Perhaps your addiction is influencing your perception of your coding performance. The best way to test this is to find an impartial third party to compare your caffeine-code to code you wrote before you started drinking coffee (as in, not the code on day you DIDN'T drink coffee).