One characteristic I see in accountable managers is having the courtesy to be wrong.

Meet Jan

Jan is an executive at a previous employer of mine, who had strong beliefs about product direction. We pursued the direction. The bet didn’t pan out as we hoped. Jan stepped down.

Now, I disagreed with parts of her strategy (easy in hindsight!), but I respected Jan’s being clear on what she believed. Her position was unambiguous enough that it might be proved one way or the other, and thus Jan made herself accountable.

She had the courtesy to be wrong.

In a traditional view, you might think a management team should aim to have fewer Jan’s – her strategy didn’t work out! – but most teams have a bigger problem.

Meet Carl

Many managers perfect the art of never being particularly wrong (or right). Let’s call this style Carl.

Carl is highly agreeable, but at the end of the day, he does little that risks being wrong. His ideas are fine and useful, but it’s hard to test outcomes of his beliefs.

Unlike Jan, Carl will never be particularly accountable – and, synonymously, makes little difference to the organization. I find this uncourteous.


Before we can know an idea’s merit, we must understand its falsifiability. Jan values falsifiability where Carl does not.

When an idea is falsifiable, we can bring evidence to bear. In turn, observable evidence allows others in the organization to contribute – to analyze, criticize, and refine. The harder an idea is to pin down, the harder it is to make progress.

The best organizations aim for falsifiable outcomes first, and positive outcomes second.


A falsifiable belief is often a divisive one. I got this notion from Chris Coyne, a founder of OKCupid: divisive questions have high information content; anodyne questions, not so much.

OKCupid’s ideal was to ask its users questions with a 50 / 50 (divisive!) split of answers. This offers more signal for matching.

Bad questions (examples mine):

  • Is ice cream delicious?
  • Are puppies cute?

These questions might split 95 / 5, and are therefore uninformative.

Better questions:

  • Should you correct people on grammar?
  • Who did you vote for?

In these cases, we might imagine something closer to a 50 / 50 split, and therefore having a fighting chance of learning something.

As managers, we should ask ourselves if we are making assertions that anyone could possibly disagree with. If an idea is about as divisive as puppies, we might be working with a Carl.

Thanks to Bret Copeland and Scott Carleton for feedback on drafts of this essay.