No, Scientists Didn’t Disprove Conspiracy Theories with Math
Support more videos like this at patreon.com/rebecca!
Hey viewers, most of you have probably been around long enough now that it’s time for a test! Take out your pens.
#1: What do we do when we read about a scientific study that conforms perfectly to what we already hoped is true?
Is the answer:
- Share it on Facebook
- Forward it via email to your entire extended family
- Use it as justification to dig up past fights with ideological opponents
- Get super fucking critical about it
The answer is D, of course! And so I hope that none of my skeptical friends have already shared a news story about how a scientist has mathematically proven that most conspiracy theories are impossible to be kept hidden for long. Right? You didn’t share it, right? Look, I haven’t had time to make a video about it until today, I’m sorry. Go delete your Facebook posts. I’ll wait.
I have to say at the outset that I spent most of last week deconstructing that terrible evolutionary psychology study, so I had nothing left over for this one. Luckily, my friend Martin Robbins in Merry Old England was on the lookout, and he wrote a very clear explanation of why this is bullshit. Link is in the doobly-doo.
Basically, the scientist in question is Dr. David Grimes, who claimed to look at various conspiracy theories that have been exposed in order to determine how long that took relative to how many people are in the conspiracy. In theory this could be a fun little exercise. In practice it is an utter mess.
First of all, he uses three data points: the NSA surveillance program, the Tuskegee experiments, an an FBI conspiracy involving falsified forensics tests. Those three points don’t form any recognizable pattern, and to figure out the number of people involved in each one Grimes appears to have just guessed.
So we have only three data points for conspiracy theories that have been revealed. Can you guess what other data we’re missing? That’s right: conspiracy theories that haven’t been revealed. This is, essentially, the toupee fallacy. The toupee fallacy refers to a person claiming that they’ve never seen a good, natural-looking toupee. The person assumes any bad-looking hair is a toupee, and doesn’t consider all the natural-looking toupees that they never noticed because, well, they looked natural. They have a large amount of missing data.
By necessity, Grimes can only study the toupees he’s noticed, so by definition, all those are going to be bad toupees.
You guys know I’m know fan friend to conspiracy theorists — check out my video on how the attacks in Paris actually happened for more info on that. Don’t read the comments. Actually, read the comments. There’s an interesting sociological study to be done there.
But agreeing with the conclusions of a study doesn’t mean the study is a good one. So while I agree that the more people involved in a conspiracy and the more time passes the more likely it is that the conspiracy will be revealed, I don’t think Grimes has discovered the mathematical model for it. In fact, as Robbins points out in his article, Grimes commits a very basic math error at the start that throws off his entire model. It’s not good science, and it’s not going to win you any arguments with conspiracy theorists.
One last thing I’ll add to Robbins’ commentary: he points out that PLOS one should be ashamed of themselves for publishing this and of their peer reviewers for missing all the problems with the study. I agree, but I’ll add that the fact that the study was published in PLOS One is part of what made it so easy to be found lacking. PLOS One is an open-access journal, which means that anyone can go there right now and review the data. Were it locked away behind a paywall, it’s less likely that it would have been discovered to be bullshit. Compare that to the bad evo psych paper I discussed last week, which took all the scientists on Skepchick with journal access a full day to even find a copy of online.
I guess another way of putting this is that the more people who get to see a bad scientific paper, the less of a chance it has of propagating. Now if only someone could make a mathematical model about that.