AGENCY

You are the managing partner of a 650-lawyer firm, and the ground is moving. Every quarter the machines can attempt more of what your associates used to do, and the gap between what they can attempt and what they get RIGHT is where your whole problem lives. You set one stance against that frontier, how much of the output senior lawyers actually check, how much junior work you reserve so associates still learn to do it, and what you spend on capability and on being able to price your own work. Then five matters walk in, and the clients have their own opinions: one will not let a machine near its files at any price, another has already priced in the efficiency it assumes you have.

The defects you do not catch do not announce themselves. They surface two or three periods later as a letter from opposing counsel, long after the decision that caused them, and the mechanism is convex, so the last increment of aggression costs far more than the first. Meanwhile the bench is a stock: the promotion floor tightens every period whether or not you fed it. Four advisers each tell you the truth about their own department and something misleading about the firm, and none of them can be satisfied at once. The debrief names the firm you actually built and the finding is worth arguing with: the judgment firm beats the efficiency firm on the balanced scorecard under every mandate, and the efficiency firm wins profit per partner outright.