Andrew Ferrier’s Blog

So let’s imagine you’ve sent a document out to a bunch of people for their review and comments, and you’re currently going through the process of updating it with their corrections, clarification requests, and so on. How do you know when the document’s done? Well, how long is a piece of string?

I’d propose the following rule of thumb:

Take the number of valid review comments you receive from a particular person, and call that a. Take the number of times you recieve comments that require you to point someone at a part of the document they’ve missed (let’s assume it is well organised), and call that b.

Divide a by b - this is the review quotient.

As this number tends towards zero, you know that the document is becoming ready. For most purposes, 0.1 is probably a good threshold for determining when to stop making further changes. It’s close enough to zero that the S/N ratio is now overwhelming you, and your time is better spent on something else.

Whilst trying to cross the road in Hursley this morning, I was struck by how the cars arrived in a long bunch that prevented me from crossing. Someone pointed out that in this case it was probably caused by the traffic light at the end of the village. This got me to thinking: is it important to have cars bunched? If they had been evenly distributed along the road, is it possible that they would be just close enough never to give me a gap large enough to cross? However, it also seems obvious that too much bunching would cause further jams at intersections where bunches came together. Is there an optimum bunching level for any given section of road, and if you plotted it against the width/intended capacity of the road, is there a correlation?

Has anyone done research on this? I guess it might come under the larger umbrella of queueing theory.