“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Jay Friedman, chief operating officer at Goodway Group.
As the attention on Mark Zuckerberg’s recent testimonies dies down, the conversation is shifting instead to what should have been addressed.
From an ad tech perspective, one would say a lot was missing. And as more people look under Facebook’s hood, the underlying practices that led to the data privacy issues we know about are coming to light.
Some of these issues raise more questions than answers. If members of Congress had asked me to help them prepare their questions for Zuckerberg, here is what I would have suggested.
1. Will Facebook continue to collect data from the ‘like’ button, given these recent data privacy issues and the upcoming General Data Protection Regulation (GDPR)?
One major point of discussion after the hearings has been Facebook’s Audience Network and its practice of following non-Facebook users around the internet. This concern has extended to many questioning whether or not Facebook will continue these practices, especially with the upcoming GDPR. I think we’re conflating two issues here.
I’m not sure if Facebook’s Audience Network is in conflict with GDPR, since Facebook collects this data most often on-property, where it has consent to do so. Facebook is simply buying ad space on another site to leverage that data and doesn’t share the data it has with the publisher in order to place the ad.
The bigger question relates to the “like” button and the “log in with Facebook” feature. It seems Facebook knows no bounds in collecting data on sites outside of Facebook where the “like” button is hit or users sign in to other properties with Facebook’s ID authentication feature.
If users hit a “like” button on The Guardian, they likely do not realize Facebook is capturing their data. I would surmise that if given a choice, users wouldn’t permit Facebook to capture that data, because it was never their intent to interact with Facebook in the first place.
2. To what extent is natural language processing and AI used to parse people’s posts, and to what extent is this used in targeting?
A few months back, I posted a sarcastic comment on Facebook about a consumer-facing wholesaler. I don’t post on Facebook very often and I visit this wholesaler maybe twice a year, spending about $50 each time (it has really good deals on socks – trust me). However, from my post, Facebook may interpret that I am interested in this wholesaler.
Facebook likely has some of the most sophisticated AI algorithms at work, but I wonder if even its AI team has developed AI that can parse through posts that deliver no value but are still included in its algorithmic understanding of me as a person.
Is Facebook’s understanding of each of us actually all that good to begin with, based on what we post?
3. Does Facebook, or any app connected to Facebook, listen to people’s conversations when they are talking to others on the phone or in person?
I think the hearings were a missed opportunity to discuss a phenomenon that I and others have experienced: ads appearing for products just discussed in conversations. I think this would threaten the company in a bigger way than Cambridge Analytica. Does Facebook eavesdrop on our conversations?
Even if Facebook isn’t doing this, a third-party app could be, with or without Facebook’s knowledge.
4. How much does Facebook weigh location data and ecommerce pixel data from advertisers to inform targeting, as opposed to the content generated by users or their off-platform activity?
Facebook recently announced new privacy protections ahead of GDPR, including how the company will get consent to collect data on any posts that are created by users for their friends only. Facebook will still collect data on all posts marked as public or visible to “friends of friends.”
The focus now moves to parsing out the value of Facebook’s data. Ecommerce pixel data from Facebook’s advertisers should, of course, only be used for that advertiser’s purposes, but it is likely mass-aggregated to build bigger models. How much does Facebook weigh this data, along with persistent location data that users don’t likely think about, to inform targeting versus the content a user is generating?
Taking this a step further, what about users’ off-platform activity that they likely don’t even know is helping Facebook build models?
5. What data will Facebook have to give up that could harm targeting or anger users who didn’t know it was being collected?
In all likelihood, Facebook’s changes to its privacy settings will not severely limit Facebook’s pool of data enough to make a difference. But is there a chance that Facebook will have to give up some data that could potentially harm targeting? And could there be a public backlash once users learn about some of the data types that were once collected?
6. How does Facebook use data from scanned images to target ads, and who is this data ultimately shared with?
Another interesting recent development has been a class-action lawsuit against Facebook that has just been given the green light. It claims that Facebook’s photo-scanning technology violated an Illinois law by gathering and storing data without consent. The resulting question: How does Facebook use this data to target ads?
But let’s take this further. Does Facebook use its photo-scanning technology to notice when people are sunburned – or sunburnt multiple times – to note that they may develop skin cancer? If so, who do they share that information with?
It’s unfair to suggest it is sharing that information with insurance companies, but could a third party be getting that data and sharing it with insurance companies? What natural rights do we believe we have that Facebook could be violating? Not just with ad targeting, but also with data sharing?
Senators, I’m happy to share my schedule with you if you need help preparing for another Zuckerberg appearance. I’m usually pretty booked a few weeks out but might be able to move some things around.
Follow Jay Friedman (@jaymfriedman) and AdExchanger (@adexchanger) on Twitter.
