Most communication about uncertain topics defaults to point estimates: "The project will take three weeks." "There's about a 70% chance of rain." "Sales will probably hit $2M this quarter." These statements are deceptively precise — they collapse a distribution into a single number and discard all the information about uncertainty in the process.

A better approach, borrowed from statistics, is to express uncertain beliefs as ranges or intervals: "I'd estimate three to five weeks, with three being optimistic." "I'd be surprised if sales came in below $1.5M or above $2.5M." This seems like a small change. It isn't.

What confidence intervals communicate that point estimates don't

A confidence interval (or more precisely in everyday use: a range reflecting your genuine uncertainty) communicates three things a point estimate cannot:

Your actual uncertainty level. A range of "three to five weeks" versus "two to eight weeks" conveys something specific and actionable about how much slack people should build in. A point estimate of "three weeks" conveys nothing about uncertainty — it sounds definite even when it isn't.

Your epistemic state, not just your best guess. Saying "I expect $2M, but I wouldn't be surprised by anywhere from $1.5M to $3M" tells people how to use your estimate. It signals: plan around $2M, but don't be caught off guard if it's elsewhere in this range.

What would actually surprise you. The edges of a genuine confidence interval reveal your reasoning. When you have to state what would be surprising, you're forced to examine your assumptions rather than just picking a plausible-sounding middle estimate.

It makes you more credible, not less

The counterintuitive finding from research on expert communication: expressing appropriate uncertainty increases perceived competence and trust, it doesn't reduce it. The exception is domains where false certainty has become the social norm — some medical settings, some management cultures — where expressing uncertainty can be misread as weakness.

But in most knowledge-work contexts, confident uncertainty — "I'm confident my range is right, even though I don't know exactly where in it we'll land" — signals sophisticated thinking. It shows you understand the limits of your own knowledge. This is a marker of expertise, not ignorance.

Building the habit

The mechanical change is straightforward: whenever you make an estimate about something uncertain, express it as a range (even if you only share the range internally). Ask yourself: "What would have to be true for this to come in much higher than I expect? Much lower?" These questions surface your hidden assumptions and expand your estimate appropriately.

The harder part is calibrating the width of the range. Most people give ranges that are too narrow — they express 80% confidence intervals that only capture the true value 50% of the time. The fix is the same as general calibration training: track your ranges against eventual outcomes and adjust the width until your stated confidence matches your actual hit rate.

Start with low-stakes predictions you can verify quickly: how long will this meeting run, how many emails will you receive today, how long before this task is actually done. The feedback is fast, the stakes are low, and the patterns you observe will transfer to higher-stakes estimates.