Why data portability is not good enough

Data portability was one of the genuine achievements of the last decade of privacy law. You can request your data. You get a file. That is more than existed before, and it should not be dismissed.

It is also, as a mechanism for actual control, close to useless. It is worth being precise about why, because the reasons point at what would work.

What you actually get

A zip archive. Inside it, JSON in a schema nobody documented, CSVs with opaque column names, media files with numeric names, and a large amount of material that is technically your data and practically unreadable.

Three problems follow.

• It is a snapshot. It was accurate at the moment of export and starts decaying immediately.
• It is one-directional. You can get it out. There is rarely anywhere to put it that will accept it, and no obligation on anyone to.
• It is a copy, not a move. The original stays exactly where it was, doing exactly what it was doing. Nothing was transferred except bytes.

The switching-costs framing was the flaw

Portability was designed to solve competition: if users can leave, incumbents must compete. That is a reasonable regulatory objective and it explains the shape of the remedy.

But most people do not want to switch platforms. They want the data to do something for them, which is a different requirement entirely, and portability was never designed to deliver it.

The result is a right that is exercised rarely, produces a file that sits unopened, and satisfies a policy goal without changing anyone's situation.

What control would actually require

• A destination. Somewhere the exported material becomes usable rather than archived.
• Continuity. A live connection rather than an annual snapshot.
• Comprehension. Something that can read forty inconsistent formats and turn them into answers.
• Revocability. The ability to withdraw access from the source, not just take a copy alongside it.

The fourth is the one law came closest to and the one that matters most, and it remains the weakest in practice.

What changed the calculus

The comprehension problem is now solvable. A capable system can read a messy heterogeneous corpus and answer questions from it. That was the missing piece: for the first time there is a reason to exercise portability that produces something better than a folder.

The remaining constraint is not technical. It is whether the destination you move your data into treats it as yours. Exporting from a platform that claimed rights over your material into a product that also claims rights over your material is not portability. It is a change of landlord.

This piece revisits an earlier essay from our Medium archive: Why data portability is not good enough (https://medium.com/prifina/why-data-portability-is-not-good-enough-we-should-aim-for-a-free-consumer-a0644196505b).