Acemoglu & company on privacy externalities – is there really harm in inference?

Acemoglu and co-authors have this really impressive theory paper out on data externalities and why these mean that data markets will probably be inefficient, why there will be too much data shared, and why people who care a lot about privacy might end up demonstrating a willingness to sell data at a low price.

For all its dazzle, I have a problem with the notion of privacy that lies at the heart of the analysis, and this post tries to explain why I think their notion of privacy s insufficiently specific to address the policy problems we face around privacy and competition. The reason I think it’s worth insisting on this is that I worry about our making our case against GAFA power a privacy case. There is great social good to be had from the responsible use of digital trails, and the more we talk up privacy harms, the harder we make it to use data for good.