“Fake News” is a symptom of Bad Epistemology

Over the last few weeks, the topic of “Fake News” has exploded on the internet.

In some respects, I love that this is being discussed, it’s immensely important that we discuss how we avail ourselves to information, and how that ends up influencing our worldviews. However, I kind of wish there’d be a more in-depth discussion about Epistemology in general, as I think divorcing the topic of “Fake News” and Epistemology, is at best, treating the symptoms, and at worst, promoting Bad Epistemology, which might lead to symptoms just as bad as (or worse than) “Fake News”, down the line.

The main issue with the discussion is the focus on what constitutes a credible source. While it’s certainly helpful to treat different sources differently, I fear the rise of a credible/not credible view, where some sources get tagged as credible, while other sources get tagged as bunk. If you like to read Vox, you might tag Breitbart as “not credible”, and if you read Breitbart you might tag Vox as “not credible”, and so, we end up with two warring tribes that refuse to listen to each other, each with their own echo chamber. I don’t find this very interesting however, and I don’t think Bad Epistemology is the major reason the echo chambers are so echo-y.

The real issue is actually way more subtle than opposing tribes refusing to listen to each other, and oftentimes this view on “source credibility” – let’s call it Source-ism – has very little to do with politics or partisanship. The issue is that an over-reliance on whether the source is “credible” or not can lead to evaluating things based on form rather than content, a feature that led to the whole “Fake News” thing to becoming a problem in the first place. Indeed, the Stanford History Education Group recently came out with a report showing that young people have a hard time figuring out the difference between a regular article on a news site, and native advertising. This shouldn’t be very surprising, if native advertising didn’t give off a credible vibe to a lot of people, it probably wouldn’t be used to the extent that it is.

But wait a minute, it really sounds like I’m giving in to Source-ism myself here by implying native advertising isn’t as credible as regular articles on news sites. If one should evaluate things based on content rather than form, isn’t it irrelevant whether an article is an ad or whether it’s a regular article? Yes and no, and that’s kind of why I’m claiming the problem is subtle, and kind of hard to talk about in concrete terms, because there’s always so many caveats.

A news site has its own agenda, incentives or biases (or at the very least, the reporters of said site do), this is simply inescapable. There’s no such thing as “unbiased reporting”, if you’re selecting what specifically it is you’re going to report, you’re adding your bias to it. It’s obviously possible to further one’s agenda by adding in things in disingenuous ways (or just blatantly making stuff up), so all news sites, and all articles, aren’t equal with respect to truth content, even if all are biased. It’s also possible to have an agenda, and to have that agenda because you’re trying to be intellectually honest. In those cases, certainly the agenda itself shouldn’t count as a point against that person and whatever it is that person is trying to say.

So what about native advertisements? Isn’t it possible that whoever paid for the content genuinely believes that whatever it is that they’re trying to push is of value, and that you would be better off if you availed yourself to it? Yes. It totally is. How likely is it though? At the end of the day, you have a finite amount of time, and you have to choose what to read in a world where there’s simply too many options. And of course, this is where it all goes wrong, again. This is a great way for the Vox reader to rationalize his or her reasons for not reading Breitbart, or for the Lakoffian linguist to not read Chomsky. The lyf so short, the craft so long to lerne. You simply can’t engage with everything, and the outgroup is likely wrong anyway, if you had an infinite amount of time you might read whatever nonsense it is they’re spouting.

Ah, yes, and despite my snark, it’s not like I don’t do that myself. In fact, you actually do have to rely on heuristics, even if those heuristics can lead to you being a bit more wrong in some cases. Hopefully though, if you’ve picked good ones and tuned them, you’ll end up being less wrong, overall.

But isn’t there something obvious that I’m missing. Wouldn’t it be easier if we all relied on “Real News” rather than “Fake News”? You’re not a Vox reader, or a Breitbart reader, and if you detect obvious signs of “partisanship” you throw up in your mouth. My answer to that would have to be: Again, yes and no. You might not be able to detect just how biased your “Real News” source is (or perhaps more importantly, what incentives it has), but that probably won’t fit into this blog post, so let’s save that for some other time. Let’s assume though, for the sake of argument, that your news source actually manages quite well to report about the world as it is, do you think it’s possible for a contender to copy its form, and come off as just as credible (or slightly less credible, but still good enough for a lot of people to “fall for it”)? I do. I think that’s exactly how native advertising or “Fake News” works, and that has been the point of this entire post. While you might be wise enough to be able to identify “Fake News” currently, don’t count on that being the case for all time, unless you’ve already thought about Epistemology a bunch (and if you have, you probably realize how thorny a subject it can be).

Also, trying to stay away from partisan sources can end up hurting you in the end. Bias can be corrected for, noise, you can’t do much with. If we call 1067 people on their landline phones and ask them what they’re going to vote for in an upcoming election, the margin of error is 3%, which means that for a given candidate, the percentage you’ve gotten by conducting your poll, will in 95% of the cases not deviate more than 3 percentage points, and is more likely to be closer to the percentage you’ve gotten than to actually be a whole 3 percentage points off. However, you might notice that you ended up with a lot of old people, much more than you think are going to actually show up to vote (old people might just be more likely to still answer polls through landlines). With a good estimate of what the demographics are going to look like on election day, you can try to correct for that, and plenty of other biases that might have slipped into your poll.

The same thing goes for news sources. A “credible news source” might miss out on some important things, or simply report on tons of irrelevant things, while a “less credible news source” might pick up on some of the stuff that the “credible news source” missed. The “less credible news source” can be compared to other news sources, and its agenda can be revealed in that way, and oftentimes corrected for. Since I feel like this is getting a bit too theoretical I’ll use a real example that I kind of didn’t want to use (it might make this post come off as less credible to some): RT is one of the news sources that people point to when mentioning “Fake News”, and they certainly have an agenda, one aligned with the Russian state. However, while they might in many instances report false information, they also have a way bigger incentive than prestigious institutions to lend a voice to western dissidents. Such as, for instance, Julian Assange, who runs an organization that certainly has an agenda, but still provides the world with tons of valuable information. Is the correct route here to tag RT as “not credible” and simply dismiss it? I don’t think so. Simply factor in their agenda and evaluate the content of what it is they are reporting. Compare to other sources. Look for primary sources. What’s the probability that the WikiLeaks e-mails are real for a given leak? What’s the probability of the leaker being the Russian government? Does it matter with respect to the truth content of the leak?

RT and WikiLeaks are quite political examples though, I mentioned earlier that Source-ism didn’t really have anything to do with politics, and it doesn’t, it just happens to be what’s on my mind as I’m writing this. Source-ism shows up everywhere. I recently had a conversation where I linked a Quora answer that I thought was a pretty good summary of a research topic, and sure enough, the person I was chatting with laughed at me for linking him Quora, so I linked a scientific paper outlining the topic instead (same points, just summarized in the Quora version). My intention was for him to read it and evaluate it, not treat it as if I was trying to link to something authoritative, indeed, that’s what I think the cruder versions of Source-ism amount to, an appeal to authority. This is not how reality operates, it doesn’t care about your prestige. What’s true is true, and it’s really hard to model it correctly, and there’s just so much information that we lack, and Source-ism really doesn’t help us. You might think you’ve gotten Epistemology (a real life friend once told me that it’s “retard stuff”), my recommendation to pretty much anyone is to keep thinking about it.