The Practice Of Systematically Questioning Your Assumptions

Cognitive scientists call it "tacit knowledge" — things we know how to do or believe to be true without being able to articulate exactly what we know or why we believe it. Michael Polanyi's formulation: "We can know more than we can tell." Most of this tacit knowledge takes the form of assumptions — background beliefs that shape perception, inference, and decision-making without surfacing to conscious attention.

The invisibility isn't laziness. It's efficiency. The brain runs on pattern matching. Once a pattern is established, the brain stops questioning it and starts using it as infrastructure. This is how expertise works — experts don't consciously reason through fundamentals every time; they've automated those judgments. The automation is valuable and also limiting. Automating an assumption means you're running on it without monitoring it.

Assumptions become dangerous at three points: 1. When circumstances change and the assumption hasn't updated 2. When the assumption was never actually validated to begin with 3. When the situation is novel enough that inherited patterns don't apply

The 2008 financial crisis is partly a case study in assumption failure. The entire risk architecture of mortgage-backed securities rested on the assumption that US housing prices wouldn't decline nationally and simultaneously. This assumption had been true historically. No one in the system — borrowers, lenders, rating agencies, regulators — meaningfully questioned it. When it failed, the models built on it failed catastrophically.

There are several structured approaches to making assumptions visible. Here are the most useful:

The pre-mortem. Described by Gary Klein, the pre-mortem is a structured exercise where a team imagines that a plan has already failed — completely and catastrophically — and works backward to explain why. This framing releases people from the social pressure to be optimistic about the plan. It surfaces the assumptions that people are worried about but haven't said out loud. The result is a list of the assumptions that are actually load-bearing and most likely to fail.

Assumption mapping. For each assumption you've surfaced, place it in a two-by-two matrix: confidence (high/low) on one axis, impact on outcome (high/low) on the other. High impact + low confidence assumptions are your critical risks. High confidence + low impact assumptions can be set aside. The mapping gives you a prioritized list of what actually needs to be validated before you commit.

The "what if I'm wrong?" test. For each core assumption, run through the consequences of it being false. Not "could it be wrong?" — that's too easy to dismiss — but "if I stipulate that this is wrong, what changes?" This forces you to understand how dependent you actually are on the assumption, rather than just noting that it exists.

Assumption reversal. Take a core assumption and deliberately state the opposite. Then take it seriously. "Customers want a better version of what exists" reversed becomes "customers want something completely different from what exists." Is the reversed assumption possibly true? What evidence would you look for? This is a generative tool for surfacing alternatives you wouldn't otherwise consider.

First-principles thinking is the most demanding version of assumption questioning. The goal is not just to list assumptions but to decompose them all the way down to the irreducible — the things that don't rest on anything else.

In physics, first principles are axioms: statements so fundamental that you treat them as known. In business or strategy, "first principles" means something more like: remove all the conventional wisdom, all the inherited practices, all the "we do it this way because we've always done it this way" — and ask what you're actually left with.

The discipline requires you to distinguish between: - Evidence — things you know to be true because you've observed or tested them - Reasoning — conclusions you've reached through logic from evidence - Assumptions — things you're treating as true without specific evidence - Social facts — things that are true because everyone agrees they are (prices, norms, categories)

Most thinking is a mixture of all four, unreflectively combined. First principles thinking tries to separate them and rebuild the argument from the evidence up.

The cost is time and effort. First-principles analysis of a major strategic question can take weeks. The payoff is that you occasionally discover that everyone in an industry has been optimizing around a constraint that isn't actually a constraint.

Organizational psychologists have a concept called "the normalization of deviance" — the process by which practices that would look alarming to an outsider become invisible to insiders through gradual acceptance. Diane Vaughan used it to explain the Challenger disaster: the engineers had seen O-ring anomalies so many times without catastrophic failure that the anomalies stopped registering as dangerous. The deviance had been normalized.

This is assumption accumulation at scale. The insiders weren't negligent — they were operating within a belief system that had gradually incorporated risk assumptions without anyone explicitly deciding to do so.

Outsiders don't share these accumulated assumptions. A new hire asking "why do we do it this way?" is doing assumption archaeology. A consultant's value is often less in their expertise than in their outsider perspective — they ask questions that feel naive from inside the system and devastating from outside it.

The strategic implication: deliberately seek outsider perspectives before major commitments. Not to defer to them, but to use them as assumption detectors. Where does the outsider push back? Where do they ask "why?" about something you've never questioned? That's your list.

Most intractable disagreements aren't about conclusions. They're about different underlying assumptions. Two people can reason flawlessly from their respective premises and arrive at completely opposite conclusions. The argument about conclusions never resolves anything; it just produces heat.

The productive move in persistent disagreement is to map backwards: "We disagree on conclusion X. What assumption would you need to make for X to be correct? What assumption would I need to make?" Usually you find that the disagreement is about empirical questions (what is true about the world?) or value questions (what should we optimize for?) — both of which are actually discussable, unlike the conclusions.

This is one of the most practical applications of assumption-questioning. Not applied to your own planning, but applied to conversations where you and another person are talking past each other. Stop arguing about the conclusion. Go upstream.

Systematic assumption questioning doesn't require hours of structured analysis every day. It requires a habit of pausing before commitments and asking: what am I taking for granted here?

The most useful moment to practice this is when you feel certain. Certainty is the signal that your assumptions have gone invisible. When everything seems clear and the path forward seems obvious — that's exactly when you're most likely to be standing on unexamined ground.

One concrete practice: keep an "assumption log" alongside your project notes. Every time you make a significant decision, spend five minutes listing the assumptions it rests on. Revisit the log monthly. Watch which assumptions were validated by events and which weren't. Over time, you'll develop a calibrated sense of which types of assumptions tend to be solid in your domain and which tend to fail.

The goal isn't paralysis. You'll never have all the information. The goal is to know which bets you're making, so you can monitor them and adjust when reality starts pushing back.