Post-Currencyism 2: Allocation

NOTE HXA7241 2014-01-19T12:14Z

But does this make sense as an economic mechanism? How can it allocate without currency?


The economic problem is fundamentally about this: using information to connect supply to demand. The critical element is information: it is not exactly money and markets that solve the economic problem, it is the information they convey.

The core is the allocation decision: which demand should the limited supply go to, to make best use of it for the economy overall? So the information needed is that which helps make that decision.

We want information about things, but the primary information we want for this basic problem is about their supply and demand (the features of goods themselves are only important in so far as they are wanted and available – only in their supply/demand aspect). And since each agent can be both a supplier and demander they can ‘plug together’ and form a secondary set of information: a network of interactions and relations of agents' supply and demand.

And this is just the information the ‘network-oriented’ economic system has. It knows what is supplied and demanded, and the networks of previous interactions. It knows the outputs of each agent given particular inputs, and where those outputs are then fed into as inputs elsewhere, and so on. That is all the basic information needed for allocation decisions.


But good information is not just about ‘coverage’ but ‘structure’ – the kind of information – its particular abstraction. There can be plenty of low-level detailed information but if it lacks the right structure it is difficult to see anything in it or use it. What currency does is provide a good, simple, useful abstraction; can this network-oriented system also do that?

Price abstracts not over ‘value’ but over supply, demand, and, perhaps most notably, time. Thinking in terms of the idea of ‘value’ is misleading and confusing: really, our notion of ‘value’ here comes from the system that currency itself enables, so trying to define currency in ‘value’ just goes in a circle.

Price (at its core) is set not simply from supply/demand, because those are not known continually, instantly, and for every distinction of item. It is set from what it was recently and what it is expected to be. So price offers a kind of blurring/filtering of supply/demand: where supply/demand is really changing moment-to-moment, in what would seem like a stream of noise, price smooths that into a broader pattern. That is its useful abstraction.


The abstraction offered by the network-oriented system is network structure: similarity and comparability of sub-regions of the network. Allocation decisions can be made by referring to similar earlier cases in the networks of interaction.

The simplest example would be deciding to give to A because A has asked in the past, or to take from B because B has given in the past – the present decision resembles the past network, though it has only one link. A deeper example would look more deeply into the network: supply more to C because C supplies more in turn to others – that is, C is comparatively more productive, and so a better choice to give resources to.

This leads to an idea of ‘node-rank’ like PageRank. A measure of productivity of an agent can be estimated from its outputs less its inputs. And so allocation can favour higher node-rank agents. But like PageRank this could be adjusted with various other information, from the network or maybe outside it.


What is the relation of the network-oriented system to a price/currency system? It seems to be a superset.

Price is made from supply and demand of a particular good: ingredients the system has – so whatever function price performs can also be done by the network-oriented system.

Price is a summation over parts of the network, whereas comparison of those actual network parts accesses more information. With more detailed information, this network-oriented system enables a wider range of possible cooperative structures.


The system seems to have no sense of the commensurability of ‘value’: you cannot obviously compare things, and make a relative priority of requests. But is that not necessary for an economic system?

No, because different requests are not really comparable/substitutable. The question the agent asks is somewhere between: “I want this particular thing, can I get it”, and “these things are available, which do I want”. The agent's relation to the economic system is in asking: “what is available to me” (for getting goods that is, but also “what should I contribute” in the other direction).

All the system needs to do is answer that question; it does not need to suggest any commensurability. Money ‘externalises’ the calculation needed to answer into a form of single comparable ordering. It gets agents themselves to answer their question by looking at prices. But the network-oriented system ‘internalises’ this and answers by looking at network relations.