How to Predict Anything

Hey folks,

We have two great podcast episodes that dropped in the last few days. The first is a conversation between my charming and wise friend and colleague Michael Kovnat with British science writer Tom Chivers about how Bayesian statistics can empower us all to better predict the future. It's an enlightened view of reality that I find useful and clarifying.

Listen on Apple or Spotify, and let us know what you think in the comments below.

Meanwhile, we also dropped this morning my conversation with philosopher Emily Austin about the benefits to all of us of prioritizing pleasure. Does that sound self-indulgent and unaligned with our collective interests? Not if it's done right. Epicureanism turns out to be a far more useful philosophy than its ancient rival, Stoicism. It was also prescient, surprisingly well-aligned with the science of human flourishing. It speaks to me — I may be, in fact, an Epicurean.

Listen on Apple or Spotify, and let us know what you think in the comments below. I would love to discuss any of the above with you, Michael, and our brilliant guests.

And now ... I bring you ... Michael Kovnat !

How to Predict Anything

Are there things that you’re embarrassed to be ignorant about? You know, those things that everyone else in the world seems to understand, but you, for whatever reason, just don’t? It’s like you missed the day in school when those things were being taught.

I have several. Imperial fluid measurements are a big one. My brain just refuses to remember how many pints are in a gallon, or how many ounces are in a quart. I have to look it up, every single time.

There’s a ton I don’t understand about finance, too. I still really don’t get how bonds work, for example, even though I own some. And while I’ve read a number of articles on the topic, term vs whole life insurance hurts my head. Maybe these things seem like no-brainers to you, but then again I'll bet there are things you’re bewildered by that are obvious to me.

One thing that falls squarely into my realm of confusion is probability. People are always throwing around things like “there’s a 10% chance of rain” or “the Chiefs are favored 3-to-1 to win the Superbowl.” I feel like it's crystal clear to everyone else what this means, but I’ve always been a bit baffled. What are we doing when we try to quantify the future like this? After all, things either happen or they don’t. They don’t 10% happen. What does it really mean to say that Biden has a 50% chance of winning the presidency, as fivethirtyeight.com is currently saying? I know it doesn’t mean that half of all voters say they’ll vote for him (because that number is somewhere between 38% and 47%, according to the polls they sample). Does it mean that if the election were held 100 times he would win 50 of them? But how does that make sense when we’re dealing with a singular event like an election? If Biden does win will we be able to say whether the fivethirtyeight forecast was good or not?

So I decided to finally clear up my confusion and read a book on the topic, a book called Everything Is Predictable: How Bayesian Statistics Explain Our World. And then because that wasn’t quite nerdy enough for me, I sat down for an episode of The Next Big Idea podcast with the author, British science writer Tom Chivers. Tom helped me get clear on the idea that when we assign probabilities to future events, we’re not saying anything deep about the truth of the election, or the Chiefs, or the weather. Rather, we’re making an assessment of our own ignorance, based on our prior expectation of a particular outcome plus the new information we have. In that sense, probability is a subjective science -- a measure of the mind more than of the world.

Along the way, Tom opened my eyes to how this way of thinking -- pioneered by an obscure Presbyterian minister in the 18th century -- offers insights on everything from drug trials to murder trials, from AI to psychedelics, from how evolution by natural selection operates to how human consciousness itself works -- our brains, after all, are thought to be primarily prediction machines. Tom thinks Bayesian statistics is a theory of (just about) everything,

Have a listen to our conversation here, and let me know what you think in the comments below. I’m 80% sure I’ll get back to you.

— Michael Kovnat

Over on the Next Big Idea Daily podcast, we heard a lot of big ideas this week, like:

How to master the art of changing minds, from author Michael McQueen.

How social prescription is changing health care, from journalist Julia Hotz.

How citizens have pushed back when presidents have usurped power, from political scientist Corey Brettschneider.

How apathy is killing democracy and what we can do about it, from writers Samantha Sage and Emily Amick.