Rethinking Information Disorder via Argumentation Theory Lens

A short note about something I realized just now.

I was thinking about the Information Disorder types chart. For those not familiar it looks like this:

In general the horizontal arrangement of this chart has tended to be read as intent, which is in line with the larger report. That is, as you move from left to right on this chart you move from unintentional and sloppy to outright lies, from a category some call “misinformation” to one some call “disinformation”. And this is somewhat problematic, because reading intent is hard. First, it’s difficult or impossible to know the intent of a person. Second, the the intent of an initial source may be different than the intent of someone who is sharing. And so on. Wardle and Derakhshan, the authors, know this of course, but it’s always been an issue.

At the same time, when you look at this progression it seems like it represents some factor of intensity or clarity of transgression. We see fabricated content as different than satire, even when we don’t know why a person might be sharing it for certain.

One way to think about it is through the five points of argument failure that I’ve talked about in the past.

In the above chart, instead of talking merely about a claim, we have evidence (grounds) that is introduced into an often unstated argument. That evidence can fail a couple of ways, but there is one big division we can point out. Circle 1 (relationship of evidence to reality) denotes cases where the relation between the evidence and reality is broken. The defining characteristic of a Type 1 failure is that the evidence is so fabricated, misrepresented, or manipulated that it is invalid evidence for any argument. Type 3 (relation of evidence to argument) failures are a bit different. This is a case where when the full story of the evidence is known it is not good evidence for this specific argument.

Take an example from an elections context. Someone creates a fake video of ballots being burned. One way to think about this sort of thing is that the content is intentionally fake, and therefore more severe. But another way to think about it is that there is you can judge whether it is good evidence without knowing the argument for which it is being marshalled. It’s fake after all, I don’t need to know what the argument is to judge its validity as evidence.

On the other hand, consider a video such as the one that led to the harassment of Ruby Freeman and Shaye Moss in 2020, where they were seen pulling out security bins of ballots and counting them as part of normal procedure. That was described to people as being “suitcases of ballots”. This video was far more damaging, and just as dishonestly used as fake videos, at least by some. But in this chart it would fall in the middle, under false context.

My point is not that the chart is wrong, but that the chart may capture something a bit different than simple intent. The difference between a faked video of ballot burning and a video of pulling out security bins is that in the second case you need to know the argument being made to assess the evidence. If you were making the argument, for example, that the process was underfunded and understaffed and you referred to those boxes as “suitcases”, it probably wouldn’t be a big deal. It could be good evidence of that, maybe. But if in the context of talking about election security you neglect to mention these are security boxes, properly sealed and stored, then this is misleading.

When seen this way, you can read a chart like this, starting on the left with evidence that is good enough for some arguments to evidence that is good enough for no arguments.

In other words, it’s a gradient of the necessity of knowing argument context to judge validity of evidence, NOT a gradient of the relation of malicious intent to a claim. This also explains why so many have found this useful in moderation contexts — it’s not so much that intent determines action — intent is unknowable. But rather, the more that validity judgments are grounded in an implied argument context, the less cut and dry such decision may be, as they rely on being able to demonstrate an often unstated argument context into which the evidence is being marshalled.