That's not why. Bosses aren't trying to fuck you over. If they're aligned with the company, they're trying to maximize A - B, where A is what you contribute, a

That's not why. Bosses aren't trying to fuck you over. If they're aligned with the company, they're trying to maximize A - B, where A is what you contribute, and B is what you cost. For most of the industrial era, it's been considered easier to reduce B than to increase A. Why? Many reasons, but the fundamental one is that most of the work was concave, which is what OP describes when he has cost per unit approaching asymptote m, which is equivalent to performance approaching 1/m from below, with more learning. He's focusing on concave, commodity work and he's right about that stuff. Machines are also taking over the commodity work, leaving only the convex for humans. See: http://michaelochurch.wordpress.com/2013/03/14/gervais-macleod-9-convexity/ http://michaelochurch.wordpress.com/2013/03/14/gervais-macle... With convex labor, the managerial assumption that A is at a maximum falls flat, because no one even knows what that "maximum" might be. With concave work, management could assume that A was near its maximum value and occasionally fire a slacker. What this means is that, for convex work, A-maximizing is a worthier goal... but almost no one has figured out how to manage that kind of work. [ETA: a more cynical view is that someone else gets a bonus/credit for reducing B, but the worker himself gets all credit for improving A.]