How many museums make expensive decisions without articulating the underlying assumptions as testable predictions? How often does your institution quantify uncertainty when making consequential decisions?
The more you accept that every decision is a bet, the stranger it is to find organizations being unintentional when placing bets. It’s easy to skip naming what we think will happen and how we’ll know if we succeed. Writing simple, falsifiable statements that reflect what you’re actually doing can be terrifying — but it’s also how organizations learn.
I’ve been building a decision quality system at Museums as Progress, and Douglas Hubbard’s framework for quantifying uncertainty has become central to how I’m thinking about strategy work. His book, How to Measure Anything, demonstrates that quantifying uncertainty improves decision quality dramatically. When organizations assign probabilities to outcomes, they think more clearly, learn faster, and improve their judgment over time.
Today’s letter explains the framework and why it matters for cultural institutions facing consequential resource decisions.
Hubbard’s Framework: Uncertainty Is Measurable
Most people are systematically overconfident. When Hubbard asks people to provide 90% confidence intervals for uncertain quantities, they turn out to be right only 50-70% of the time. So, not much better than a coin flip on average. Their ranges are too narrow — they’re usually overconfident, which shows up in the workplace as underestimated project timelines, overestimated returns, and committing to initiatives based on certainty that isn’t warranted.
Hubbard’s solution is deceptively simple: assign probabilities to outcomes instead of using vague confidence language. When you say “I’m 70% confident this will work,” you’re making a falsifiable claim that can be checked later. The specific number matters less than cultivating the discipline of quantifying your beliefs. (You’ll need to be operating in an environment where people feel comfortable making claims they know may be incorrect for the team’s sake. Some would say that kind of culture work needs to be done before you adopt an approach like this; I think adopting approaches like this is the thing that builds the culture you want, but save that debate for another day.)
The goal isn’t perfect prediction — it’s calibration. If you say you’re 70% confident about 100 things, roughly 70 of them should happen. Track this over time, and you’ll find patterns: areas where you’re reliably accurate and others where you consistently over- or underestimate. That feedback loop is valuable.
What makes this useful for cultural institutions is that it forces assumptions out of the shadows. Every resource decision contains implicit bets about what will happen as a result. Hubbard’s approach makes those bets explicit and testable. You can see what you’re actually betting on, assign probabilities, and learn whether your judgment about those types of decisions is reliable.
What It Looks Like in Practice
Consider a mid-sized art museum facing a 30% budget reduction. Leadership is weighing whether to consolidate community engagement work into curatorial staff responsibilities while maintaining capital commitments and core operations. The decision feels necessary given constraints, but what are they actually betting will happen?
Making implicit bets explicit might surface assumptions like:
Curatorial staff can sustain partnership activity at 60-70% of dedicated team levels
Community engagement outcomes won’t decline measurably if facilitation shifts to part-time attention
Protecting capital commitments now will generate future capacity that offsets current cuts
The revenue gap won’t persist if programming costs decrease
Quantifying uncertainty means assigning probabilities to each assumption: How confident are we that curatorial staff can maintain partnerships — 70%? 40%? The number matters less than the discipline of stating a belief that can be checked later.
Tracking outcomes means documenting predictions with evaluation dates, then checking what happened. If curatorial staff maintained 40% of partnership activity rather than the predicted 60%, that’s valuable information. Not because the decision was wrong given constraints, but because the pattern — how much partnership activity depends on dedicated capacity — informs future choices.
The framework doesn’t judge whether the decision was correct. It offers a method for learning from decisions regardless of outcome.
Why This Matters for Museums
Budget constraints force impossible choices. When resources get tight, institutions make tradeoffs between curatorial capacity, community programming, research, and operations. Without explicit predictions, organizations miss opportunities to learn from their decisions. “We thought it would work” becomes “We estimated 80% probability that curatorial staff could sustain partnerships at previous levels, and here’s what actually happened.” The difference matters if you’re invested in improving judgment over time.
Questions worth tracking across the sector:
At what point do blended staffing models (curatorial + community work) break down?
Does physical infrastructure (renovated spaces, new buildings) generate community connection outcomes at levels comparable to dedicated programming staff? (More concrete example: To what extent does the new multi-million dollar art park depend on at least a few dedicated staff to realize the outcomes we’re hoping for?)
When resources get tight, which functions do museums consistently protect and which do they treat as discretionary?
What We’re Learning
I’m building this into our practice at Museums as Progress — tracking predictions, documenting calibration patterns, and learning what types of judgments I’m reliable about and where I’m consistently off.
It’s uncomfortable. Assigning probabilities to beliefs means creating a record of being wrong. But that discomfort is the point — it’s how learning happens.
I’d welcome the conversation if you’re curious about experimenting with this approach in your institution. Just reply to the email or leave a comment on the post. I’m documenting what I’m learning and am interested in others participating.
What matters is building organizations that learn from their decisions rather than swinging from one crisis to the next.