Web 2.0 may indeed be waning as a meme, as Crosslink Capital's Peter Rip suggests . Technology evangelists and journalists have probably vented all there is to say about dynamic web pages and online social behavior, the two key ingredients of this round of internet projects. And they have moved on . Financial interest appears to be shifting, if anything, back: to first-generation web companies such as Webex and Opentable which are mature enough to command billion-dollar-plus valuations. But it would be ironic if web hype were to be deflated, not because of these real trends, but because of a blogging venture capitalist's foolhardy reliance on dubious Alexa numbers.

The hypothesis — that Web 2.0 is "over and out" — rests on a few charts from the popular Amazon-owned measurement service. According to Alexa, sites such as Techcrunch, Michael Arrington's highly influential tech news and reviews title, are on the decline. His reach, according to the Alexa chart, has dipped from about 0.30% of web users at Techcrunch's peak in November, to about 0.22% last month. Other specialist web news sites have shown similiar falls.

Seemingly conclusive. Except that Alexa is totally unreliable. It has long overstated the audience for the geekiest sites because the readers of those sites, who include many web marketers, are more likely to have Alexa's toolbar installed on their machines, and so show in the company's sample. But, now, not even the trend lines can be trusted.


A source at Alexa says all US sites are being handicapped, because of an explosion in use of their toolbar in China and India. It's not clear how much this is to do with genuine growth in internet use in those countries, or downloads of the unit's toolbar by Asian spammers. In any case, the effect is to dilute the reach of US sites as measured by Alexa. Only the most rapidly growing, such as Wikipedia, are holding their own.

The proof is easy. Techcrunch, bravely, publishes its own traffic stats, as measured by a Sitemeter tracker. The site continues to grow, reaching a record 3.4m pageviews in January during the tech trade show season, nearly 60% above its level of August 2006, as the chart shows below. Since November, Alexa's estimate of Techcrunch's audience has declined by 26%; its actual traffic has increased by the same percentage. Alexa is wrong.

Despite the methodological amateurishness, Rip's item seems to have touched a nerve; the post is the talk of other weblogs and news sites. Expect to see a batch of print articles on the coming backlash.

For the hype to be punctured so accidentally would be poetic. It was Alexa's inflated numbers for geek sites that sustained the hubris of entrepreneurs, and the wide-eyed enthusiasm of the press covering the new tech boom. Now — largely because of Alexa and China, the twin obsessions of many web entrepreneurs — some of that confidence will evaporate. This is like one of those stockmarket corrections, much overdue, triggered in the end, not by a change in the fundamentals, but by a seemingly trivial glitch in the trading system.