Preferences don't get aggregated: Turning other people's plans from obstacles into affordances
Social choices must work the same way individual choices do: the social entity has a set of preferences ranking alternatives from most preferred to least preferred, and it must have a choice function selecting the most preferred alternative to be the one that is implemented. If social choices didn’t work this way, we’d need a separate theory of choice for collectives rather than for individuals, and some kind of empirical way of determining whether a system is a collective intelligence or an individual one, a distinction that probably doesn’t exist.
Presumably, the preferences of the social entity are in some way based on the preferences of the members of the collective. We’d also hope that any group we’re a part of is somehow bringing our preferences together in such a way that we’d all be happy with the results as individuals while also allowing us to leave in peace with each other. So one major focus of research on social choice theory has been to figure out ways in which the preferences of individuals can be aggregated into group preferences.
The main conclusion is that you can’t aggregate preferences (at least not without violating desirable conditions). If you know that Alice wants pizza for lunch, and Bob wants hamburgers, and Claire wants tacos, then you do not know what the Alice-Bob-Claire collective wants for lunch. There’s just no nice way to aggregate their preferences.
Instead of thinking about aggregating preferences, we can think about a social order in terms of rendering people’s plans mutually compatible with each other. For example, if there’s one tomato avaiable for purchase, and I plan to buy a tomato, this is mutually consistent with everyone else’s plans in the market only if no one else plans to buy a tomato. It doesn’t matter whether their preferences are aggregated so long as when one person pursues their own plans, that fits with other people’s plans instead of contradicting them.
When people are poorly coordinated, other people’s plans are obstacles to your own plans. Maybe you plan to buy a tomato, but lots of other people plan to buy tomatoes as well, which makes it hard for you to fulfill your plan. A cognitive glue like the price system helps people achieve mutually compatible plans by updating their perceptions of how available resources are in their environment based on how other people’s plans are changing—so if tomatoes were highly available before, but then many more people plan to use tomatoes such that they become much less available for your plans, you will perceive that tomatoes are less available and therefore plan to use something else. The result is a shared model of everyone’s plans, and social preferences are just the parameters of the shared model. So social preferences can be constructed not by aggregating preferences but by the processes of a cognitive glue that render people’s plans mutually compatible.
Normally, other people’s plans are obstacles to one’s own, but a cognitive glue helps to turn people’s plans from obstacles to each other into affordances for each other. For example, a baker’s plans to bake bread in exchange for other goods and services can be exploited by you to get the bread you want by providing the baker with whatever good or service you produce—shoes, for example, if you’re a cobbler. Their plans are tools that you can use to achieve your own plans. In particular, by changing your plan to better fit with their plan, you can get more of what you want, such as bread.
So the reason we don’t know the lunch preferences of the Alice-Bob-Claire collective isn’t because we haven’t found a clever way to aggregate their preferences but because preference aggregation is the wrong idea. We need to study the process by which the gruop preferences are constructed—a process consisting of the members of the group adjusting their plans to one another to turn obstacles into affordances.