Oh but I'm sure all the shady people honoured the terms and conditions Zuck had, right? In that case, gow can there be any shady people, as Zuck said, they're strict about protecting their users' data! /s

Any chance you can expand on this? What data, in particular, did you obtain that would make people feel uneasy?

From what I remember, you had access to the data that would be visible to a users' friends by default: photos, posts, friends list, but not private messages.

None of that data is actually particularly scary by itself, although I wouldn't enjoy the thought of it all being out there.

What Cambridge Analytica was bragging about was their big data model promising to distill users' data to a psychographic profile.

There's this story about 30 "likes" (of books / artists/ movies/ etc) being enough to predict peoples' sexual orientation with higher accuracy than the coworkers sharing a desk with someone.

The large publicity for such "big data" methods in the last years may have contributed to the change in public opinion, and (part of) the reason Facebook is just now getting its 15 minutes of infamy.

I’m going off memory too, but I think that besides the name and profile photo of your friends (and email address?) this API didn’t return much else. And I also think I remember that the amount of your data that would turn up in a friend list API call was limited in some way by your own privacy settings.

In our circle around that time people were calling out apps that required these permissions and recommending that users not authorize them if it was not obvious why the app would require that kind of data. John Gruber comes to mind as one of those people. These calls for caution were not in the mainstream though.

Edit: I believe this API contained info like favorite tv show or band if those fields were filled out in your profile, but again I’m fairly sure that your own privacy settings could have limited this if you had them set to draconian levels. So in other words, yes this API was limited to publicly available data, more or less.

I've written up a number of these issues. Here's one where you can use Likes to determine quite a bit with minimal effort: https://caseysoftware.com/blog/social-apis-for-social-evil

I also worked on a quasi-popular facebook app in a similar time period and you got a lot of data but the specifics would change from month to month so it was sometimes more and sometimes less.

The worst, for me, was posting on behalf of other people and getting replies to your app like "I'm so glad to hear from my favorite grandson after so long!".

Ironically, considering the current kerfuffle, I remember my biggest frustration being The Guardian, as they'd spam my feed with 'friend X read article Y on The Guardian'.

Why would you post on behalf of other people?

Because it was allowed and it is an easy way to spread an app virally? "John Doe via ViralApp: [This app is great! Get your own Virus today by clicking here!]"?

Yeah I mean their API usage policy for many years was basically a gentleman's agreement to only fetch and keep what you needed.

but... you were forced, basically, to have access to more than was needed in the first place. just to implement fb login, you had to ask for permission for friends list. you couldn't not ask for it, regardless of whether you used it or even wanted it.

Was the data anonymized ?

I had the contrary experience. I always felt that FB gave me less information via the API about the people around me.

I feel all this discussion around CA is non sense because everybody knew that was possible and we are arguing about it only once someone big took an advantage. Is like leaving you door wide open and finding that someone stole your stuff. On the other hand a more limited API makes it unusable for fair purposes.