While listening to an older episode of econtalk with Nassim Nicholas Taleb, I was reminded of the closing lecture of Robert Sapolsky's "Human Behavioral Biology" course (Sidenote0: full course videos available here, definitely recommended if time allows) (Sidenote1: I heard of the course from this post by Matt Perryman)
But first, back to my statement. When I say "I don't know" what I'm really saying is "I don't know an EXACT solution for this PARTICULAR problem". However, that doesn't stop me from doing something useful in the situation. (note that doing nothing is also one of the options)
More genereally, we may not know what precisely to do, but we can still have a decision framework that converges upon a desirable solution.
As Slavoj Zizek points out, when faced with complex problems (context: Occupy Wall Street), the first thing to do is not to focus on finding alternative "solutions", but rather, to ask better questions.
That is precisely what is echoed in Saplosky's last lecture (listen to it! but if you're pressed for time, skip to the "Why does it have to be so complicated!" section),
which I roughly paraphrase:
some of you will have in 30 seconds to decide with somebody who is an ER who has taken a vast amount of pills while trying to kill themselves -- is it them, is it their disease, is there even a boundary, do I give the command (against their will) to have their stomach pumped?
The question there isn't "what is the right solution?", but something approximating "what is the framework I need to adhere to so that the least harm will result?" It's the context-specific knowledge and experience that allows one to formulate these decision frameworks and thus be an expert. It's the courage and confidence to execute on these frameworks that makes one great.
Note that we shouldn't be fixated on the dramatic, time-constrained scenario that Sapolsky gave. Many equally difficult scenarios play out over months or even years (politics, personal relationships, business deals, the mexican drug war, etc...).
The key remains using past experience to formulate a cost model, and then to derive a framework that minimises cost, and then having the faith to execute on that model (with emphasis on the word "faith" -- courage before clarity).
easy example: My cost model for going out drinking on New Year's Eve revolves around the greatest possible cost -- Dying in a Drunken Car Accident. This is a binary outcome. Thus I don't go out drinking on New Year's Eve -- the upside is not worth the cost
Unfortunately, the concept of "Anti-Knowledge" is intuitive only to a few, and I was fortunate enough to be one of them.
All I can recommend is to confront an important personal problem, that does not demand an urgent solution -- you have time to think it over. Use the time to ask the million and one "if I do X, then the worse outcome is MaybeNotSoCrazyAfterAll( X )" questions, compare and constrast outcomes, and then hopefully through a mechanistic process, see the path of least cost, wait a few days, gather courage, and charge forward without regret.
Just don't expect your spouse to follow....