Tuesday, January 03, 2006

Glorifying Sophistication

I think I read some article over the holidays referring to a "sophisticated device", probably related to the military, but I can't remember. It got me thinking, though, about how we tend to glorify sophistication in the things we engineer. Why do we equate "sophisticated" with "better", especially in harsh applications like warfare or safety critical applications like medicine or aviation?

It seems to me that in an environment where reliability, maintainability, and ease of use are paramount, "sophisticated" is bad. I would tend to prefer "dirt simple" as a positive adjective. Perhaps the obsession with sophisticated fanciness is what leads to bloated, expensive aircraft like the F-22. Contrast that with the A-10, one of the cheapest but most powerful and reliable warplanes ever made. Perhaps the best example, though, is the infamous AK-47 rifle, one of the most simple, reliable, easy to maintain, and thus ubiquitous weapons on the planet.

Of course, there are plenty of counter-examples where sophistication has its merits. Computer chips are the first thing to come to mind. But integrated circuits can also be seen to have simplified computing because they are cheap to mass produce, have no moving parts, and are easy to replace. Another example might be GPS, which uses Einstein's theory of relativity. However, GPS is also a system that has overall reduced the complexity to the user, by reducing the complexity of navigation, removing the need for ground-based systems, and proving more reliable than any other system.

Perhaps the metrics that we should use when evaluating the "goodness" of some technology, especially in critical applications, is "net simplicity". This metric would accumulate the simplicity or complexity of the entire life-cycle of technology, from manufacture, use, repair, replacement, and disposal. A new technology should have a net lower complexity than its predecessor, but the individual complexity distribution may shift. Take AM and FM radio as an example. FM is more complicated to understand and produce than AM, but when equipment mass produced, it is just as simple to build and maintain, and actually uses less power than AM. Or, in the case of software than analyzes some data, it may use more difficult or complicated algorithms in the back end, but to the end user it makes the data easier to understand. A good spreadsheet that takes a large table of data and plots it into an easy to interpret graph is a pretty basic example.

So, the conclusion from this rant is that sophistication is not a bad thing, but it should not be an end goal in and of itself. Instead, the goal should be net simplicity. The sooner we realize that, the better we can engineer things that will actually work reliably in the field instead of being finicky lab toys or expensive wastes of money.


Post a Comment

<< Home