Let me try to rehash the process, using the next release as an example. Let us say Dries branches 5.0 a few weeks from now. The makes HEAD (technically trunk) open to development again. At that point, there is no need for 5.0 to have a code name, since it has a version number that will not change. The next release is expected to be, most probably 5.1. But it can be 6.0 as well depending on how much new features make it in before the code freeze. From 2 weeks from now until the code freeze (6 months or more), the "next release" has no definite name. For that period of 6 months, the code name will be used to reference that release. Any references in documentation, issues, mailing lists, ...etc. will not reference a name (bikeshed) that will never again be used in the future for any other release. There would be a page on drupal.org listing release names and the corresponding version numbers. If some documentation still references bikeshed, and someone is reading that in a year or more, it is easy to know that bikeshed became 5.1. On the contrary, being referenced as HEAD or CVS or "the next release" is a name that gets reused every release. then the next release gets another unique code name (say peace and banana for amusement value [note, I am not advocating this, just a humorous example]), ...etc. Note that the same is true even for Windows (e.g. Longhorn became Vista, Chicago became Windows 95, ...etc.). Microsoft abandoned release numbers after Windows 3.1 and chose year numbers, then cutesy marketing names. Apple and Intel use code names as well, although for different reasons. Examples here: http://en.wikipedia.org/wiki/Project_code_name I hope it is now clear why code names have value and when they will be used in the development cycle the most.