Thread by @AlexanderAbdo

Yesterday, @ProPublica reported that Facebook has modified its site in a way that prevents journalists from studying political ads on the platform.

It’s an important story that shows Facebook’s hostility to journalism about Facebook.

Let me explain. /1

propublica.org/article/facebo…

First, it’s essential that journalists and researchers be able to study Facebook’s platform, so that we can all answer some basic questions: Is Facebook distorting public discourse? Is it deepening our social divisions? Is it being highjacked to manipulate our elections? /2

These are existential questions for society, but FB hasn’t been forthcoming with details or data. And even if it had, we’d still need *independent* journalism and research to verify FB’s claims.

Why? Remember this story about Volkswagen? /3

bbc.com/news/business-…

Second, studying Facebook is hard. The platform relies on machine-learning algorithms to decide what to show you and in what order. Studying Facebook’s effect on public discourse means studying those machine-learning algorithms. /4

But machine-learning algorithms are black boxes. To study them, you need the ability to collect their *outputs*.

It’s like Plato’s Cave. We can’t see the object of our study directly, so we study its shadows. Here, the shadows are the algorithmic outputs. /5

Third, Facebook’s terms of service prohibit the kind of data collection that would be necessary to study its algorithms. There are good reasons to block data aggregators, but FB’s terms apply equally to bona fide journalism and research carried out in privacy-preserving ways. /6

That’s why we (@knightcolumbia) proposed that Facebook adopt a “safe harbor” for journalism and research on the platform that complies with safeguards against abuse.

We’re still in negotiations with FB, but so far they’ve resisted our proposal. /7

nytimes.com/2018/08/07/us/…

Fourth, in the absence of a safe harbor for their work, journalists and researchers have gotten creative in collecting data they need to study FB. They’ve asked the public to voluntarily contribute the data, using browser plugins that you can install on your own computer. /8

ProPublica has a political ad tool. @kashhill at @Gizmodo has one for studying FB friend recommendations. @mozilla also has one for political ads. /9

gizmodo.com/keep-track-of-…

blog.mozilla.org/netpolicy/2018…

Facebook has told ProPublica and Gizmodo that these tools violate the terms of service and should be taken down, even as it has acknowledged the public interest in the reporting the tools have generated. /10

gizmodo.com/facebook-wante…

That’s the background for Facebook’s recent decision to kill ProPublica’s political-ad collector with an update to the site. /11

In defending the decision, Facebook told ProPublica that the update was “routine” and focused on preventing “ad blocking and ad scraping.” @alexstamos, FB’s former chief security officer, has made a similar argument.

I don’t buy that, and you shouldn’t either. /12

What’s wrong with Facebook’s explanation? There are two key problems. (Sorry for another list.) /13

First, FB says it's blocking ad blockers, but it’s worth dwelling on the fact that this is a commercial interest. FB is entitled to protect those interests, but we should recognize that, in a balance b/w important journalism and commercial interests, FB chose the latter. /14

How significant is that commercial interest? How many ad blockers actually took advantage of the capability FB just disabled? Was it a real problem?

The answers to these questions would help explain how much Facebook values the journalism it just obstructed. /15

Second, it’s key to note that FB's change isn’t about privacy.

FB has prevented *users* from automatically downloading ads shown to them and the explanations for why they’re seeing those ads. FB is limiting user control, not respecting some privacy decision made by users. /16

FB says that ad scraping can expose users’ information to malicious actors, but this explanation seems bogus.

Based on our own conversations with FB, the concern seems to be that malicious actors will convince users to download malicious browser plug-ins. /17

This is pretty far-fetched. If someone convinces you to download a malicious browser plug-in, you probably have a lot more to worry about than the malicious scraping of the advertisements you see.

And yet, ad scraping seems to be all that FB's recent change targets.

Why? /18

The upshot is that FB is controlling what we learn about FB. It says it values journalism, but it obstructs journalism in favor of curated data sets it is slowly releasing.

That’s not a commitment to transparency. That’s a commitment to controlling the narrative.

END THREAD