I forget what publication I found this in.
I know a systems analyst (in those days he was called a management engineer). He almost had a nervous breakdown trying to apply scientific management principles to the Manhattan Project during World War II. He tried to functionalize and standardize and specialize the work of mathematical physicists. He tried to set up standards. He tried to develop an appropriate appraisal form.
His first difficulty, the most traumatic, was that he found it impossible to describe the work. He couldn't even produce a job description. He went into the cubicle of a theoretical mathematician and said, "All right, what are your tasks, duties, and elements?" The fellow said, "What's that?" He went through it again. The fellow finally answered, "Well, what I am trying to prove is that locally compact sets are not dense in themselves in Hilbert space." The engineer asked, "That's a duty? Show me these things." The fellow said, "I can't because they don't really exist, They are just abstract ideas which we invent." The systems analyst exploded. "You are working with something that doesn't exist? Com on now, tell me your tasks, duties, and elements."
Eventually, the engineer went to the second page in a book, published by the U.S. Employment Service, which tells how to analyze a job. It says, "If you don't make any sense of what the incumbent says, watch what he does." So he began to observe.
Well, the mathematical physicist did only three things. He drank coffee in the office, he looked at books, and he wrote on the blackboard. That's all he did. Obviously, my analyst was getting nowhere.
He went to the man's boss, who really wasn't his boss because it turned out that the man really didn't have a boss. "What's this fellow doing?" he asked. The boss said, "We don't know shat he is doing. If we knew what he was doing, we wouldn't have him doing it." Astonished, the analyst asked, "Do you mean to tell me that you don;t know the tasks, duties, and elements of this subordinate?" The bos responded, "Hell, no. That's why we've got him doing it. He's the only man in the country who understands this sort of thing."
So the analyst went on to the next part of the interview form. "now tell me," he asked the boss, "how do you know when he is oint the job well? What are the criteria?" The boss answered, " I haven't the slightest idea." When the analyst persisted, he finally said, "Well, I guess he is doing it well when he tells me so." "Do you mean to tell me," shouted the analyst, "that you depend on a subordinate to tell you when he is doing a job right?" The boss stood his ground. "That's absolutely true. He's the only person in the country who can understand the proof of these theorems. If he says he's doing it rihgt, he's doing it right. That's why we have him doing it."