I might agree with Marvin's sentiment IF there were an accepted nomenclature for the concept addressed in the article. Yes, at a high level, it's just normalization. But when discussing the specifics of many to many relationships, there is no generally agreed upon name for them. "Decoupling" is maybe a bit unconventional, but actually makes quite a bit of sense; after all the process of normalization is really that of decoupling-or removing-functional dependencies in the data...

Besides, as useful as SSC is, and has been over the years, anyone who has frequented it for any amount of time should realize that most of the articles are written by members of the community and as such, should not be considered "authoritative" which is not to say that they can't be useful or even quantifiably correct. Steve has always had more of an open door policy to contributions, which means that just about any lunatic can get published...;-)(speaking primarily of myself who used to be a fairly active author on SSC)