The economist Ronald Coase argued in 1937 that the existence of firms requires an explanation and provided one, which has become classical. The existence of firms requires an explanation because it is always possible to organise production by outsourcing each of its phases, ie. searching for a seller of that phase in the market. This is theoretically more efficient than running a small command-economy, that is a firm with employees and material assets whose productive uses are centrally allocated by the firm managers. Hence we should expect that firms that outsource everything until they cannot be called “firms” any longer will outcompete the cumbersome central-planners to extinction.
Why then are there firms, and huge ones at that? Tu put it simply, Coase argued that markets are not smooth, but they come with “transaction costs”: a seller should be sought, information about its services/goods obtained, prices negotiated, etc. Firms emerge when these transition costs can be avoided by internalising some of the productive phases, thus outcompeting productors that opt for markets. As a result, the economy is organised by an admixture of planned economies (firms) and markets.
One way one can look at the so-called “sharing economy” (e.g. Uber, AirBnB, etc.) is from the standpoint of Coase´s theory. Transaction costs (esp. search costs) have been dramatically lowered by digital networks, and as a consequence very “light” firms with no hardware (no cars, no hotels, etc.) and few employees have emerged, liquidating older, more planned, economies. The “sharing economy” is simply an economy that is organised a bit more with markets, and a bit less by managerial central-planners.
One fact that is less appreciated is that the very same dynamic is at play in some forms of cognitive production.
Consider any cognitive problem, let us say a medical diagnosis. By definition, for any particular diagnosis, there must be a person or a group of person in the world that is able to make it more accurately and faster than any other: let us call it “the top team”. But what happens in fact is that a particular group of people appointed by hospital managers on the basis of qualifications will try to make the diagnosis. This is quite inefficient, as it is very unlikely that that particular group is the top team for any particular case. But of course, they are the best placed: they are physically in the hospital or they can be called up in the middle of the night from the affluent suburb where they comfortably dwell, whereas the unknown top team is, well, unknown. In technical terms, there are transaction costs in the search for the top team. That is why there are expert teams, and why we pay them a salary even if they mostly remain idle. That is why hospital managers appoint expert teams instead of painstackingly looking for, literally, the best in the marketplace for any particular case. Expert teams are always a second-best in terms of knowledge, but they are usually the first choice all things considered.
But of course, transaction costs are diminishing even for the search for expertise. That is ultimately why crowdsourcing is emerging in biomedicine and, more generally, in science. For many tasks and probably for the overwhelming majority of them, for instance for standardised tasks as diagnosing a seasonal flu, the top team will be only marginally better than the alternatives. For seasonal flus, even the top team in the world will not be good enough to make it efficient to incur in the transaction costs involved in its search, no matter how low the costs will become. But for extremely complicated cases crowdsourcing may help – and the less it costs, the more it helps. Illnesses that are recalcitrant to diagnosis are one such cases. The absence of the top team might mean long suffering or even death for an undiagnosed patient. That is why platforms as CrowdMed are emerging, and delivering promising preliminary results.