people who have actually seen/read/played/heard the work think of it?" To determine that, I could exclude all the "Haven't seen/read/played/heard" ratings, and take the average, rather than the sum, of the remaining ratings to see what the average person who is familiar with the work thinks. There is still an issue with this, though, which is that some works might have almost no one who is familiar with them, and then the average would be based on an extremely small sample size and therefore unreliable. For this market, I will only consider options that at least ten people gave a score (Excellent, Good, Okay, or Bad) to. I will resolve to the number of categories for which the favorite, based on this average, is different from the favorite based on overall score.

If there's a tie when using any method, I'll count it as a full change unless both methods result in the exact same options being tied (in other words, I'm not bothering with "partial agreement" like I did in the dwarf planet market). 

If I add any others, they won't count for this market. 

The survey is officially out! You can take it here: 