Meta: purpose of this doc – lay out case to potential donors, spark discussion […], clarify in longer form what Manifund is, [attract collaborators/advocates?]
In Manifund’s 2023 retrospective, we said, "Our initial thesis was that grant applications and screening largely can be done in public, and should be."
Economic Experimentation
Manifund's sister company, Manifold, runs experiments with prediction markets. Because Manifold runs on play money, and because users get to participate on the market-creation side as well as the prediction side, it’s arguably the best playground in the world for experimenting rapidly with financial mechanisms. Because official currency isn’t involved, both Manifold and its users (and the two in combination) can run experiments rapid-fire without jumping through many hoops. For instance, Manifold has tried various types of automatic market-making algorithms, mechanisms for multiple-choice questions, short selling, limit orders, contests, leagues, even a dating app spinoff. And users have made bots, non-question questions like “___ stock (never resolves)”, markets that serve as bounties to incentivize particular events to happen, questions with meta resolution criteria like “Will the total volume of YES trading on this question be more than x mana?”. Manifund is different, because lots of real money is moving through our programs. But we're still small and fast-moving ([example of grant turnaround time]), and our ethos is similar: we aim to be an incubator for lots of small- and medium-scale experiments testing out funding mechanisms for charity and public goods.
Effective altruism originally asked the question, “What properties make a charitable project useful, efficient, and valuable?” Manifund is an attempt to extend that question in two ways:
We delegated donor funds to a set of regrantors each of whom chose, based on their domain knowledge and networks, how to allocate their budget across projects.
We hosted the 2024 grants program for Astral Codex Ten, a blog written by Scott Alexander.
We created a platform for projects to sell shares in any future funding or prize winnings they might receive. (Read more here.)
We’ve experimented with two models of impact certificates: Our initial stab at impact certificates allowed investors to buy certificates immediately. For the ACX impact certs, we implemented a way for contributors to better coordinate to crowdfund a project: they make offers to buy impact certificates, and if there are enough offers by a prespecified end date to cover a founder’s minimum costs, the buy orders all go through.
Is there anything general we can say about these kinds of experiments?