Chasing peak performance is expensive and all too often turns out to be fatal...
Graphics compiled while flying from Dohar to London, text written in a Frankfurt coffee shop and dispatched to silicon.com via a free wi-fi net at 30Mbps.
Optimising things is now something of a management fetish. The modern credo is cut the slack, take out as many people as possible, reduce all waste and focus on just-in-time delivery at minimum cost and maximum profit.
A degree of optimisation makes sense in most situations. We shouldn't be wasting anything. But overdoing it is always ruinous. Over-optimisation crashes people, departments and complete companies. When a power outage or the sickness of one employee is enough to halt or seriously impair the business, then the end is near.
The rarity of those events is not an argument that holds water either. For example, you sometimes hear that air conditioning is unwarranted because it never gets hot enough, or that computers don't need replacing and the old software or systems will do, or that your department can cope with a doubling of its workload.
That is the sort of thinking behind the condition of the UK's roads and airports during last winter's big freeze. Everything stopped because provisioning for winter was seen as unnecessary and because a huge cost saving was to be had.
So by saving a few million in salt, de-icer, snow-ploughs and other equipment, the country lost billions in trade and working days, not to mention wasted fuel. Without exception, underinvestment and underprovisioning always incur costs that far outweigh any savings or other benefits.
It's not just that the economics of the situation are all too obvious, it is that the physical principles are well established over millennia and make the credo of optimisation so absurd.
For example, would you like to travel to work in a family car or an F1 racing machine? The F1 would get you there at great speed now and again, while the family car would do it more modestly day after day, year after year. So, performance really does equal brittleness. It's axiomatic, and well proven.
How does all this look in reality? All systems, man-made or otherwise, follow a bathtub curve of reliability. Early failures and deaths are followed by a long period of stability - good health and near failure-free operation - to be followed by a wear-out and death phase.
As we increase the performance and efficiency...