I've always found that finding a similar scope problem and how long it took is the best predictor of how long the new problem is going to take.