[mbroshi]

This is something I have thought about a lot.

I have had coffee daily, gone months without any caffeine, and currently average about one cup of coffee every week or two.

I think the best approach is experimental and data-driven. Everyone has different genes, life circumstances, sleep quality, etc., so it's hard to deduce too much from others' anecdotes. Once a week, I do a version of the Beck Depression Inventory[0] to measure my mood over the previous week. For me, daily coffee definitely causes me to be slightly depressed. I have found a cup when I really need it (which is about once every week or two) is ideal. But, whatever it is you want to optimize for (lines of code, happiness, sleep quality, etc.), make sure you are measuring, and then start experimenting!

[0]: https://en.wikipedia.org/wiki/Beck_Depression_Inventory