Some premeditated crimes (e.g. first-degree murder, bribery) can be deterred by threat of punishment, but when people are making the snap-decision to commit a crime of opportunity, they aren't carefully weighing the risk vs reward.
If your point is that people (at least those who aren't scientists or statisticians) round small probabilities to zero, then I agree, and it's something that I think people need to be more aware of when considering deterrence or other behavior. However, I don't think you or the article made this point clearly. The article would be so much shorter and simpler if it was condensed down to that one observation.
It's quite reasonable to say that punishment only has an effect when it is clearly linked to an action, but that sounds like a completely different message than saying punishment doesn't work (implicitly assuming it almost never occurs).
It still has death sentences and the dubious record of an incarceration rate higher than any other country.