I want to promote the Motley Bot Challenge, a challenge I set up to create bots that accurately forecast diverse questions on Manifold. Here's the main market with the full details of the challenge: https://manifold.markets/CDBiddulph/will-there-be-a-manifold-bot-that-m
I'm not necessarily interested in increasing the number of traders; I specifically want to attract people who will build high-scoring bots for this challenge. In other words, I want to find interventions that will increase the odds of good bots being built. If traders in my main market know about this intervention and trade rationally, a good intervention should correspond with an increase in the main market's probability.
Anyone can suggest their own interventions by adding them to the options above. Although only one intervention will be used to resolve this market, I will send Ṁ50 to anyone whose intervention I use!
Resolution
After this market closes in three weeks (February 11), I will choose a single intervention from the options above, according to some mysterious criteria. I will do that intervention as written, announce it in the comments of the main market, and send the person who suggested it their Ṁ50 reward. After that, I will not do any other unexpected interventions on the market for two weeks.
Two weeks after the intervention, I will figure out how many percentage points the main market has moved since then. This might straightforwardly be the difference in probability between "now" and "two weeks ago," but if I see any suspicious spikes at the end of the two-week period, I will use the probability before the spikes.
I will resolve the intervention I used to 10x the positive change in the main market's probability. For instance, if the market moved from 66% to 74% two weeks after I did intervention X, option X would resolve to 10*(74-66) = 80%. All other interventions will resolve to N/A.
Post-resolution
If I decide to use your intervention, but I use it after the single intervention I used to resolve the market, I might modify your suggestion - for instance, if you suggest spending Ṁ1,000 on something, I might spend Ṁ500 instead. Even in this case, I will still send you a Ṁ50 reward (as mentioned above) for the inspiration.
Valid interventions
To be eligible, an intervention must cost me no more than $10. It shouldn't involve falsely manipulating the main market - I want to actually increase the expected score of bots in this challenge, not just the nominal probability of the market.
As explained below, all of my suggested interventions involve "sweetening the deal" for bot creators by providing opportunities to claim mana if they make a high-quality bot. However, I'd also be interested in other categories of interventions I could do:
Promoting the challenge outside of Manifold
Creating resources to help prospective bot creators
Anything else you think could help!
Explanation of my interventions
The interventions I started the market out with all involve creating a mana incentive for potential bot creators.
My thought is that the main market is at a high enough probability (66%) that making a really good bot and betting that market up wouldn't be incredibly profitable. If there were a highly-liquid market sitting at a low probability (like 10%) which would resolve to a high probability if someone made a high-scoring bot, this would provide a high potential upside for prospective bot creators.
Unless stated otherwise, any option that mentions "boosting" will use the default boost options - half of the mana going to ads and half to subsidies. I'd be interested to see if people think other ratios of ads to subsidies would do better - feel free to add your suggestions as options in this market. Note that the main market is currently saturated with ad boosts, so adding more wouldn't immediately do anything - I think the campaign had Ṁ4,000 of funds to start with on January 2, and it's still at Ṁ1,520 as of January 21.
One option mentions buying Ṁ500 of NO in a market. My idea here was that I would use an alt account to buy Ṁ500 of NO, artificially lowering the price of that market to incentivize people to bet against me if they think they think they or someone else can build a high-scoring bot. I won't do this if I find out that it's not allowed.
Explanation of "Spend a total of Ṁ1,000 to create and boost a market that resolves to the profit of the top-scoring bot minus 100"
This market would work similarly to the main market (quoted below):
I will resolve the market based on this maximum profit as follows:
Resolves NO if the profit is zero or negative.
Resolves to X% if the profit is ṀX for some X between 0 and 100.
Resolves YES if the profit is Ṁ100 or more.
The difference is that the market will resolve NO if the profit of the top-scoring bot is less than Ṁ100, YES if the profit is Ṁ200 or more, and (X - 100)% if the profit is ṀX for some X between 100 and 200.
For instance, if the top-scoring bot made Ṁ160, the market would resolve to 60%. Note that this corresponds to a 16% return relative to the maximum amount of mana a bot may invest.
@gpt_news_headlines By forcing the flat fee of 1 per market you're elminating the intelligence. This is not a binary problem and making it so means that the bots aren't thinking.
The main market was at 68%... and now it's at 61%. This market resolves to 0%. The market I boosted only got a total of 4 traders 🤦
I don't necessarily think the change in probability would have ended up positive anyway, but the main reason for the precipitous drop is @colorednoise selling their YES position and putting M500 into NO near the beginning of the 2-week period. Curious what your reasoning was here @colorednoise!
I boosted this market: https://manifold.markets/CDBiddulph/what-will-be-the-score-of-the-bot-t
We'll see if the main market is actually 3 percentage points higher in two weeks