But what percentage of them are active sites and what are parking pages / domain squatting? It's been shown in the past that a single domain parker moving their hosting can have a vastly disproportionate effect on these numbers while affecting almost no real end-user traffic.
As it stands, this statistic is vastly too crude to be of any value.
There are lots of statistics in the survey. They do talk about "Active" sites. I just thought that this one particular comment was the most surprising, so stuck it in the title. It's still a pretty impressive statistic, even if it does include parked pages.
I don't know this factually for this specific instance, but often times, the big moves seen in "hostnames hosted" are caused by domain name registrars changing platforms. For exaple, GoDaddy serves an ad billboard page for every expiered or un-configured domain registered with them. This results in a huge number of these pages, so when/if someone like GoDaddy moves from, say, IIS to Apache (just an example), you'll see a big swing in the charts.
These kinds of statistics are interesting to look at, but they don't mean much to me. I select a web server based on suitability to my needs, not "what are all the cool kids using".
As it stands, this statistic is vastly too crude to be of any value.