Finally, science knows why right-wingers get suspended from social networks
Finally, science knows why right-wingers get suspended from social networks
Plus does OpenAI infringe copyright, or not?
The US election is nearly upon us. I know: you’re sick of hearing about it. I know I am, because rather than parcelling their presidential election into a neat six-week event where politicians offer policies and then people vote for (or not for) them, they’ve allowed it to metastasise into a monster that goes on literally for years. It’s exhausting, and the frenzy on social media as the Event Itself approaches has only intensified in the past eight years.
In observing all the madness, one tends—from across the Atlantic—to think that there’s a certain inequity between the two sides. One looks a hell of a lot more rational and, if we’re honest, truthful than the other.
Things like this:
We occasionally get the same sort of thing in the UK, but I can’t immediately think of an example of a politician taking credit for a big project that they specifically voted against. Still, hooray for Community Notes, right? Apart from the fact that Notes aren’t that great at timely or targeted updates, as the Wikipedia entry points out:
In November 2023, the Atlantic Council conducted an interactive study of Community Notes highlighting how the system operated slowly and inconsistently regarding Israel and Gaza misinformation. In one example, an image originally received a Community Note but continued to spread regardless receiving over 3 million views after a week. Hundreds of viral posts from the notes public database were analyzed and according to researchers fast-moving breaking news wasn't labeled. Across 400 posts of misinformation, a note took on average 7 hours to appear, while others took 70 hours. The analysis however did show that over 50% of the posts received a note within 8 hours, with only a few taking longer than 2 days. The study included 100 tweets from 83 users who had signed up to X Premium in the past 4 months, along with 42 tweets from 25 accounts that were reinstated by Elon Musk, including Laura Loomer.
But your feelings about the disparity in truthiness between the two sides has now been confirmed by a study published recently in Nature: “Differences in misinformation sharing can lead to politically asymmetric sanctions”. Here’s part of the abstract:
We first analysed 9,000 politically active Twitter users during the US 2020 presidential election. Although users estimated to be pro-Trump/conservative were indeed substantially more likely to be suspended than those estimated to be pro-Biden/liberal, users who were pro-Trump/conservative also shared far more links to various sets of low-quality news sites—even when news quality was determined by politically balanced groups of laypeople, or groups of only Republican laypeople—and had higher estimated likelihoods of being bots. We find similar associations between stated or inferred conservatism and low-quality news sharing (on the basis of both expert and politically balanced layperson ratings) in 7 other datasets of sharing from Twitter, Facebook and survey experiments, spanning 2016 to 2023 and including data from 16 different countries.
So: more likely to be bots, more likely to be suspended (at least, pre-Musk), more likely to share low-quality content. (The first and last probably go together frequently.) The researchers, from MIT, Cornell, and Yale University, note that the natural outcome of this is that if you actually have a working system for suspending people who create trouble, then you’re going to be suspending the pro-Trump crew more than the pro-Biden (well, pro-Harris now) crew. Which is exactly what was observed. Which led to more complaints from the pro-Trump crew, including Trump himself, about pro-Trump people being suspended for expressing themselves, which has the potential to spiral out of control unless someone tamps down all the madness.
The raw numbers though are pretty dramatic:
First, accusations of political bias are based largely on anecdotes or salient unique cases, such as the suspension of former President Donald Trump. Our data allow us to evaluate these claims more systematically. Indeed, we find that accounts that had shared #Trump2020 during the election were 4.4 times more likely to have been subsequently suspended than those that shared #VoteBidenHarris2020. Specifically, whereas only 4.5% of the users who shared Biden hashtags had been suspended as of July 2021, 19.6% of the users who shared Trump hashtags had been suspended.
Those seem to me really big numbers, both the difference and the raw percentages who were suspended. But then you take a look at the difference in trustworthiness of the content they share. In the diagram below, from the paper, the lower number is better (ie sharing more trustworthy content).
You can see the dramatic difference here. As much as anything this is quite depressing: why has the right wing in the US fallen into consuming falsehoods, or at least is doing it more often than the other side of the political aisle1? I don’t have a quick answer, but it’s evident that the toxic combination of the First Amendment and the removal of the fairness doctrine under the Reagan administration in the 1980s meant that Mad Junk could proliferate. And one of the first lessons of social warming is that outrage spreads faster and further than normal content (even the positive stuff), and that outrageous lies are much easier to find and spread than outrageous truth. Quite why it would be the right wing that would fall prey to this process is less obvious, but it’s notable that Trump’s strongest support comes from those without a university/college education:
In 2020, according to CNN’s exit polls, voters with a college degree accounted for 41% of the electorate and they supported President Joe Biden 55% to Trump’s 43%. Trump got the support of about two-thirds of White voters without a college degree, but he lost White college-educated voters.
(Note too, wearily, that education level was the strongest predictor for vote choice in the Brexit referendum.)
One might thus draw a connection between lower education level and choice of candidate—though of course there are still that one-third of white voters with a college degree who did vote for Trump. (People like Bill Ackman, one supposes.) The CNN story linked above points out that the swing states are, well, swing states have hit that Goldilock point: “not skewing too much [towards] college-educated voters and not too much non-college”, according to one expert in the story.
But that, unfortunately, means that we are left biting our fingernails. Even so, we do have the faintly comforting thought that the people who are backing Trump are more likely to be wrong about things, sharing junk, incapable of discerning between the truth and untruths.
That however isn’t going to be a huge amount of help if things go pear-shaped after Tuesday 5th. But the Nature paper is an entertaining read. I commend it.
Glimpses of the AI tsunami
(Of the what? Read here. And then the update.)
• Former OpenAI researcher says the company broke copyright law. Well, maybe, but OpenAI doesn’t agree, and neither do its lawyers.
• AI search engines infused with “scientific racism” in search results. The crapification of the web continues.
• AI analysis could stop the spread of health misinformation. Great if true. But it always depends on the training set.
• Runway releases new AI facial expression motion capture. Act One, as it’s called, “allows users to record video of themselves or actors from any video camera — even the one on a smartphone — and then transfers the subject’s facial expressions to that of an AI generated character with uncanny accuracy.” We are going to see so much faked film of real people where their facial expressions have been captured from somewhere else.
• You can buy Social Warming in paperback, hardback or ebook via One World Publications, or order it through your friendly local bookstore. Or listen to me read it on Audible.
You could also sign up for The Overspill, a daily list of links with short extracts and brief commentary on things I find interesting in tech, science, medicine, politics and any other topic that takes my fancy.
• I’m the proposed Class Representative for a lawsuit against Google in the UK on behalf of publishers. If you sold open display ads in the UK after 2014, you might be a member of the class. Read more at Googleadclaim.co.uk. (Or see the press release.)
• Back next week! Or leave a comment here, or in the Substack chat, or Substack Notes, or write it in a letter and put it in a bottle so that The Police write a song about it after it falls through a wormhole and goes back in time.
Then again, from the broader European perspective, the two political parties in the US both look like unhinged rightwing projects where the only difference is of degree. Though some of the stuff the Democrats are doing is definitely quite loopy in the other direction.