I've just found out that Paul Christiano and Katja Grace are already buying certificates of impact.
In these writings we propose the creation of a sub-field of knowledge, Moral Economics, and provide in broad strokes its characteristics. We assemble previous writings that can be considered prospective subfields. We also discuss several concepts that have been developed in recent years that are instrumentally useful for Moral Economic thinking, and propose several new ones that could also bear fruit if the study of this area develops further in the future.
Certificates of Impact, Doing It Right - Giles Edkins
Moral market failure: how COIs might help
Problems with COIs, and their solutions
Implementing COIs
Agential Identity in Moral Economics
The intuition behind this I take from conversations with Duncan Germain, Stuart Russell, Davi Mamblona, Geoff Anders, where I got the intuition that the announcement of a field is a great way to generate a memetic attractor around an idea which permits people to look at it in a new light, as has happened to parkour, pilates, hiking, planking, memetics and others. For instance Bostrom starts his Hail Mary paper:
In the field of superintelligence control studies, which focuses on how to ensure that a hypothetical future superintelligent system would be safe and beneficial, two broad classes of approaches to the control problem can be distinguished: capability control methods and value selection methods.
In seven words and a stroke of genius, he created a precedent that I find of immense value. He created a field of study. I have already cited this passage in papers and writings of mine, and now others too can refer to superintelligence control as a field. My conversations with Stuart Russell (and many grad students) have the goal of creating an AI safety engineering course at Berkeley, to set the precedent for some of the 1500+ universities that use his AIMA textbook to also have that course. You can imagine how happy I was to see the first words of Bostrom’s paper. I literally grinned for some ten seconds. Academics are strange.
There is ample room to consider other items in the moral economic cluster that are worth exploring but were not thought of by any of the cited texts or editors of this writing. Taking a glossary of economics, I can see within Moral Economics parallels with interesting and different properties to the following concepts:
Accelerator
Bank
Bond
Moral Capital Gain
Conditionality
Cost of Job Loss
Credit
Moral Debt (see e.g. Lakoff 1997)
Inflation
Deflation
Depreciation (see e.g. Haste consideration above)
Dividend
Effective Demand (is it approximately unbounded? unlike classical demand?)
Externalities
Finance
Gross Moral Domestic Product (How much moral good did an entity generate)
Inequality (who are the Moral 1%?)
Interest
Labor extraction
Macroeconomics
Microeconomics
Multiplier
Moral Poverty
Profit
Public Goods
Moral recession
Shares
Supply-constrained (more frequent in moral economics)
Value Added
All of these would behave in interesting and different ways in altruistic economics or moral economics - go back and try yourself to envision the differences. However, economics is not my field of expertise, so I can only show some of the pointers but not develop the concepts in full. This is a job for economists. As Bertrand Russell would say, one job of philosophy is to ask questions about what we don’t know and create ways for that to become knowable. Much of philosophical progress was the creation of what today are considered sciences. In Moral Trade, Toby Ord has set the ground for this to happen once again, I suggest we follow suit and begin exploring this unknown scientific territory.
I'm interested in moving moral economics forward in a different way: by creating some kind of online "moral market" and seeing what your happens.
There are two possible systems I could implement:
I'll describe the points-based system here, as it's the one I've thought through a bit more. I presume it theoretically diverges from a certificate of impact system, but I haven't thought through exactly how.
Users have points. The total number of points in the system is 1 billion.
At any time, a user with nonzero points can make a request that somebody else donate to a particular charity in exchange for some of those points.
Fundamentally that's the only mechanic that I'm imagining right now. Other bells and whistles can be added, such as prediction markets, or other goodies that you can purchase with points such as volunteer time.
The requests stay on the table until someone takes them up, so a (new or existing) user can acquire points by seeing which requests are currently active and donating to one of the relevant charities.
Why would anyone want points? Points can be used to influence other people in which charities they give to, how much and when. (Although if everyone agrees that points are worthless then this leverage disappears).
What are some use cases? A charity is running a fundraiser, and its supporters all want each other to donate to the fundraiser as soon as possible, so that the charity's staff aren't tearing their hair out. If any of these supporters have points, they can use some of them to encourage other supporters to donate early, by raising the points-per-dollar value of the charity.
Any other use cases? Moral trade might be possible - donating to a charity becomes slightly more attractive if you get points in return, and those points are some reflection of how much other people like the charity. I don't know how this would play out in practice though.
Trading points sounds like a lot of work. Yes it would be! Possibly enough to wipe out the value gained by moral trading. So the system would need one other major feature: automatic trading.
How does automatic trading work? Each user assigns a subjective utilons-per-dollar-donated value to each charity, as well as a value to holding onto the cash themselves. The system calculates a utilon-per-point value somehow. It can then automatically set the donation request price to be (utilon per dollar of charity divided by utilon per point). The system can also make suggestions to the user to donate when utility of charity + utility of points you'd get back > utility of holding onto the money.
Aren't you glossing over some things here? Yes, several.
How is this any use to me if I'm not a consequentialist and don't believe in utilons? I haven't thought about that yet.
This is all just chit-chat, and we're never going to see this happen, right? Wrong. I'm working on it here, although it's currently little more than a login page and a couple of database tables. Development help welcome! https://github.com/edkins/moral-market
Oh one other thing - I think the trickiest part of this system will be verifying whether someone has actually donated to a charity at the time they said they did. Every charity does it a different way.