Linda Linsefors recently wrote a post making the case for impact purchases. I've always been somewhat skeptical of impact purchases, at least as they are currently applied. I see two key problems with impact purchases: firstly that the market of sellers is far larger than the market of buyers and secondly that many people would have done those projects anyway.
Buyer's Market vs. Seller's Market
How much impact do these purchases have on your decision to pursue a project? If you think you are almost certainly going to be funded, then it'll make a huge difference. If it is only a small chance, then it'll make almost no difference at all. The problem I saw in a past impact purchase (can't find the link) was that there were way too many people trying to sell impact with all of them chasing a vanishingly small amount of money. This meant that projects either had a small chance of having their impact purchased or that the amount of money provided was small compared to the effort that went into the project.
One potential solution to this is to limit the size of the market. In the impact purchase I saw, the buyer was willing to consider almost anything. While this ensured that high-value projects weren't excluded from consideration, it also made the imbalance between the impact for sale and available capital as large as possible. For this reason, I'd suggest defining a narrow market - for example AI safety or wild animal suffering only.
However, the market might still be far larger than the available capital. One way to limit it further would be to limit the purchase to a certain group of people. For example you might only include alumni of Harvard or people who attended an AI safety camp. This results in the exclusion of a lot of talent, however it's arguably better to motivate a few talented people a lot than a much larger pool a little. You could limit it further by only considering projects that occurred in a certain time period.
Regardless of whether you use an impact purchase or provide funding normally, there's always a chance that the project would have taken place regardless. However, impact purchases greatly increase this problem since the person already did the project without knowing whether or not they would be funded.
Let's imagine a near optimal scenario where if the project succeeds you have a 90% chance of being funded. Even a 10% chance of not be funded or receiving minimal funding could be a deal-breaker in many circumstances. But of course, this doesn't include the chance of the project failing. And considering the market dynamics above, the chance of the project being funded reasonably could be extremely small. This tends to suggests that the projects that get funded might only be those that would have and could have occurred anyway. As mentioned before, narrowing the field might make incentives strong enough for more of the difference to be counterfactual.
Another issue is that only people who knew about the impact purchase during a project could have been influenced by it. Limiting it by category of person can help here, as say if it's only open to alumni of program X, the organisers of that program might be able to contact those eligible to let them now. Another possibility would be to make anyone eligible if they register as potentially interested during a particular period. Further, if we limit it to, say, the next two years, then we exclude all projects that were completely before the announcement and which hence couldn't have been influenced.
One possible response is that even if these projects aren't counterfactual, people who create impact deserve funding as a matter of fairness. This seems like a reasonable position, but notice that it is in tension with utilitarianism. Another argument is that people who had an impact in the past are likely to spend the money to have an impact in the future. This might be the case, but if this is the primary vehicle through which these purchases have an impact, it might be worthwhile crafting a mechanism that more narrowly focuses on this.
Another issue with impact purchases is the potential to harm relations between people, lead to bitterness or to demotivate people. People tend to be very personally invested in their projects and tend to have quite a high evaluation of the impact of their projects. Buyers will often only value projects at a fraction of what the seller thinks that it is worth. This may lead to the seller feeling unappreciated or insulted. This is compounded if the seller put a lot of time and energy into the project with the expectation of some kind of reimbursement. Even though people are rejected from grants as well, there's a difference between someone deciding another project is more promising and someone explicitly telling you that your project isn't worth anywhere near what you think it is. I don't have a suggested solution to this problem.