I agree this would be better — then the funders would be able to fund Alice’s project for $1 rather than $10. But still, for projects that are retroactively funded, there’s no surplus-according-to-the-funder’s-values, right?
I think there is still surplus-according-to-the-funders-values in this impact market specifically, just as much as there is with regular grants. Retro funders were not asked to assign valuations based on their “true values” where maybe 1 year of good AI safety research is worth in the 7 figures (though what this “true value” thing would even mean I do not quite understand). Instead, they were asked to “operate on a model where they treat retrospective awards the same as prospective awards, multiplied by a probability of success.” So they get the same surplus as usual, just with more complete information when deciding how much to pay.
I can’t speak for Zach, but that’s not the meaning of “surplus” I had in mind above.
Suppose Funder’s bar is 10 utils per dollar. In standard operation, it will buy some work at 10 util/$, but will also have the chance to buy at 11 util/$, 12 util/$, and maybe even a little at 20 util/$. By surplus, I meant the extra 1 . . .2 . . . 10 util/$ that were above the funding bar. Because Funder was able to acquire utils on the cheap in these transactions, it can afford to acquire more utils within its budget.
If Funder bought impact certificates on a 10 util/$ basis, it wouldn’t have any surplus of this type. The fix may be that Funder should buy at its average weighted util/$ basis, which is higher than its funding bar. Of course, this requires the funder to figure out its average weighted util/$ in the relevant cause area, which might or might not be easy.
Ah, hooray! This resolves my concerns, I think, if true. It’s in tension with other things you say. For example, in the example here, “The Good Foundation values the project at $18,000 of impact” and funds the project for $18K. This uses the true-value method rather than the divide-by-P(success) method.
In this context “project’s true value (to a funder) = $X” means “the funder is indifferent between the status quo and spending $X to make the project happen.” True value depends on available funding and other available opportunities; it’s a marginal analysis question.
Bob has a project idea. The project would cost $10. A funder thinks it has a 99% chance of producing $0 value and a 1% chance of producing $100 value, so its EV is $1, and that’s less than its cost, so it’s not funded in advance. A super savvy investor thinks the project has EV > $10 and funds it. It successfully produces $100 value.
How much is the funder supposed to give retroactively?
I feel like ex-ante-funder-beliefs are irrelevant and the right question has to be “how much would you pay for the project if you knew it would succeed.” But this question is necessarily about “true value” rather than covering the actual costs to the project-doer and giving them a reasonable wage. (And funders have to use the actual-costs-and-reasonable-wage stuff to fund projects for less than their “true value” and generate surplus.)
(This is ultimately up to retro funders, and they each might handle cases like this differently.)
In my opinion, by that definition of true value which is accounting for other opportunities and limited resources, they should just pay $100 for it. If LTFF is well-calibrated, they do not pay any more (in expectation) in the impact market than they do with regular grantmaking, because 99% of project like this will fail, and LTFF will pay nothing for those. So there is still the same amount of total surplus, but LTFF is only paying for the projects that actually succeeded.
I think irl, the “true value” thing you’re talking about is still dependent on real wages, because it’s sensitive to the other opportunities that LTFF has which are funding people’s real wages.
There’s a different type of “true value”, which is like how much would the free market pay for AI safety researchers if it could correctly account for existential risk reduction which is an intergenerational public good. If they tried to base valuations on that, they’d pay more in the impact market than they would with grants.
This is ultimately up to retro funders, and they each might handle cases like this differently.
Oh man, having the central mechanism unclear makes me really uncomfortable for the investors. They might invest reasonably, thinking that the funders would use a particular process, and then the funders use a less generous process...
In my opinion, by that definition of true value which is accounting for other opportunities and limited resources, they should just pay $100 for it. If LTFF is well-calibrated, they do not pay any more (in expectation) in the impact market than they do with regular grantmaking, because 99% of project like this will fail, and LTFF will pay nothing for those. So there is still the same amount of total surplus, but LTFF is only paying for the projects that actually succeeded.
What happened to “operate on a model where they treat retrospective awards the same as prospective awards, multiplied by a probability of success.” Can you apply that idea to this case? I think the idea is incoherent and if not I want to know how it works. [This is the most important paragraph in this comment.] [Edit: actually the first paragraph is important too: if funders aren’t supposed to make decisions in a particular way, but just assign funding according to no prespecified mechanism, that’s a big deal.]
(Also, if the funder just pays $100, there’s zero surplus, and if the funder always pays their true value then there’s always zero surplus and this is my original concern...)
There’s a different type of “true value”, which is like how much would the free market pay for AI safety researchers if it could correctly account for existential risk reduction which is an intergenerational public good.
Sure. I claim this is ~never decision-relevant and not a useful concept.
That’s a valid concern. The traditional form of surplus I had in mind [edit: might not be; see my response to Rachel about the proper util-to-$ conversion factor] there any more. However, the funder probably recognizes some value in (1) the projects the investors funded that weren’t selected for retrofunding, and (2) aligned investors likely devoting their “profits” on other good projects (if the market is set up to allow charitable reinvestment only, rather than withdrawal of profits—I suspect this will likely be the case for tax reasons).
If those gains aren’t enough for the retrofunder, it could promise 100% payment up to investment price, but only partial payment of impact over the investment price—thus splitting the surplus between itself and the investor in whatever fraction seems advisable.
Hmm. I am really trying to fill in holes, not be adversarial, but I mostly just don’t think this works.
the funder probably recognizes some value in [] the projects the investors funded that weren’t selected for retrofunding
No. If the project produces zero value, then no value for funder. If the project produces positive value, then it’s retrofunded. (At least in the simple theoretical case. Maybe in practice small-value projects don’t get funded. Then profit-seeking investors raise their bar: they don’t just fund everything that’s positive-EV, only stuff that’s still positive-EV when you treat small positive outcomes as zero. Not sure how that works out.)
the funder probably recognizes some value in . . . aligned investors likely devoting their “profits” on other good projects
Yes.
If those gains aren’t enough for the retrofunder, it could promise 100% payment up to investment price, but only partial payment of impact over the investment price—thus splitting the surplus between itself and the investor in whatever fraction seems advisible.
Surely this isn’t optimal, there’s deadweight loss. And it’s still exploitable and this suggests that something is broken. E.g. Alice can do something like: write a bad proposal for her project to ensure it isn’t funded in advance, self-fund at an investment of $10, and thereby extract $10 from the funders.
At least in the simple theoretical case. Maybe in practice small-value projects don’t get funded.
Scott Alexander has stated that: “Since most people won’t create literally zero value, and I don’t want to be overwhelmed with requests to buy certificates for tiny amounts, I’m going to set a limit that I won’t buy certificates that I value at less than half their starting price.” I’m not sure exactly what “starting price” means here, but one could envision a rule like this causing a lot of grants which the retrofunder would assign some non-trivial value to nevertheless resolving to $0 value.
Surely this isn’t optimal, there’s deadweight loss.
It’s impossible to optimize for all potential virtues at once, though. I’m actually a bit of a self-professed skeptic of impact markets, but I am highly uncertain about how much to value the error-correction possibility of impact markets in mitigating the effects of initial misjudgments of large grantmakers.
One can imagine a market with conditions that are might be more favorable to the impact-market idea than longtermism. Suppose we had a market in upstart global health charities that resolves in five years. By default, the major funders will give to GiveWell-recommended charities. If a startup proves more effective than that baseline after five years, its backers win extra charitable dollars to “spend” on future projects (possibly capped?) -- but the major funders get to enjoy increased returns going forward because they now have an even better place to put some of their funds.
And this scheme would address a real problem—it is tough to get funding for an upstart in global health because the sure-thing charities are already so good, yet failure to adequately form new charities in response to changing global conditions will eventually mean a lot of missed opportunities. In contrast, it is somewhat more likely that a market in longtermist interventions is a solution in search of a problem severe enough to deal with the complications and overhead costs.
And it’s still exploitable and this suggests that something is broken.
I do agree that collusion / manipulation is a real concern here. By analogy, if you’ve looked at the history of Manifund’s sister site (Manifold Markets), you’ll see a lot of people concocting very clever ways to manipulate the system. I am pretty skeptical of majority self-funding for this reason. This whole idea needs to be empirically tested in play-money models and low-stakes real money environments before it is potentially ready for prime time. It is too underspecified for an investor to make big moves in reliance on (unless the project is something they almost would have funded absent the impact cert) or for retrofunders to strongly bind themselves to.
I also think that as a practical matter there has to be some quality pre-screen before projects go on the market. Otherwise, the coordination problems with too many projects for the funding available will get severe. If offers are too spread out as a result, then few projects will get enough to launch. And as a current micrograntor, I can already see that dumping too many lightly screened on a marketplace is going to disincentivize people making the effort to screen and then evaluate projects. Going back to manipulation, you’d have to manipulate your project enough to not get funded, but not so much as to fail the pre-screen.
I guess one final defense to manipulation is that retrofunders are on the honor system. If there is reason to believe that someone manipulated the system, they are not actually bound to buy those certificates. That option would need to be exercised rarely or the system would crumble . . . but it is there.
I agree this would be better — then the funders would be able to fund Alice’s project for $1 rather than $10. But still, for projects that are retroactively funded, there’s no surplus-according-to-the-funder’s-values, right?
I think there is still surplus-according-to-the-funders-values in this impact market specifically, just as much as there is with regular grants. Retro funders were not asked to assign valuations based on their “true values” where maybe 1 year of good AI safety research is worth in the 7 figures (though what this “true value” thing would even mean I do not quite understand). Instead, they were asked to “operate on a model where they treat retrospective awards the same as prospective awards, multiplied by a probability of success.” So they get the same surplus as usual, just with more complete information when deciding how much to pay.
I can’t speak for Zach, but that’s not the meaning of “surplus” I had in mind above.
Suppose Funder’s bar is 10 utils per dollar. In standard operation, it will buy some work at 10 util/$, but will also have the chance to buy at 11 util/$, 12 util/$, and maybe even a little at 20 util/$. By surplus, I meant the extra 1 . . .2 . . . 10 util/$ that were above the funding bar. Because Funder was able to acquire utils on the cheap in these transactions, it can afford to acquire more utils within its budget.
If Funder bought impact certificates on a 10 util/$ basis, it wouldn’t have any surplus of this type. The fix may be that Funder should buy at its average weighted util/$ basis, which is higher than its funding bar. Of course, this requires the funder to figure out its average weighted util/$ in the relevant cause area, which might or might not be easy.
Ah, hooray! This resolves my concerns, I think, if true. It’s in tension with other things you say. For example, in the example here, “The Good Foundation values the project at $18,000 of impact” and funds the project for $18K. This uses the true-value method rather than the divide-by-P(success) method.
In this context “project’s true value (to a funder) = $X” means “the funder is indifferent between the status quo and spending $X to make the project happen.” True value depends on available funding and other available opportunities; it’s a marginal analysis question.
Actually I’m confused again. Suppose:
Bob has a project idea. The project would cost $10. A funder thinks it has a 99% chance of producing $0 value and a 1% chance of producing $100 value, so its EV is $1, and that’s less than its cost, so it’s not funded in advance. A super savvy investor thinks the project has EV > $10 and funds it. It successfully produces $100 value.
How much is the funder supposed to give retroactively?
I feel like ex-ante-funder-beliefs are irrelevant and the right question has to be “how much would you pay for the project if you knew it would succeed.” But this question is necessarily about “true value” rather than covering the actual costs to the project-doer and giving them a reasonable wage. (And funders have to use the actual-costs-and-reasonable-wage stuff to fund projects for less than their “true value” and generate surplus.)
(This is ultimately up to retro funders, and they each might handle cases like this differently.)
In my opinion, by that definition of true value which is accounting for other opportunities and limited resources, they should just pay $100 for it. If LTFF is well-calibrated, they do not pay any more (in expectation) in the impact market than they do with regular grantmaking, because 99% of project like this will fail, and LTFF will pay nothing for those. So there is still the same amount of total surplus, but LTFF is only paying for the projects that actually succeeded.
I think irl, the “true value” thing you’re talking about is still dependent on real wages, because it’s sensitive to the other opportunities that LTFF has which are funding people’s real wages.
There’s a different type of “true value”, which is like how much would the free market pay for AI safety researchers if it could correctly account for existential risk reduction which is an intergenerational public good. If they tried to base valuations on that, they’d pay more in the impact market than they would with grants.
Oh man, having the central mechanism unclear makes me really uncomfortable for the investors. They might invest reasonably, thinking that the funders would use a particular process, and then the funders use a less generous process...
What happened to “operate on a model where they treat retrospective awards the same as prospective awards, multiplied by a probability of success.” Can you apply that idea to this case? I think the idea is incoherent and if not I want to know how it works. [This is the most important paragraph in this comment.] [Edit: actually the first paragraph is important too: if funders aren’t supposed to make decisions in a particular way, but just assign funding according to no prespecified mechanism, that’s a big deal.]
(Also, if the funder just pays $100, there’s zero surplus, and if the funder always pays their true value then there’s always zero surplus and this is my original concern...)
Sure. I claim this is ~never decision-relevant and not a useful concept.
That’s a valid concern. The traditional form of surplus I had in mind [edit: might not be; see my response to Rachel about the proper util-to-$ conversion factor] there any more. However, the funder probably recognizes some value in (1) the projects the investors funded that weren’t selected for retrofunding, and (2) aligned investors likely devoting their “profits” on other good projects (if the market is set up to allow charitable reinvestment only, rather than withdrawal of profits—I suspect this will likely be the case for tax reasons).
If those gains aren’t enough for the retrofunder, it could promise 100% payment up to investment price, but only partial payment of impact over the investment price—thus splitting the surplus between itself and the investor in whatever fraction seems advisable.
Hmm. I am really trying to fill in holes, not be adversarial, but I mostly just don’t think this works.
No. If the project produces zero value, then no value for funder. If the project produces positive value, then it’s retrofunded. (At least in the simple theoretical case. Maybe in practice small-value projects don’t get funded. Then profit-seeking investors raise their bar: they don’t just fund everything that’s positive-EV, only stuff that’s still positive-EV when you treat small positive outcomes as zero. Not sure how that works out.)
Yes.
Surely this isn’t optimal, there’s deadweight loss. And it’s still exploitable and this suggests that something is broken. E.g. Alice can do something like: write a bad proposal for her project to ensure it isn’t funded in advance, self-fund at an investment of $10, and thereby extract $10 from the funders.
Scott Alexander has stated that: “Since most people won’t create literally zero value, and I don’t want to be overwhelmed with requests to buy certificates for tiny amounts, I’m going to set a limit that I won’t buy certificates that I value at less than half their starting price.” I’m not sure exactly what “starting price” means here, but one could envision a rule like this causing a lot of grants which the retrofunder would assign some non-trivial value to nevertheless resolving to $0 value.
It’s impossible to optimize for all potential virtues at once, though. I’m actually a bit of a self-professed skeptic of impact markets, but I am highly uncertain about how much to value the error-correction possibility of impact markets in mitigating the effects of initial misjudgments of large grantmakers.
One can imagine a market with conditions that are might be more favorable to the impact-market idea than longtermism. Suppose we had a market in upstart global health charities that resolves in five years. By default, the major funders will give to GiveWell-recommended charities. If a startup proves more effective than that baseline after five years, its backers win extra charitable dollars to “spend” on future projects (possibly capped?) -- but the major funders get to enjoy increased returns going forward because they now have an even better place to put some of their funds.
And this scheme would address a real problem—it is tough to get funding for an upstart in global health because the sure-thing charities are already so good, yet failure to adequately form new charities in response to changing global conditions will eventually mean a lot of missed opportunities. In contrast, it is somewhat more likely that a market in longtermist interventions is a solution in search of a problem severe enough to deal with the complications and overhead costs.
I do agree that collusion / manipulation is a real concern here. By analogy, if you’ve looked at the history of Manifund’s sister site (Manifold Markets), you’ll see a lot of people concocting very clever ways to manipulate the system. I am pretty skeptical of majority self-funding for this reason. This whole idea needs to be empirically tested in play-money models and low-stakes real money environments before it is potentially ready for prime time. It is too underspecified for an investor to make big moves in reliance on (unless the project is something they almost would have funded absent the impact cert) or for retrofunders to strongly bind themselves to.
I also think that as a practical matter there has to be some quality pre-screen before projects go on the market. Otherwise, the coordination problems with too many projects for the funding available will get severe. If offers are too spread out as a result, then few projects will get enough to launch. And as a current micrograntor, I can already see that dumping too many lightly screened on a marketplace is going to disincentivize people making the effort to screen and then evaluate projects. Going back to manipulation, you’d have to manipulate your project enough to not get funded, but not so much as to fail the pre-screen.
I guess one final defense to manipulation is that retrofunders are on the honor system. If there is reason to believe that someone manipulated the system, they are not actually bound to buy those certificates. That option would need to be exercised rarely or the system would crumble . . . but it is there.