They're not being rigorous; it's just simpler to explain in terms of the data model. The data model is very simple and clean (and most of git's benefits flow d

They're not being rigorous; it's just simpler to explain in terms of the data model. The data model is very simple and clean (and most of git's benefits flow directly from the model). The UI is awful... confusing and inconsistent. The engineering is brilliant: informative error messages and hints; fast; great features (e.g. a better diff, with --color-words; logical argument parsing). The only significant things missing are deliberate design choices inherent in the data model. e.g.: not storing renames explicitly (which BitKeeper did), so you can't easily search for changes in a file that was renamed/moved. It could be simulated with out-of-band info (like tags), but isn't.