Data in a Database Is Not Knowledge
How organizations mistake memory for understanding
Most organizations believe knowledge is something you can store.
Put data in a database.
Add the right fields.
Make them required.
Run reports.
Understanding is supposed to appear on the other side.
It doesn’t.
David Deutsch (physicist, author, father of Quantum Computing) explains why. He argues that knowledge does not come from data. It comes from explanations that survive criticism. Data doesn’t create understanding. It can only test ideas that already exist.
Without an explanation, data just sits there.
A database is excellent at storing symbols.
It is terrible at storing meaning.
To make information fit into a system, we strip away context. We compress reality into dates, categories, percentages, and statuses. This makes information easier to process and harder to understand.
Deutsch defines knowledge as information that plays a causal role in keeping itself true. A row in a database doesn’t do that. It doesn’t explain itself. It doesn’t push back when it’s wrong. It doesn’t improve under scrutiny.
It just persists.
And persistence gets mistaken for truth.
Most enterprise systems demand answers before understanding exists. The software requires a close date, a confidence score, a status color. Reality is still uncertain, but uncertainty isn’t allowed. So people guess. They enter something that lets them move on.
The database fills up.
The dashboard looks precise.
Nothing gets more accurate.
From a Deutschian perspective, this is backwards. Precision without explanation is not knowledge. It’s a frozen guess. And once frozen, it gains authority simply by being stored, aggregated, and visualized.
This is why more data rarely improves decisions. And why better dashboards often make things worse. You’re not converting data into knowledge. You’re locking assumptions into systems and mistaking durability for truth.
At this point, someone will ask how to “optimize” this insight for a platform or a workflow or a dashboard.
That impulse is the problem.
Most so-called optimization — for software, for metrics, even for Substack — is just pattern-matching dressed up as certainty. It feels actionable, but it isn’t knowledge in Deutsch’s sense. It’s advice that works until it doesn’t, because it was never grounded in explanation to begin with.
Knowledge lives in people.
In their ability to explain.
In their willingness to be wrong.
In their capacity to correct.
Databases don’t do that.
They remember.
They don’t understand.
And memory is not knowledge.
That version should read clean, self-aware, and just sharp enough that the meta lands without slowing the piece down.