Dunning Krugered
Greetings patrons,
It’s time for another research update. Today, I muse about three things:
- Two years on Patreon
- Interested in Discord?
- The Dunning-Kruger effect as auto-correlation
Two years on Patreon
On a whim back in March 2020, I setup a Patreon account. The idea was that it would be nice to earn some money from blogging, and that maybe … just maybe, it would turn into a full-time income.
From the start, I’ve had mixed feelings about the Patreon model. What Patreon sells is the chance for creators to earn a living doing what they love. That opportunity is obviously a good thing. And for me, Patreon has worked well. My support is growing steadily, and I’m extremely grateful for that.
On the other hand, the crowdfunding model is basically a recipe for inequality. All crowdfunding sites sell a dream — the dream that anyone can earn a living doing what they love. But the reality is that only a tiny fraction of creators will ever earn a living from crowdfunding.
The reason that crowdfunding generates inequality is simple: it creates a positive feedback loop between fame and fortune. The more famous you are, the more funding you attract. Now, this feedback operates in many aspects of life. But crowdfunding implements it in the purest form.
For this reason, I’ve always had reservations about crowdfunding. Because of these qualms, I started studying Patreon immediately after I joined it. (I wrote a script to scrape Patreon’s public data.) The data that I gathered are pretty shocking. The distribution of patronage is unbelievably unequal. By my estimates, the Gini index for the number of patrons per creater is about 0.97. (The maximum Gini index is 1.) Put another way, the top 1% of creators have about 65% of all patrons.
When I first looked at these numbers, I thought that the problem was at the top. And in part it is. On Patreon, the top creators have thousands of patrons. That sounds like a lot. But when you look at the actual incomes of top creators, they’re not that large — at least when judged against the wider backdrop of capitalism. Suffice it to say that there are no Patreon billionaires.
So if no one is getting obscenely rich on Patreon, how can inequality be so great? The answer lies at the bottom of the distribution. On Patreon, the vast majority of people earn nothing. In other words, they create an account, but never attract any patrons.
Now, for most people, this outcome is not a big problem because they have other jobs. In other words, they join Patreon with the dream of quitting their day job, but it never pans out. That’s no tragedy. But it would be a tragedy if we made crowdfunding the core model of reimbursement. In that case, the vast majority of creators would literally starve.
So as someone who studies inequality, I can say with certainty that crowdfunding alone is not a good way to fund artists (or, in my case, scientists). What is a good model is to combine crowdfunding with a universal basic income (UBI). What the UBI does is put a floor on income, which stops starving artists from … starving.
For me, this is not just a theoretical exercise. You see, shortly after I joined Patreon, I started receiving a UBI. Well, the government didn’t call it that. What happened is that when COVID struck, many Canadians (including me) lost their jobs. Fortunately, the Canadian government stepped in and created a relief fund that was extremely easy to collect. In fact, it was so easy that the relief fund worked effectively like a UBI.
And so in March 2020, I found myself being payed to stay home and work on my research. Pandemic aside, the experience was quite lovely. (If you were a reader at the time, you probably noticed my blog output increase. That’s because I wasn’t working a second job to make ends meet.)
In the time since, the Canadian government has ended the relief benefit. Fortunately, I started a nice postdoc at York University. So for the moment, I’m quite happy. Still, if tomorrow the government implemented a modest UBI, I’d jump at the opportunity to leave academia permanently. And I’m guessing that many researchers would do the same. After all, no one gets into science for the money.
To end this reflection, I’m happy to be on Patreon, and looking forward to the day when it allows me to be a fully independent researcher. But I know that crowdfunding cannot work as a general income scheme unless it is supplemented by a universal basic income.
Interest in Discord?
Now that I’ve ranted about the problems with crowdfunding, let’s switch gears and talk about things that you, as patrons, might like.
A few months ago, a patron suggested that I start a Discord server to facilitate discussion among patrons. Discord, if you’re not familiar, is a private messaging system that anyone can use to host a group conversation.
The idea intrigues me. On Discord, you could talk to other people who support my work — people who are (presumably) like-minded. And of course, I would be part of the discussion. To that end, I’m interested to hear your thoughts about this idea.
My opinion is that Discord is probably worth doing, but only when I have a sufficient number of patrons to facilitate an interesting discussion. In my head, the magic number would 100 patrons. Or maybe to be clever, I should wait until I hit Dunbar’s number (150).
Anyway, let me know your thoughts about using Discord.
The Dunning-Kruger effect as auto-correlation
Enough about me. Let’s talk psychology. Have you heard of the Dunning-Kruger effect? It’s the idea that unskilled people tend to overestimate their competence. Or, as John Cleese puts it, ‘stupid people are so stupid that they have no idea how stupid they are’.
The Dunning-Kruger effect was first discovered by David Dunning and Justin Kruger in 1999. Since then, its become part of pop culture. And you can see why. If you’ve ever interacted with an idiot, the Dunning-Kruger effect seems self-evidently true. But more importantly, there have been lots of confirmations of the phenomena — so many, in fact, that the evidence seems overwhelming.
Or so I thought.
Then I read this article by Jonathan Jarry, which led me down a fascinating rabbit hole. (By the way, I live for a good research rabbit hole.) When I came out the other end, I was convinced that the Dunning-Kruger effect does not exist. It is a statistical artifact — a stunning case of auto-correlation.
I’m going to lay out the reasoning in an upcoming post. But suffice it to say that the Dunning-Kruger effect amounts to taking two random variables, x and y, and showing that x correlates with y - x. In other words, you’re correlating x with itself. Once you work through the reasoning, it becomes clear that the Dunning-Kruger effect is an artifact of auto-correlation. What’s surprising is how long it took (about 20 years) for people to realize what was going on.
Although I’m not a psychologist, I find this result fascinating, mostly because auto-correlation has been a major theme of this blog. I’ve written extensively about how attempts to test theories of value (both Marxist and neoclassical) run afoul of auto-correlation. (Examples here, here, here, here, and here.) That’s because economists inevitably end up testing theories of value by correlating two forms of income. But these forms are always connected by an accounting identity, which renders the correlation trivial.
Given this theme, I was fascinated to see that the Dunning-Kruger effect was guilty of the same thing. I’m working on a post where I break down what’s going on. But in the mean time, here’s some reading on the topic:
(Notice how the last title is effectively suggesting that psychologists are enumerate. I love that this made it past peer review.)
Until next time
That’s it for this update. Thanks for supporting my work, and happy spring (which shows no sign of appearing here in Toronto).
Cheers,
Blair