Is Personal Data an Extension of the Self?

The AOL leak is normally filed under the heading of failed anonymisation, a cautionary tale for engineers who believe that deleting a column deletes a person. That reading is too modest. The far stranger fact is that the reconstruction was accurate. The queries carried a woman's insomnia, her debts, her garden, her grief, and they described her more closely than her neighbours could have managed. She had authored none of it as a statement about herself, and she could not have withheld it without leaving the century. Her data was not a possession she had misplaced. It was a description of her that existed outside her, with its own life and its own traffic.

So the question worth asking is not who owns it, but what it is. Is personal data an extension of the self — an organ of identity, continuous with memory and intention — or is it something a self merely leaves behind, like a footprint in setting cement?

What the extended mind actually requires

Andy Clark and David Chalmers argued in 1998 that cognition does not stop at the skull. Their test was functional rather than anatomical: if an external resource is reliably available, automatically endorsed, and easily accessible, the boundary of the cognitive system should be drawn to include it. Otto's notebook of addresses, in their example, is not a tool he consults; it is part of the machinery by which he remembers.

Run a behavioural profile through the same three conditions. Availability: exhaustive, continuous, better than your own recall. Accessibility: instant, and increasingly worn on the wrist. Endorsement: absent. You cannot correct the profile by remembering differently. You cannot decline a memory the system insists you have. The notebook answers to Otto; the profile answers to whoever holds the pipeline. An external store that fails the endorsement condition does not extend a mind — it represents one, and representation without standing is the oldest political problem there is.

Locke, labour, and the involuntary production of property

The proprietary argument is seductive because it borrows Locke's prestige. In the Second Treatise, mixing one's labour with what is unowned creates title. You produce your data by living; therefore, by the same logic that turns gathered acorns into property, it belongs to you.

Two things go wrong. First, the labour is largely involuntary — Locke's labourer is an agent who elects to work, whereas scrolling, walking, and breathing inside a sensor field are not choices in any sense that survives scrutiny. Second, data is never produced alone. Every query is a joint production of your curiosity and another party's interface; every step count is co-authored by an accelerometer. There is no clean moment at which your share can be separated from the machine's.

The proprietary instinct — it is mine, so I should be able to sell it — is not a defence of the self but a transfer of it. A functioning market in personal data presupposes that persons are partially commodifiable, and a society that accepts the exchange rate has already conceded the larger argument. Margaret Jane Radin saw this coming a decade before the platforms arrived.

If you found this valuable, consider supporting our work.

Three ways of drawing the boundary