This is when programming really clicked for me: I learned that if the problem isn't well understood and solved before you sit down to write a line of code, then its not time to code yet. Model the problem space first.

And as another commenter chimed in, its true that some problems only emerge once you start writing the code. But that's alright, its part of the process. The planning isn't so much to figure out every problem ever, but to model the known problems so that programming the solutions can begin. I go back and forth between white-boarding and coding as things emerge.

I always tell myself that if I'm sitting at the keyboard and I don't know what to type, it's time to go back to the whiteboard because I'm not understanding the problem.

edit: clarity