Some here are worried that their data will still be used to train (ads) models: you can easily remove a large sample from your training set and still have a great predictive model. This would most likely not be hard to implement (modulo edge cases like whether you keep parameters trained on someone who has since decided to enroll in the training program).
> orders of magnitude more than Le Monde or even Facebook is getting from ads per visitor
No, Ads per visitor were roughly $18/m/u lately (quoting an anon on Twitter: check the financial reports), so modulo FX, €10 would be about 50% less. It would be a mistake to make it prohibitive. There’s a big debate on whether people picking the option would be valuable users or not. I suspect not (for reasons that rely on my experience working on the platform, so I won’t go into too much detail: explore the Ad tools to get what I mean).