I was about to observation that the python code as posted in the blog missed an important optimization, reordering some of the tests to: for b in range(1,

I was about to observation that the python code as posted in the blog missed an important optimization, reordering some of the tests to: for b in range(1, 27) for n in range(1, 27) if b**3 + z**3 + n**3 + o**9 == 1997 for k in (3,4) for e in range(1, 27) This reduces the number of innermost tests and on my laptop it runs about 18x faster than the blog's solution. That was: for k in (3,4) for b in range(1, 27) for e in range(1, 27) for n in range(1, 27) if b**3 + z**3 + n**3 + o**9 == 1997 Then I see that @eutectic did the same optimization in his Haskell code. Nice!