Manifold is mulling a new product, perhaps hosted at manifold.love, which takes aim at the dating market.
The site would feature one prediction market per dating profile that asks: "Who will I date for 6 months?"
In a first, we would leverage crowdsourced betting to surface the top matches for each person, based on real predictions of how long their relationship would last.
Read my full writeup here! Excerpts are below.
Super interested in what the Manifold community thinks about this proposal. Let's iterate on the design to make it as strong as possible.
At Manifest, Robin Hanson urged us to find a high-value use case for prediction markets. He suggested markets on which candidate to hire. But markets on who should date might be even higher value (and more fun)!
Thesis
Matching people into relationships is a killer use case for prediction markets because:
It is fun for traders, which means people will be motivated to bet and thus create accurate forecasts
Accurate forecasts of relationships are super valuable
Betting on love is fun
Aella’s and Shelvacu’s markets got a giant amount of traction (the biggest two at Manifest!)
Proof School’s original popular markets were betting on relationships continuing
Austin’s OG markets on getting a girlfriend and marrying Rachel were some of the site’s most popular
People can’t look away. They even will bet on people’s relationships when they don’t know them personally. It’s too fun.
Prediction markets are an ideal solution to matchmaking
Prediction markets allow crowdsourced betting on who should date who with feedback from actual outcomes. Those bets will produce accurate forecasts, going on Manifold’s strong track record.
Matchmaking is a forecasting problem, so let’s use the most powerful forecasting tool!
How it might work
Users opt-in to create dating profiles
One main prediction market per person:
“Who will I date for 6 months?”
Free response market constrained to dating profiles on Manifold.
Resolves to next such person you date for 6 months
Subsidized liquidity (M2000?)
Created when you first fill out your profile.
Custom UI for browsing and betting on dating profiles
A table view with one row per person, standard info in columns, link to full profile (or opens in sidebar on desktop)
Select two people and bet on their relationship success
For each person, show the market’s top three matches (and let you bet on them directly from the table view)
Filtering, sorting, pinning potential matches via checkbox, leaving notes
Principles
Matchmaking is betting, and everyone can play matchmaker
Empower matchmakers
Give them the UI and tools to be effective
Public dating profiles with high information drive better crowdsourced decisions
Kickstart a flourishing ecosystem of fun secondary dating markets
The hope is that by doing the main matchmaking through markets, people will be more willing to create lots and lots of secondary markets on aspects of their dating lives. And they will be hilarious and super engaging
E.g. Will we kiss tonight? Where will we go for our date? Will our relationship last two weeks? Will he propose this year? How many children will we have by 2030?
Privacy where possible (Not 100% on this yet)
Dating profiles won’t by default be linkable and shareable. You can only find them in the table with everyone else
Allows people to feel freer to include more about themselves
Dating profiles will not by default be linked to your Manifold account. Can use a pseudonym
Encourage evolving community sophistication at matchmaking
Matchmakers will learn what questions to ask daters (comment section on dating profile)
Daters will fill in more information over time based on what is helpful
With DM’s (private 1:1 chat), traders can ask about character flaws too sensitive to broadcast publicly
Resolution criteria
To resolve YES, manifold.love needs 1000 daily active users on average over 7 days.
An active user must perform at least one of the following:
- Profile create and edits
- Bets (from people without profiles too, but has to be on manfold.love domain)
- Comments / endorsements
- Messages
I reserve the right to modify this / add more actions that I feel are primary.
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ24,773 | |
| 2 | Ṁ17,008 | |
| 3 | Ṁ15,581 | |
| 4 | Ṁ10,408 | |
| 5 | Ṁ8,073 |
People are also trading
I'm not surprised the outcome here. It takes some work to create a profile and the potential users need to be convinced that there's people here worth creating profiles for.
Short of hitting some kind of viral exposure, via a meme or a wildly popular video/article, it wasn't going to happen that fast. For some real world contrast I checked up on a different dating project that started 5 years ago and only recently hit 300,000 registered users... advertising yourself as open source and privacy focused doesn't exactly draw in users.
Manifold love definitely has better keywords than that to draw potential users in with, though I could see a potentially large amount turned away if the advertising makes people think they have to play a stockmarket minigame to participate.
The end of an era. It's Valentine's Day and Manifold Love is now a few months old. But it did not hit the 1000 daily active users goal (which was a bit ambitious I'll admit).
However, the trajectory for the app has actually been very good. We've been growing almost exponentially, and the features have become more solid over time, including compatibility questions like OkCupid, liking, and shipping.
And! We've just launched the THIRD major iteration of prediction markets with the aim to make them as useful as possible at informing you of who you should date.
Each market costs $100 USD which all goes into a subsidy so that there will be plenty of mana to win by playing matchmaker.
Go check it out! https://manifold.love/markets
@JamesGrugett You should use data on second/third dates to train a proper ml model that predicts chances of a successful relationship, I bet it will be better than the human predictors pretty quickly - manifold.love has all the advantages that make AI models better at forecasting and none of the human advantages
@colorednoise Don't you need a large amount of data to train such a model? If someone does figure this out, it's also possible to create a trading bot to bet appropriately on matches!
@JamesGrugett you are charging $100 for a chance to be matched with 49 people? Yeah, that will sure encourage adoption
I don't understand the cost here, nor the guarantee.
I understand Tinder costs a crapload, and Bumble is shit too. And than all the others match group owns.
I have literally been on the site for 4 months and never got matched.
Free Tinder ~100 Matches/day (Its tinder)
Free Bumble ~19 Matches/day (Mid quality)
Free Facebook Dating ~12 Matches/day (Very high quality matches)
So what if Creators started charging Manifold to post their Markets, is that kind of the same thing?
@SirCryptomind If you are regularly getting matches on those other sites, you might not be the target market.
Signed, 0 matches [with non-scammer non-spambot humans] per year on all the platforms you listed.
@JamesGrugett you can train a model on social networks (scan photos and determine how long people are together). Though, they can object to such usage of information...
@CodeandSolder Manifold Love is free to use for liking and messaging. Only prediction markets cost money now, but you get it back if you don't go on a 3rd date, so it's like a deposit.
@colorednoise @AnT I just read an interesting piece where someone tried to create an AI model predicting if matches after running some speed dating events: https://dkras.substack.com/p/dating-experiments
MdL in Time today:
https://time.com/6694798/manifold-love-dating-markets/
Anyone wanna bet it'll be enough to drive 7k new users in the next 10 hours?
@Sparr It's worse than I thought. 6 signups in the last 24h, which if anything is below trend. Any effect of the Time article is in the noise and not visible at all.
@MartinRandall yes but it's an average over seven days, so 1000 BoL attendees wouldn't be enought in its own right.
@MartinRandall Only if they do the relevant tasks to show on the graph, I think. James linked this graph for the relevant metric in another comment https://manifold.love/stats
@mkalbert What we need is a three-man resolution council to validate the authenticity of all 1000 profiles. Jack, Conflux, Martin... you around?
This is closer than you might think! We're at 150 daily active users today. Less than an order of magnitude to go!!