When we started Flood Magazine in July, I made the conscious decision to do as little coding as possible, with the reason being that it’s 2010, after all, and putting content online should be easy with the tools already available.
In the end, I chose a simple site generator called Jekyll, which is so minimal in functionality that it would be a stretch to call it a content management system. You basically save plain text files onto your hard drive, and Jekyll converts it into static HTML. This process suited us nicely for about two months until it became apparent that we needed a different solution to accommodate more writers.
I asked a few people for suggestions, and unsurprisingly, almost everyone recommended WordPress. By some measurements, WordPress owns more than half the CMS market share and powers 12.4 percent of all websites. It’s open source, easy to setup and most people already know how to use it.
WordPress should have been the obvious choice, but because I’m such a snob, my search didn’t end there. I spent nearly a month exploring other non-PHP publishing platforms because, you know, PHP isn’t sexy. I was so fixed on using Radiant CMS, which is written in Ruby, that I wasted an entire week customizing and hacking it, only to give up in frustration.
Then I found this article written by Paul Biggar, founder of NewsTilt, who, in hindsight, realized that they should have built a platform around WordPress instead of rolling their own. “I reasoned that the platform was the core of our technology, and we were a technology company, and smart technology companies needed the flexibility that comes from writing the core of their platform themselves,” he wrote. The decision to forgo WordPress was just one of the many mistakes that plagued his start-up. A few months after NewsTilt went live, Paul shut down the company and returned to the investors the remainder of their money.
Today, I’m happy to announce that we have released the second version of Flood Magazine, which runs off a slightly modified version of WordPress 3.0.1 on a LEMP stack. Our code lives on GitHub, and we use Capistrano for deployment. Everything is running stably, and I think it was the right decision to have something up and running rather than get bogged down in the technology.