There are lots of surveys and reports on the popularity of various content management systems. Are they reporting genuine popularity or downloads?
Modularity confuses people
Some CMSs are highly modular and the core download contains very little that changes from week to week. Most of the changes are in optional add-on modules that may change as often as twice per week. The base download might change once every few months and record only 10,000 downloads per week. A popular add-on module undergoing rapid development might be downloaded 100,000 times per week.
Now look at an all-in-one CMS where everything is in one big download. To get the latest version of an optional module, you have to download the whole package. Instead of an add-on module recording 100,000 downloads and a base package recording only 10,000 downloads, you have the base package recording 100,000 downloads.
I prefer the modular approach where you download only what you need, the core software remains stable, and you do not have to constantly refresh the whole Web site just to get one tiny update for one module used on only a few sites. The popularity statistics do not look as good for the CMS but the workload for the Web site maintainer is far less.
New versions distort the figures
The major CMSs release major new versions every two or three years. When a new version is out, downloads skyrocket for six months until the new version becomes stable. If none of the competing CMSs have new versions, then the CMS with the new version records many more downloads than everyone else. A year later the high scoring CMS drops down in popularity, a different CMS releases a major new version, and a different CMS hits the top of the popularity polls for a year.
Blogs or Web sites?
Wordpress is a CMS used for a lot of blogs. Drupal and Joomla are CMSs used for full Web sites. While all three can be used for blogs and all three as CMSs, the historical difference gives Wordpress a lot more downloads than the other CMSs.
Blog sites distort reality
If you want to focus on blogs and look in detail at the various ways to create a blog site, look at cPanel, Fantastico, and similar products. A lot of cheap hosting services aim at the blog market with Wordpress installed through the hosting service control panel. Using those services, every blog is a separate download for Wordpress.
Now compare a simple blog site using Drupal. You download Drupal once then switch on the option that gives every user a blog. You can have 20,000 blogs running from one Drupal download.
You can use either approach with most of the top CMSs. The important difference is the most common approach. Wordpress is commonly used one way and Drupal the other way. Joomla gets a bit of both. For Drupal you would have to measure the number of blogs using Drupal, not the number of downloads used to create those blog sites.
Multisite creates more sites per download
The Drupal multisite feature lets you create a lot of Web sites from one download of Drupal. Some other CMSs have an equivalent option but the equivalent options in those other CMSs are used less frequently than the Drupal multisite option. The number of Web sites using Drupal is far greater than the number of downloads used to create those Web sites.
I currently run over 200 Web sites from 12 copies of Drupal. That type of multisite use is common and makes surveys of downloads useless.
Domain module creates many more sites per download
The optional Domain add-on module lets you create multiple Web sites using one download of Drupal. While the multisite approach is good for a diverse range of Web sites, the Domain module approach fits Web sites with a lot of common features. The Domain module approach lets you set up tens of thousands of similar Web sites with very little work. This brings you back to the problem of counting Web sites instead of downloads.