The line about twenty years of experience being one year repeated twenty times is old enough to be furniture. It is also vague about the mechanism. Something fails to accumulate, but what?
Not exposure. The person with twenty years has genuinely been present for hundreds of decisions. Not outcomes either, in most cases. The bid was won or lost. The hire worked out or did not. The expansion either paid for itself by the fourth year or it sat there.
What fails to accumulate is the record of what the person believed beforehand. And without that record, the outcome teaches nothing, because there is nothing for it to correct.
Memory is not a storage device
Most people assume they remember their own predictions. They do not, and the failure is not a matter of ordinary forgetting. Memory actively revises the prediction so that it matches what happened.
Baruch Fischhoff established this in the mid-1970s in a series of studies with an elegant design. Participants estimated the likelihood of various outcomes before an event. Afterward, once they had been told what actually occurred, they were asked to recall their own earlier estimates. They could not. Their recalled estimates had drifted systematically toward the known outcome. People who had given something a thirty percent chance remembered giving it sixty, and they were not lying. The revised figure was what they sincerely recalled believing.
Consider what this does to a career. An irrigation district board in the Magic Valley votes to line four miles of canal rather than defer the work another season. Six years pass. Conveyance losses fell, though a run of wet years muddies the picture, and a pump station failed in year three for unrelated reasons. Ask any board member now what they expected at the time and they will tell you, with complete sincerity, roughly what happened. The ones who opposed the project will recall having had specific reservations that were in fact never voiced. The ones who supported it will recall a confidence they did not have.
Nobody in that room is dishonest. The board simply has no access to its own prior state, and so the six years produced no learning. If the same question arises again, they will approach it exactly as they did the first time, minus nothing, plus a feeling of seasoned familiarity.
The record has to exist before the outcome
The correction is unglamorous. Write down what you think will happen, before you find out.
Four things need to be on the page. The decision itself, stated plainly. The reasoning, compressed to the two or three considerations that actually drove it rather than the full defensible list. A prediction specific enough to be wrong, with a rough probability attached: not “this should improve efficiency” but “conveyance losses down at least fifteen percent by the third season, and I would put that at seventy percent.” And the thing that would tell you that you had erred, named in advance.
That last item does most of the work, and it is the one people skip. A prediction with no stated failure condition cannot be falsified after the fact, because you will find a reading of events under which you were basically right. Everyone does. Naming the disconfirming evidence while you are still uncertain is the only moment at which you will name it honestly.
Judge the decision, not the result
There is a second failure that runs alongside the first, and a written record is what makes it visible.
A decision and its outcome are different objects. A well-reasoned decision can produce a bad result, because the world contains variance that no reasoning eliminates. A reckless decision can produce a fine result for the same reason. Organizations that evaluate people on results alone are, over any short period, largely rewarding and punishing luck. They are also teaching everyone watching that the reasoning does not matter, which is a lesson people learn quickly and act on.
The correct test is whether the decision was sound given what could have been known at the time. That test is only available to someone who wrote down what was known at the time. Without the record, hindsight supplies the answer, and hindsight always finds the outcome to have been foreseeable.
This produces a distinction worth keeping. Some of your worst results followed good decisions and require no change in method. Some of your best results followed poor decisions and should frighten you. A leader who cannot sort their history into those piles is not learning from experience regardless of how much of it they have.
Why almost nobody does this
The practice costs perhaps ten minutes per significant decision and a half day of review each quarter. Measured against what it produces, it is close to free. It remains rare, and the reason is not oversight.
The record’s principal function is to prove you wrong. It exists to catch the moments when you were confident and mistaken, and to hold them where you cannot revise them. Nothing in the ordinary operation of an organization rewards a leader for assembling that evidence about themselves, and a great deal in the political life of an organization punishes it. The written prediction can be quoted back. The remembered one cannot.
Which is the actual reason twenty years so often becomes one year repeated. Not laziness, and not a shortage of feedback. A rational unwillingness to keep the only document that would make the feedback usable.
Edited by Patrick J. Wolf, PhD