It’s a well-worn cliché that the perfect is the enemy of the good, but it’s nevertheless true, and particularly when it comes to the development of or implementation of computer systems. My experience with chalk bears this out.
In 1987 I was sent to Hungary by my employer to adapt an MRP system for a television manufacturer at a factory about 80 km from Budapest. An MRP (manufacturing resource planning) system plans the production of finished goods by breaking them down into their components and calculating the quantities of each that are needed, and by when, to meet customer orders. A television has several hundred parts and is manufactured from semi-finished components that are themselves manufactured or bought. Computer systems use databases, and they are generally fast, but, depending on the data that are stored in them can become ‘clogged’, especially if the same item of data gets stored too many times.
Ours was one of the first MRP systems to be implemented in Communist Eastern Europe, and the engineers at the factory were excited by the idea that they could capture the exact quantities of each material and component that make up a television, including, as it happens, the minute quantity of chalk that is consumed when an inspector marked each component as good enough to be used. That way, they could calculate the amount of chalk they needed to order from their chalk supplier. It seems, though, that chalk can be found dozens and dozens of times in a television (back then, anyway!), even if there is much less than a gram in total. So much chalk, or rather so many instances of it, that the MRP system ground to a halt, got clogged with it, when calculating the total amount needed for the manufacture of several thousand televisions. Chalk was the most-used component of all.
‘Much easier to buy a few boxes a month, and forget about the calculation,’ we said, but there was huge disappointment amongst the engineers that we could not be precise about how much was needed.
Judgements about what is ‘material’ are essential in systems, in business, even in finance. There comes a point where academic accuracy just isn’t important, isn’t worth the effort. That point comes where the cost of accuracy exceeds the cost of inaccuracy. Chalk is cheap.