The frighteningly insightful Om Malik addresses a question this week that has been rolling around The Browser's head recently: Why is it so hard to get reliable figures about Web traffic? He specifically goes after Alexa, with some arguments we've heard before (that the Alexa tool bar doesn't capture Firefox users), and some that are new to us (that Alexa apparently goes down for substantial periods, up to a day a month).

But the problem is actually much bigger than that, and has begun to move from the griping stage to something approaching an industry revolt. Anybody who publishes a big site like ours can tell you that there is a massive discrepancy between the audience we measure and the audience reported by professional outfits like Nielsen/NetRatings and comScore Media Metrix - and not a discrepancy, from the publishers' point of view, in the right direction!

On some level, there is a legitimate philosophical debate about what should be measured: Web publishers tend to emphasize page views, mostly because that number comes out sounding really big. By contrast, the yardstick firms focus on unique visitors, arguing that measuring page views brings in all sorts of noise, like counting pop-ups or even pages viewed by robots. It's also worth remembering that arguments about how to properly measure audiences are hardly unique to the Web business.

But there's another dynamic at work here worth thinking about. For all of the arguments about how information wants to be free and how the Web makes our world more transparent, it's pretty ironic that very basic data - like how many people visit which sites - are available only to those willing to pay.

Malik suggests that a kind of consortium of Google (GOOG), Microsoft (MSFT) and Yahoo! (YHOO) could come together to address traffic, an approach that strikes The Browser both as unlikely - too much proprietary info at stake - and oddly corporatist. Web 2.0 advocates in particular like to talk about outside disruptive technologies (think Napster) exploding entire industries; imagine how disruptive it would be to find a technologically sound way of measuring Web audiences, and then making that data freely available. Is anybody working on that?