I don't think there's an easy answer. I think, however, it falls into well-defined pilot projects. Specify an idea or a route, or multiple ideas and routes, and assign them likelihoods of success and what you would need to know to see if they work or not, and then imagine managing the bouquet of routes as an investment portfolio. The only secret I have--and I am no genius at finishing projects--is what you're doing is project management/route management, not time management. Each route is a GTD project. It has a goal, a possible outcome. Do a weekly review, review all routes, compare them against each other consciously. Ask yourself "What would I need to know to decide if route X is a good idea?" then ask "how much work is it to find that out?" then "is this a worthwhile investment, given the possible payoff if it is a good idea, relative to other paths?" the weekly review will help you ask these questions consciously...and answer those questions in an on-going way, as new information emerges.

You want to separate the decision-making levels, but not neglect those levels.

It's easy to go down rabbit holes for too long if you aren't comparing.

All that said, going down any route is better than none. You DON'T need to decide FIRST if route X, the first route, is the best before you begin. It's iterative...and you need to embrace the iterativeness of it. The way of doing that is committing to the fact that you are learning and evaluating in an on-going, weekly-review kinda way.

what do you think, Urban? I'm trying to figure this out myself.