Zuckerberg’s Hearings — Between the Lines

The Facebook CEO had a heated hearing, with lots of stories and rumors around what the platform actually collects. Some criticized Zuckerberg for dodging the inquiries, while others expressed their fury over questions that were not asked, or the limited technical knowledge of the lawmakers.

In this piece, I’ll go over some of the questions asked and the answers given, and will try to offer answers from publicly known facts about Facebook.

Does Facebook track users across devices?

Sen. Roy Blunt asked that, but Mark could not give an immediate answer.

Turns out, Facebook does that, and advertises it too as part of their services: Measuring Conversions on Facebook, Across Devices and in Mobile Apps

Now here is the tricky part; if Facebook tracks users even if they are not using Facebook. That is also something Facebook does — but not like some spyware. We’ll get into that shortly.

Does Facebook listen to your audio to extract information for advertising?

Zuckerberg gave Sen. Gary Peters a straight “No” on that.

There are many speculations that Facebook listens to your mobile, like people talking about cat food and getting ads for it a while later. But Facebook does not. It doesn’t need to.

Doing something like that would require a speech to text engine that would constantly record and analyze what people have said. Consider that this would require a constant audio upload stream to Facebook servers where all the sound files must be stored and analyzed. As discussed in Wired, storing that amount of information would require 33 times more storage than what it currently possesses.

Does Facebook track you even when you are not on their website or app?

Rep. Debbie Dingell brought up the topic on the Facebook Pixel and Facebook Like buttons, while Rep. Ben Lujan brought up the issue of “shadow profiles” and that Facebook collects data of people who are not even Facebook users. Facebook works with data brokers who offer your profile information to the giant, so it’s not black magic.

Facebook Pixel is installed on millions of websites and apps and they don’t even show the FB logo. But through there, it gets your browsing history — what links you clicked on or what you added to your shopping cart. The trial version of many apps live on ads — if the ads come from Facebook’s Audience Network or if the ad provider exchanges info with Facebook, it can learn what apps you are interested in.

As for the data from the brokers, despite being masked by an algorithm, it can still be matched against Facebook account information. That’s how the other half is covered. This data can cover things like your salary, car preference, home size, political affiliations, and spending habits.

Face recognition

This one is no secret, so there were no “questions” around it — it’s a matter of fact that FB runs face recognition. Zuckerberg defended that by bringing up the “Chinese competitors”. While you can opt out the feature, it is on by default. As such, it can scan millions of people and find out relationships, even from people in the background.

Can Facebook do object detection? Like if you are drinking tea or coffee? There are no evidence of this, while it could be a viable next step. Besides, the technology is still evolving.

Location tracking

This was another fact, taken for granted. The FB app connects to your phone’s GPS and tracks your location. It turns out, even anonymous location tracking can be extremely revealing, let alone one that knows who you are!

It’s not just limited to the stores you went to (which reveals a lot of your shopping habits), but with the use of AI, location tracking can reveal where you live, where you go to work, and in some cases even what you are doing. In fact, Google used AI and location tracking to detect illegal fishing. Combine location tracking with a timestamp and it can also reveal whom you have met, what kind of habits you have and what sort of hobbies interests you (it could track you to your gym, right?)

Deleted accounts and data

This one was one of the hotter topics, brought up by Reps. Dean Heller: and Cory Gardner, to which Zuckerberg promised to “follow up”. Until we get that answer from Facebook, here is a fact from traditional database development practices; user data does not get deleted. It only gets “soft-deleted”, i.e. put in a state where it’s not accessible for the public but still sits there in the storage.

Reason? Besides extracting data, the technical side is that users have many other information tied to them. To remove a user account entirely, the system has to go through all related nodes and cut those as well, which in turn could trigger other database actions. In bigger systems, some of these cases might even be missed, and cause bugs and errors in the system. This effort is simply not worth it, so it only gets soft-deleted.

Besides this, “historic data” is necessary — for instance, a user should not be able to register for a 14-day trial system twice — even if they ask the admin to delete their account.

Is Facebook selling our data?

No. Zuckerberg stressed that his company “sells ads”. That is how they make their profit. In fact, Facebook’s platform is especially lucrative for advertisers as it allows them to pinpoint a very specific demographic, while costing less than Google Ads.

The other thing that makes Facebook potentially dangerous is that it is a closed system: a system that can see your current interactions, has control over the content it shows you, and can measure the results of those things is a perfect fit for human behavior optimization. That’s what makes it especially lucrative for political ads. I have discussed that topic in more length here.

Closing thoughts

We’ve tried to give a fair overview of what Facebook might know about you, and how it knows it. But honestly I don’t believe Mark deserved what he got. From simply a data collection perspective, Google dwarfs Facebook. They were just lucky that Zuckerberg became the bullseye for privacy protection movements. In fact, Google has similar (if not more) access to the same techniques and similar resources as discussed above. Let’s hope they learn their lesson from what happened to Facebook.