The 37signals blog Signal vs. Noise pointed me to the Wikipedia entry for the waterfall model of software development. Sure sounds a lot like the ADDIE model to me. I wonder if there is any connection. Well...no matter...they are not relevant any more.
Having spent a lot of time working on a Ph.D. where I studied software development processes and now having built a services company based on building quality software and eLearning Solutions, there's still a lot that is relevant in both the Waterfall model and in ADDIE. We don't really use either model directly, instead we have a modified version and a process that's highly flexible to meet the needs of the particular project.
37signals tells you that all your clients (internal or external) care about is the end result. Get there as quick as you can. Get rid of intervening documents that just get in the way of getting to the end result.
I understand where they are coming from, but I would point out one really important thing - and in fact, when you are in a service business it's the most important thing ... The client's expectation of the end result.
A service business is ultimately all about mutually setting expectations and meeting those expectations with your client. I would claim that the real value in the Waterfall Model and the ADDIE model is successively more granular definitions of expectation.
I agree that these models are bloated. That we need streamlined versions of the models to be able to handle the speed in which projects need to occur. But, if you are iterating from the start through successive definitions of expectations and attacking risks early, you are destined to fail to meet expectations on some projects.
You can get a white paper on this topic that I wrote from my company web site: Innovative Projects and Software Development.