b.cognosco

There's an interesting article in the current Atlantic Monthly about the likelihood that stricter security systems will make Americans less safe than they'd be otherwise. In it, one of the foremost experts in the world on cryptography talks about how certain systems fail badly -- by which he means that when they fail, as they inevitably do, the consequences of the failed system are worse than the consequences of never having had such a system in the first place. The full shutdowns and total rescreenings of passengers following airport security breaches are one of his examples -- when these systems fail, they do so in a way that is extremely costly and painful. Interesting in its own right, and unquestionably true. It got me to thinking about the broader implications of failing badly -- that a really well-designed system fails well, and that that might be a source of compelling competitive advantage in business -- that one's systems are designed such that the inevitable failures in any given component are isolated and prevented from causing more widespread and costly failures. While I haven't even finished the article yet, it seems that systems that fail well are designed so that components are modular, with responsibility for only one small area of functionality, and that surrounding systems are designed to catch and isolate the consequences of any failure.

Now, neither Scott nor I have really thougth this through, but it seems likely that failing to adequately consider the consequences of failure is at the heart of all sorts of major screw-ups -- ranging from California wildfires started by the Forest Service to the Worldcom debacle.

I wonder how often a failure point is observed but not noted by someone in a company, and I wonder if the open use of klogs within a company could broaden exposure to those failure points, thereby hardening the system and improving competitive advantage?