I just sold half of a blog post

April 2015

I’m excited to report that 50% of the impact of my donation matching literature review has just been purchased in the first round of Paul Christiano and Katja Grace’s impact purchase!

This means that Paul and Katja now own 50% of the altruistic consequences of the post, so you should consider it only 50% as much of a good deed as you previously did.


Paul and Katja are purchasing “certificates of impact,” which are a philanthropic instrument they invented. It’s like funding someone to do something, except that instead of giving them money before they do the thing, you give them money afterwards.

Why is that better?

It helps to make an analogy to normal tradeable goods.

Suppose I’m buying a shirt. What I generally do is look at all the shirts that are available and then buy one. (Contrast this with the funding model that’s normal for philanthropy, where I would pay someone money for a shirt, and then they would make it.)

The normal market model puts most of the downside risk of shirt production on the seller. If my tailor runs out of cloth and can’t make my shirt after all, then my tailor doesn’t get to sell it so my tailor loses the money. This aligns incentives better: my tailor is likely to try harder to make a saleable shirt if they haven’t gotten paid for the result yet. (For instance, Kickstarter products might experience fewer production delays and failures if sellers were only paid when their product shipped.)

An even more important aspect of the normal market model is that it makes it much easier to decide what shirt you want. This isn’t such a huge deal for shirts, where it’s not so hard to predict ex ante how good the future shirt will be. It’s much harder to do that in philanthropic funding!

There’s a much more in-depth explanation of why an impact purchase market would be helpful on the website.

How did the process work?

I applied through the form on Paul and Katja’s website, giving a short description of the blog post, why I thought it had good consequences, and the minimum amount of money for which I would undo the post’s impact.

Then Paul and Katja made their own evaluation of how much each applicant project did, and ran a complicated yet nifty auction process that ensured they would purchase the maximum amount of impact possible while ensuring that I was incentivized to report the true minimum price at which I would be willing to sell the post’s impact.

(The structure of the auction is such that my stated price for the post has no effect on what price Katja and Paul buy it at, only on whether or not it gets purchased. It’s pretty cool.)

Isn’t this basically Paul and Katja throwing money at you for no reason? Why would you name a minimum price greater than zero?

It’s not free money for me, because I’ve now been better incentivized to do things that Paul and Katja think are good for the world.

Well, actually, right now, since Paul and Katja are the only buyers of impact purchases and the market is very thin, I think there’s some free money to be had. For instance, if Paul and Katja discontinue the program at the end of the year and nobody else steps up, then the incentives will go away, and nobody will really care that much that once they bought 50% of one of my posts.

Also, because there are so few entrants right now, impact is getting sold at a substantial markup–according to Paul and Katja’s impact evaluations, the price at which they purchased impact from sellers this round was almost double what they could have gotten from donating to GiveWell top charities. But I don’t expect that to continue.

In the long term, though, impact purchases don’t involve any money-throwing at all.

It’s easy to think they would, because at first it seems like Paul and Katja aren’t really buying anything. The right to claim 50% of my post’s impact doesn’t seem worth that much because people might not play along. For instance, people might still give me close to 100% of the social status benefits from doing something altruistic.

But actually, the more important thing Paul and Katja are buying is an incentive for me to do more stuff they’d want to buy in the future. On this view, the point of giving them a claim to the post is to make sure that I can’t sell more than 100% of it. (If I could do that, it would distort the market instead and lead to people getting too much incentive to do things.) The public notice that they own half of the consequences is simply a mechanism of enforcing this.

So in the long term–if the impact purchase market grows–there’s a very good reason for me to name a nonzero minimum price for my post, which is that I might think I could sell it later, or to someone else, for more money.

How was the experience? Would you do it again?

Yes! I think the impact purchase program is really cool, and could fill an important gap in what EA projects get funded/properly incentivized to get done. The application was very quick. (And on top of that, I got paid.)

There’s one round of purchases each month, and the next deadline is in a bit over a week on April 25. You should check it out and apply!

Enjoyed this post? Get notified of new ones via email or RSS. Or comment:

email me replies

format comments in markdown.

Jeff Wu

I really hope more buyers pop up and this idea succeeds. I think it’ll be mostly a social problem, rather than a technological one. Like I still find myself giving you most of the credit for the post. And even if you sold “100%” of your post, you’d still surely have some status gain from it. (Of course, you deserve some, if nothing else, for slightly lowering the price of altruistic deeds.)

Curious - is that note you added on the sold post what they bought? Or did they buy some abstract certificate that was issued out of thin air?

The certificate issuing and preventing double-spending problem seems similar to the problem cryptocurrencies solve. But maybe the community is small enough that it’s easy to simply keep track, for awhile.




I think it’ll be mostly a social problem, rather than a technological one. Like I still find myself giving you most of the credit for the post.

I tried to say this in the post, but I think that the social problem is actually pretty orthogonal to whether impact purchases work. The important thing that Paul and Katja are buying isn’t the status boost from claiming credit for the post–indeed, if they’re buying altruistically, they shouldn’t care about that at all. Instead, the main point of their purchase is its incentive effects.

By creating a market for impact purchases, they indicate to me and other people that one can get paid for doing things that have good long-term impact on the world. With a deep enough market, I can imagine people doing lots of things they wouldn’t have otherwise done, like working full-time on a bunch of small low-hanging-fruit projects instead of being employed by an EA org.

And I think this goes through even if the social problem doesn’t get solved and the original creators keep all the credit for the project to themselves. The only effect from that would be to slightly distort incentives in favor of selling things in the impact market, since the seller would get both money and altruistic credit for the purchase. But I think you could compensate for this by decreasing your estimate of the altruistic good done by the project you’re buying. And anyway, social status effects already produce much more severe distortions than this in other ways.



I just don’t get how they actually evaluate value, or what should be considered. The full transparency of applications, combined with the unsure valuation, create a weird situation where:

  1. I have a number of things I could potentially apply with

  2. And yet I have no idea if these would be laughed at or seen as very valuable and funded.

And I don’t want to submit 20 (hyperbole) different applications ;)



@Anonymous: there’s a spot on the application where you explain the consequences of your thing and make an argument for why they’re good/how good they are. If you can make a reasonable argument, nobody will laugh at you even if they disagree.

Also, it seems like they only publicize their valuations for the winning projects (check this with Paul or Katja if you’re worried). So the public won’t see if one of your projects gets evaluated poorly.


Paul Christiano

@Anonymous: You can mention in the application that you don’t want it to be published if we don’t fund it (we’re happy to withhold any info that people ask us to withhold, though we’ll mention the existence of unpublished applications). I’d err on the side of submitting things, except insofar as it’s annoying to spend the time on the application. You could also submit a few things this month and see how it goes.

@Jeff: We don’t really expect to get 50% of the social credit. We are mostly buying the altruistic impact, the part left over if the project creates 0 skills and receives 0 social recognition (like an anonymous donation). I expect we will get a non-zero amount of social credit, but it’s just gravy.

We’ve actually had a handful of people express interest in buying; in the short term I think it is more likely that the bottleneck will be sellers. I hope we’ll open it up to a two-sided auction in the third or fourth roundhttp://www.benkuhn.net/impact-purchase#.


Jeff Wu

Ah, that makes sense.

I was imagining a long-term goal might be for it to become acceptable for rich folks who don’t care too much about effectiveness to contribute to a certificate-buying fund for status reasons. I dunno if that would be very valuable



Does the money really give people a further incentive to do good things? EAs are giving away excess money to charity anyway, so one EA giving another EA money doesn’t seem like it’d be motivating. If someone bought the impact for something that I did, I would just donate that money to charity. And if the buyers weren’t doing this program, that money would presumably go to charity as well.



@Gina: that’s might be true for a lot of people on the current margin, but:

So I think the incentive effects still exist to some extent right now, and will get a lot stronger in the future as the market grows.


Tom Ash

I found this helpful. I got that it’s partly about creating an incentive and not about literally transferring ‘credit’, though I think it’d be a good idea for the http://impactpurchase.org/ to explain this a little more clearly :) However, I’m still unsure about how much it will actually incentivise people in the current situation (partly because of the thinness and possible evanescence of the market, as you say). How incentivised do you feel Ben? How likely are you to undertake additional EA projects as a result of it?

I’m hesitant about encouraging .impact people to solicit many unsolicited applications (or doing so myself), given the time cost and the difficulty of estimating their prospects given these currently depend purely on Katja and Paul’s values and epistemology. I did submit something like someone’s donation to GiveWell’s recommended split of charities in case it provided a helpful benchmark (though I asked if this was so, and picked a highly leveraged example which fell into a ‘project’ so the system wouldn’t just get flooded with straight donations if that’s now what Paul and Katja want).