Steve made one very good point in his article, which you don't want to lose as this thread drifts away from joins to nulls.  The article was not about how boolean logic works with nulls, (which WAS explained quite nicely by Lutz).

A Where clause in an outer join is a filter, applied after the join takes place.

After that sinks in, all the other logic makes complete sense.  It also explains why you sometimes see SQL Server not use an index where is "seems like it should" due to a field being mentioned in a where clause.