Formal opinion · 2017
Prédictions, chiffrement et libertés
[Predictive analytics, encryption and civil liberties — Conseil national du numérique (French national digital council) opinion, 2017]
Opinion of the Conseil national du numérique (French national digital council) on predictive uses of data and on the protection of encrypted communications.
Of the whole series, this is the opinion that has aged best, and it needs saying without self-satisfaction: having been right is not the same as having had influence.
The 2017 position rejected the deliberate weakening of encryption, on the grounds that a vulnerability introduced for legitimate authorities remains a vulnerability for everyone. The argument has never needed revising. It had to be served up again, almost unchanged, during later European debates on access to encrypted communications and during comparable French legislative attempts.
The “predictions” part has held up less well. The 2017 analytical framework reasoned about systems trained for a defined task, whose use and scope could be debated. It says nothing about general-purpose systems whose uses are not fixed at the design stage. This is not an error, it is a limit of its time, but it makes the first half of the opinion less useful than the second.
One uncomfortable question remains: did this opinion change a single decision? I have no proof that it did. An advisory opinion that states the right position without shifting the actual trade-off feeds a body of doctrine, which is not nothing, but the two should not be confused.