That's not the sort of variance that was discussed in the article, where an explicit mapping function is applied, not an implicit subtying relationship.