Mark McSpadden

Last week over on the CU Loop, Andy LaFlamme (a cool CU guy and active member of the Banktastic community) writes that Sometimes It’s Just Better to Start Again and while I’m a little late to the party, but I thought I’d chime in and share some thoughts on the situation.

Developers face similar problems when developing applications to the puzzle analogy Andy uses. We build piece by piece and often when stepping back, realize it’s not fitting as well as it should.

Some of this can be avoided using best practices, but part of it is just due to the fact that you learn so much during the process, that there is no way you could have known putting A and B together wasn’t going to be as good as putting A and C together.

A relatively fresh concept in our space is that, instead of wiping everything out and totally rewriting the application, replace your existing application chunk by chunk.

Back to the puzzle

The problem that I have with starting again from scratch is, well, isn’t that how you got to where you are right now? You started with a blank canvas, some focus, and some vision and then started laying down pieces. Sure, this time you have a better focus and better vision, but honestly, when you are done with this “new” puzzle you are going to know more and therefore think your puzzle still isn’t fitting together right.

Not only will you be in a very similar spot in the “end”, but you’ll also have wasted a ton of time putting some pieces back exactly as they were. Looking at your puzzle you know some pieces fit. Why spend the time taking those pieces apart and putting them back together?

But what if those pieces have a better fit? It’s a valid question, and one you can certainly address after the puzzle as a whole gels pretty well. The goal is to produce a whole puzzle, not the optimal union between any two individual pieces.

A Different Kind of Puzzle

The more I think about it, the more I like thinking of organizations, marketing efforts, and web applications as photo mosiacs instead of the traditional puzzle. Now, I honestly have no idea how you actually create one of these (apart from using cool software features), but the wikipedia entry mentions replacing a chunk of the actual main photo with a smaller photo of “appropriate average color” and that I can grasp.

The goal is find a small photo that fits the general color requirement of the space. Doing this over and over produces an end result that is close to the original photo or idea. I’m sure you could spend forever on certain sections, polishing and improving, but only after that general concept is reached do you actually see improvement on the end product.

I’m starting to ramble but I think I’ve made my point and you won’t find me totally starting over anytime soon. :)