Lexicon Entry

Long tail

Long tail. The structural distribution in consumer markets in which the dominant share of attention goes to a small number of well-known items, while a larger number of less-known items collectively account for the rest. Curation operates against the long tail; aggregation surfaces it.

On the concept

The long tail describes a statistical pattern that holds across most consumer markets: a small number of items receive the dominant share of attention, sales, or use, while a much larger number of items collectively account for the remaining share. Plotted with item rank on the horizontal axis and attention on the vertical, the resulting curve has a steep head (the most-popular items) and a long tail (everything else, falling toward zero).

The pattern was given its current name by Chris Anderson in a 2004 article and a 2006 book, but the underlying structure is older. The classic Pareto distribution, the Zipfian distribution that describes word frequency, and the head-and-tail structure of best-seller lists are all closely related. The long tail describes consumer markets, attention markets, and most cultural distributions.

Why it matters for curation

The long tail matters for curation because it sets up two distinct editorial challenges, with different correct answers.

In the head of the distribution — the small number of well-known items — the editorial challenge is comparative ranking. Most readers know that MyFitnessPal exists; most readers can name the major office chair brands; most readers have heard of the major literary novels of the year. The editorial work is to rank these credibly and to explain the ranking with reasons. Curation in the head is comparative judgment work.

In the tail — the much larger set of less-known items — the editorial challenge is discovery. Most readers do not know about the better third-party Obsidian plugins, the better small-press novels, or the better mid-tier office chairs. The editorial work is to surface these credibly and to explain why they are worth the reader’s attention. Curation in the tail is discovery work.

A serious recommendation publication needs to do both. Top Picks Report’s lists deliberately mix head-of-distribution items (where the editorial value is in the ranking) and tail items (where the editorial value is in the surfacing). The Aeron is a head-of-distribution chair; the Branch Verve is closer to the tail; both deserve editorial attention.

On the algorithmic-recommendation argument

The argument that algorithms can replace curation is, in part, a long-tail argument. Algorithmic recommendation systems are good at the tail — they can surface items the reader would not have known about — and weak at the head, because head-of-distribution items are popular for reasons that algorithms do not weigh well (institutional reputation, editorial endorsement history, peer-reviewed evidence). The result is that algorithmic recommendation is most useful for surfacing the unfamiliar and least useful for choosing among the familiar.

Curated publications, conversely, are most useful for choosing among the familiar (the head) and only sometimes useful for surfacing the unfamiliar (the tail). The two practices are complementary; the conflation of them, as we have written elsewhere, is the category error.