I've been using Movable Type as my blogging system of choice for several years now -- not just because it's open source but because it's a good program with great features. And yet the newest revision, version 5, feels like it falls far short of what could -- and needs -- to be done.
When work started on version 5, I'd hoped the folks at Six Apart (the company behind Movable Type) would take the time to correct many of the longstanding problems with MT. These are issues that people have cited again and again, and in many cases have used as reasons to ditch MT entirely and go with the far more popular WordPress (which, for all of its faults, seems to provide its users with features they want.)
Installation and upgrading. Getting MT set up for the first time with no existing install isn't too bad, but getting it upgraded to the latest version requires a good deal of file-copying calisthenics. The situation's made all the trickier if you have plugins that drop files in weird places, and you don't want to mistakenly overwrite something. It gets even more complicated if you have a cgi-bin directory that isn't world-readable, since MT tries to put some world-readable files there by default.
Live publishing. MT's main focus has been on static publishing -- generating regular HTML, which can be served up quickly but has to be republished every time you make a change. The program's live-publishing option -- which generates pages as you request them -- suffers from two major problems. One, it's coded in PHP, which puts it at odds with the rest of the program (written in Perl). Older plugins, many of which are in broad use, won't work with the PHP live-publishing system, so those users are stuck with rebuilding static pages over over again. A third-party plug-in attempts to replace the PHP-based publishing system with a Perl-based one, but I had little luck getting it to work well.[Note: I've since found that MT5 has much broader support for dynamic publishing options, so I'm tentatively taking this off my gripe list.]
Unwanted features. E.g., the Zemanta plugin, which didn't work with a lot of configurations (especially non-English ones). The effort involved to add that could have been far more productively directed into any number of other things people had been asking for. This doesn't seem to be a new problem: at one point while talking to folks on the development team, I saw other features in progress which were technically impressive -- but on second thought they seemed more gimmicky than truly useful.
Now I'll back up a bit. The changes that are in MT 5 make the program more than worth the upgrade -- a redesigned UI that makes a bit more sense than the last one, for instance; performance improvements; some new template-language functions. I also don't plan on jumping ship to WordPress any time soon, which would mean ditching my entire investment in MT -- not financial, exactly, but the work I've put into making the templates behave as I want.
I do, however, plan on giving Melody -- the open source fork of the Movable Type core -- a good, hard look. I'm hoping the development team for that project will be able to listen that much more closely to their audience.
InformationWeek has published an in-depth report on new software models. Download the report here (registration required).