Drink for Better Coding: The Ballmer Peak Is Real, Study Says (2012)
observer.comIf this is true, I suspect (anecdotally) that it is only true for solving small problems that don't require serious thought, patience or diligence.
In that experiment, the problems required 10-15 seconds of thought to solve.
Programming well can take 10-15 minutes just to rebuild a fresh mental model of the system you're working on.
It even mentions that drinking boosts creativity but not working memory. Coding is far more dependent on working memory than creativity.
The reaction time property was the one my mind latched onto, and I think may be the crux of the benefit (at least, has been for me.)
If I need to do a lot of rote programming, CRUD, etc, drinking can REALLY HELP, because I suddenly feel both a lot grumpier to be working on something dry and repetitious, (and my brain seems to be running around less getting distracted because there's less brain functioning). It also helps with choice paralysis for things for which... honestly, it hasn't really mattered. For certain classes of tasks, the same thought process that excels in designing architectures, seeing pitfalls, and thinking critically can get more in the way, and a "just get it done" hack is needed. For tasks that fall under these categories, I wonder to how much falling back more on "muscle memory" is a benefit than letting more of our conscious thoughts intrude.
Anyway, just some thoughts, not at all scientific, wouldn't intend for the above to be interpreted as such :)
A few years ago, I had a player in my tennis team who had a small beer before playing matches.
It's also real for Arcade games and fighting games.
cc: Human Resources