Hacker News new | ask | show | jobs
by horsawlarway 1443 days ago
I disagree - I think that "reasoning about moral intuitions" is completely useless if you're attempting to reason about them in utter isolation and with the assumption that the subject is omniscient.

It's like economists assuming perfectly reasonable actors in markets, or physicists assuming a perfectly spherical object that ignores wind resistance.

They're toy problems that don't match reality at all, and the value is dubious at best as anything other than a very gentle intro to the subject.

Here's a thought experiment - How many people do you think would actually make the choice they state they will make if you present them the trolley scenario in real life with no warning? People who say they will pull the lever are fooling themselves.

1. They won't know how to read tracks

2. They won't know how the lever works

3. They don't know for sure that anyone will die: those five people might be able to move off the tracks just fine themselves

4. If they do pull the lever they're almost certainly going to get arrested or troubled by the legal system, because they fucked with shit and someone died afterwards (the courts won't give a shit that "they thought five other people might have died!").

5. For all they know the trolley operator can stop just fine, why would anyone be about get hurt?

6.... on and on.

Basically - you're setting up an impossible framework, the results (even if you get them) are useless because they're only valid in that impossible framework.

If the results of the real world never match the results of the framework you've set up, what is the value of that framework? It's just a shitty model with bad reproducibility. We have lots of those.

2 comments

They did this experiment in real life https://www.youtube.com/watch?v=1sl5KJ69qiA . It's not 100% perfect but it is still an amazing test.
So what's your alternative, throw your hands up and proclaim ethical reasoning to be impossible, nothing is true, everything is permitted, and it doesn't matter what you do?

All models are wrong but that doesn't make them useless.

I much prefer: "All models are wrong but some models are useful". For example - Flat earth is a model that is both wrong and useless.

My claim is that the trolley problem is useless. It asks people to make a guess about how they would behave, but that guess is predicated on a set of initial conditions that are impossible to fulfill (omniscience is a bitch to get in real life).

How does gathering all that incorrect data help you? What ethical reasoning are you trying to tease out here?

Here's one you might love - "if the earth is flat and you reach the edge, would you jump?"

Now lets just categorize everyones answer to the that question.... and: Hold the phone! The earth isn't flat? It doesn't matter? It turns out this question has basically no relevance to anything!

Is it fun? Sure. Is it useful? I have doubts.

I think the alternate is to not try to be a railroad operator in an emergency when you might make the emergency worse or take on huge liability.

If you are a professional trolley network controller then you have the judgement and the duty to operate/not operate the lever. I think that few people would question the ethics of a professional operator flipping the switch to save the most people.