I was recently asked about whether or not the code from a market test Alpha should continue on into the offering. I don’t know the answer, but here’s my thoughts. Intuit doesn’t have a doctrine that I know of, just some history. Prototype code is almost never used in a commercial launch, but Alpha versions and such have gone on to be full products. Quicken and QuickBooks were retooled after about the third version. My own general view is that if the code is reasonably well done, there’s no reason you can’t continue with it, using an emergent architecture method. Too many offerings have been sidelined by trying to get the architecture “right” the first time. But if the code seems untenable now, it might be a good idea to start afresh on it. My prototype version of Keyword Search would never have worked in a real product, so it was tossed before I even started on a publishable version. Echominder was a prototype that turned into a launch, and we started down the retooling path after a few months. It probably would have been a better product if we had started over on it before launch. Engineer “excitement” can also be a factor - if a group is excited about a recoding, you can get a much more engaged team, and better results - that’s what happened with Intuit Labs - the innovation.intuit.com codebase was tossed out, and it’s a better product, with faster iteration potential, than it would have been otherwise. So… use your instincts, don’t over-architect, and take the time if it’s necessary. Don’t take the time if it isn’t.
-
I posted to intuitlabs.com
Throw away your code?
http://intuitlabs.com/blog/archive/2009/05/throw-away-your-code/
May 5 2009, 11:30pm | Comments »
1