Accuracy of browser statisticsWednesday, June 11, 2003
One piece of "information" is pounded into web designers over and over and over: Microsoft's Internet Explorer is completely dominant. To prove this point, the statistics are wheeled out with increasing regularity. It looks convincing, the numbers look sizable. But...
Bjorn Jacke has an interesting article about how web browser statistics lie and tell us nothing. There are some commonly known grey areas, such as User-Agent spoofing, but Bjorn has brought some new ideas to light, such as IE's propensity to open multiple connections to a webserver and download partial content.
Googling around for more info on the accuracy of browser statistics, I found a similar article by Jeff Goldberg, "Why web usage statistics are (worse than) meaningless", although an old article it does cover the issues of caching proxies and some of the more common myths of stats.
There's a succinct quote by Michael Olski on the Chimera mailing list: "That IE has 90+% of usage today, in no way assures that that will still be true tomorrow, or next month, or next year. That's the key to standards compliance!". This highlights the danger of using current web stats to justify non-standard design decisions - it is bound to fail in the future, its simply a matter of when.