Wednesday, December 22, 2004

Signal and Noise

My father used to tell me when I was growing up that if I was going to do something, I should do it right or not at all. I'm sure that we've all been told this at one time or another. In general, I would say that it isn't bad advice. It's another way of saying don't be half-assed in your endeavors.

Sometimes, though, advice like this can get you into trouble. The problem, as I see it, is the subjectivity in the phrase "do it right." Often times it is easy to divine what this means. But, the more complex the problem is, the more difficult it is to determine what the right way is. And since we are often ready to abandon efforts when we decide that a certain path is not the right one, it's fairly important that our criteria for right and wrong be appropriate for the situation.

Let's look at an example from the world of engineering (yes -- I know, let's all contain our excitement).

In Chaos: Making a New Science, James Gleick describes a problem that IBM engineers were faced with during the infancy of computer networking. At the time the engineers were attempting to transmit data between computer systems across telephone lines. Unfortunately, telephone lines were (and are) quite noisy. That's OK for us humans since a dropped word here and there won't disturb us; we can generally fill in the missing pieces from the surrounding context. However, computers cannot recover dropped data bits in the same way. Dropped data is simply lost, often corrupting the entire datastream. Clearly this is an important problem for the engineers to resolve.

To solve this problem, they initially attempted to raise the signal strength to drown out the noise. They also investigated using quieter transmission lines. Both of these strategies achieved a degree of success. However, they still were faced with occasional bursts of "spontaneous noise" that, despite their best efforts, they could not successfully eliminate or even explain. Eventually they came to the conclusion that they would never be able to completely eradicate noise from the system. In other words, they accepted that noise was a fundamental component of the system that they had developed. Once they faced this reality, they switched tactics. Instead, they used standard signal strength and transmission lines while employing an error correction algorithm that would resend data that had been lost in transit. It meant that more data had to move through the system for the same amount of information to be transmitted. But, it was much cheaper and, most importantly, extremely effective. In fact, this identical strategy is still used today in modern data transmission systems (TCP/IP networks, CD/DVD players, etc.).

Now, let's get back to my father's advice.

If the engineers had been too rigid in their interpretation of their goal, if they had insisted upon sending noiseless signal, they would never have achieved the ultimate goal. If they had interpreted "doing it right" as completely eliminating noise, and if they had followed my father's advice to the letter, they might have decided to do it "not at all." Fortunately, they were able to take a more expansive view of the problem.

I think there's a lesson to be learned here.

As a society, we are constantly faced with important, large-scale problems of staggering complexity. The systems we employ to address these problems are frequently just as complex. And in all of these systems, noise exists. No matter how hard we try or what resources we bring to bear, we will never eliminate this noise. It is a fundamental component of these systems. Instead, we should accept the inevitability of noise and focus on the larger problem at hand.

Conservatives traditionally trash the welfare system because it can be abused and because it creates disincentives in the communities it is attempting to assist. To them, these issues justify its abolition. However, these problems are the noise of the welfare system. I freely acknowledge that welfare abuse exists. Likewise, I acknowledge that it creates a mild disincentive in some individuals. But the existence of this noise does not indict the system as long as it is achieving its central goal -- averting the most extreme consequences of impoverishment. Strategies should be developed to correct this noise, but its existence alone does not argue against the system at large.

In any substantial military action, there will be civilian casualties, collateral damage, and war crimes committed by all engaged parties. No amount of effort will successfully eradicate them. They are the noise of the war system. Does the existence of this noise mean we should never enter military conflict? No. But it does mean that we have to accept them whenever we apply military force. We should do all we can to reduce their consequences, but we must understand their inevitability. Therefore, any honest attempt to justify military action must include these factors in the calculus. Whatever the goal, it must be worth the infliction of atrocity.

Noise exists. Always. Accept it. Expect it. Work with it. And don't let it distract you from doing it right, whatever it is.
Weblog Commenting and Trackback by