On Tue, Feb 25, 2020 at 9:06 PM Volker Krause <vkra...@kde.org> wrote: > > Given how little practical experience we have with this, I'd be cautious with > publishing unreviewed raw data though. And that isn't just theoretical. >
Quite. Publishing the raw data is a bad idea from a privacy perspective. Anonymisation is not magic, let alone easy. Just like in crypto, figuring out some construction that you cannot break yourself is never very difficult but figuring out constructions that *others* cannot break is rather harder. Besides, it gives "them" another data set to hone their skills on, which may not be such a good idea in 2020. Aggregate data should be safer, as long as we have more than 1 user :) Regards, - Johan
