The Long Tail is about how the mass market is turning into a million niches. The term refers to the yellow part of the sales chart at left, which shows a standard demand curve that could apply to any industry, from entertainment to services. The vertical axis is sales, the horizontal is products. The red part of the curve is the "hits", which have dominated our commercial decisions to date. The yellow part is the non-hits, or niches, which I argue in the article will prove equally important in the future now that technology has provided efficient ways to give consumers access to them thanks to the "infnite shelf-space effect" of new distribution mechanisms that break thought the bottlenecks of broadcast and traditional bricks and mortar.
The two big points of the Long Tail theory are these: 1) The yellow part potentially extends forever to the right; 2) The area under that line--the market it represents--may become as big as the hits at the left.
The Wikipedia entry on the Long Tail does an excellent job of expanding on this.
The shift from hits to niches is a rich seam, manifest in all sorts of surprising places. This blog is where I'm going to collect everything I can about it.
As the Wikipedia entry on LT emphasizes that it is simply a popular name for a "long-known feature of statistical distributions (Zipf, Power-laws, and/or Pareto distributions)". Accordingly, it applies to various aspects of all phenomena, including software production.
LT is fundamentally based on the dialectic between "convention and invention": when a process becomes very convenient and affordable through commoditization, this enables more unique needs to be met through invention.
A great software example of this that I just ran across is machinima: "a new form of filmmaking that uses computer games technology to shoot films in the virtual reality of a game engine." (See also the Wikipedia definition of machinima.) In this instance, the "convention" is a commoditized "game engine" and the invention is the variety of scenes that individual "artists" create with it. Red vs. Blue (movies created with the Halo engine and images) is the most popular example of machinima, but check out the growing number of examples, including an MTV machinima video at machinima.com.
Mass customization is how large corporations can address the LT, but mass innovation (aka mass amateurization) shifts control from the seller to the buyer: the buyer customizes the product or service without the participation or even the approval of the vendor, cf., the recent NY Times article on the relationship between Nike and sneakerheads who customize its sneakers. (See also my collection of mass innovation links.)
The most WS-* related example that comes to mind is Greasemonkey: "a platform for running scripts that inject new functionality into web interfaces. If you're a UI designer, this might frighten you. What it means is that any kid with a bright idea and a knack for DHTML can create a new interface for your site, and it will probably be better than yours. (There's a lot of bright kids out there in the world.) Why should you get paid when the bright kids will do your job better for free? The key to survival will be going meta: design for the bright kids. Create a flexible, modular set of APIs and a well-documented example UI or two that shows how they are used. Learn from Amazon and release your grip on the end-user experience. But developments like Greasemonkey disrupt more than just job descriptions: they disrupt business models too. For example, I will never see a Google AdSense ad again, thanks to a handy Greasemonkey script." [Greasemonkey Stole Your Job] Greasemonkey enables anyone to "reprogram" Firefox to filter and transform any and all web interactions ON THE CLIENT SIDE. Now the "look and feel" of a site in no longer in the hands of the site owner, but in the hands of the browser owner or any intermediary.
BTW, my fellow blogger Tom Murphy has a nice link to Long Tail thinking applied to specifically to software.