Making it up as you go along

On his blog, Seth Godin poses the question:
"Just wondering: Is there any other way to make it up?"

Yes, you can plan it all out in advance.

While this approach begins with a goal or problem to solve, once the plan is in place, it really becomes about fulfilling that plan. The problem is, inevitably you will encounter something you hadn't considered, or the requirements will change (or more likely, finally be delivered.) Then, due to the sunk costs (in time) of making the plan and getting buy-in from your associates, you are put in a position of trying to adapt the plan to support these things, even though the approach may no longer be the best. The further along in a project you are when this occurs, the more likely you are to just "make it work," regardless of the implications.

On the other hand, making it up as you go along means you are generally focused on the goal or problem the whole time. You are free to adapt as needed as long as you get there in the end. The downside here, is that there is no clear end product. This gives both you and your client permission to add/change features at will, probably complicating, and almost certainly prolonging the project.

The planning approach frequently leads to "what were they thinking?" product designs. The improvisation approach often results in vaporware or unprofitable, never ending projects.

For me, I work best when I make it up as I go along, and I almost always end up with a good result. Unfortunately, if it's work for hire, it is almost a requirement to create a safe, predictable result rather than something that might be better, but isn't what the client was expecting. This feels wrong though, and is probably why I don't do much freelance work. My day jobs are usually planning driven, but good bosses will see the potential and let me run with ideas while they try to sell them back up the chain. Regardless, at the end of the day I go home and do projects for myself, perhaps my toughest but most rewarding critic.