I'm glad you're aware of Carmack's association with the fast inverse square root (hopefully you also know that it doesn't originate from him) but the joke here isn't the denigration of the code itself.

The joke is that the original need for a fast inv sqrt is moot at compile-time.