The article sounds like Dijkstra in 1988. My bet is another 22 years won't change anything.
Also, software isn't really construction. It's built by the people who design it, for the most part. Imagine a skyscraper (or car, bridge, etc) built entirely by engineers. Yikes.
In fact techniques exist to produce software with near-zero defect rates (primarily relying on extremely thorough code inspection, IIRC), but they're far too expensive for commercial use.
You don't expect a mechanical engineer to know the chemistry of a steel melt, and you don't expect a chemist to know about engines.
Yet in CS we either teach the detailed chemistry and expect the student to pick up the engineering themselves - or we teach just engineering without explaining what a metal is