While I've had various things to do with the WordPress community back as far as 2003, it finally time to completely part ways: I have become less and less happy with the direction that WordPress has been taking over the past year, though the issues are really far deeper and go back, in some cases, to the very origins of the project.
Ever since the project started, Matt Mullenweg has progressively become more and more protective of the source base, especially since WordPress.com launched, making it progressively harder to get any code into the repository. For example, having discussed a bug at great length with a number of people including Matt, and written a patch for it, taking a great deal of my time, Matt then changed his mind and on his own decided that it wouldn't get committed. There have been plenty of occurrences of Matt single-handedly making decisions — if there is a secret cabal where decisions are really made can you at least stop claiming that the development community has any say in anything?
The entire project seems to now be run in such a way so that Automattic has software with as many features as possible to profit from, with little regard for any bugs or any features that are mostly invincible to the end-user.
It took over two years for WordPress to have Atom 1.0 support added. Why? Matt wasn't happy with the patches, despite there being plenty around for a long time which were bug-free. When Atom 1.0 support was added, for some reason or another, the comment feeds use a different pipeline (actually, they use a different set of string-concatenation strings) and was outputting absolute rubbish — |link|@content instead of @href; |link|@type contained the blog name, not a MIME type; |updated| and |published| contained RFC 822 dates, instead of RFC 3339 dates (which are totally and utterly different). Also, it is possible to get invalid bytes into the feeds, which under XML is a well-formness error, and must therefore cause a fatal error (luckily for WordPress, out of the major browsers, only IE/Win actually obeys this). So, when a patch finally gets committed it is too much effort to visit the Feed Validator before release? Why be so overly protective of the source base if you let such rubbish in anyway? Also, as one final note on the subject of Atom 1.0 support, Matt said,
this is a enhancement, not a bug, despite Atom 0.3 being an obsolete I-D, a series made publicly available for comments before publication, and is liable to totally change. Any change of a draft is a bug in any older implementation.
While some may argue that the above is merely an enhancement, surely nobody would argue that a
critical bug that caused IRIs to be stripped should be fixed in a plugin? This really seems to be the case.
Furthermore, I have serious issues with WordPress's
focus on aesthetics, web standards, and usability. If there is a focus on aesthetics, why is Kubrick the default theme, even though there are far better templates available? If there is a focus on web standards, why did I even write the above paragraph? Why does by default WordPress use a transitional DOCTYPE? Is WordPress still transitioning to standards? Why is WordPress.org served as it SHOULD NOT be (i.e., XHTML 1.1 served as text/html — even XHTML 1.0 would be better!)?
Lastly, the thing that finally made me think that I should totally get off WordPress (my blog had for a while been running the security-fixes-only 2.0 branch, to avoid the chaos and the insane bugginess of later releases) was that after having decided to have 120 day release cycles, including one month after a feature freeze, Matt went and commited what is one of the largest changes in several years to WordPress within the feature freeze, causing the entire release to be pushed back. This, IMO is the ultimate example of Matt focusing on maximising WordPress.com features and profits without caring about affects it might have to the open-source project.
So, where does the future lie? Habari. Habari is developed under the Apache meritocracy model, so it should be far harder for some benevolent person in a position of power turning against the wishes of the community as a whole. Also, Habari is developed for modern web hosting environments, and makes use of open standards, and is not afraid to go against the de-facto standards of the day, such as XHTML superseding HTML, provided reasoning is given. Due to Habari's design, using PDO prepared statements, and an XML serialiser for what XML is outputted, it is far less fragile than something patched together over the years everytime something breaks such as WordPress.
Also, due to Habari's organisation, it is possible for someone completely new to the community to step in and say something that will totally change the direction of the project (after, as with many things, it has been voted upon), something that cannot be done by people who have been around WordPress for a long time, yet alone someone who is completely new.
Many of the largest contributors to Habari were WordPress contributors previously — some of them well known within the community — who left WordPress often for reasons similar to the above. As far as I can see, since a number of them walked away from WordPress, WordPress has gotten progressively worse and buggier. These are people who experience with working with blogs. They know what past mistakes have been made. Above all, they are willing to change their opinions if you give them good reasoning.
For the first time since 2005 (when I got through three designs in ten months), my site has been redesigned. Unlike the earlier designs that were thrown together relatively quickly, this one has taken over a year: minimalism, when done properly, is no easier than something more visually complex. The colour scheme has changed around five times throughout the course of the design, and the number of images has varied from zero to two (yes, that's the most complex the design got).
More will eventually be posted about the design, and the inspiration behind yet. Alas, the future will hold different challenges to those in the past, and any future designs will therefore be different to what this is.