True enough - but code that humans can't read is far more difficult to debug, to profile, and to optimize. Efficiency can be improved after the code is working, but improving readability on a prematurely-optimized codebase frequently involves `rm -rf`.