Wired on Friday: Between its founding in 1993 and 2001, Excite went from being one of the top search engines on the net, and the darling of investors, to auction in bankruptcy court.

Back when it was on top, it found it had a few very popular search queries - such as sex, MP3s or Britney - and millions of far less popular queries every day. It understandably focused on how to make money advertising on those few highly popular terms.

Joe Kraus, co-founder of Excite, now sees that this approach doomed his business. On his weblog, he writes: "We couldn't figure out how to make money from 97 per cent of our traffic. We couldn't figure out how to make money from those queries asked only once a day."

The editor of Wired magazine, Chris Anderson, has spent his free time over the past year exploring an economic oddity he calls "the long tail", after a graph showing the phenomenon.

Essentially, Anderson is studying nanoeconomics, or small numbers of purchases by a few customers - niche markets of dozens instead of mass markets of millions.

The market wisdom Kraus knew was formed in the world of physical sales. Shelf space is scarce and reserved for the things that sell the most.

This has a cyclical effect; the most popular things become more popular because they are the most available and obscure properties quickly become impossible to find. A few hits make up most sales and the misses - the lion's share of all products made - make up a trailing percentage at the end or aren't available at all.

While physical availability is often seen as a sign of quality, many quality products simply never enter the virtuous cycle of popularity.

In the nanoeconomic world of Anderson's long tail, some markets exist without physical constraints and everything becomes equally purchasable.

Everything is just as findable, whether it's a hit or a miss. Popular pieces still sell well but we can buy any obscure thing that our quirky selves need. We all have quirky selves - unmet interests that most people can't pursue in a mainstream market.

Increasingly, our idiosyncrasies are finding their analogs online and claiming their portions of our disposable income. The long tail operates most strongly among online businesses, which don't have the trouble of shelf space.

A physical Amazon.com bookstore would be the size of a shopping centre at 2.3 million titles. This explains why Amazon's biggest real-world competitor in the US, Barnes & Noble, carries 130,000 titles.

How do Amazon's customers respond to their increased choice? The statistics are illuminating: between 25-50 per cent of their sales come from outside the 130,000 physically available books. (No one knows for sure the exact number, in part because no one can say for sure which books are available in physical retail.)

Most companies doing business in the long tail aren't seeing percentages like Amazon's. Those newly available items may make up closer to 20 per cent of overall sales. But still, a fifth of sales isn't something most companies are willing to drop on the floor.

So the new opportunity of the long tail in online sales is beginning to interest traditional businesses that once focused on making mass market wonders.

In some ways, the long tail has been with us for much longer than the internet.

"The Sears catalogue was the long tail of the Kansas farmer," says Anderson, referring to the mail-order catalogues that opened the US mid-west to the bounties of the big city stores in the early 20th century.

Anderson sees that Google in particular ate Kraus's lunch in a way that no one could have foreseen. Through Adwords, a service in which it sells ad space by the penny associated with individual search terms, selected by advertisers from a web-based application, it democratised not only who would see ads but the process of becoming an advertiser. It discovered millions of people that would have never considered themselves advertisers, and connected them with millions of niches never before reached with ads.

There's evidence that the long tail changes the market, flattens out the stars and allows more products to make some money. The sales curve looks less severe when revenue is drawn away from the top, and may still be more in total sales.

But taking advantage of the long tail requires both the hits and the niches. Perhaps, most importantly, it requires the ability to treat consumers as individuals, something much more doable with technology that creates customised interfaces, and connects niche customers to each other.

Anderson plans to release a book on the long tail next year, which will probably be a hit. Kraus has learned his lesson, and is back with a new company, JotSpot, this time focusing on the long tail of software needs. With any luck, he'll be the next small thing.