People adopt the most common variant around them at a rate higher than its frequency alone would produce, which makes groups internally similar and mutually different.
Boyd and Richerson's 1985 models distinguish merely following the majority from disproportionately following it. The second is what generates and holds group boundaries: a variant slightly more common in one group becomes overwhelmingly common there, while the neighbouring group locks in the alternative, with no difference in merit between them.
In practice
- Two offices in the same company settle on incompatible conventions within a year, and each regards the other's as obviously wrong.
- A practice at 55% adoption is at 90% two years later without any new argument for it.