I cannot see why people use WordPress over the likes of Concrete5 or a full CMS on any other platform outside of WordPress being accessible for entry-level developers. For this reason, a full rewrite of WordPress to be solely a basic CMS would probably have these novice developers flock to the system, allowing WordPress to scale back down to doing what it does best.
I run a very modestly profitable website that gets about 1k visitors per day. I'm not even an "entry-level developer", I'm not a developer at all, with only the most rudimentary ability to kludge together a little JS and php.
I chose Wordpress for the site, even though it's not a blog. In fact, creating a system where posts weren't displayed in chronological order actually took a few hours of blindly hacking at various WP functions.
The reasons I chose Wordpress over an existing CMS:
* I already know how to use WP
* There is a vast array of FLOSS themes and plugins to meet virtually any need available
* Wordpress is well supported by shared hosting and managed VPS providers (through cpanel etc.)
* It's relatively easy to get free high-level support through IRC because so many people know WP
Most of these reasons are inertia based. It would be a vast undertaking to create a CMS version of Wordpress that actually gained traction because, awkward as WP is, it has an enormous library of free existing solutions to take on virtually any problem. This quality, which the author of the article complains about, it the same reason WP would be so hard to kill. Even if your product is better for a given purpose, people aren't going to give up the advantages of WP I outlined above just for an incremental improvement in UX or speed.
Now I'll be the first to tell people there are certain things that you should build from scratch, but the examples of what people build with WordPress can be pretty breathtaking even to me:
The point I disagree with the most is that users won't need training to use it. I hear this all the time with absolutely nothing to back it up, other than "I haven't trained a user to use it".
Well, I have, and the users for this company struggled a lot with how things will work. I've trained users on numerous scripts, including Umbraco, Sitecore, Concrete5 and bespoke CMS's and every one of the others were far easier to both sell to a client and train. Umbraco and Concrete5 are far better for users than WordPress.
But that only includes *.wordpress.com subdomains, and our highest traffic blogs almost always invest the money to have their own domain. We have a tag to track those in Quantcast, and it's currently at 129.7M people in the US, which would place it between Facebook (143M) and MSN (98M). (Blogger might have a similar boost into the top 10, but I don't see any others in the top 50 that could have so many mapped domains.)
Of course we cache, with a publicly available WP plugin called Batcache.
I agree that using caching is normal, but the difference is the total reliance on caching. I wouldn't be impressed with google's scalability either if it were 99.9999% static pages being served up.
I think we agree that "scaling" small (sub-RAM-size) amounts of data to a largely logged-out and cached audience is easy, but I think you think of WordPress.com as much smaller and more static than it actually is. My apologies if I'm misunderstanding your point of view.
If you want to see standalone sites in the top 100 running WP besides wordpress.com, check out time.com, umbrellanews.com, celebuzz.com, and large sections of nytimes.com, cnn.com, and people.com. If you were to spider the top million Alexa domains, you'd find about 17% of them on WP:
I realize lots of people run wordpress. That doesn't support the claim that it easily scales to the traffic demands of a top 10 site.