Did Facebook hack Science? Cheating academics and regulating transnational digital utilities

Has Facebook hacked Science? I wouldn't call the 2023 Science and Nature studies of Facebook's impacts "a debacle" like this otherwise-excellent overview of the present state of Facebook's play does. That link is to an article in The Conversation by professor Timothy Graham of Queensland. I obviously liked the original studies, as can be seen by the fact I feature them prominently in this blog. 

But it was entirely evident to me as soon as they came out that what those studies showed was: 

• Facebook can have positive impacts, and did while being watched. 
• So it must be watched. Constantly. 

Yet Nick Clegg said at the time that Science had shown Facebook is safe, and so such expensive research was no longer needed. This is in the worst case a lie, and in the better case a standard fallacy of the digital era. No one should ever trust AI, or any digital product, because they can always be updated and changed. Rather, the procedures scientists like Andy Guess initiated should be viewed as pilot studies for what must be continuous monitoring of sizeable social media networks as a part of their regulation. 

Note that the vulnerability of digital products to change and misuse is significantly raised if governments compromise cybersecurity by banning encryption. Then you are worrying not only about the deployers of the product, but anyone else who has managed to hack into it, including criminals and hostile governments.

Speaking of not trusting AI (or any digital service), here's an article I wrote for the UN's university some years ago, titled No one should trust AI. Why not? It's just not the kind of thing you trust. It's not another person. It's a digital product – you regulate products. Governments have to enforce liabilities when those products are inadequate. That's a big part of what the EU's digital governance legislation – including but by no means limited to the AI Act – is about. Facebook might even be more than a product. It might be a transnational digital utility. We need to learn to regulate those too. More about that in coming years...

But in short, scientific results on digital products only hold for the periods when they're being observed. So constant observation should be a part of ordinary regulation. 

A view from the top. Ironically, when I've visited facebook, I've never taken pictures of people.

BTW, here's a collection of all the papers I can find (and a book I found that found a lot more) about whether there's a link between social media use and polarization. Science failing to find correlations between social media use and affective polarisation.