I’ve read an interesting article, written in 1989, about ‘herd
behaviour’. You are probably aware of the piece anyway, but just in case I thought I’d mention it. Here’s a section that looks like it relates to current controveries, even though it largely predates it:
Suppose you have some curve between the extreme of this opinion and the extreme of that opinion. You have some indefinite, statistically quite insignificant distribution of opinions. Now in that situation, suppose that the refereeing procedure has to decide where to put money in research, which papers to publish, and so on. What would happen? Well, people would say, “We can’t really tell, but surely we shouldn’t take anybody who is out here. Slightly more people believe in this position than in any other, so we will select our speakers at the next conference from this position on the opinion curve, and we will judge to whom to give research funds,” because the referees themselves will of course be included in great numbers in some such curve. We will select some region there to supply the funds.
And so, a year later what will have happened? You will have combed out some of the people who were out there, and you will have put more people into this region. Each round of decision making has the consequence of essentially taking the initial curve and multiplying it by itself.
Now we understand the mathematical consequence of taking a shallow curve and multiplying it by itself a large number of times. What happens? In the mathematical limit it becomes a delta function at the value of the initial peak. What does that mean? If you go for long enough, you will have created the appearance of unanimity.
The full article is here:
28th July 2005