For one, Beeminder offers a great way to collect the sorts of data that lead to better planning in the future. It also gives real-time feedback about whether you're achieving what you'd set out to do by putting in the expected level of effort. If not, you can either adjust your plan while it's still in the early stages, or change how much energy you're putting towards it.
I like the idea of back-planning. From my personal experience, I suspect that the primary benefit I'd get from it is applying my freshest, most detailed planning to the last stages of the project - the ones where I actually have the least information about the process. Often when I make step-by-step plans, the first stages are very detailed. As I move along in the project, though, there often comes a point where many of the steps are too tied to previous aspects to fully flesh out the details. That's where I start hand-waving and making baseless guesses, and that's also usually the point when the plan breaks down. But the end of a project is where all of those possibilities converge to a clearly defined outcome again, and there are definite steps that need to happen to get there - steps I would have never reached working my way from the beginning of the project.