In the simple 5 (track A)-vs.-1 (track B) trolley case, if I steer the trolley on the track with the 1 man (Jones), his death is in no way a means to the salvation of the 5. Thus, my action is best described as steering the trolley not onto B but away from A; the death of Jones is unfortunate, but is foreseen though not intended. If Jones were able to break free of his bonds and escape in time, so much the better; then all 6 men would be saved.
Now let’s look at the evil colonel case. Unlike the situation with Jones in the previous case, if one of the 10 men I am motivated to shoot dodges the bullet, I’d have to shoot him again. The death of group 1 is a definite means to saving group 2. Hence double effect prohibits it.
There is another problem with utilitarianism in the colonel case: adopting this theory ends up enslaving us to evil men. Anyone will be able to say, “Do this evil, or I will do something even worse,” and utilitarianism would seem to command us to obey.
We might say, I feel the colonel’s demonic presence; he is laughing at my predicament which he engineered, mocking me, saying that I am damned if I do, and damned if I don’t. I don’t want to satisfy him, make the devil happy. This is a reason to abstain from killing.
But if this situation arose on its own, not having been maliciously contrived, then my decision is to that extent easier.
The 3 criteria are jointly sufficient to justify an action; but they need not be necessary. So, killing the 10 even without satisfying double effect may be permissible, but the case is much less clear-cut.