I've just spent 2 days banging out a prototype of a genetics learning object. I love working fast. I have to get an interface up and running in order to really understand the underlying interaction design.
I generate a lot of kludge before I get my data structures figured out. I find it's easier to re-implement a simple kludge than to modify an elaborately planned architecture that didn't work out exactly.
Maybe it would be better to just sit back and attend meetings and wait for a consensus before writing a line of code. Tick. Tock. Tick. Tock. How many circles can I draw on one page of legal pad?