Biden's age and the problem with the misinformation cope

The tendency to dismiss challenges to liberal orthodoxies as "misinformation" is costly.

On Friday morning, I woke up to a consensus among commentators that Biden’s debate performance had been terrible. I then saw the evidence of this in the short clips circulating on social media. The clips are bad—catastrophic, even. To understand how representative they are, I watched the entire debate.

Biden should drop out. This is not because he is the worst candidate. Even focusing just on the debate itself—the whole debate, not just the short clips—Biden’s contributions were generally more coherent and accurate on issues of substance and policy. Trump had the advantage of not appearing old, frail, and forgetful, but that was about it.

Nevertheless, one of the unfortunate things about politics is that you have to act in the world that actually exists. Democracy is not about truth, justice, or fairness. It is about winning votes from citizens who are mostly uninformed and pay very little attention to issues of substance or policy.

In that context, the undeniable spectacle of somebody who looks and speaks like Biden is politically suicidal. As many have pointed out, voters are not just evaluating Biden now but his likely progression over the next four years. Moreover, Democratic Party strategists should be considering the high likelihood that Biden will generate even worse clips over the next few months of campaigning.

These points have been made elsewhere, especially by Nate Silver, whose insight on this topic was vindicated by the debate and its aftermath. (The same cannot be said of his critics). In this uncharacteristically brief post, I will reflect on how this debacle connects to one of this blog's recurrent themes: “misinformation”.

How not to think about misinformation

In December, I had a strange Twitter debate with leading misinformation researcher Sander van der Linden on the topic of Biden’s age and misinformation.

The context of the disagreement was a scientific article on “Visual misinformation on Facebook”. The article aims to challenge an empirical consensus that misinformation is relatively rare on social media by measuring the prevalence of misinformation in images (e.g. memes) rather than text. It finds that 23% of sampled political images on Facebook contain “misinformation” and that visual misinformation is 5-8 times more prevalent among right-leaning political images than left-leaning ones.

Those are pretty striking findings. How did the researchers define and measure misinformation? As is increasingly common among misinformation researchers, they use an extremely expansive definition, which includes placing “facts in a misleading context”.

Here is one of the images they coded as “misinformation”:¹

In my view, classifying this as “misinformation” is absurd. Not only is it a joke, but the joke expresses a reasonable (even if hyperbolic) evaluation of Biden’s cognitive decline.

In response, Van der Linden suggested I believe in “alternative facts” and argued the meme should be classified as misinformation because “it plays into the well-documented false narrative that Biden is senile (notably fact-checked as false)”.

In the debate, we saw how problematic this approach is. As I have argued repeatedly (e.g., here, here, and here), the problem with highly expansive definitions of misinformation—those that focus not just on very clear-cut falsehoods and fabrications (e.g., literal fake news and deepfakes) but content that might conceivably be “misleading” in some way—is that there is no reason to think misinformation experts are well-positioned to apply such definitions reliably or impartially.² Instead, their judgements will—like everyone else’s—be corrupted by bias, partisanship, wishful thinking, and more.

The presidential debate vindicated this assessment. Against a bizarrely popular view that worries about Biden’s age are merely a product of “misinformation” and right-wing echo chambers on social media, it turns out that this view was itself the product of highly misleading narratives circulating around liberal echo chambers.

More importantly, the debate also illustrated how the liberal establishment’s approach to “misinformation” can be very costly. It might be emotionally comforting to classify all challenges to preferred liberal orthodoxies of the day as “misinformation”. It might even serve short-term propagandistic purposes. However, it ultimately makes you look silly in the eyes of a less biased public and erodes trust in social science and liberal media outlets.

Most importantly, it compromises your ability to make good decisions. If elites within the Democratic Party had spent less time dismissing worries about Biden’s age and more time facing its reality, the current political situation would be much less dire.

1. The authors claimed they did not code raw images but included their surrounding “context”. However, in the paper itself they claim that they included the image in their analysis because it fits the “misinformation theme” of depicting “Joe Biden as senile”. They don’t provide their data or say more about what “context” is necessary over and above whether the image fits the theme (or what it even means to fit the theme). They also classified as misinformation “images that targeted Joe Biden’s son Hunter Biden” and “suggesting that Democratic candidates endorsed violence.”

2. Importantly, there was a lot of incontrovertible, clear-cut misinformation in the debate itself, both from Biden—he is *still* pushing the false claim that Trump referred to neo-Nazis as “very fine people”, for example—and in pretty much everything Trump said.