We can leverage Section 230 to limit algorithmically driven disinformation

Mark Zuckerberg. Photo (cc) 2012 by JD Lasica.

Josh Bernoff responds.

How can we limit the damage that social media — and especially Facebook — are doing to democracy? We all know what the problem is. The platforms make money by keeping you logged on and engaged. And they keep you engaged by feeding you content that their algorithms have determined makes you angry and upset. How do we break that chain?

Josh Bernoff, writing in The Boston Globe, offers an idea similar to one I suggested a few months ago: leverage Section 230 of the Telecommunications Act of 1996, which holds digital publishers harmless for any content posted by third-party users. Under Section 230, publishers can’t be sued if a commenter libels someone, which amounts to a huge benefit not available in other contexts. For instance, a newspaper publisher is liable for every piece of content that it runs, from news articles to ads and letters to the editor — but not for comments posted on the newspaper’s website.

Bernoff suggests what strikes me as a rather convoluted system that would require Facebook (that is, if Mark Zuckerberg wants to continue benefiting from Section 230) to run ads calling attention to ideologically diverse content. Using the same algorithms that got us into trouble in the first place, Facebook would serve up conservative content to liberal users and liberal content to conservative users.

There are, I think, some problems with Bernoff’s proposal, starting with this: He writes that Facebook and the other platforms “would be required to show free ads for mainstream liberal news sources to conservatives, and ads for mainstream conservative news sites to liberals.”

But that elides dealing the reality of what has happened to political discourse over the past several decades, accelerated by the Trump era. Liberals and Democrats haven’t changed all that much. Conservatives and Republicans, on the other hand, have become deeply radical, supporting the overturning of a landslide presidential election and espousing dangerous conspiracy theories about COVID-19. Given that, what is a “mainstream conservative news site”?

Bernoff goes so far as to suggest that MSNBC and Fox News are liberal and conservative equivalents. In their prime-time programming, though, the liberal MSNBC — despite its annoyingly doctrinaire, hectoring tone — remains tethered to reality, whereas Fox’s right-wing prime-time hosts are moving ever closer to QAnon territory. The latest is Tucker Carlson’s anti-vax outburst. Who knew that he would think killing his viewers was a good business strategy?

Moving away from the fish-in-a-barrel examples of MSNBC and Fox, what about The New York Times and The Wall Street Journal? Well, the Times’ editorial pages are liberal and the Journal’s are conservative. But if we’re talking about news coverage, they’re really not all that different. So that doesn’t work, either.

I’m not sure that my alternative, which I wrote about for GBH News back in June, is workable, but it does have the advantage of being simple: eliminate Section 230 protections for any platform that uses algorithms to boost engagement. Facebook would have to comply; if it didn’t, it would be sued into oblivion in a matter of weeks or months. As I wrote at the time:

But wouldn’t this amount to heavy-handed government regulation? Not at all. In fact, loosening Section 230 protections would push us in the opposite direction, toward deregulation. After all, holding publishers responsible for libel, invasions of privacy, threats of violence and the like is the default in our legal system. Section 230 was a regulatory gift, and it turns out that we were too generous.

Unlike Bernoff’s proposal, mine wouldn’t attempt to regulate speech by identifying the news sites that are worthy of putting in front of users so that they’ll be exposed to views they disagree with. I would let it rip as long as artificial intelligence isn’t being used to boost the most harmful content.

Needless to say, Zuckerberg and his fellow Big Tech executives can be expected to fight like crazed weasels in order to keep using algorithms, which are incredibly valuable to their bottom line. Just this week The New York Times reported that Facebook temporarily tweaked its algorithms to emphasize quality news in the runup to the election and its aftermath — but it has now quietly reverted to boosting divisive slime, because that’s what keeps the ad money rolling in.

Donald Trump has been crusading against 230 during the final days of his presidency, even though he doesn’t seem to understand that he would be permanently banned from Twitter and every other platform — even Parler — if they had to worry about being held legally responsible for what he posts.

Still, that’s no reason not to do something about Section 230, which was approved in the earliest days of the commercial web and has warped digital discourse in ways we couldn’t have imagined back then. Hate speech and disinformation driven by algorithms have become the bane of our time. Why not modify 230 in order to do something about it?

Comments are open. Please include your full name, first and last, and speak with a civil tongue.

9 thoughts on “We can leverage Section 230 to limit algorithmically driven disinformation

  1. Thanks for your reply, Dan. I’ve always found you to be a reasonable voice.

    I’ve never said that MSNBC and Fox News are equivalents. The fact that I mention them in the same context doesn’t mean I think they are the same.

    My main point is that if Fox News gets free ad space to pitch liberals, it’s not going to put the wacked out nutty stuff in those ads. In general, this will generate some more moderate content on that site. And I think it would be healthy to give CNN or MSNBC a chance to pitch content intended to attract conservatives. (As it says in the op-ed).

    My proposal will slip neatly and quickly into the current system that these sites have. By contrast, regulating algorithms is uncharted territory. And removing all targeting will significantly erode the sites’ utility — and is the proposal most likely to generate the most intractable opposition. It could never come to pass. My proposal, on the other hand, could be implemented by the FCC on day one of the new commissioners’ terms.

    I appreciate the debate. There is no one answer. But my hope is that my proposal will move things forward more effectively than the shouting, screaming, and Section 230 bashing that pervades the current discussion.

  2. Steve Ross

    Good idea, at least to start with. Two points:

    1. Facebook, Google, Bing, Yahoo, Apple, can and do massively nudge the AI algorithms by tuning preferred search terms, such as “illegal aliens ” rather than “asylum seekers” (which is what most of these folks are, legally.

    2. A better term for AI is ML, for machine learning. The routines are powerful for categorizing data records, but just provide a gloss to the goal of getting a profitable suggestion out there.

    This is because bots and term-shading and paid influences and on and on put their thumbs on the scale just as live catagorizers do. My own continuing AI (mainly an ML technique called random forest) study of broadband in rural areas gives very different conclusions with small data adjustment. It proves that the data from FCC itself is not predictive, and not worth collecting under current regulations. Medical AI algorithms shaping diagnosis and treatment decisions usually under-treat Blacks because they have not been receiving ideal care in the first place.

  3. This is the second Section 230 piece I’ve seen today.
    The first is Dean Baker (Center for Economic Policy Research) here:
    https://cepr.net/getting-serious-about-repealing-section-230/
    where he claims repeal of 230 “would force Facebook to restructure its ad selling process. If people were posting ads that could potentially cost it millions of dollars from losing a libel suit, it would almost certainly see a need to review ads before they are posted.

    This is likely to mean hiring thousands, or even tens of thousands, of people to review ads for potentially libelous material. These additional hires will be a major expense to Facebook. Also, it will have to turn down many ads, because they do contain potentially libelous material. That means both considerably higher expenses and lower revenue from fewer ads. “

    1. Dan Kennedy

      There’s no question that Facebook couldn’t function at anywhere near its current level of profitability if Section 230 were repealed or substantially modified. Fine with me.

  4. Why not simply enforce current antitrust law and break up Facebook and other gigantic tech monopolies? Seems to me their sheer size and scale, combined with information moving at the speed of light, is the (pun intended) BIGGEST factor in making them anti-social networks. Break up Facebook and Google, at the very least.

  5. JR Stephens

    “Liberals and Democrats haven’t changed all that much. Conservatives and Republicans, on the other hand, have become deeply radical…” Only a true liberal, living in his ivory tower-with no view of the real world-could pen that statement. Please remind me who protested illegally, looted, burned and destroyed our cities this spring and summer. Oh yes, that doesn’t count, because they were progressive, liberal Democrats responding to injustice. Two wrongs STILL don’t make a right!

Comments are closed.