The accountant's wrong count had one advantage over the modern dashboard: it was small enough to be caught.
The person who disputed the number was often standing nearby. The transaction it described was recent enough to be remembered. The discrepancy between what the clay said and what had actually moved through the warehouse could, with effort and some argument, be identified and corrected. The error had a small radius.
The modern equivalent has a much larger one.
A dashboard presents information with the confident authority of good design. The chart loads cleanly. The number is current. The format implies rigour. And everything it shows is, in some narrow technical sense, accurate — it is a faithful representation of what was entered into the system that produces it.
What it cannot show is the quality of what was entered. It cannot signal that the metric it displays was defined three years ago for a different product. It cannot indicate that the category it uses was reorganised, and that this year's figures are not comparable to last year's for reasons nobody thought to document. It cannot note that the assumption embedded in its calculation is contested, or that the dissenting view in last quarter's analysis was not reflected in the final summary.
It presents what it was given with equal confidence regardless of whether what it was given was worth giving.
This is the polite lie. Not an active deception — no one intended to mislead. A structural one. The medium presents information with a clarity that the information itself may not deserve, and the recipients of the dashboard have limited means of knowing when the presentation is outrunning the reality.
The discipline this requires is not distrust of dashboards. They are often useful. It is the maintenance of a prior question: what are the conditions under which this metric was defined, and do those conditions still obtain?
Most organisations do not build that question into their relationship with their own data. They build systems that surface information quickly and accurately, and they treat that capability as intelligence. What they have built is retrieval. The intelligence is the thing that happens afterwards, when someone asks whether what was retrieved is still the right thing to be looking at.
The clay recorded the wrong count with perfect fidelity. The dashboard does the same.
Someone has to be responsible for knowing the difference.