Over the last couple of weeks I have been working on getting TwoNeeds back on track. After six months or so of active development we had a rather complex but solidly designed and scalable back end and UI for most of the important features. However, with lots of changes to the feature set the pieces didn't really fit together in any coherent way. It all worked technically and we could keep making progress, but it would never become a good product that way.

When that happens, the only real solution is to throw out the code and start over. While it may seem like a lot of wasted effort, building the first version was necessary to get a proper understanding of how the app should be built. It is immediately clear what pages are needed, and its possible to do some stuff with jquery that wouldn't have fit into the initial prototype.

I've been doing most of the work myself so far - partly because the rest of the team is busy with other projects, and partly because much of the detail gets designed as I build it - writing full specs would probably take just as long as the initial build. I should be able to delegate more as the work progresses - it's quite easy to have someone make specific changes to a component, but describing an overall vision is pretty hard.

I'm also noticing that a lot of the work is on the less interesting stuff. In order to build the interesting features, we have to build a lot of stuff that has been done before in a way that supports the new stuff that will be added later.