It's Christmas, 2012 as I write this, and as is so often the case when the year winds down and the paying customers grow scarce, I think about the immense technical debt that my website represents.

I registered vivtek.com in September of 1997. Vivtek.com is older than BoingBoing, older than Slashdot, older by far than Google or Facebook or Twitter. It's only a month younger than Sluggy freaking Freelance and it's older than User Friendly. And when I started it, I had no idea whatsoever how to build a website.

In my defense, neither did anybody else.

I chose AOLserver as the infrastructure because my main paying customer at the time (OK, my only paying customer at the time) was running AOLserver. AOLserver runs Tcl for dynamic content, and at the time the "active page" wasn't really a going thing. The model was CGI-like; you'd register a URL on the server against code, and that would run and return HTML. Hand-written HTML that had never heard of CSS; CSS had been defined in 1996 but most browsers didn't support it yet.

OK. So pretty quickly I realized that the URL could still map onto the file system, and I put tags into my pages that would be interpreted on the fly whenever a page was loaded. This worked pretty well because AOLserver is actually pretty fast, and my traffic was never horrible. Initially, the tags were just things like the footer to the page, and later the navigational sidebar, and they were in files directly next to the content, but shared.

But then I wanted to write content in Perl. So I wrote tags that called Perl scripts and inserted the content. And I wrote tags that organized the wftk manual I started rewriting. And I just kind of lost interest in maintaining it all.

So there it sat. Some of the Perl scripts are not friendly; if a bot hits a number of pages with Perl scripts in rapid succession, the load on the server can stop my mail arriving. And hey - now Github serves static content.

So I'm moving it all into static content. Some of it won't make sense. Eventually I'll have to rewrite any content I want to keep as a JSON script or something and hit it from JavaScript on the static pages. Because it's become crystal clear to me that Github will do a much, much better job of keeping the static content available than I will.

(Until ten years from now, when Github will have disappeared, and I'll be sitting at a desk in a dark room at Christmas in 2022, revising all this again.)