For a long time we believed the trick was asking the right person. Twenty years of measurement say otherwise: what counts is how many different angles are in the room.
Philip Tetlock tracked experts for twenty years and collected twenty-eight thousand forecasts. The result: on average they barely beat chance, and the most famous ones did worse than the rest.
But they are not all alike. Tetlock borrows an image from Isaiah Berlin. The hedgehog knows one big thing and bends everything to it. The fox knows many and borrows from each. The foxes win, clearly.
The lesson is not about intelligence. It is about how many ways of reading a situation you are willing to put on the table.
Kapari does not gather its voices for volume. It gathers them so that no camp is missing.
We measured it on one case, drawing panels of each size 400 times from the same pool of simulated voices. What collapses with few voices is not precision, it is coverage: at ten voices, about one draw in four contained all six camps; at thirty voices, 94 to 95%.
In other words, too few voices are not slightly wrong about everything. They drop an entire family, and you cannot see that it did.
In a company decision, the trouble rarely comes from the people you thought about. It comes from the team you left out of scope, or the customer you assumed was locked in.
The voices exist so those people are in the room before the day itself.
And you keep control of the weights. You state who counts most, in plain terms, and the result shows it.
Tetlock, P. E. (2005), Expert Political Judgment. Tetlock, P. E. and Gardner, D. (2015), Superforecasting. On the robustness of simple weights: Dawes, R. M. (1979), "The robust beauty of improper linear models in decision making", American Psychologist.
All seven theories on one page: the map of this series
Kapari measures its audience with Google Analytics, to know how many people read these pages. Nothing is stored on your device until you accept.