This is actually a comic exaggeration of a well-known principle in learning psychology called the Yerkes-Dodson law, which states that for optimal learning, you need a specific level of arousal -- not too low, not too high.

http://en.wikipedia.org/wiki/Yer...



Programming can be viewed as continuous learning. If what you're doing is not challenging you, pushing your envelope a bit, it is boring. If it is too challenging, it is overwhelming. So programming tasks that elicit the most productivity have characteristics of learning tasks. Which means the Yerkes-Dodson law applies. Which means you need to get to the right arousal state as an enabling condition. Alcohol is one way to get there.

In fact this is generally true of all creative work, as discovered by Mihaly Csikzentmihalyi in his classic work on the "flow" concept.

The interesting question is whether this normal distribution might sharpen almost to a Dirac delta for programming.

There is one reason to believe it might: the 10x programmer effect. When you are in the zone, you are almost an order of magnitude or two better than when you are grinding along. So yes, you might actually have a Y-D curve that looks like the xkcd cartoon if the 10x effect also kicks in.

Anecdotally, I'd say this is true. I haven't programmed in years, but I recall a few such "optimal stimulation" phases when I got insane amounts done in a few days. The cause was alcohol a couple of times, or just general bootstrapped stimulation in other cases.

I see the same phenomenon with my writing. I've done some of my best (i.e. 10x) writing in the last few years while optimally drunk.