sc Stefan Coburn

Clues, Not Truth

October 30, 2026 · 4 min read

Observations, data, numbers, and statistics do not come packaged with truth. Humans take these raw materials and construct a narrative. The data constrain the stories one can tell with them, but the constraints are far looser than most people believe. In a world awash in data, one can easily invert the usual process: select the desired narrative first, then find the data to support it.

I believe in objective truth, but I also recognize the reality of motivated reasoning. Humans are primed to double down on evidence that favors their beliefs and to minimize or discredit data that challenge their dogmas. With this in mind, data carry both utility and danger.

This truth was never lost on Clayton Christensen. The Harvard Business School professor often remarked that data do not magically exist as objective truth. Humans build the categories, select the metrics, decide what to count, and ultimately create the narrative. I once heard a funny story. A man arrives at the gates of heaven, where an angel welcomes him and offers to show him around. As they tour this heavenly abode, the man notices a striking absence: there are no numbers, spreadsheets, or datasets anywhere. He asks his guide why. The angel replies, “Data lie. Nothing that lies can be in heaven. So if people want data, I tell them to go to hell.”

Data as a Rorschach Test

Psychologists administer Rorschach tests using inkblots that carry no inherent meaning. Humans are pattern-matching, pattern-making machines, so we readily turn random noise into a coherent message or story. Test subjects look at these symmetrical blots and freely associate their meaning. It is not so different from lying on the grass and finding shapes in the clouds: even in random reality, we find what we seek.

So it is with data. And what humans almost universally seek is confirmation of their preexisting assumptions and beliefs.

I recently watched this play out in real time. During a San Francisco Giants broadcast, the camera found a mother returning to her seat with a baby strapped to her chest and her arms full of food while her husband sat and waited. The announcers supplied the dialogue, crowned her “mom of the year,” and joked that she probably wanted to smack him. The internet did the rest. Millions of viewers took the clip as evidence of a lazy, inconsiderate husband, and some urged her to leave him.

Days later, the wife published her own account, and the story nearly reversed. One of her husband’s oldest friends had recently died, and the game was her gift to him, an emotional escape. Earlier in the evening he had offered to hold the baby; she kept her in the carrier. When she returned with the food, he offered to take it and asked whether she wanted to eat first. She told him to go ahead. She had eaten before the game; he hadn’t. The footage was real, but the narrative was invented. The data were incomplete, and the announcers, and millions of viewers, were all too quick to fit a story to them.

At its best, science provides a framework and a bulwark against convenient confirmation, through standards (p-values), methods (double-blind trials), and adversarial replication. But science is a human endeavor. These tools have strengthened it against our drive to fit data to our biases and assumptions, but they have not made it impervious.

Proceed with Caution

A hypothesis supported by data is better than a guess supported by none. The tension between the utility and the danger of data can be partly resolved by fighting confirmation bias directly. Seek out the data that could refute your presumption. What would an intellectual opponent put forward to dismiss your narrative or support a competing view?

The world is complex, yet we seek comfort in simple stories aligned with our culture, upbringing, faith, friend group, profession, and so forth. Reality is under no obligation to fit our preconceived notions. To understand things as they really are, we must not delude ourselves into believing that data are self-explanatory truths. They are clues that help us fill in our mental map, but the map is not the territory. We are motivated to see the landscape as we want it to be, not always as it really is.

And so I must, as we all must, exercise the muscle of epistemic humility. I must practice seeking out the data that conflict with my worldview, dogmas, and preconceptions, and remain open to both new data and new interpretations.

Sources

← All Essays