What are your priors?

Understanding privilege through the lens of Bayesian probability.

In any given week on Twitter you can find an activist, or a scholar, accusing an entrenched white journalist, or their editor, of having unexamined privilege.

The response by these journalists is well worn.

It’s generally a three fold sarcasm-driven process. Step one: undermine the concept of privilege. Step two: make fun of the idea that they have it, whatever it is. Step three: belittle the person accusing them of having it.

These tactics are infuriating to me, and I used to feel Twitter rage when I saw this play out. And I used to spend energy trying to figure out how to communicate with those journalists. I wanted to explain to them, rationally, and step by step, why the activist were generally right.

The problem, I thought, was that we weren’t communicating well enough. We weren’t articulating ourselves as well as we need to. And maybe that’s sometimes true.

But now, when I watch this happen, I just feel embarrassed for the journalist. They tend to be older, whiter and straighter. They tend to be men. And they’re often habitually sardonic.

I want to consider now, as a slight sidestep, Bayesian methods of assessing the probabilities of truth claims.

For many of us who care about stats, and everyone ought to, we’re often unwittingly applying frequentist methods. If you’ve studied some stats, then you know how this goes.

Suppose you have a bag full of 1000 white stones and 1000 black stones. And you pull stones out of the bag, one at a time, and after ten pulls you have ten white stones. What are the chances of pulling out another white stone?

Generally speaking, in a frequentist model, we would first assume that the bag of stones is randomized. Getting that out of the way, we would then calculate the odds of pulling out a black stone:

# white stones / (# black stones + # white stones)
= 990/1990 = %49.75

In other words, it’s still still roughly 50%, with black stones being very slightly favoured.

Alternatively, we could apply a Bayesian method. In this approach, we’re less casual about simply assuming the bag of stones is randomized.

Allow me to replay the framework: you have a bag full of 1000 white stones and 1000 black stones. And you watch your assistant pull out ten stones, one at a time, and after ten pulls you have ten white stones. What are the chances of pulling another white stone? Put another way: what probability can I assign to the hypothesis that the eleventh stone will be black?

On the Bayesian model, I need to ask myself, what are my priors?

In this case, my prior is that the chance of pulling a black stone is the same as pulling a white stone.

But what are the actual chances of pulling out ten white stones, if the bag is properly randomized? It’s roughly one in a thousand.1

Indeed, by the time the fifth or sixth white stone was coming out of the bag, most people would start to re-examine their prior.

Put another way, pulling ten white stones out of the bag is evidence that the bag is not randomized. It would seem skewed somehow. Like, maybe the assistant put the white stones in first and the black stones second, and didn’t shake the bag very well or very long. Maybe the assistant has something against black stones.

Intentional or not, in the real world, randomization is rarely perfect.2

If I watched ten white stones get pulled, I would reasonably conclude that the chances of pulling another white stone are better than pulling a black stone; not worse, as the frequentist model suggested.

So this brings me back to the role of privilege in journalism and beyond.

Take, for example, a claim made by a woman that they were sexually harassed by a coworker. What is the likelihood of this being true?

Journalists that believe that the world is fundamentally just, or that rape stats are skewed in favour of feminists, will assign a lower probability to the allegation.

Privilege shapes us. Privilege shapes our priors.

Is the journalist immersed in the research? Are they up on the data? Journalists that don’t find ways to challenge their privilege will find themselves saying stupid shit.

Andrew Coyne once dismissed an entire report which claimed that our federal criminal justice system was systematically discriminating against Indigenous People.3

Matt Gurney, also an editor at the National Post, called rape chants at UBC silly. He also called the investigation into the rape chants silly. Because, you know, teenagers! It’s not a big deal.4

Here’s Jonathan Kay writing that to “the nation’s great credit, racism, sexism and homophobia have become rare in Canadian public life.” Regrettably, he’s now the editor of The Walrus.5

It is perhaps unfair of me to focus on journalists. The art of saying stupid shit is everyone’s purview.

For me as well; I’ve had my fair share.

So it’s not clear to me that I need to focus on journalists. I do so because they have a professional responsibility when it comes to truth claims.

I don’t know. I guess all we can do is keep encouraging folks to re-examine their priors in light of new evidence.

One interesting conundrum is that when journalists are challenged on Twitter to examine their privilege, they often retreat to a kind of defence and attack framework. It’s my hope that later they take a deep breath and do the work.

I like to believe that re-evaluating one’s priors is what makes a journalist a researcher and not just a mouthpiece for men’s rights activism or white supremacy. Journalists have an opportunity to be role models. Some do this very well, and others do not.

Progress is being made. I think.

I wept when I read Desmond Cole’s essay, “The Skin I’m In: I’ve been interrogated by police more than 50 times—all because I’m black” – not because it’s news to me.

I wept because he was published. I was moved because it’s beautifully written and persuasive and people were sharing it, and I take it as evidence of progress. And it’s part of the public record now.

I hope Coyne and Kay and Gurney and all of us with white skin privilege are reading it.

"... racism, sexism and homophobia have become rare in Canadian public life (says Jonathan Kay, FALSELY)

  1. The chance of this happening is the same, roughly, as flipping a coin 10 times and getting heads: $latex (\frac{1}{2})^{10} $
  2. Everything is kind of clumpy and connected and messy.
  3. On Twitter, Coyne tweeted out a link to the story saying, “Most sensible comments here are Atleo’s. ” In the article, Atleo only commented on the importance of working Indigenous folks to “better reintegrate offenders into society.” The author of the report, meanwhile, had considerable, scathing criticisms – as were the other commentators in the article. I have always been confused by this kind of brash, racist, dismissal of the expertise in the report. Because Coyne deleted all of his tweets, the original is gone (https://twitter.com/acoyne/status/309831462553784320) but others remain: https://twitter.com/BarrieFriel/status/309841095909597184.
  4. Ashley Csanady assures me that Gurney once wrote a beautiful article on fatherhood and feminism. Well, we can’t find it. Until then, all we have is this winning example of Matt Gurney’s views, and this softer, nicer, more thoughtful version of Matt Gurney’s views.
  5. Why is the Walrus so boring? And so white?