However, even more importantly, data could be used in more broadly beneficial ways than providing discounts or similar purely economic benefits. Cooperative ownership would enable data to be used in ways that conventional companies would not have an incentive to do.
To illustrate the limitations of innovation by conventional enterprises, let’s imagine that Facebook would start using data to develop a new feature that would make it easier for users to organise live events, and as a result, people would spend more time going to events at the expense of being on Facebook. The feature would harm shareholder profits as users would spend less time looking at ads, and it would therefore be removed even if users would find it beneficial. In other words, there is a conflict of interest between owners and users of the platform. If, on the other hand, Facebook were a cooperative owned by its users, it could simply decide that the benefits of the feature to members would exceed the costs. A conflict of interest between users and owners can be dismantled by making users into owners.
For Pentland, cooperatively owned “community data” could enable what he describes as “people in a community asking each other questions’‘. To understand what he means, let’s imagine a scenario where a person notices their neighbour has their arm in a cast and asks them what happened. He replies by describing how he fell on a bicycle because the roads are more slippery than usual and recommends warning others about it. With a data cooperative, similar knowledge could be derived in a more sophisticated manner by detecting that an increased number of residents in the community have been in cycling accidents, which can prompt them to organise better pavement gritting. Pentland’s ideas about “community data” are not just theoretical - he has been involved in creating numerous successful pilots, including one in Senegal that revealed a new way to organise public transportation with a 15% improvement in effectiveness without any additional costs.
Previously there was a mention of an imaginary credit union that would use a data cooperative for organising members to collectively bargain discounts on cat food. They might also discover that many cat owners have a challenge in finding someone to catsit for them. The data cooperative could enable them to form mutual aid circles where people catsit for each other: if a member looks after another member’s cat for a week, they could have their cat being looked after by some other member for a week. For capitalist firms, there would be no incentive to do this. Because people make reciprocal favours instead of paying each other, there are no financial transactions to profit from. Credit union members, on the other hand, might simply decide that it is in their interest. In fact, the credit union could help members to start an independent cat-sitting cooperative for this purpose with an online platform, which would be relatively effortless to build using existing free, open-source software such as Wordpress. Because the platform would be owned by the users, the incentive would be to lower the transaction costs to a minimum, instead of the almost opposite incentive conventional platform companies have: to maximise transaction commissions.
Community data can be produced, and is in fact often more effective to produce, in a privacy respective manner. For example, Pentland points out that one major existing example of community data, that of government censuses, are able to tell the number of residents and their average age, income, etc. in a given area without knowing any personal details of any individual resident, let alone having to create an equivalent of a phonebook like directory with a detailed profile of every resident. I won’t go into much technical details about his solutions for ensuring privacy, but those interested can look into “Open Algorithms”, or “OPAL” for more details. Briefly summarised, the key mechanism behind OPALs is that the data is never moved or copied from the “data vault” and “algorithms are moved to the data” instead. The current norm is the opposite - collecting a lot of data into one location where algorithms are run on it. Instead, with OPAL, each personal data vault runs the algorithm independently and returns an answer that is aggregated in a way that does not risk the privacy of any individual. This somewhat resembles other forms of community data, such as the national census mentioned previously.
Pentland also makes the case on how this could tackle the “cold-start” problem in which incumbent tech giants possess a big advantage over new competitors due to having access to more data. With data cooperatives, the data would no longer be hoarded by the monopolistic giants, but would allow equal opportunity for all individuals and businesses to present their ideas on how it could be used for the benefit of the members.
This radical restructuring of the key source of power for the currently most powerful companies in the world could help usher in an era of more accountable and equitable technology companies - spearheaded by cooperatives. The cooperative movement needs to avoid trying to simply catch up and copy its capitalist competitors and instead realise that it can radically restructure the future course of technological progress in ways its competitors can’t. For cooperatives to explore and establish themselves in new and emerging sectors, new forms of financing for cooperatives need to emerge. Pursuing this is what drives us at Coop Exchange, and if you think it’s a goal worth pursuing, join us in this endeavour by subscribing to this newsletter.