Over the weekend, I had problems upgrading installs of both WordPress and Movable Type. The latter you may have noticed if you stopped by here at any point.

I find the implications troubling, so I’m going to indulge in a bit of an essay. The non-web-geeks amongst you should probably skip this post.

Movable Type

Movable Type has a very fresh and pleasant back-end interface, but futzing around with upgrades is a clunky process. There are lots of files to move – even a minor upgrade involves a complete reinstall – and when something goes wrong it’s not always obvious where to look for diagnostics, nor where to turn for help. There are all kinds of problems:

  • The documentation is written to describe the system, not to explain its use. Where explanatory pages exist – and they often do, it seems – they’re usually not linked to each other. I think this is mostly because the docs have been in a state of considerable flux of late, making it hard to weave them together in the subtle ways which turn out to be useful.
  • The docs are still incomplete, and there are areas where it’s not clear if the docs – and particularly the comments, where much of the solid advice lies – are current, or dating back to MT3.x.
  • The footer of MovableType.org pushes one towards these forums… yet the ‘current’ forums, per my understanding, are here instead… and often the most helpful responses are on the MT consultant’s mailing list ProNet. Don’t get me started on the wiki.
  • Baffling website plurality: MT’s schizophrenic nature as open-source code on the one hand, and a platform for corporate consulting services on the other, leaves one baffled. Recent reorganisations have helped, but it’s still a right mess.
  • Not only is it a mess, but bits of the various MT websites keep falling apart. This weekend, it was search at the Documentation site (still broken as I write this), but at least logins and posting were working at the forum.

To be fair, these are all areas of current work for Six Apart, the company behind MT. And the pace of change has been ferocious recently, with lots of good stuff happening in the last few months, from the Mid-Century templates I’m using here, to Action Streams, to the startling performance improvements of MT4.2. It’s OK, though not ideal, that the docs should be catching up; that the structure of the websites is still confused is more concerning.

However, it’s remarkable that I’ve been using MT here for six years. Database upgrades have been, that I recall, seamless, and template backward-compatability is exemplary. There’s lots about MT to like.

So my complaint is that it’s merely ‘good’, and still, after all this time, has some way to go before it’s ‘great.’ Recent rewrites may have improved publishing performance, but the application itself stills ‘feels’ slow, and I’ve no clear idea of why using it under fastcgi on my server is unstable. I tried, briefly, at the weekend, and while it’s dramatically faster, it still throws 500 errors at seemingly-random intervals. I might try again.

I’m left enticed by some of MT’s promise, impressed by the flexibility and relative ease of hacking its templates (and not having to deal with database gubbins in the process), but suspicious that life could, nevertheless, be better.


The standard answer to ‘I need a website’ is, it seems, ‘WordPress.’ It’s a terrific package. So what are my beefs with it?

  • Updates. Updates updates updates. Sure, it’s highly visible for hackers, and an obvious target for attack. Hence, lots of security patches. Fine. But such fatalism hides two issues: first, the nagging suspicion that having database access in the page templates is plain bad application architecture, and that can’t help. Second, that frequent updates are a pain in the arse for users. They may be the price one pays for ‘free,’ but it’s too high a price.
  • Updates that break things. Of all the WordPress installs I’ve done – about thirty or so, I guess – only one is now (a.) current and (b.) still running. I’ve lost count of the times my database has failed to update properly, or new stuff has broken old plugins in a catastrophic way, or something else funky has happened. WordPress, like Windows, appears to have a half-life of about 18 months. Perhaps less.
  • Even the working install I have (which is less than two years old) has database issues, in that a few releases back the default text encoding was changed without, it seems, any note in the upgrade docs. This bit me at the weekend, when I tried to refresh the config file. While the fix was easy – back out the configuration changes – I’m now left with a non-standard database format. Updating the database looks painful and risky; but if I don’t, I’d give the install another six months before it craters.

While most of the WordPress installs I had that went really bad were a while back, and it certainly seems more stable than it used to be, the bottom line for me is that I just can’t be bothered with it. Keeping up-to-date is simply more effort than I’m willing to put in.

Again, I’m left with feeling like there should be something better.


My obvious recourse should probably be a hosted service, likely at WordPress.com or TypePad. Trouble is, keeping current with at least one web publishing system is pretty much essential for the work I do, where I may have to lob up a prototype or even deploy a site at short notice.

For example, just as we were convening the judges for SciCast last year our main server collapsed. I don’t have any control over that, but the problems were clearly serious: I built out a judging site in Movable Type more-or-less overnight.

I’m willing to invest a little time to be able to do that sort of thing… but perhaps not as much investment as currently.

Sandvox? RapidWeaver? I even own a copy of the latter, but I’ve never really used it – partly because I’ve a suspicion I prefer the former. I should play with these, though – they may be useful for quick lash-up jobs.

Expression Engine, perhaps?