TL;DR bet on which word will be the most likely to be the first word in a valid haiku.
Background here:
The first word will be chosen from this market with a weighted random number based on their market weights at the time of market resolution.
Edit: The first word will be chosen from the top 20 scoring words in this market, with a weighted random number based on their market weights at the time of market resolution.
All markets will resolve when either 1) a valid haiku has been created or 2) the process has failed to produce a valid haiku, in which case the “failure” options in each market will pay out.
I will not trade in these markets. I reserve the right to modify the text of the markets to avoid unintended consequences/gaming, preserving as much as possible the original intent and mechanism.
1,000
3.00
@Joshua Actually yeah that would be another good way to rescue this horror! I think top 10 or top 20 should be ok.
Actually I have a pretty bad feeling about this entire experiment. The rare words not bet on are too numerous. Short of an insane amount of mana spent for word choice, it's likely that the haiku we end up with either falls apart partway through or is just aesthetically ugly/random. I've bet a substantial amount on "Haiku fails" accordingly
I think a better way to go about this would be to have a nomination phase for each word (rather than nominating words for the entire vocabulary; a bountied question would suffice), followed by a much smaller multiple-choice market to choose the next word. I think Manifold Plays Chess went quite successfully off this model.
As an example if we used the probabilities from this market we'd get atrocities like these:
Poop future cherry
Refrigerator. Putin
A sky water of
Moonlit bayesian code
From in away the to sky
Sadness we cherry
Interview cherryProfit moonlit the moonlit
Love from cherry treesPredictive who then
Of singularity brings
To if how how it
(For these I used some basic Python coding to make a list of random words chosen with the given probabilities, and then greedily picked the first item from the random choices that fit in the haiku structure. For all but the first I had to skip a word.)
EDIT: added a quoteblock for clarity (why is the Manifold text editor so counterintuitive?!)
@duck_master Fair point about the wide word choice. I expect that in markets for subsequent words, they will be much more concentrated on words that are a plausible next choice. Because this is the first word, the choices are much more dispersed. Let’s see how it goes. The “haiku fails” choice is there for a reason!
@Ansel You do make a good point, but it doesn't address the underlying issue, which is that the nomination phase and the choosing phase should be separated just because there are so many possible words to even consider.
@duck_master Can you explain a bit more about what you mean? The nomination and choosing of each word are separate markets currently
Never mind, I got you- separate vocab markets for each word. It might work. Anyway there’s a lot of bounty invested in the current setup so I’d rather keep it going as designed. I might put a cap on the number of words going forward and choose a random selection from the vocabulary
@Ansel I'd mean that first you'd create a "nomination" bountied question for the next word, where users submit potential choices for the next word only (similar to what you did for the haiku's vocabulary). Then, you create "choosing" multiple choice market to choose among the choices submitted on the bountied question. This reduces the risk of having lots of words from the long tail of inappropriate words that would cost too much mana to bet down.
