>I'm not sure that follows from this article.

I can see how. In the example given, the strength of coding speed is created via a bias against careful review of edge cases. When it works (most of the time) , you increase your coding speed and reduce review of edge cases even more, until something blows up

The interesting insight from the article is that a coder is not an inflexible monolith - they can vary the expression of a "strength/weakness" pair (strength/weakness being a misnomer at this point in the argument) to suit the circumstances