Except that they didn't even quite get that right. The article seems to contrast git and hg's versioning schemes, when in reality they're both hashes of the content (the non-unique hg version numbers referred to are merely local aliases).