Agree with parent and gp. Not only that, the article lacks an Econ 101 understanding of microeconomics. Most markets have a few big firms because of the pervasive effect of economies of scale[1]--generally, bigger firms have lower cost per unit. That's why 1-3 is common in many markets. (For an interesting discussion of why it's socially advantage to have N>=2, see the notion of "deadweight loss" that happens when there is only one firm[2]).
Article is right in one way: software is very different from physical goods. The marginal cost of each additional unit of software is 0. And there's evidence (such as appears frequently on HN) that a bigger firms do not necessarily produce software more efficiently. I've always suspected the software industry is more dominated by network effects[3], which has a similar effect on the market as returns to scale: a small number of firms & high likelihood of natural monopolies.
I think a big difference though is geographic segmentation. Pepsi could dominate the Southwest and Coke could dominate the Northeast; there's no reason why that would be true of amazon vs. buy.com.
We do see that with internet companies only for national or language boundaries, e.g. baidu in China, gumtree dominates in the UK vs. craigslist in North Am, Yahoo beats google in some asian markets, Orkut is popular in Brazil, etc, etc.
Article is right in one way: software is very different from physical goods. The marginal cost of each additional unit of software is 0. And there's evidence (such as appears frequently on HN) that a bigger firms do not necessarily produce software more efficiently. I've always suspected the software industry is more dominated by network effects[3], which has a similar effect on the market as returns to scale: a small number of firms & high likelihood of natural monopolies.
[1] http://en.wikipedia.org/wiki/Economies_of_scale
[2] http://en.wikipedia.org/wiki/Deadweight_loss
[3] http://en.wikipedia.org/wiki/Network_effect