"Regulatory capture is a form of government failure which occurs when a regulatory agency, created to act in the public interest, instead advances the commercial or political concerns of special interest groups that dominate the industry or sector it is charged with regulating." [0]
Even if he doesn't get to dictate the regulation (thus not achieving regulatory capture), it would still accomplish the mission of keeping the business alive. They will continue selling their ads and making money, regulation or no regulation.
If he specifically described what a transparency report would say, 99% of hot takes would be about how it's wrong or whatever. People's brains would turn off and the opportunity for a conversation would be squandered.
There's an opportunity here for more sincere activists to promote meaningful transparency, because now he's put it on the radar of otherwise low-tech-literacy audiences.
For me, true transparency is being able to look at code. If your code is interpretable, and it looks okay, it is okay.
When it's difficult to interpret, like with a collaborative filtering algorithm used to select newsfeed items, an independent group ought to be able to query the "baked" system or measure user behavior and come to its own conclusions. I wouldn't call this a "transparency report," it's more like an "independent analysis."
It would take a while to determine what questions to ask. But to actually answer them would take an afternoon: a database connection, really. After all, they have incredibly comprehensive user behavioral data, so you can pretty much answer anything you can think of.
There's no cost burden or competitive disadvantage. Their entire tech infrastructure is built around answering user behavior questions easily. And most of the value is tied up in the data, not the models.
The problems they're experiencing are related to their particular implementation of collaborative filtering. If outside groups could see that code (at least), it would be easier to say, run a simulation, and demonstrate what kind of user behavior (or enemy action as it were) makes 10% of people miss 99% of "benign user content" and see 1% of enemy action content.
Maybe you don't believe it's implementation related, and you think it's something innate to social media. Well, what if the newsfeed just showed content randomly? Clearly that would bury enemy action content, just because it's so relatively infrequent. It's important not to be too dogmatic with this sort of stuff, because it will obscure your ability to reach across the aisle to stakeholders like Facebook and get them to do stuff in a sincere way.
So at its most basic level, sharing the code can't possibly lead to regulatory capture as you describe it. It would be more advanced than the status quo, which is, "Whatever the hell Facebook wants you to hear."
I think this should really give a strong impression. Venezuela went from one of the most prosperous nations in the world to what is likely now the worst place in the world for a poor person to live in, and it only took a few decades. There's no reason the US is immune to the same fate, and it started long before socialism with things like regulatory capture.
"The Venezuelan economy has a profusion of restrictions and impositions on businesses, particularly in the areas of labor and new business formation. Only the big, well-established companies can afford the high regulatory burden. This is consistent with George Stigler’s (1971) insight on regulatory capture, which abets government creation of monopolies stymieing competition. Further, the high levels of informal or underground economic activity in Venezuela are consistent with existence of a high regulatory burden (see Diaz and Corredor 2008). Unsurprisingly for such a setting, corruption is rampant." [1]
[1] https://econjwatch.org/articles/venezuela-without-liberals-t...