Thread by @evanlapointe

Generalizations are unintelligent by their nature. Here’s how to translate any generalization into actual, usable truth.

Start with this idea. Every generalization is a simplification of a statistic.

Men are taller than women.
Women are better at multitasking than men...

Behind those statements are statistics. They might look like...

...the average man is x’y” and the average woman is a’b”

And behind those averages are distributions, often normal distributions...

If it’s a normal distribution, we need to look at minimum and maximum values.

That’ll tell us how wide the distribution is, which will give us a sense of how “normal” average is. If there’s a wide distribution of heights, maybe only 5% of men are within a few inches of average.

So that means if we start parading around an idea of what “average” people are like, we might be talking about 5% of people, not 95% of people.

Have you ever been in a meeting that summarized the average customer? You were probably excluding 95% of customers in that discussion.

So behind every generalization is a statistic, and behind every statistic is a distribution that represents actual reality.

Where teams fail is in not understanding distributions. They make decisions based on the statistic or the generalization itself.

Here are some examples:
- what does the market need?
- what are most developers, designers, or salespeople like?
- what conditions will our product be used in?

We might generalize the market and fail to see segments we can address precisely. We might generalize applicants and fail to hire the best people. We might generalize conditions and fail to realize that people use our product in dozens of contexts.

The worst kinds of generalizations are about human nature. Those are the generalizations that cause the largest and highest frequency of issues. Here’s why.

When we say “human nature,” we are often talking about risk tolerance, creativity, political or bureaucratic behavior, honesty, ulterior motives, integrity, incentives-driven behavior, and stuff like that.

Human beings are dramatically different from each other in every one of these regards. Some humans are risk averse, but 5% at the high end of the distribution are extremely risk tolerant. Some humans have good intentions but the bottom 5% are psychopaths and sociopaths.

When we say crazy generalizations like “believe good intentions,” “people are risk averse,” and “that’s because of our incentives,” we make it highly likely, if not guaranteed, that we will make catastrophic mistakes.

We practically guarantee that we will not believe the right people (generalizing their behavior and input as worse or less valuable than expected) and that we will believe the wrong people (generalizing their behavior and input as better or more valuable than expected).

So what should we do about it? It’s actually simpler than it may seem...

Remember that behind generalizations are statistics.
And behind statistics are distributions.
What are those distributions?
Group them into segments.
Make decisions at the segment, not generalization level.

This will help you observe the market more clearly.

This will help you make vastly superior hires.

This will fuel vastly more intelligent discussion.

This will enable vastly smarter decisions.

This will power vastly higher quality and precision in execution.

That’s it. Your teams will feel genius compared to yesterday.