Scoble points out July survey results showing an increase in IIS hosts while Netcraft's August results tell quite a different story.

Before the marketing folk get too excited about perceived gains in market share and glorious triumph over the evil competition (note: the evil competition is always "the other guy") it should be pointed out that these surveys don't actually measure market share - they measure what operating system (amongst other things) each host that they survey is running.

For instance, Netcraft rightly points out that Verisign is in the process of moving a lot of hosts from Interland (who use IIS) to somewhere else (who uses Apache). Does this mean that Microsoft lost thousands of customers? Nope, it means that Interland lost one customer - Verisign. Microsoft didn't lose much, if anything. These numbers are not all that meaningful when it comes to measuring market share.

We do need a number that the marketing folk can get excited about though. Numbers promote competition and competition promotes innovation. I like innovation and I'd encourage all of the vendors to start releasing stats that more appropriately describe the market share of the respective platforms. Personally, I'd love to see a number that quantifies, in absolute dollar amounts, the percentage of ecommerce that each platform is responsible for faciliating. Or maybe, total HTTP gets...or how about security patches per second? Segfaults per month?