Is there a way to objectively determine which is better? Kill injure person and damage a car and a tree or kill several people. After reading this story about people preferring to save 10 people, but only if it is not themselves in the car.


Essentially it makes sense to me to not want to sacrifice yourself by default. What if there are no pedestrians but the car runs into the tree or other car anyway?

I know this situation is very hypothetical and I guess that's the problem with this question. Hypothetically it makes sense to favor the many over the few. But not in every situation.

Is there a simple logical way to deduce this though?

  • 1
    – E...
    Jun 29, 2016 at 17:00
  • No, there isn't. Ethics is inherently about personal choices and preferences, even if such choices are regulated by a group one belongs to they still do not become "objective". In a recent survey of (analytic) philosophers 68% chose switching in the trolley problem, i.e. sacrificing one for five. Curiously 24% chose neither switching nor non-switching, whatever that means philpapers.org/surveys/results.pl
    – Conifold
    Sep 28, 2016 at 0:20
  • To notice, the article is not about the dilemma itself, but about the curious reaction of respondents, who want the cars to be programmed to minimise victims, except in the case they would personally be in the smaller group of possible victims. Oct 31, 2016 at 11:52

1 Answer 1


To be frank, that is slashdot, and most articles there are idiotic. They claim that there is a dilemma, but it doesn't exist. The goal of creating a self driving car is to make it safe. To avoid all collisions. The scenarios described do NOT happen. Nobody, ever, in the worlds history, has been in a situation where they had the choice between killing themselves in an accident or killing ten pedestrians. And the chances are ten times slimmer in a self driving car that is programmed to drive carefully and avoids a situation like that. If you think otherwise, post a link to a case.

But since you are posting on philosophy, you should have thought a bit more about this. This isn't about you deciding whether to die to save ten lives. It's about making a decision how cars should be built. Let's say the whole situation happens X times. There are X drivers, and 10X pedestrians. If you are ever involved in the situation, then it is ten times more likely that you are among the pedestrians. So it is in your own best selfish interest that in this kind of situation one driver should die and not ten pedestrians, because that kind of design makes it ten times more likely that you survive. In addition, drivers in a car are very well protected. Especially in a self driving car that would predict the accident and can optimally control seat belt, airbag and so on and can optimally slow down.

Now you might try to modify the computer of your car. That has three problems: One, you don't have the slightest chance to pull this off. Two, if you pull it off and ten people die, you are going to jail for a very long time. Three, if my wife is among the victims, I'd make you wish you had died instead.

Dave: Your maths is wrong. It doesn't matter how often you drive and how rare it is that you are a pedestrian. If people never walk then the number X of accidents in the equation is small. Maybe very small. It doesn't change the fact that you are ten times more likely to be the pedestrian in such an accident. Maybe the one time in summer when the wheather was beautiful and you ate a bit too much and decided to have a walk, or the one time when your car broke down and you had to walk, you can't get around maths.

Isaacson: Please think about it and please think very very hard before you tell me my math is wrong. In every single situation where there would be a choice between one driver dying or ten pedestrians dying, there will be ten times as many pedestrians involved. If nobody ever walks, then clearly there is no chance to ever run into ten pedestrians! And the whole argument is about general rules, not individual cases.

  • If there is as statistical correlation between frequently riding in a self driving car, and not being a pedestrian (plausible but might not realize) then you don't get the 10x gain in safety implied by your calculation. To put it more bluntly, and selfishly, if you never walk on/near the street because you have a cool self-driving car, you are never on the pedestrian side of the equation. You need a veil of ignorance type of idea to get your calculation to work our.
    – Dave
    Jun 29, 2016 at 21:27
  • @gnasher729 No, Dave's maths is correct, you are making a common statistical error. What you have deduced is the probability of picking from a random group a person who has been a pedestrian in an accident. That indeed would be 10 times greater than your probability of picking from such a group a person who had been the driver in such an accident. But being either a driver or a pedestrian is not a random factor, it is a set of largely known factors.
    – user22791
    Sep 28, 2016 at 6:44
  • For proof, consider the theoretical person who never drives, there is zero chance they will be involved in an accident as a driver. Ten times zero is zero, but that is not the probability of their being involved in an accident as a pedestrian.
    – user22791
    Sep 28, 2016 at 6:46

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .