> We'll require developers to not only get approval but also sign a contract in order to ask anyone for access to their posts or other private data.
Sounds like they'll make the developer agreement legally binding so they can take legal action for violating the ToS.
Obama's campaign said they were scraping for academic purposes, downloaded a bunch of data and then lied about deleting it? There are multiple levels of B.S. when it comes to Cambridge Analytica and Facebook. Only one finds comparison in the 2008 or 2012 races (on either side).
What Facebook did with the Obama campaign was far worse. No authorization was required - they gave them free, full access to the entire social graph. Didn't like Obama and didn't authorize an app to access your and your friends' info? Too bad, your information was still given to them and used to help get him elected. That wasn't a problem for anyone in the press though, because of course, Obama was a Democrat.
edit: I misspoke. There was an app required, but the Obama campaign accessed 190 MILLION profiles, versus 50 million involved in CA. According to [1]:
"The campaign boasted that more than a million people downloaded the app, which, given an average friend-list size of 190, means that as many as 190 million had at least some of their Facebook data vacuumed up by the Obama campaign — without their knowledge or consent...A former [Obama] campaign director, Carol Davidsen, tweeted that Facebook "was surprised we were able to suck out the whole social graph, but they didn't stop us once they realized that was what we were doing.""
[1] https://www.investors.com/politics/editorials/facebook-data-...
you are doing the old classic false equivalence... the deliberate use of an obama sanctioned facebook app that asks you to press a button to share the get out to vote (probably for obama) with their friends is 100% different than Cambridge University creating psyche profiles on people to ostensibly use as a fun way to find out your profile. then to sell that harvested data to a different campaigning firm Cambridge Analytica to use to micro target ads to them for political purposes unbeknownst to them is way different. It's about expectations: one expects it to be used in a political fashion because they opted into it and the other is not expecting it to be used to target themselves and their friends later.
So how is that a false equivalence?
permission for one.
> The campaign called this effort targeted sharing. And in those final weeks of the campaign, the team blitzed the supporters who had signed up for the app with requests to share specific online content with specific friends simply by clicking a button. More than 600,000 supporters followed through with more than 5 million contacts, asking their friends to register to vote, give money, vote or look at a video designed to change their mind.
intent and deliberate action up front by the user of the facebook app for two not some shady psych profile that gets sold later and used behind the scenes to target them without any link to the campaign. that's absolutely a false equivalence.
Also, that number 189 million is absolutely likely to be way too high and is likely bullshit. Say I have 190 friends with a significant overlap with say 50 friends who all know each other and are likely also friends with each other. it simply doesn't follow that every single friend has 190 unique friends to get a graph of that size even though the average across all users might be 190 friends. have a few more of those for different groups (business, whatever) and you have a much lower number. coupled with the fact that you have no idea what the average number of friends the people that opt into that app have if you are just comparing it to the overall average number of friends for the entire set of facebook users you possibly have a lower number still.
Which were the same permissions that the users of the Kogan app gave. Remember that the issue here revolves around the fact that friends of the users of either app never gave permission for their data to be used. The only difference is that Obama did this to about 4X as many people.
Also, that number 189 million is absolutely likely to be way too high and is likely bullshit
According to their own campaign manager, they obtained the entire US social graph, and Facebook knew about it and allowed it to happen. The 190 million number may actually be low.
you're still wrong but at this point, with all the evidence showing just how different the situation is, i suspect you want to be for whatever reason.
For hiring this isn't as touchy a subject, but surely there's a qualitative difference in these two interactions. One is based in an above-biard interaction. The other in subterfuge and hidden motives.
I think you are the one missing the point.
> The OFA app determined who should share what with who by running all the extracted profile data (including that taken from opted-in users' oblivious, perhaps anti-Obama friends) through their psychographic models.
Yes, this was a political app, downloaded by people motivated to get obama to be the president. They shared that info willingly to the obama campaign. The obama campaign used that information to suggest "hey man, we need help in texas and you know someone that might be able to help us. Can you do us a solid and send them some of this info?" Nothing I have seen suggests that they did anything untoward with the information voluntarily and directly shared with the campaign. They used it to suggest people to share the message to. And they did any action with the direct permission of the people using the political app.
> It's identical, except, for "the good guys".
Not even a little bit. The Koger fellow used the data he harvested from those psych profiles and also got the friends information that the people taking those "tests" doubtfully wanted them to have. If I take a silly test that is a facebook app I would not expect them to datamine my friends to later sell that information off to a political campaigning firm to later use for a political campaign that they may not have wanted it to have. If trump had used an app that did the same as obama I would have had no issue with it. he didn't, and I do.
Those people who were opted-in by proxy, i.e. the friends you sold out, may not have wanted the Obama campaign or Cambridge Analytica to get that info, but they never had a choice!
Both apps (well, OFA for sure, CA allegedly) took data legitimately provided to them, used it to feed predictive models, and then actioned marketing around exploiting those learnings.
>We released this tool for the Obama Campaign in August 2012. Over the next 3 months, we had over a million supporters authorize our app, giving us potential access to over 200 million unique people in their social network. For each campaign activity (persuasion, volunteering, voter registration, get-out-the-vote), we ran predictive models on this friend network to determine who to target. We then constructed an “influence graph” that matched existing supporters with potential targets. For each supporter, we provided a customized list of key friends with whom to share different types of content. [0]
Literally the ONLY difference, other than the political leanings, is Koger's app was then "acquired" by CA, in breach of Facebook's TOS. Which again, is something that happened probably ALL THE TIME in the pre-2014 wild west of Facebook app mining.
[0] http://www.rayidghani.com/what-can-non-profits-learn-from-so...
bullshit. plain and simple. you are falsely making those two equivalent.
Do I have a right to be upset about this or not?
https://twitter.com/mbsimon/status/975231597183229953
"I ran the Obama 2008 data-driven microtargeting team. How dare you! We didn’t steal private Facebook profile data from voters under false pretenses. OFA voluntarily solicited opinions of hundreds of thousands of voters. We didn’t commit theft to do our groundbreaking work."
We can all argue about whether facebook should even allow apps the kind of data they do, but the crux of what CA did was to re-sell data they collected from Facebook, right? This is where the two aren't the same, as far as I'm aware, but like I said, I'm not super in the know.
No. An independent developer had an app many years ago, recorded data from that app, and then sold that data to CA years after the fact. That developer did in fact violate Facebook's developer platform policies by selling the data, but CA had nothing to do with the app or what was represented to the people using it when they installed it.
Why didn’t Hillary win? I’m sure she had access to similar tools?
I think someone needs to take a look at the Hillary campaign and contrast it to identify exactly how effective Facebook data is
How is that not exactly the same thing, just done on a much larger scale by Obama? I get it, this is mostly a lefty crowd here on HN. But I just can't stand hypocrisy on either side. The fact that hundreds of commenters in here want to defend one guy and castigate another for doing the exact same thing based solely on the political affiliation of each is disgusting to me.
Source? All I can find is a scheme where they prompted people to sign in to a campaign site and grabbed the friend data that way. Icky as hell, but it's not obvious why that's "even worse" than CA.
What distinguishes CA and the Trump campaign is how well the data was used. CA staff apparently just did a better job at using the data to manipulate people. Their canvassing app. Using bots on Facebook and Twitter. Plus entrepreneurs from Eastern Europe or whatever. And maybe they went further, discouraging potential Clinton and Sanders supporters fron voting.
But anyway, the key point isn't whether CA or Obama/Clinton got more data from Facebook, or whether Facebook willingly helped Obama/Clinton. The key point is how well voters can now be manipulated. It's another level in the problem that money buys political power. And it's arguably how AI could do the same.
https://twitter.com/mbsimon/status/975231597183229953
"I ran the Obama 2008 data-driven microtargeting team. How dare you! We didn’t steal private Facebook profile data from voters under false pretenses. OFA voluntarily solicited opinions of hundreds of thousands of voters. We didn’t commit theft to do our groundbreaking work."
I've seen no hard data showing that any of this did Trump any good. Apparently not a single study has been done as to whether people either failed to vote or changed their vote based upon fake news or the use of this data.
I do follow fringes of online anarchism and anomie, and I was surprised to see so much support for Trump, when the polls were showing him so far down. And much of that support was so odd that I thought it ironic. But whatever.
I'd like to see such a study, for sure. But I'm not optimistic. Arguably, many who voted for Trump would be no more forthcoming to researchers than they were to those doing the election polling. There's just too much polarization, I suspect.
By the way, this is something that would have been stopped on any other app long before they ever accessed 190 million profiles. So Facebook essentially gave them the entire US social graph, that they wouldn't have allowed any other app with only 1 million users to have.
Fair point. But wondering if anyone asked them to delete it, and then went about verifying it later? I am guessing nobody bothered too hard because as the person in charge of the strategy said "Facebook were on our side"
https://twitter.com/cld276/status/975568208886484997
---
They [Facebook] were very candid that they allowed us to do things they wouldn’t have allowed someone else to do because they were on our side.
---
It's not like academics are any morally superior to anyone else anyway so I don't think Facebook should be handing that stuff out to anyone.
Source?
"Facebook was surprised we were able to suck out the whole social graph, but they didn't stop us once they realized that was what we were doing." - Carol Davidsen, Former Obama Campaign Director
The extraction of Facebook's social graph is the element I agree "finds comparison in the 2008 or 2012 races" [1]. But Obama's campaign didn't mis-represent its identity nor intentions to Facebook. Kogan and Cambridge Analytica did. Obama's campaign didn't lie about deleting its data when asked to (which, to my understanding, it wasn't). Kogan and Cambridge Analytica did. Moreover, Obama's campaign reported its backers to the FEC; Kogan and Cambridge Analytica do not.
As for Obama's campaign not being asked to delete data, all apps were required to delete data that they came into possession of under the pre-2014 policy. Those were the terms they agreed to when they deployed the app - especially one that was allowed special access to the entire US social graph (where no other app would have been). So if Kogan violated the policy, so has the Obama campaign.
Obama: "Give us a list of your Facebook friends, and we'll help you contact them about voting for Obama." While the Obama campaign did collect data about friends, this data was voluntarily given to them by users. If the Obama campaign had your Facebook data, it was because one of your friends knowingly and voluntarily gave it to the Obama campaign.
Kogan/CA: "Take a free personality quiz!" The people taking the quiz likely had no idea that they were supporting a political campaign or helping Trump win. You could have your data given to CA even if nobody in your friend graph supported it.
Also, that 190 million figure is probably inaccurate, because there is probably significant overlap in people's friends lists.
It was given to them by the 1 million users that authorized the app. Not the other 189 million (or more) users.
Also, that 190 million figure is probably inaccurate, because there is probably significant overlap in people's friends lists.
According to their own campaign manager, they "sucked out the whole of the social graph" with Facebook's blessings. So they may have actually had more than 190 million, since there are more US users that on Facebook.
Citation needed. Everything I've seen indicates Obama's campaign didn't violate the rules: https://twitter.com/mbsimon/status/975231597183229953
"I ran the Obama 2008 data-driven microtargeting team. How dare you! We didn’t steal private Facebook profile data from voters under false pretenses. OFA voluntarily solicited opinions of hundreds of thousands of voters. We didn’t commit theft to do our groundbreaking work."
Sure....coming right up! How about a statement from Obama's former campaign director saying that they "sucked out the whole of the social graph" with Facebook's blessings?
That doesn't prove what you are saying. "sucked out the whole of the social graph" is a non-technical statement that doesn't have an explicit meaning, nor is it specified what user data they were scraping.
You might think “well that’s egregious, but the rules at the time allowed it”. Unfortunately, even that isn’t true. Obama’s app did not play by the rules. The Facebook developer platform rules at the time stated, essentially, that you could not use the data you gained access to through your apps for any purpose outside of the operation of your apps. In other words, you weren’t allowed to create an app that claims it just sends a message to your friends about how wonderful Obama is, and then take the friend data you gain access to through that app and use it for the targeting of political advertising and/or campaign strategy, which is by all accounts precisely what they did. So they broke the same rules that Kogan did, just on a much larger scale, and with Facebook’s tacit approval.
Gleaming insights from that data and then using that data for targeting on FB is/was the point.
Creating an app that collected the data under false pretenses, transferring that data to another 3rd party and then claiming you never had that data makes it a different situation.
How do I know this? I developed apps on the platform from 2007-2014 and was always reading the rules and any changes they issued because I didn’t want to violate them and have my app banned. They were exceedingly clear on this issue. Sadly, Obama’s app was allowed to violate these rules and received no ban.
Unless I'm reading this interface totally wrong
http://www.jonloomer.com/wp-content/uploads/2014/09/create-f...
But regardless, it was still a developer TOS violation to export user IDs of people that hadn’t personally authorized your app and use those IDs in any custom audience even back when Obama did it. In other words, you weren’t supposed to grab user IDs of friends of your users and use those in custom audiences. In fact at one point, in an attempt to enforce this policy, Facebook stopped returning friend user IDs, and instead gave proxy user IDs that were meaningful only within the API, but couldn’t be used for custom audience targeting. Then they got rid of the target by ID option altogether.
I know a lot of folks have been desperate to make a false equivalence with the Obama campaign's social media use, but it doesn't pass the sniff test:
https://washingtonmonthly.com/2018/03/21/no-obama-didnt-empl...
https://www.investors.com/politics/editorials/facebook-data-....
"Facebook was surprised we were able to suck out the whole social graph, but they didn't stop us once they realized that was what we were doing." - Carol Davidsen, Former Obama Campaign Director
They used the exact same technique to access data from 190 million profiles - and about 189 million of those were without explicit authorization. Exactly the same technique, just with 4X the reach that CA had.
If you know of any evidence that Obama's campaign broke either the law or FB's ToS, please let me know.
That is simply not true. Facebook rate-limited apps for exactly this reason. Additionally, although it was technically possible, the TOS did not allow mass-scraping of friend data for any purpose, much less political purposes. Facebook monitored and routinely banned apps long before they ever accessed the data of even a few million friend profiles, let alone 190 million of them. That's why, according to Obama's own campaign manager, Facebook was "surprised" but then decided not to ban them - which they would have done to any other app.
the data was used for political microtargeting (in violation of FB's TOS)
Which is precisely what Obama did.
If you know of any evidence that Obama's campaign broke either the law or FB's ToS, please let me know.
See above.
Was it bad management ?
Was it not using Facebook data effectively ?
Was CA that pivotal? Was it fake news or better microtargetting?
If both sides used it why did one side lose?