I talked to a guy from the company Medtime which is making personel planning for the health care sector, i.e., a program for the process of coordinating who is on shift when, on a hospital. Nice product and it looked good too. The task of planning the duty rosters for hundreds of people, taking into account many rules and conditions is extremely complicated and their product looked like a necessary tool to do it.

But someone still has to do it. Someone and not something. There is a human being administrating the planning tool, who talks to the people having special wishes such that "I would really prefer not to be on nightshift with Jim again anytime soon" or "Can I skip Mondays?" and whatever. It is a natural job for a human: It requires good intuition and understanding of human relations.

It is a difficult but not impossible task to teach a machine to do a similar task. But it requires data to substitute the experience and knowledge of the human administrator. And thats the point: As we record more and more data, more and more tasks are possible to solve by machines. In that sense, a full featured machine culture is a successor of the "data recording age". If you are afraid of the machines taking over, on the other hand, the game is to keep data recoding to a minimum.