Predicting for Good: Charity Prediction Markets
Austin notes: This was our application to the FTX Future Fund. Due to time constraints, we haven’t collected as much feedback as we would have liked; please let us know what you think about our proposal!
Your project
Does this proposal fall under one of the areas in our "Areas of Interest" page? If so, what area does it fall under? (It's fine if the answer is no.)
Epistemic Institutions
Please describe your project in under 100 words.
Prediction markets work best when people can wager something they value. But past attempts at real-money prediction markets have been shut down by the CFTC.
Manifold Markets is building a charity prediction market (CPM). People can place bets using real money, but are required to donate all proceeds to charity. Since the money never returns to users, CPMs allow real-money betting while being consistent with financial regulations.
FTX Future Fund can allow us experiment with CPMs by providing a matching pool for user donations. This will increase participation in prediction markets, provide a testbed for real-money prediction markets, and boost charitable donations.
If the project has a website, what’s the URL?
Please describe what you are doing very concretely—not just goals and long-term vision, but specifically what you are doing in the next few months.
Manifold Markets is launching its first CPM next month with the goal of assessing demand and working out technical details. People will be able to use Manifold dollars to place bets and send winnings to their preferred charity from a pre-selected list. Manifold dollars will be converted into donations at face value (so $1 USD = $1 sent to charity). Additionally, we will subsidize these contributions using a matching quadratic funding pool of $10,000.
After running this CPM for roughly one month, we will increase the amount of prize money, offer a larger set of approved charities, and experiment with other questions and matching mechanisms.
After the second round of prize money is used, Manifold will solicit partnerships with charitable organizations, allowing them to match user donations in exchange for decision-relevant forecasts.
By subsidizing charitable donations, we hope to vastly increase the number of people who participate in prediction markets. After working out the difficulties of managing a large, active prediction market, we will explore the applications of prediction markets for institutional decision making, impact markets, and information bounties.
What’s the case for your project?
The primary reason to fund charity prediction markets is that they allow people to legally trade on prediction markets using something they value. Other reasons CPMs could be a good idea:
Prediction markets reward people for being right about the future. People who are often right about the future may be good at deciding which charities are good.
Prediction markets can help charities make better decisions. For example, the AMF may post a question like “Will there be >600k malaria deaths in 2022”, or more specifically “Will AMF’s follow-up survey show that 40% or more of bednets are still in use after 3 months?”
Prediction markets can act as donor lotteries. Donors who win have more reason to carefully research where to send their money; donors who lose can save time deciding among charities.
Prediction markets are fun! Donors may donate more to have fun, compared to what they would have donated otherwise.
Donors can invest money back into the CPM platform, to fund development or offer donation matches to new traders.
Donations to charity are less subject to diminishing marginal returns compared to personal wealth. For example, I’d personally prefer a guaranteed $5k rather than a coin flip between $0 and $10k, but I’m ambivalent about which of these gets donated, according to expected value theory
Prediction markets can train donors to get into the mindset of putting down probability estimates on beliefs, and thinking rigorously about the future.
Donors may not have to pay taxes on earnings in a CPM, the way they don’t for any earnings in a donor advised fund.
[optional] If you have further materials for us to read (e.g. documents you've written for other potential funders), you can link to them here. You can also upload them below.
Sam’s original writeup on charity prediction markets
If you have further materials for us to read (e.g. documents you've written for other potential funders), please link to them here. You can also upload them below.
Manifold Market’s Seed Round Memo (WIP)
How long have you been working on this project, and how much has been spent on it?
We have been working on Manifold for the last 4 months, and this specific project for 1 month. We have not spent any money on it yet, but have set aside $50k for initial subsidizing of CPM’s. This comes out of an early $200k grant made from EA Funds’s Long-Term Future Fund.
What has been achieved so far?
We have determined how to implement CPM’s on the Manifold platform, written materials to introduce CPM’s to a general audience, and spoken to several people working in charitable organizations such as OpenPhil and Rethink Charity to assess the relevance of CPM’s to the grantmaking process.
Do you have any reservations about your project? Is there any way it could cause major harm? If so, what are you going to do to prevent that?
Scaled up, CPM’s may become a focus for political groups. In the best case, this would draw money out of politics and into charity. In the worst case, CPM’s could direct more money towards divisive charities. However, the first few tests of CPM’s will avoid this problem by using a pre-approved list of charities.
Finances
How much funding are you asking for?
$1-2 million
[optional] It can be helpful to provide a budget range: a minimal, mainline, and maximally ambitious version.
Low - $500k
Mainline - $1 million
Ambitious - $10 million
Can you provide a rough budget? (Either in text, or as a linked spreadsheet.)
Iteration 1: 3 rounds, 50k/round (150k total) to test different subsidy mechanisms. Each round would last ~1 month.
Iteration 2: 2 rounds 100k/round (200K total) to assess demand and identify technical problems with more users. Each round would last ~2 months.
Iteration 3: 1 round of $1 million over a longer period, to assess market properties at scale. This would last ~6 months.
Conditional on ambitious funding:
Iteration 4: 10 rounds, $1 million/round researching the use of CPM’s for institutional decision making, impact markets, and information bounties.
Is this a non-profit or for-profit?
Non-profit
If this is a for-profit, please explain the terms of the investment. (We’d strongly prefer that you keep it simple by using a standard YC SAFE. More complicated structures may take longer to review and/or be held to higher standards.)
We’ve already separately secured funding via the Future Fund Regranting Program, as a for-profit investment in Manifold’s main operations.
By when do you intend to have spent the money?
In roughly one year.
Will you ask for more money at that point? If so, how much might you ask for?
Yes. Assuming users are enthusiastic about CPM’s, we would ask for $5-10 million to distribute the following year.
What will it look like if your project has gone poorly / just OK / well at that time?
Poorly: Few people use the matching funds, few new users join in order to participate
Just OK: People use the matching funds, users join in order to participate, but users cash out quickly rather than try to grow their donation via good forecasting.
Well: People use the funds, many new users join, and people actively bet in an attempt to increase their donations.
The metrics we’d track, along with our goals:
Number of users participating: 1k for our first round; 10k in a year
Amount of money donated by users: $10k for our first round; $1m in a year
Enthusiasm of users: Such that users would be vocally upset if we stopped supporting this
Effectiveness of user donations: Comparable to decisions made by OpenPhil
Public press: Familiarizing charitable donors with prediction markets, and prediction market enthusiasts with effective causes
How much $/yr could this project plausibly productively use at maximum scale down the road? Please explain your answer.
Once technical details are worked out, CPM’s are only limited by the total amount of charitable giving (~$500 billion/year) and the number of users interested in participating.
Even in the short term (~3 years), CPM’s could quickly become relevant for EA organizations and donors, and direct a significant fraction of EA funding, on the order of $100 million/year.
In the longer term, CPM’s will likely increase the volume of money directed towards charitable causes. In addition, CPM’s can be used to identify new charitable causes and reward work in these areas. With this combination of increased giving and better giving opportunities, it seems plausible that CPM’s could direct $1 billion/year, within 5 years.
About you
Who is applying? Please list your name(s) and location(s), as well as your LinkedIn(s) or personal website(s), if you have them. If this is a new project, please include this for all founders.
Founders:
Austin Chen, San Francisco, CA. https://www.linkedin.com/in/austinch/
James Grugett, Austin, TX. https://www.linkedin.com/in/james-grugett/
Stephen Grugett, Austin, TX. https://www.linkedin.com/in/stephen-grugett
Consultants:
Sam Harsimony, US. Writes the eponymous blog Harsimony https://harsimony.wordpress.com/
Sinclair Chen, San Francisco, CA. https://www.linkedin.com/in/clarity-chen/
What, if anything, is your involvement with the effective altruism community? (50 words only)
Austin has taken the GWWC pledge; after taxes and rent, donations are his biggest expense. He reads the EA Forum obsessively and posts occasionally.
Sam has been active in the effective altruist community for several years. He posts regularly on the EA Forum, and has written for GWWC.
If this project is less than one year old: for each of the people listed on the project: what’s the most impressive thing you’ve done? (50 words each, only)
Austin honestly thinks Manifold is pretty impressive (it’s only been 4 months since we started!); if he had to choose something else, going from outside consultant to tech lead at Streamlit in 3ish months.
Sam uses his blog to highlight original, impactful ideas. He invented CPM’s, was the first to use the term “retroactive public goods funding”, and has proposed a simple, politically feasible approach to land value taxation which avoids the use of complicated appraisal systems.
Other
If you had any other ideas you considered applying with, please list them — we might be interested in them, too. (100 words only)
We’d like to build Impact Certificates (aka Retroactive Public Goods Funding), natively in Manifold Markets. Public goods projects (e.g. open-source software; high-quality blog posts; speculative research) could issue shares on our platform. Investors could buy these shares, in expectation of a future payout from a philanthropist or community oracle.
See also: https://manifold.markets/Austin/will-manifold-implement-retroactive
Is there anything else we should know?
We’re still sad to have been rejected from the FTX EA Fellowships. Can we visit y’all in the Bahamas anyways?