• Running the scripts (which included FTP to the server) to publish a new article took a while, so it wasn’t something you could just quickly do.
• The scripts were only available on one machine, so publishing could only happen from there.
• When my comments form was turned on, it got a bunch of comment spam.
• If I wanted any new features, I had to do them all myself.

I thought I’d take a look at some actual blogging software to see what was available. I ran across an extensive comparison (though I can’t remember where that was now) where the person doing the comparison decided on using WordPress. I had thought about trying to use a .NET application, since I’m trying hard to learn .NET deeply right now, but there was a telling comment about .Text (the only .NET blogging tool considered) in the comparison. The reviewer said something like, “.NET programmers are more excited about .NET than actual working software.” While a sweeping generalization, I sense a bit of truth in there. Also, as a programmer, it is good to be exposed to lots of things, so a PHP application seemed good for technical diversification.

I created my own WordPress theme, both in order to preserve my site’s look as well as to learn more about WordPress. Then I needed a way to migrate my content. I don’t have a ton of articles, but certainly more than a manual approach would work for. I tried adding a couple of articles by hand, and quickly discovered that the WordPress editor treats all whitespace as significant, unlike HTML’s text flowing regardless of line breaks.

My first attempt at an automatic import was to transform my content to WordPress’ native export/import format. I abandoned that after a little while because it was based on RSS, but added a bunch of stuff that wasn’t really relevant for my content. I realized that I might as well just do an RSS import, which was easier.

The significant whitespace characteristics I noticed in the editor carried over to imports as well, so my conversion program had to strip out all the non-significant linefeeds out. I also discovered that if there were XML namespace declarations where the importer didn’t expect them, then it would think that there weren’t any posts to import. It would just say, “All done! Have Fun!”, which was kind of infuriating until I figured it out. I think I’ve heard someone say that blogging software is only as good as its import and export functions, and I almost gave up on WordPress at this point.

Once I got my content imported, I was happier about WordPress when I discovered Ryan McGeary’s WP-Syntax Plugin. I had thought about doing something like that myself at some point. I did make one small tweak to the plugin, though. GeSHi’s default syntax colorings were a bit too much for me, so I switched the GeSHi option over to use CSS styles for the highlighting, and chose fewer and more subtle colorings. I’m sure there are lots of other interesting plugins that I’ve yet to discover.

There have been some glitches (bug-like behavior), some manual work required even after my automatic import, and still some things to figure out, but I’m optimistic that WordPress will work out.

My program, ByChance, works like this. You supply an image that represents the ideal in survivability. Once you’ve done that, you click the “Start” button. The program starts with a black image with the same dimensions as the “fitness” image you picked. It then proceeds to pick a pixel in the image at random (around 200,000 possibilities, depending on the shape of the image). Then it picks a random color (24 bits, so there are around 16.8 million possibilities) for the random pixel and tests the survivability of the new picture. If the new picture is more like the “fitness” picture than what was there previously, the new picture, with its single different pixel wins. If not, what was there before wins. The program displays the winning picture, and also keeps a running total of the number of generations that have happened. A generation is every time a pixel changes and the fitness test is done.

Early Screen Shot

Later Screen Shot

I’m certainly not an expert in evolutionary theory, but it seems like the program serves as an interesting analogy. I should point out that evolution doesn’t have a perfectly ideal goal that it is shooting for like the fitness image. Given the forces of nature around us and all the aspects of our environment, there is pressure toward a survivable form.

Some of the choices I made in writing the program were arbitrary:

• Why not make the image size random too?
• Why start with black? Why not start with white or a completely random starting image?

Maybe something a little different would be a better analogy of our best understanding of what really happens. This experiment is kind of related to the infinite monkey theorem, but with not so many permutations and with pressure toward correctness.

After running the program several times, I’ve noticed a couple of things. The image becomes recognizable really fast. The image becomes perfect some time later than my patience allows for, but I expect it would take a really,really long time.

If you would like to run ByChance, you can download it from here.