Breadth vs. match: a dilemma for Economy App (alert: economics inside)

In a recent thread, Matt and I debated at some length the applicability of Economy App to an economic subsystem like Edgeryders. Edgeryders works by patching along (that is, matching skillset with task) three main kinds of commodities: in-kind donations (people donating effort to other people’s projects); outright hired labour (people selling effort to other people’s projects); and in-kind investments (people investing effort in other people’s projects, with compensation to come when and if such projects take off.

In order to make this work, Matt and the EA crowd have dreamed up a number of hacks. You can hire someone on EA by offering money as money and requesting effort; donate by offering effort and requesting “thank you vouchers”; invest by offering effort and requesting contingent money vouchers, that take value zero if the project does not take off and some positive value if it does.

To an economist, this way of reasoning is heavily reminescent of general equilibrium theory. Mathematical economists have been investigating under which conditions you can reach a market-clearing equilibrium for all goods and services at once ever since the 19th century. As they did so, they quickly found out that they needed to differentiate goods not only according to their physical nature (you can’t exchange apples and oranges in the same market, unless you use one as the numerary for the other), but also according to:

  • space: skiing gear in the Swiss Alps has more value than a Caribbean island.
  • time: an umbrella has more value on a rainy day than on a sunny one.
  • contingency: a ton of wheat next year will have more value if crops are scarce than if they are abundant.

The contingency provision is of particular significance because it invokes probabilistic math into the picture. Arrow-Debreu-McKenzie needed it to derive their equilibrium theorems. I would also add

  • identity: I would be happy to work for the Electronic Frontier Foundation, but not for the Climate Change Denial alliance, even if salary and working conditions are identical. 

Now, all of these things pose problems to EA just as to any market-clearing system. On the one hand they make clearing at all possible: no one wants to be forced into an exchange system that does not take into account all of these things, and does not give parties full control on what they exchange for what. On the other hand, they constrain the algorithm and therefore make it more difficult to close deals.

Now, in principle EA solves it all through the expedient of the numerary. Through a number of interesting hacks, most of these can probably work in the context of Edgeryders. But a numerary does not clear markets per se: what clears markets is that prices can adjust to equlibrium (and that an equilibrium even exists). Two items can be similar and have identical price, but not be identical: for example, bicycle A is more lightweight but I like the gear in bicycle B better. Or I want somebody to write blog posts for me, but I think Alice is a better match for what I have in mind than Bob. What mechanism do you have in mind for letting me accept a deal that implies Alice-the-blogger, while rejecting a very similar (but, to me, different) deal that implies Bob-the-blogger?

Almost forgot…

… I would very much welcome [akurnya]'s opinion on this post and the Economy App for Edgeryders thread!

Your orders, please?

Regarding the bloggers example: our algorithm will not match you with anybody whosoever offering copywriting when you need copywriting. Instead, you look through the list of offers (search function, filters, categories help you). And for every offer, you decide if you’d like to order it for its price tag, taking into account (as you said) space, time / urgency, contingency and identity. For every item you want to get you are supposed to create an “order group” in Economy App and put all your mutually alternative orders in there. Once the algorithm gets one of these alternative orders executed for you, it automatically deactivates the others. (This mechanism is to ensure the barter algorithm has enough flexibility to make deals.)

Now, about the price adaptation mechanism: This will in the first version be informal, like on eBay. Which means, if sellers realize that they do not get a reasonable amount of orders for their product, they might consider lowering the price to increase demand. This of course missed capturing the “conditional expression of interest” from those who just did not order because the price was too high, while everything else was acceptable. So to improve on this, there will later be a “bid price” mechanism: An order automatically becomes active once the seller lowers the price below a threshold that a buyer entered as a bid. And the seller may see bids beforehand to evaluate the interest in the product.

Did we miss anything important in this approach?

No obvious flaws

No, not at all! The reason why I recalled Arrow-Debreu at all is that it is recognized that time, place, and contingency specification of commodities place a high cognitive load on market operators. They managed to prove existence, uniqueness and optimality of a market-clearing equilibrium, but only at the cost of assuming complete information on all commodities across time, space and states of the world, and no computational friction. At this, behavioral economists shook their head and embarked on a completely different trajectory. It seems that the same problem applies to EA: commodity specification shrinks the pool of offers available to close any deal… but this can’t be helped, we just need to hope to achieve critical mass in at least some of these dimensions. Anyway the flaws, if any, are not in the design of EA, but in its need for a large number of users to become viable. In the Edgeryders implementation this might actually be easier than in general, as the services being exchanged are not going to be many.