Understanding the Bradley Effect

I've finally come across some decent reporting on the issue in today's NYT.  Confirming Nate Silver's analysis of the Democratic primary, the conclusion is that this phenomenon is no longer in play for polls with good sampling methods.  After several days of watching TV punditry mangle their descriptions of the Bradley Effect (especially when they talk about "the reverse Bradley Effect"), this piece is quite refreshing.  

But, while we're at it, let me get this off my chest: As the term is conventionally used, the Bradley Effect is only about lying to pollsters.  It is not a cover term for all aspects of racial demographics in electoral polling.  OK?