Manifund: Prediction Markets but Charity
Got a free $500 to invest in impact certificates, so I have to figure out how this works now.
Usually, I dump money into our german Give Well instance, where I don’t have to think that deeply about charities. Re-granting impact certificates is kind of the opposite.
What You Buy
Here is Manifund’s example of impact certificates:
A research team proposes a project to develop a forecasting model to prevent pandemics. They ask for $5,000 to work on this project.
Investor Ivan offers to buy $3,000 of impact certs at a $6,000 valuation; other investors contribute the rest, and the project is successfully funded at this higher valuation.
Three months later, the forecasting model proves to be effective in predicting the trajectory of an upcoming pandemic and helping hospitals take action.
The Good Foundation values the project at $18,000 of impact, offering to buy up all of the outstanding certs.
Since Ivan owns 50% of the project's certs, his stake has tripled in value from $3,000 to $9,000; he sells them for $9,000 to The Good Foundation, netting a $6,000 profit. (Important note: for legal reasons, profits on Manifund impact certificates can currently only be used to donate to charity and can't be cashed out in the normal way.)
Step 4 is the magic one for me. This hypothetical Good Foundation essentially buys shares and destroys them. What do they get out of the deal here?
ACX discussed multiple models of impact certificates and has a section on What Is Being Sold? It seems to be mostly bragging rights but I don’t understand how one can meaningfully buy them. Everybody knows that Good Foundation did neither do the actual work (like the research team in the example) nor did they provide the necessary capital (like Investor Ivan). In theory, we could introduce a gag order for Ivan, so he must not claim any bragging rights. Still it is visible that he made money picking charities and Good Foundation does not.
What does Good Foundation really buy then?
Being a non-profit, the Good Foundation does not care that much what they receive for their money but what the impact of spending the money is. Now consider what happens if you buy a shirt in a store: Of course, you get a shirt but what is the impact of the sale?
Primarily, the store has money to buy a replacement shirt. The impact of your money is that another shirt is made. Likewise, Investor Ivan again has cash to invest in other charities and if he made profit he is incentivized to do so. The impact of Good Foundation buying up all outstanding certs is that unknown other charities get funded. They give up some control.
Secondarily, the store ensures the availability of shirts by storing them. Ivan takes the risk of the outcome since he holds shares for a while, so he acts as insurance for Good Foundation. They pay a premium for this since they only buy successful projects. Having this risk allocated to a third party (Ivan) could increase the “liquidity”. In other words, more charities get funded. This might be the core benefit of a platform like Manifund to society? Not superior resource allocation but more funding because they can be treated as profitable investments by dump money.
Thirdly, your choice of store and shirt has an impact. At least statistically, more shirts and stores like these will appear and fewer of your alternatives. The invisible hand of the market adjusts to the demand. There is a little switcheroo here. Investor Ivan does not assess “the impact of the forecasting model”. He bets on “what Good Foundation will think the impact of the forecasting model is”. ACX calls Good Foundation an “oracular funder” and concludes on this aspect with “guess this one is up to the final oracular funders.”
Here is how I would express what happens with impact certificates involved: Good Foundation outsources the choice what to fund to people like Ivan and only steers them through vague “values.” They might describe their values in advance but ultimately they concretely express them by buying (and destroying) shares. Ivan acts as a matchmaker between proposals and funders with certain values.
Manifund does not seem to understand itself as a matchmaking service but rather an impact assessment service. Kinda funny since they are closely related to Manifold.love which is all about matchmaking. This activity seems similar to a startup targeting certain venture capitalists. Maybe this aspect is implicit so far, because oracular funders on Manifund all share Rationalist values or because this is just intuitive thinking in Silicon Valley.
Ok, so much about my understanding of the theory.
Manifund in Practice
For the ACX Grants 2024, Manifund offered me $500 to re-grant. Seems like a nice opportunity to try impact certificates in practice. That isn’t much money for most of the proposals, so I’m just a little fish. My plan was to split it into five times $100, so I have to pick five proposals.
I guess I immediately made a mistake: Instead of apply for an impact certificate, I donated to a project. Did I act as oracular funder here or is it just an alternative provided by the platform? Anyways, that was a hundred bucks for hackathons without any certificates involved.
Ok, these are my four picks now. I didn’t really spend time on the valuation and just selected something slightly above the minimum.
Building software to demystify land use regulations, Offered $100 @ $16K valuation. I have no clue about YIMBY/NIMBY in California. However, the valuation of 15k seems so cheap for buildings. It only takes a second anecdote like the brother mentioned there, which can be traced directly to this MVP. (7 days left and $14.740 short of funding)
News through prediction markets, Offered $100 @ $600 valuation. Ok, I’m biased here since The Base Rate Times has some similarity to my newsletter. I love their aggregation diagrams. (7 days left and $62 short of funding)
Distribute HPMOR copies in Bangalore, India, Offered $100 @ $1.3K valuation. HPMOR certainly had an impact one me and having some colleagues in Bangalore, I think it is a great target. There seems to be a focus on AI safety with this proposal but the book advertises rationalist thinking in general, so I consider the value potential is higher. Even with a low success rate, it should not be hard to achieve the 1.3K valuation. (7 days left and $50 short of funding)
Quantifying the costs of the Jones Act, Offered $100 @ $11K valuation. Maybe I’m a sucker for doing such research. The impact of a repeal could easily be billions. Of course, a single paper will not achieve this but it could start something. The protectionist aspects around the Jones Act make it plausible to me that funding through “normal science grants” might not happen, so this could fill an important gap. (7 days left and $7752 short of funding)
Kudos to
for already asking my questions in the comments there. 😉Given the funding gaps, maybe I should reallocate in a few days so at least two of them meet their funding goal. Until then, maybe one of you wants to pick up the shares instead.
Apparently, I have not actually “bought” impact certificates yet. I only “offered to buy” them so far because this is an auction. That makes sense initially because if the price immediately jumps much higher, the difference should bring additional funds to the proposers and not to the quick traders. Trading shall happen later to track the price changes as a project develops. Expect more on this in a some months.
Chess.com organizes a prediction contest about the FIDE World Championship.
The german edition is about state elections. I don’t think it’s that interesting for an international audience. In short, we also have alt-right problems in Germany.
Probably until next week, charitable readers! 😊