Map Meets Territory

This post was originally an email sent to the Better Questions Email List. For more like it, please sign up – it’s free.

——
What is the territory?
——

In our last email, we discussed the fact that we rarely come to a decision without any data.

We have pre-existing beliefs about how likely or unlikely certain outcomes are.

These beliefs are called priors, and they influence our decision-making both before…

(“Based on how many Toms I know, what percentage of men do I think are named Tom?”)

…and after the fact…

(“Google says 95% of men are named Tom, but that can’t be right, because I haven’t met that many Toms.”)

Whether our priors reflect reality depends on how representative our experiences are.

Because of this, our priors can be more or less accurate, even when based on real experiences.

You can think of priors as quick approximations used to help make guesses about complicated problems.

In this way, they are much like maps – they help us get to where we want to go, but they are an imperfect reflection of reality (just as a map necessarily leaves out huge amounts of detail).

This is why:

“The map is not the territory.”

Alfred Korzybski

If priors are our map…

Then what is the territory, exactly?

Take a hypothetical woman – we’ll call her Jill.

You don’t know much about Jill…only that her friends call her “a bit wild, and very outgoing.”

I’ll ask you to make a guess about Jill…

Is it more likely that Jill is:

• An administrative assistant?
• Or a personal trainer?

Now that you’ve read last weeks email on Bad Priors, you may have a sense of how you make a guess like this…

You’ll consult your priors and compare the administrative assistants and personal trainers you’ve met to your image of Jill.

And herein lies a problem.

Human beings love narratives, and when presented with striking (but perhaps misleading) information we use those narratives to help us make decisions.

In this case, Jill’s outgoing nature seems to make her a perfect fit for personal training – she’ll like talking to clients and have lots of energy.

Meanwhile, her wild side seems to make her a poor choice for a quiet office setting.

This seems to make sense…

But we’ve been strung along by the narrative of who Jill is…

And we’ve ignored the base rate.

The base rate is the likelihood that some event will happen, based on the average outcome of similar events.

If 1 out of 100 people in your high school drop out, the base rate of dropping out at your school is 1%.

If 10 people each year in your town are killed by escaped laboratory mice driven by an endless thirst for revenge, and your town has 10,000 residents, then the base rate of being killed by MurderMice is 10/10,000 (or .1%) .

Let’s think about this applies to Jill.

Google tells me there were about 374,000 personal trainers in the US – let’s assume that’s low and round up to 500,000.

Meanwhile, there are over 3 million administrative assistants in the US.
Even if we assume that Jill’s personality doubles her chance of becoming a trainer…

(something I’d be very unsure about)

…she’s still 3 times more likely to be a secretary.

Humans love stories.

And because of that, we tend to put far too much weight on “individuating data”…

Characteristics we recognize, patterns we’ve seen in our own lives.

We consult our priors, notice patterns, construct narratives from those patterns, and then use those narratives to predict what will happen.

In doing so, we ignore the base rate – and, perhaps, an uncomfortable truth:

It is the average that is predictive, not the individual.

Knowing the average number of deaths on airplanes is more predictive for us than a friend telling us about their dramatic near-death experience…

The average outcome of investing in high-risk, high-volatility stocks is more instructive for us than the story of a neighbor who made his billions investing in a startup that breeds MurderMice.

If our priors are the map…

The base rate is the territory.

Wherever possible, we need to not simply consult our own priors…

…but learn to consult the base rate as well.

This is a simple principle in theory, but incredibly hard to apply in real life.

The pull of narrative is incredibly strong in us. We are trained to look for the particular, for the individuating, for the specific.

We think in stories, not in averages.

But averages help us make better decisions.

Asking “what happened before?” is just as, if not more valuable, then asking “what seems like it’s going to happen now?”

When our priors lead us away from understand the base rate…

our map diverges from the territory…

and our decisions become more and more inaccurate.

And sooner or later, we look up from the map…

And find ourselves in uncharted territory.