Let's talk about outsourcing. Not about the usual arguments for and against, which have been masticated to a pulp so gooey that it's now suitable fodder for the daily newspapers. Let's talk about outsourcing and the next generation of IC design managers.
There are three skill sets in the repertoire of a fine manager. The first set is purely administrative: the knowledge, patience, discipline and humor to make sure the reports get filed, the budgets submitted and approved, and the weekly status reviews turned into an accurate project chart. These are all skills that anyone who has wrested a degree from an engineering school can readily learn.
The second set of skills is more personal, and involves leadership. Some people, through a combination of choices and styles, bring about cooperation, creativity and positive feelings in their groups. These skills are probably learned by imitation rather than by analysis, and come from growing up with role models who exhibit them or working under managers who exercise them. They are often grounded in a personal sense of well-being and self-sufficiency that comes from either successful childhood or successful therapy both rare enough.
The third set of skills is intellectual, but more deeply based than the mechanical skills of project administration. These skills allow the manager to construct a correct predictive model of the entire design indeed, the entire project based not just on the design documentation but on personal experience. These are the skills that allow an architect to look at the structure of a piece of IP and predict that timing closure at 266 MHz will prove impossible. Or to intuitively grasp that what appears to be a minor implementation change in one corner of the chip will turn into a relayout on the other side. These skills come from having learned, firsthand, the detailed tasks that go into the design.
These last two skill sets come from participation. When we outsource, we send away not only nasty, difficult tasks but also opportunities to learn, opportunities that would have formed our next generation of managers. That is the real loss: not that someone else learns to do it cheaper, but that we destroy the environment in which the next generation of managers could develop.
Ron Wilson covers microprocessors, programmable/reconfigurable logic and the chip design process. He can be reached at firstname.lastname@example.org.