Inversion

Thinking Tools2.3 Establishing the RequirementsAlso known as Invert, Always Invert

Popularized by Charlie Munger (Poor Charlie's Almanack, 2005), after the mathematician Carl Gustav Jacobi's maxim "invert, always invert."

Turns a goal around to ask how it would fail, lists those failure modes, and treats each one as a rule to avoid.

A billing migration turned around: how to make it fail, and the rule each failure gives

Each line is a way the team could guarantee the migration hurts customers; under it is the rule that blocks it. Tick the rules that apply to this project (the auditors have already come and gone, so one does not), and the panel under the list collects them as the plan's rules. Illustrative.

Turned around: how would we make sure the billing migration hurts customers?

  • Cut over on the first of the month, when invoices run
    • Cut over mid-month, between billing runs
  • Skip the parallel run to save two weeks
    • Run one full billing cycle on both systems and match every invoice
  • Let the vendor map the old discount codes alone
    • Finance signs off on every discount code mapping
  • Switch the old system off the same night
    • Keep the old system readable for ninety days
  • Tell the call center the morning after
    • Brief the call center a week ahead, with a script for billing questions
  • Schedule the cutover inside the auditors' visit
    • Keep the cutover out of the audit window

Cut over on the first of the month, when invoices runNo rule kept here yet.Rule: Cut over mid-month, between billing runs

Skip the parallel run to save two weeksNo rule kept here yet.Rule: Run one full billing cycle on both systems and match every invoice

Let the vendor map the old discount codes aloneNo rule kept here yet.Rule: Finance signs off on every discount code mapping

Switch the old system off the same nightNo rule kept here yet.Rule: Keep the old system readable for ninety days

Tell the call center the morning afterNo rule kept here yet.Rule: Brief the call center a week ahead, with a script for billing questions

Schedule the cutover inside the auditors' visitNo rule kept here yet.Rule: Keep the cutover out of the audit window

An example

The finance team at a regional insurer had to move customer billing off a fifteen-year-old system by quarter end. The project plan said "smooth cutover" and little more. At the kickoff the finance director asked a different question: if we wanted this migration to hurt our customers, what would we do?

In twenty minutes the whiteboard held eleven answers. Cut over on the first of the month, when invoices run. Skip the parallel run to save two weeks. Let the vendor map the old discount codes alone. Switch the old system off the same night. Tell the call center the morning after. The team struck the five that could not happen on this project (the auditors had come and gone) and turned the other six into rules, among them one full billing cycle run on both systems, finance sign-off on every discount mapping, the old system readable for ninety days, and the call center briefed a week ahead. Two of those rules were not in the plan.

Why it works

Success has many routes and most of them are vague; failure is concrete. Asked how to make a project succeed, people offer aspirations: communicate well, test thoroughly. Asked how to make it fail, they name specific acts with dates and owners, because the ways things break are easier to picture and closer to experience.

Each failure, turned around, becomes a rule that can be checked: either the parallel run happened or it did not. Inversion also gets past a team's attachment to its plan. People defending a plan will not volunteer its weak points, but asked to sabotage it on paper they find them quickly, and nobody is accused of anything, since the question is hypothetical.

The thinking behind it

Charles T. Munger Charles T. Munger, vice chairman of Berkshire Hathaway, made inversion one of his standing habits, in talks such as his 1994 lecture at the USC business school and in Poor Charlie's Almanack (2005), edited by Peter D. Kaufman. His advice was to ask what would produce the outcome one wants to avoid, and then avoid it. He took the phrase from the mathematician Carl Gustav Jacobi's maxim, invert, always invert, and moved it from mathematics into decisions. He offered it as an investor's maxim, one model in a latticework drawn from many fields, from talks and without research behind it.

How to run it

  1. State the goal in one plain sentence. Include the date and what success means to the people affected, such as no customer billed wrong.
  2. Turn it into a question about failure. Ask what the team would do to guarantee the goal is missed, and give people permission to be inventive.
  3. List failure modes fast, without judging. Specific acts beat categories: "skip the parallel run" is useful, "poor testing" is not.
  4. Strike the ones that cannot happen here. Keep anything with a plausible path, including the entries that would embarrass someone in the room.
  5. Turn each survivor into a rule. Write it so anyone can tell whether it was kept, and give it an owner.
  6. Put the rules into the plan. Each rule becomes a task, a gate, or a sign-off; a rule nobody owns is a wish.
  7. Rerun the list when conditions change. A new deadline, vendor, or team member brings new ways to fail.

Where it goes wrong

The list stops at the obvious. Teams name the failures they have already planned for and feel reassured. Push for the uncomfortable entries, such as a sponsor who loses interest or a vendor who overpromised, by asking what nobody in the room wants to say.

The rules stay vague. "Do not rush the cutover" cannot be checked; "one full billing cycle on both systems, signed off by finance" can. Rewrite any rule two people could disagree about.

Inversion replaces the plan. A list of things to avoid says what not to do, never where to go, and a project that dodges every failure can still miss its goal. Use the rules as guardrails on a plan that already says how it will succeed.

When to use something else

  • Premortem. When the plan is settled and each person should imagine, alone and in writing, that it has already failed and why.
  • Failure Mode and Effects Analysis. When there are many failure modes and they need ranking by severity, likelihood, and how easily each would be caught.
  • Second-Order Thinking. When the danger lies in how others react to a plan that works exactly as designed.

Where it sits in SOLVE eX

Go to the source

Back to the catalog