A long tail distribution is one in which the majority of the events in the distribution are attributed to a relatively small number of items. This is also referred to as the Pareto principle (after Vilfredo Pareto, an Italian economist who devised the concept in 1906) or the 80/20 rule.
In the case of online book sales for example, only 20% of the books sold will be "hits". This is the same for music, movies, mobile phone apps, TV shows and games. The other 80% of things will be in the "tail" of the distribution, which, as the name suggests, is very long.
The point Anderson made in the book is that providing the methods of production and distribution are essentially free (which in essence they are with things digital), then it doesn't really matter that something in the tail only sells one copy because if you have enough things in the tail, you still end up making a lot of money. So for Amazon, iTunes and Netflix, providing huge catalogues catering for every niche interest imaginable turns out to be very profitable.
Of course, the people producing the music, books or movies in the tail will all presumably have day jobs because they won't be able to directly make a living out the sales of a few copies of their work - but they are writing and playing for reasons other than making a living.
It turns out that in universities, academic publication also follows a long tail distribution. A relatively few academics produce a lot of work each year and the majority (80%) produce very much less, perhaps 1 or 2 outputs a year.
As a consequence of government funding approaches and global university ranking schemes, universities have been encouraged to look at the quantity of overall output from their institutions. This has caused some universities to focus on the "short head" part of the distribution, imagining how good it would be to expand that section by having every academic be a "hit" and move into the head of the distribution.
By focussing on the head of the distribution however, they have missed another approach that, like Amazon, Apple and other online industries focuses on the long tail.
The long tail in academic terms represents a whole range of people who produce a modest amount of research around an almost equally large number of research topics. The benefits of this are that the range of research that is carried out by a university is broad and diverse. This should factor into the overall quality of the teaching that the university carries out, which is also usually broad in coverage. It also factors into the potential impact and social engagement ability that the university is able to bring to bear.
From the perspective of a university worried about performance in ranking or government assessment exercises, the issue is not having a tail in the first place but that the tail is not sufficiently long and is related to the number of staff that are employed. The answer here is not how to get rid of people in the tail or somehow to convert them into superstar performers, but to extend the tail by various means. Two ways of doing this are already employed by most universities although they probably don't realise how important they are. The first involves increasing collaboration with other academics in other universities. The second is by increasing the number of people who can publish, use the university by-line and not cost anything, e.g. visitors and other adjunct appointments.
As every other industry has shown, it is impossible to increase the number of "hits" you have beyond the 20% without unsustainable investment. This doesn't leave universities with much other choice than focus on the tail and instead of making it shorter, they should be striving to make it longer.