The Friday after Thanksgiving is the traditional start of the holiday shopping season in the US, and retailers have now created a bizarre ritual in which people are lured to the stores at 5am with the promise of extreme bargains that sell out almost immediately. The spectacle of screaming mobs and fisticuffs over cheap DVD players is, I suppose, a testament to Americans' love of a good deal, but it also makes me wonder: why don't the many people (your truly included) who despise the Christmas shopping chaos buy more of their gifts online?

It's certainly true that online shopping continues to grow rapidly, with forecasters predicting a 20 per cent to 25 per cent increase this year, but it still accounts for well under 10 per cent of all retail sales. Judging by the mail at my house, it's hardly made a dent even in the catalogue shopping business – of which it is a kind of electronic extension – and predictions that the internet would bring about the end of the shopping mall and the department store and the car dealer now look, at best, radically premature.

To understand the promise and limitations of online retail, it's illuminating to look at the fortunes of the Amazon.com. Founded in the mid-1990s as an online bookseller, the company soon raised its sights and set out to become an online superstore, the Wal-Mart of the web. Jeff Bezos, its youthful and appealing founder, was nothing less than Time Magazine's Person of the Year in 1999, a kind of proxy for the many monumental changes heralded by the internet.

By most measures, Amazon has been an extraordinarily successful company, with revenues of more than $8 billion last year and a current market capitalisation of some $17 billion. But it's a very, very long way from being Wal-Mart (revenues: $312 billion, market cap: $200 billion), and hasn't even succeeded in dominating the book business to the extent that many expected. Its recent decision to expand its business by renting server space and other infrastructure and become a provider of web services was seen as a sign not of strength, but of just how far it has fallen behind Google in seizing the opportunity of the internet.

The fundamental issue for Amazon has always been a simple one: margins. In theory, online retail should have much better margins than traditional retail, because you don't have all those expensive leases and sales assistants and other physical-store costs to deal with. In practice, however, it has turned out that the infrastructure needed to provide a good online shopping experience – both the web technology and the fulfilment apparatus – is quite expensive.

The slick interface and whizzy features of Amazon – the recommendation engine and the price comparisons and gift programs and all the rest – don't come cheap (which is why most smaller online retailers don't have them). And then there is the shipping issue. Amazon's margins and sales volume are very sensitive to offers of free shipping, and to the costs of warehousing. On products that are physically heavy, delivery costs alone can eat up most of the savings from not having those pesky stories; that's basically what doomed the big early plays in the online grocery business.

Indeed, if the norm in retailing was that the seller would deliver things to your house, people would be touting the great innovation and costs-savings of having people go to a central location and pick up their purchases themselves.

What the internet is best for is information arbitrage, i.e. the ability to gain advantage by knowing more, and thus it is far more common for people to research their purchases online than to actually buy things online. Thus it is important for retailers to have a good web presence, but in many categories, understandably enough, people still want to touch the merchandise. A book is a book, but a pair of shoes is something else.

And then of course there is the fact that for many people, shopping is a social activity, a form of entertainment. Both department stories, which are experiencing a revival in the US, and mall operators understand this very well, and thus are investing more heavily in the full experience. It's hard to see how any web interface can compete on that level.

Ultimately, the growth of online shopping will be limited by the proportion of the population that likes to go to the store. With the rise of broadband and continual improvements in the technology, online shopping will surely grow for many years to come. But it won't displace the apparently widespread desire to join the scrum at Target on the day after Thanksgiving. Personally, I'm thankful for the option.

Explore Tech & Web

• Personal Tech

• The Web

• Gadgets & Gaming