[codergirl]

I sometimes feel like I have the opposite problem. I can come up with the algorithm which requires an understanding of why it works on the problem, but then I lose my train of thought while coding it (albeit this is happening in stressful interview write-on-the-whiteboard-while-I-judge-you situations).

[abc_lisper]

finally, somebody like me :). How the hell should i scale?

I write tests, but do too much testing as the code size grows and that is very limiting.

Ohh, on losing the train of thought thing, this is the very reason why i don't use IDEs, because they freeze at random times. One technique that worked for is solving the problem first of board in a sufficient detail, or writing down as comments how you are going to solve the problem, and slowly fill in the details.