October 10th, 2004


The art of science

Yesterday I was writing a response to a entry by shesingsnow lamenting the deplorable run-time she was getting from her laptop battery and it became (again) really clear to me how much of an art Computer Science really is. Tweaking a laptop to get the most of its battery is an art. Getting wireless networks to run right certainly seems like an art -- and maybe a black art at that. Every situation, every building is unique, and the presence or absence of people (ugly bags of mostly water) can make a quantifiable difference in performance. E.G. an average size guy standing six feet away from me on the line from me to the AP will knock ~10 dB off the signal.

It seems that the old guild skill ratings -- novice, apprentice, journeyman, master -- are the most sensible ways to describe practitioners in IT. Going along with Computer Science as art, there certainly is a lot of deeply held superstition and myth. E.G. an oft-repeated but wrong idea is that the 2.4 GHz radio waves used by IEEE 802.11b/g wireless devices are absorbed by water. The frequency that's absorbed by water is a lot higher, and some who know this fact but not the next fact will use someone else's repeating of the water absorption myth as an excuse to ignore everything they say. The next fact is that, while 2.4 GHz radio waves are not absorbed by water, they do a very good job of yielding up their energy and exciting water molecules when they meet. This is why microwave ovens use this frequency range. This is also why masonry (which has a fair amount of water in it) will attenuate wireless signals more than sheetrock, and why a potted plant or a human body acts like a wireless sponge.

Of course this is all bunk. It is a science. All of the hardware operates in according to scientific laws. It's all created by human engineers; it is all understandable, measurable and knowable. Unfortunately the knowledge space is almost incomprehensibly vast. There is so much to know -- and much of it takes specialized instruments to know -- that the practitioner in the field is left to fend for himself with a thimble-full of knowledge. IMO, that's why a packet sniffer is such a powerful tool: it shows you, with certainty, everything on a given network connection. No guessing; either a packet is there or it's not.

But, how many of us have sniffers? And, how many of us know how to use them? In our organization the answer is two and two, one a journeyman, the other master. So, I suppose that under these conditions -- a nearly infinite knowledge space, minimal tools and limited training -- it probably is safest to treat the science as an art. Knowledge, amplified by instinct and logic, applied successfully make the master from the apprentice.
  • Current Music
    Mix; mostly Vangelis