First Transition

Pressure: The wrong explanation survives because it gives you something to do.

I threw up in transition one of ten consecutive triathlons.

Transition one is the changeover from swim to bike. Out of the water, run to your spot, wetsuit off, helmet and shoes on, go. Two minutes, maybe three, and for ten races that was where my body quit. Then I got on the bike, rode, ran, and finished.

I had an explanation before I had finished toweling off the first time. I was not fit enough, which was my primary fear. The swim was costing more than I had trained for, and my body was saying so at the first moment it got the chance.

So after that race was done, I got to work and trained. More swimming, more miles, harder weeks. Each race I arrived fitter than the last, and each race I threw up in transition one.

Ten times.

What kept that explanation alive was not stubbornness, and it was not a shortage of information. It was that “not fit enough” is a problem with a familiar response attached to it. I knew how to train. I could start that afternoon. Finding out what was actually happening would have meant stopping, and stopping does not feel like progress when there is another race on the calendar.

Here is what I did not notice, for a long time.

It was only transition one. On the bike I felt better, and I kept feeling better as the race went on. By the run I was fine. Every race, the same shape.

That shape argues against my own explanation, and not subtly. An endurance problem gets worse over a race, not better. I had that data ten times, from inside my own body, and it did not register. I had already decided what kind of problem this was, so the evidence got sorted into the category instead of being used to question it.

After race two, my wife Mandy said I might be getting nauseous from the open water swim, and that I might want to try dramamine.

She had watched two races. She did not have the training log, the split times, or the felt experience of the swim. She got it right eight races before I did.

Race eleven, I took dramamine.

Out of the water, run to my spot, wetsuit off, and nothing happened. I felt good. I got on the bike feeling good, which I had not done once in ten races, and standing there in transition it landed. I did not reason my way to the answer. I felt the difference, and then I understood what the previous ten had been. Turns out it was simply motion sickness induced by swimming in choppy open water.

I have thought about that a great deal, because I have watched the same thing happen in organizations I was responsible for. A recurring failure picks up an explanation early, usually in the first cycle or two, and the explanation that sticks is the one with work attached. We need more people. The team is too junior. We started too late. Each of those points at something the team already knows how to do, and each can absorb unlimited effort without ever being tested.

The diagnosis was wrong, and it had one advantage over the right one: I already knew how to act on it.

Pattern: Once you have named a problem, new evidence gets filed instead of tested.

Identify the Pattern

Watch what happens in the first minutes after something goes wrong.

An explanation forms fast. It has to; you cannot act without one. But the explanation that forms first is not always selected for being right. It is often selected merely for being available, for fitting the evidence you happen to have, and, above all, for pointing at something you can do. An explanation that hands you a familiar action feels like understanding, because it ends the discomfort of not knowing and starts the relief of working.

Once it is in place, it changes what the next round of evidence is for. New information is no longer a test of the explanation. It is material to be filed under it. Ten races of “only in transition one, better after” should have been ten refutations. They were ten confirmations that I needed to train harder, because I had already decided what the problem was and the data was arriving pre-sorted.

Medicine has studied this more carefully than most fields, because there the cost of a wrong first explanation is the health and well-being of a patient. Pat Croskerry, an emergency room doctor who has spent much of his career studying how doctors get diagnoses wrong, describes the habit of stopping the search once an answer fits, and calls it premature closure.^[1] Mark Graber and colleagues examined one hundred cases where hospital doctors had reached the wrong diagnosis, and found that stopping too early, settling on an answer and no longer weighing the alternatives, was the single most common thinking error.^[2] Errors of thinking, as opposed to errors of process or missing information, were involved in about three quarters of the cases. These were experienced physicians working problems they had trained for years to solve. Expertise did not protect them. In some readings it is what exposed them, because a deep pattern library produces a confident first answer fast.

Their work is about doctors and patients. Applying it to a leader working out why their own team keeps failing the same way is my extension, and I hold it with some care. But the mechanism does not appear to require a white coat.

Then there is the question of why Mandy saw it after two races and I did not see it after ten.

Daniel Kahneman and Dan Lovallo drew a distinction that is useful here, between what they called the inside view and the outside view.^[3] The inside view treats the problem in front of you as unique, and reasons from its particulars: my training, my splits, how the swim felt. The outside view treats it as one instance of a class, and asks what usually happens to things in that class. Mandy was not reasoning about my fitness, nor the fears I carried. She was reasoning about a category, a person who is sick right after swimming in open water, and that category has a well-known answer. Their work concerns forecasting and planning rather than diagnosis, so the application is mine. But it explains something I found humbling: Mandy had less information and a better answer, and both of those facts came from the same place. She was standing outside it.

And then the last piece, which is the one I find most useful. Why did the answer arrive at race eleven, in my body, rather than at race ten, in my head?

Dedre Gentner and colleagues have spent years studying how people extract the principle from examples. In one line of work, participants who compared two cases side by side were markedly better at abstracting the shared structure than participants who studied the same two cases separately.^[4] A single case, however carefully studied, tends to stay a case. A contrast makes the structure visible. Their participants were students learning negotiation from written cases, so the reach to a difference felt in a parking lot is mine. But I had ten cases and could not see the pattern. One contrasting case, and I saw it in the time it took to put on a helmet.

She had less information than I had, and she was looking at the symptom while I was looking at myself.

What It Costs

The effort compounds in the wrong direction. Every cycle spent on the wrong variable builds real capability at the wrong thing, and each round of effort deepens the commitment to the explanation that called for it. This is sunk cost with a training log attached. By race eight I was not just wrong about the problem; I had invested enough in being wrong that admitting it meant discounting a year of hard work.

The evidence keeps arriving and keeps getting filed. Once the category is set, disconfirming shape gets absorbed rather than noticed. You are not ignoring the data. You are seeing it and assigning it a meaning that fits. Ten data points, zero updates, and at no point did it feel like I was refusing anything.

The people who saw it stop saying it. Mandy said it once. That is what people do with advice that does not land; they do not repeat it, because repeating it feels like nagging, and because you have made it clear you have an explanation already. The outside view is available for about one conversation, and then it goes back to being polite.

The team learns that grinding is the response. When a leader meets a recurring failure with more effort, the team learns that effort is what recurring failures are for. Diagnosis becomes a job with no owner. Over time you get an organization that is very good at absorbing problems and out of practice at naming them, which is a comfortable place to be right up until the problem is one that effort cannot absorb.

Reframe the Mindset

I was stubborn and I should have listened to my wife. But turning a mechanism into a character flaw is not the whole story.

I did what a capable person does with a plausible explanation and a familiar action available. I acted on it. The failure was not a failure of effort or humility. It was structural: I had no mechanism for reopening a question once I had closed it, and every cycle of effort made the closure firmer.

There is a second reframe, and it is the one that changed my practice. I assumed for a long time that the way out was to think harder. More analysis, another look at the data, another attempt to reason about the root cause. That is exactly the wrong instrument, because analysis inside a wrong frame mostly produces better arguments for the frame. What I needed was not more reasoning about the ten. It was an eleventh that was different.

I did not reason my way to the answer. I tried something different, felt the difference, and then I understood what the previous ten had been.

Transform the Tension

Three moves. None of them require you to be smarter than you were last cycle. They require you to do something other than try harder.

Look at the shape, not the count. Before the next round of effort, describe the failure precisely: when it happens, when it does not, and what happens after. Then ask whether your explanation predicts that shape. Mine predicted worse-over-time. The reality was better-over-time. That mismatch was available after the first race, and I could have found it in two minutes if I had described the shape instead of counting the occurrences. Ten failures with the same distinctive shape carry more information than ten failures added up.

Ask what you were already told. Go to the person closest to the problem but outside it, and ask a specific question: what have you told me about this that I did not act on? Then take notes instead of defending. The outside view has usually already been offered, once, early, without much force, and it did not land because you had an explanation in hand. Going back for it is not weakness. It is the cheapest test you have.

Change one cheap thing for one cycle. Not a project, not a transformation. A variable. Pick the cheapest untested explanation and run one cycle with it changed. Decide in advance what a difference would look like so you cannot explain it away afterward. A contrast will teach you more than another retrospective, because a contrast is visible in a way an argument is not.

Shift: What have I been outworking that I have not yet stopped to diagnose?

Reflection (Questions to Ponder)

• What recurring problem on my team have I explained the same way for more than two cycles, and what has that explanation asked us to do?
• Does my explanation predict the shape of the failure, or only that it will happen?
• Who has offered me a different explanation that I set aside, and what did I set it aside in favor of?

Connection (Questions to Discuss)

• What is our working explanation for the problem that keeps coming back, and when did we last test it rather than act on it?
• What have you told me about this that I did not act on?
• If the explanation we are working from is wrong, what would we expect to see that we are, in fact, seeing?
• What is the cheapest thing we could change for one cycle to find out?

Subtraction (Things to Stop)

• Stop treating effort as evidence that you understand the problem. It is evidence that you can work.
• Stop counting failures. Describe them. The count says it keeps happening; the shape says why.
• Stop running another retrospective inside the same frame. If the last three produced the same explanation, the frame is the finding.
• Stop hearing outside advice as a judgment on your effort. Usually it is a category you have not considered.
• Stop waiting to understand before you change something. Sometimes the understanding arrives after the change, not before it.

Action (Things to Try)

• Write down your current explanation for one recurring failure and the shape of failure it predicts. Compare it to the actual shape. Note where they disagree.
• Go to the person closest to the problem but outside it and ask what they told you that you did not act on. Write down the answer before you respond to it.
• Pick the cheapest untested explanation and change one variable for one cycle. Decide in advance what a difference would look like.

Footnotes

1. Pat Croskerry, “Cognitive Forcing Strategies in Clinical Decisionmaking,” Annals of Emergency Medicine 41, no. 1 (2003): 110–120. Croskerry describes the habit of stopping the search for alternatives once a satisfactory explanation is found, and identifies premature closure as one of its consequences. ↩︎
2. Mark L. Graber, Nancy Franklin, and Ruthanna Gordon, “Diagnostic Error in Internal Medicine,” Archives of Internal Medicine 165, no. 13 (2005): 1493–1499. One hundred cases were identified through autopsy findings, quality reviews, and voluntary reports, so the sample is not random. Thinking errors contributed in 74 percent of cases and system or process failures in 65 percent; most errors had more than one cause. Stopping too early on a diagnosis was the single most common thinking error. ↩︎
3. Daniel Kahneman and Dan Lovallo, “Timid Choices and Bold Forecasts: A Cognitive Perspective on Risk Taking,” Management Science 39, no. 1 (1993): 17–31. The inside view and outside view are framed there as competing approaches to forecasting and planning; the application to diagnosing a recurring personal or organizational problem is mine. ↩︎
4. Dedre Gentner, Jeffrey Loewenstein, and Leigh Thompson, “Learning and Transfer: A General Role for Analogical Encoding,” Journal of Educational Psychology 95, no. 2 (2003): 393–408. In Experiment 2, participants who compared two cases outperformed those who studied the same cases separately at abstracting and transferring the underlying principle. Participants were novices learning negotiation from written cases; the extension to a contrast experienced physically is mine. ↩︎