I recently saw an edited clip on YouTube in which a person said something that was false but very funny in response to a question. In the comments, someone said that the person in the video didn’t respond to the question that way, and that the clip was just edited to make it seem that way. I felt somewhat disappointed that the joke had been contrived by the editor.

This was a clip from a livestream, which is a common way of consuming this content. Streamers rely on clip channels to popularize their channels. Clippers vary in how “fancy” or “slick” their editing is. Some have high production value with added graphics and sound effects, others are more or less curators of unedited stream excerpts. The point is that clippers have a certain amount of artistic license to make jokes land more effectively, insert jokes of their own, and so on. What I described in the example above, however, I thought was probably going too far so as to be putting words in someone’s mouth.

To my surprise, another commenter chimed in saying that this was incorrect, and the person in the clip really did say that in the original stream. The edited clip wasn’t misleading at all.

It was at this point that I realized that I believed this second commenter, but I had no good reason to. Before seeing their comment, I also believed the first commenter without having a good reason to. Neither “side” really presented evidence, but all I would have to do to find out the answer is watch the VOD (archived video) of the original stream. I decided not to.

The thing is, it’s not really important to me what happened in this situation. True, if the clip was taken out of context in a misleading way, I would disapprove, but I wouldn’t do anything about it or change how I act in any way. It’s just not important. What’s interesting to me, though, is that I think I always trust whoever spoke last the most. This is fallacious reasoning. It is essentially an argument from ignorance.

Assertion, testimony, and evidence

The situation I described is rare in that the truth can easily be found out for oneself. There’s no need to rely on what YouTube commenters say. Most of the time, we have to take someone’s word for it what happened. I think that humans probably evolved to mostly trust one another because sharing information was a critical skill for our hunter-gatherer ancestors. When someone gives you information, you are likely to think it’s true unless you have evidence to the contrary. If someone else tells you the first person was wrong, you’re likely to take this as evidence they’re wrong.

The reasoning goes like this:

1) The assertion of P is evidence that P is true unless there is otherwise evidence that P is false.
2) Person A asserts P.
3) From (1) and (2), there is evidence that P is true.
4) Person B asserts not P.
5) From (1) and (4), there is evidence that P is false.
6) From (1) and (5), there is no evidence that P is true.

The problem, of course, is with premise (1). Framed in this way, it may seem obvious that this premise is false. Whether an assertion is evidence shouldn’t depend on the existence or knowledge of different evidence. Moreover, an assertion is not evidence. Let’s consider how this works and why we might think this way.

We know that lying is easy. Virtually every human has lied at some point. Sure, it can be hard to get away with it, but the actual act of lying is extremely easy. It’s also very common to be mistaken. So we know that, in general, it’s possible for something someone is telling us to be false. Depending on our experience with the person, the nature of the assertion, and our own upbringing and physiology, we put more or less stock in what they say. The brain does this pretty automatically. For practical reasons, it’s necessary to quickly estimate how likely a statement is to be true, and this is not a logical process but a heuristic one.

This way of doing things was highly adaptive in the ancestral environment, but human communication in the internet age is very different, and our faculties aren’t really made for it. This is why I had my initial belief that the first commenter was correct and my subsequent belief that they were incorrect. Instead, we can try to work out what we have evidence for logically.

First, an assertion is not evidence. Let’s unpack this. An assertion of a proposition is an utterance or writing that is a statement of that proposition. “The sky is blue” is an assertion that the sky is blue. Any proposition we can imagine whatsoever can be easily asserted regardless of truth value or anything else, so the fact that something has been asserted tells us nothing about its truth value. One might respond by saying that truths are asserted more often than falsehoods, so that assertion gives us a statistical reason to believe the proposition might be true. However, even if truths are asserted more often than falsehoods, I don’t think we can use induction like this. All it tells us is that any assertion is more likely true than false. Going back to the comment example, the two commenters’ contradictory assertions would both be more likely true than false. It doesn’t provide evidence for one assertion over another, in particular for any assertion over its negation.

“But,” one might say, “someone merely telling me something can give me a good reason to believe it. If my neighbor tells me they saw a possum, I’m justified in believing them. We even use witness testimony as evidence in court.” This is true, and it’s worth investigating what is different here. There is a difference between an assertion and testimony or what might be called a report (not just talking about legal testimony). Assertion is when someone states that a proposition is true. Testimony is when someone recounts their personal experience with a situation. This can be confusing since both consist of declarative statements.

Let’s consider the example of the existence of god. Some religious apologists have said that them merely claiming that god exists is enough positive evidence that the burden of proof shifts to the atheist to disprove it. The main way I have seen atheists respond is by pointing out that the assertion of a proposition is not evidence for that proposition. Indeed, this logical move doesn’t work, and ultimately only serves a rhetorical purpose. Contrast that with someone’s testimony of their personal experience of god. This is evidence for the existence of god, and does have to be addressed by the atheist.

The fact that assertions are not evidence doesn’t mean that all statements are equally likely to be true or false. That being said, the distinction is not always clear. Rather than saying a certain utterance is either an assertion or testimony, it makes more sense to interpret it as being (sometimes) both. The extent to which a statement asserts something to be true has no bearing on whether that thing is likely to be true. The reliability of testimony depends on the person, your relationship with them, and the context.

When the commenter said that the clip was misleading, they were implicitly providing testimony that they either saw the original themselves or heard about it from a source they deemed reliable. Since whether it was misleading is strictly a posteriori, the only reasonable way a person would claim to have this information is if they believed they or someone they trust observed it. Contrast this with, for example, things people assert after claiming to figure them out a priori such as many philosophical propositions.

In this context, the testimonial value of these YouTube comments is basically zero. These are pseudonymous strangers on the internet that I have no reason to trust and are very likely to be lying or mistaken, much more so than any other ordinary conversation I might have with a person. My brain, however, is adapted to an environment where people are generally trustworthy. I (and probably most people) am naturally overly credulous towards things said online.

In particular, the way this credulity plays out means that whoever has the last word is disproportionately trusted over other, equally-(un)trustworthy strangers.