The Discipline of Diagnosing Past What Everyone Expects

Why this matters

The dangerous pressure is not always a person pushing. Sometimes it is a shared expectation that everyone in the room already agrees on: the customer, the last tech, your dispatcher, and your own first glance. "It's always the thermostat on these." "Everybody knows that brand's boards fail." When the expected answer is the consensus, testing past it feels like wasting time proving what everyone knows. That feeling is exactly the trap. The best diagnoses in a shop's history are the ones where somebody kept going after the obvious answer, because the obvious answer was wrong just often enough to matter.

Consensus is a hypothesis, not a finding

An expectation shared by five people is still an expectation. It carries the same evidentiary weight as one person's guess, which is none, until it is tested. The number of people who believe a cause does not raise its probability, it only raises the social cost of checking. Treat "everyone expects X" as a well-supported hypothesis worth testing first, and nothing more. The moment you treat it as already proven, you stop being a diagnostician and start being a stenographer for the crowd.

Where the expected answer comes from, and why it's usually close but not exact

Consensus is not stupid. It usually points at the right neighborhood, which is why it is seductive:

• Base rates: that part really does fail most often on that system, so the guess is right more than half the time. Right often enough to trust, wrong often enough to burn you.
• The last few jobs: the shop just saw three of the same failure, and now every similar symptom looks like the fourth.
• Reputation: a brand or a part earned a bad name once and keeps it long after the pattern changed.

Each of these makes the expected answer a good place to start and a bad place to stop.

The discipline: confirm the cause, do not just match the symptom

The habit that beats consensus anchoring is refusing to close on a symptom match. The expected part explains the symptom, but so might two other things. Before you replace the obvious suspect:

• Prove it with a measurement, not a resemblance. "The symptom fits" is not "the cause is confirmed."
• Ask what else produces this exact symptom, and rule those in or out.
• Look for the upstream cause that could have killed the obvious part, so you are not back in a month having replaced a victim instead of the culprit.

Cross-trade, the shape is the same: the tripping breaker everyone blames, replaced twice before someone finds the shared-neutral fault; the "bad thermostat" that is really a low charge; the pump everyone condemns that is starving from a clog upstream; the appliance board blamed for what a failed sensor was feeding it.

When testing past the obvious is not worth it

Discipline is not the same as ignoring base rates and testing everything to death. Pushing past the expected answer earns its time when the stakes or the doubt are high, not on every call:

• Do push past when the fix is expensive, safety is involved, the obvious answer has already been tried and failed, or the symptom does not fully fit the expected cause.
• It is fine to start with the odds on a low-cost, low-risk, textbook-fitting call, as long as you confirm before you close and stay ready to reopen if the fix does not hold.

The judgment is proportional: the more it costs to be wrong, the further past the crowd's answer you test.

The mental model to keep

Consensus tells you where to look first. It never tells you when to stop looking. Your signature goes on the diagnosis, not the crowd's, so the crowd does not get to close it for you.

References

• Trade-standard diagnostic practice on root-cause versus symptom-matching
• Human-factors literature on anchoring and consensus bias in diagnosis
• See related: Confirmation Bias and Anchoring in Diagnosis, Holding Your Diagnosis When You're Pushed Toward an Answer