Down into Darkness or up into Fog

IEEE Spectrum Magazine, March 2003

There are certain mysteries in life that baffle me, ordinary things that people take for granted, yet on closer inspection seem inexplicable – like why doesn’t rain come down my chimney? One such thing that bothers me is how students in engineering can continue to be trained in only four years of college in spite of the steady exponential expansion of our field. There is so much more to know every year in engineering, but only a small, fixed number of courses available in the curriculum to cover everything that needs to be known. How do we do it?

Part of the explanation is that as we add new knowledge on top of our field, old stuff falls off the bottom. I shudder to think of all the useless material that I’ve learned in my career, like that in most of the courses I took long ago in college myself. However, some of the core material – the concepts and math – endures surprisingly well through the years. So I see an expanding universe of changing knowledge surrounding a solid core of stable fundamentals.

Nonetheless, even accounting for the obsolescence of material, there really is so much more to know today than there used to be. The question remains: how do we do it? And I think the answer must simply be that we don’t. Of necessity we practice triage on the scope of our field; no one learns everything, and each of us understands a smaller fraction of the knowledge as time goes along. I think that we manage this through an implicit layering of our field.

The concept of layering is formalized in networking, for example, where all of the information needed from the lower layers is encapsulated in an interface description. The details of what transpires at the lower levels are hidden. I believe that you can look at all of electrical engineering today in that way. More and more, engineers work within a given layer and are increasingly ignorant and unconcerned about knowledge in the layers below and above them.

I imagine these layers as a staircase. I stand on a particular step, circuit design for instance. I look down the stairs and see them disappearing into blackness. I don’t even know where the bottom is. Above me the steps ascend into fog. It’s all I can do to keep clarity at my present position. Below me the steps are labeled chip design, devices, materials, physics, quantum mechanics, relativistic quantum electrodynamics, maybe even string theory – steps that descend into the most intricate and complex properties of the physical world.

Above me the steps increasingly concern applications and larger systems. Looking up I see logic design, digital signal processors, algorithms, math, software, architecture, applications, systems, economics, social analysis. In the fog I can’t see the top either. Someone else is going to have to worry about these things; I can’t handle it all.

I often think that in this staircase power radiates downwards, and that it is engineering’s destiny to climb higher and higher into the empowerment area at the top. For example, a few lines of code at the software level command all the power of the algorithms, processing, devices, and physics at the lower levels. I think too that increasingly most engineers will work at these higher levels.

However, in terms of empowerment, the converse is also true. An invention at the device level has profound implications for all the layers above. I see in some sense the weight of the world upon the shrinking percentage of engineers at the lower levels. What they do is amplified in the extreme by the larger world above them.

I imagine the individual engineer climbing these stairs towards the top, but I think the metaphor isn’t quite right. Instead, I think it’s like as you try to climb they keep adding more steps, both above and below you. I ran into a physicist the other day who was trying to take apart quarks to find what was inside. There is no bottom, there is no top, and every year there are more steps as our field gets chopped into more manageable segments.

There is a viewpoint that the engineer of the future will cross disciplines and be more conscious of social systems. That may be, and it would mean that the engineer would be working near the top of the staircase. But somebody would still have to be downstairs in the darkness of the basement doing things in the physical world.

I’m not sure how my furnace works either. I call a furnace repairman. It bothers me, though, because I still want to know everything. The ultimate truth is that it just can’t be done.