Long drag

The expression long drag (of English Long Tail , expression sometimes translated by “ long queue ”) was employed for the first time in 2004 by Chris Anderson in an article of Wired to describe part of the market of the companies such as Amazon or Netflix. It is also used in the Statistique S, often in relation to the distributions.

The long tail in the theory of the statistics

The long tail is a current expression to indicate a known phenomenon for a long time statisticians (Loi of Zipf, Distribution of Pareto, Distribution of Levy). The law most used in statistics, the normal Law, vulgarly called the bell of gauss for its form, is criticized in certain its uses for its narrow tail. In finance for example, this narrow tail corresponds to under estimate of the extreme variations, and of new models, known as with broad tail , seek to eliminate this problem. The phenomenon is also known under the names of large tails , tail of the laws of power or tails of Pareto . These distributions are in general of the form opposite.

In these distributions, a population at great frequency or great amplitude is followed by a population to low frequency or of low amplitude, which decreases gradually in a “tail”. Very often, the not very frequent events or of low amplitude (the long tail ), represented by the yellow portion of the graph, can on the whole represent a weight or a number more significant than the first part of the graph.

Such distributions are surprisingly frequent. For example, in English running, the word the is most current; other short words such as off , and and to are as very current, much more as of other current words. For example, the word the accounts for 12% of the words used, while the word barrack occurs less once on 50.000; but by accumulation, words as rare as barrack composes approximately a third of the text. These rare words are the long tail of the English vocabulary.

The long drag of Chris Anderson

The expression long drag , is used by Chris Anderson, which refers to the test of Clay Shirky " Power Laws, Weblogs and Inequality" who noticed that a relatively small number of weblogs have many bonds Web pointing towards them, whereas the long tail made up of million weblogs have only few bonds which point on them. In conference series and the article of Wired, published in October 2004, Anderson describes the effects of this long tail on the economic models present and future. He thinks that the products which are the object of a slack demand, or which have only one low volume of sale, can collectively represent a market share equal or higher than that of the best-sellers, if the channels of distribution can propose enough choice. Examples of such channels of distribution can be caught in Amazon.com, Netfix, Rezolibre.com or Wikipédia. As these examples illustrate it, the long tail is a market potential, made available by the possibilities from Internet.

An employee of Amazon describes the long tail like this: Today, we sold more books which were not sold yesterday than we did not sell books which we sold also yesterday. In the same way, the articles little read of Wikipédia have, collectively, more readers than principal articles, than one can also find on encyclopedias general practitioners such as the Encyclopædia Britannica.

The limit is derived from the graph which is plotted thanks to the data of popularity of each page of Wikipédia. In this graph, the articles of Wikipédia are laid out throughout the x-axis, and the rate of frequentation is indicated on the y-axis. In this example, the page of Wikipédia which receives the most visits would be all on the left in the green, whereas the page that you currently read would be very on the right in the yellow, like the majority of the articles of Wikipédia. The same thing occurs for the books to the inventory from Amazon: the total request for the little required articles exceeds the total request of the much in demand articles.

Relationship with the distribution and storage costs

The key factor which determines if the sales of a trade have a long tail is the distribution and storage cost. When these costs are weak, it becomes profitable to sell products little requested; when it is expensive to store and distribute, only the most popular products are sold.

A significant example is the trade of hiring of vidéos. A traditional store of hiring is limited by its length of linear of racks, which it pays in the form of rent. To maximize its products, it must expose only the titles most asked, to be sure that all space is well used. Netfix on the contrary has centralized warehouses, which enable him to lower the storage costs; the cost of distribution is the same one, whether it is for a popular film or another required little. Netfix can thus be profitable by having a catalog much broader than traditional stores. The Logistique advanced storage or of distribution thus makes it possible to rent more films with restricted public that popular films.

Cultural and political consequences

The long tail can have important cultural and political consequences. When the Opportunity cost (storage and distribution) is high, only the most required products are sold. When the effect long tail goes, the minority tastes are satisfied and the individuals have more a great choice. In a company where popularity is determined by lowest common denominator, the effects of a model with long tail can increase considerably the cultural level of a company. The Télévision offers an good example of it: each chain has a limited time of diffusion (24 hours maximum per day), therefore the opportunity cost of each time slot is high; the chains of TV thus choose the program which potentially will gather the most world. But when the number of chains increases, like their means of diffusion (chains Hertz iennes, by satellite, cable, ADSL, TNT), the choice of programs and the cultural level increase.

Some of the greatest successes on Internet openly took into account the long tail in their economic model. One can quote eBay, Yahoo!, Google, Amazon.

See too

In economy, the effect long drag is often studied in combination with the effect superstar, for example in the study of the market of the music

Random links: