The long tail

slate presents a good discussion abou the concept of the long tail:

"What are the Long Tail's limits? As a business model, it matters most 1) where the price of carrying additional inventory approaches zero and 2) where consumers have strong and heterogeneous preferences. When these two conditions are satisfied, a company can radically enlarge its inventory and make money raking in the niche demand.
...
it's important to remember that many industries don't rely on the weird economics of information products.
...
The Long Tail also sometimes doesn't work in its home category: the information-technology industries. The key issue is the question of standardization. Sometimes consumers want a diverse set of product offerings. But sometimes they prefer a standard or compatible product. Most of Anderson's examples are content firms, where product diversity is almost always a good thing. But in the information-transport industry, standardization is usually more important. Do people want 10 different types of (incompatible) Internet connections? Or just the fastest one they can get? How about 30 types of (incompatible) Ethernet cables?
...
for every diverse Long Tail there's a "Big Dog:" a boring standardized industry that isn't sexy like Apple or Amazon but that delivers all that niche content you're hungry for.