Free interactive learning tool

The Influence Diagram

Actors as nodes, arrows for who affects whose outcomes, annotated with what each is measured on. Draw it for any recurring dysfunction and the loop is visible within about ten minutes — and it is usually embarrassing.

Guess first — three years, same complaint

Every quarter, in the same words: "we never get to the roadmap." Urgent client work always wins. Four department heads have said it, two reorganisations have been run, one person has left over it. What is most likely to be true?

The diagram — click a measure to change what someone is paid for
What the structure predicts
Measured by nobody
Replace-the-people test
Swap all four for competent strangers. Does the behaviour change?
Where this bites
Reorgs

New people, same behaviour

If replacing everyone would change nothing, replacing one person changes nothing either. Most reorganisations move names between boxes and leave every measure exactly where it was.

Quality

The thing with no node

Whatever nobody is measured on loses every time, silently, and never appears as a decision. It is not losing the argument — it was never in it.

Good intentions

The metric that worked

Most harmful measures were introduced deliberately, to fix something real, and did. The damage is the second-order effect nobody modelled — which is why the person who introduced it is often the one now most frustrated.

Exhortation

Gravity beats willpower

"We just need to be more disciplined about X" is an attempt to defeat gravity by asking nicely. It holds for about six weeks. Design the slope instead of fighting it.

Incentives are gravity. You can push against them briefly with willpower and process, and the system falls back — usually within about six weeks. Before proposing a fix, annotate every node with what that person is actually measured on, then ask whether replacing all of them with strangers would change anything. If not, you have found a structure, and the person you were about to blame is the one square you can safely leave alone.