3 ms·

Many very good points. One thing to note is that in these real-life examples you've laid out, there's a non-trivial difference in effort and action between cho

Many very good points. One thing to note is that in these real-life examples you've laid out, there's a non-trivial difference in effort and action between choosing to save people (e.g changing your entire career path to earn more money) versus not doing so. In the Trolley Problem, the action can be made arbitrarily insignificant, trivial and ambiguous. For example, instead of pulling a lever, we could instead define the action as nodding one's head, or blinking twice. At a certain point of triviality, the action/inaction dichotomy and the do-something/do-nothing dichotomy loses coherency and we realize that ultimately this is a choice between two mutually exclusive options, and subjects are intentionally choosing to kill 5 to save 1. Perhaps a more fitting real-life analogy to the Trolley Problem would be as follows. You currently donate $100/month to charity A, which saves 1 life per month. A more effective charity comes along called charity B, which for $100/month is capable of saving 5 lives per year. The process to switch from A to B is easy and trivial. Despite knowing all this, you purposely choose not to switch, thereby causing four extra deaths per month. Does that make the person equivalent to a contract killer who's killing 4 people a month? Society has deemed not, probably for reasons pertaining to pragmatism (we want to disincentive real murderer) and evo psych (we didn't evolve to have disdain for opportunity cost deaths, irrespective of how important it is morally speaking).

jerf 6y ago
"Perhaps a more fitting real-life analogy to the Trolley Problem would be as follows." A non-trivial part of why the trolley problem bothers so many people is that it can't be salvaged because it is intrinsically not anything like real life. In real life you are never presented with such a binary problem. Everything is a mix of arbitrarily unbounded priorities, predictive ambiguity, and even ambiguity in the outcome. My personal "resolution" to the problem is, yes, obviously, you save as many lives as possible even if that means actively changing the outcome... but that's essentially irrelevant, because you will never in your life be faced with such a stark, mathematically-rigid choice. There's always other options... not always relevant ones or desirable ones, but always something. Even when you construct the problem as close to a real life one as possible, you end up with something more like "You just realized you're driving on ice and going way too fast. In front of you are two pedestrians crossing in front of you, one on each side of the road. You have about .3 seconds to consider this problem. Your System 1 instincts think you might be able to weave between them but there's a chance you'll start spinning and hit both. You could also try to jank to the right and go up on the curb, but you might crash yourself into that tree pretty hard and still hit one of the pedestrians. Regardless of what you choose System 2's gonna be second guessing you pretty hard for the rest of your life." You don't get "Choose A and this will happen, choose B and this will happen", it's a big mess. It's always a big mess. Even if you literally encountered the trolley problem in real life, one must consider whether your understanding is accurate, or if the person who has explained it to you is lying or has ulterior motives, and so on. (And that's ignoring "this is clearly the trolley problem and I must be on TV or in a psychology experiment"; let's pretend it's in a world that isn't obsessed with this particular problem.) Now, that is arguably the point of the trolley problem, to get down to that level and ask ethics problems. But it's dozens of layers of abstraction away from anything relevant to the real world. It's a mistake to think otherwise and stress about it as much as the gestalt seems to.
- Wowfunhappy 6y ago
  > In real life you are never presented with such a binary problem. Everything is a mix of arbitrarily unbounded priorities, predictive ambiguity, and even ambiguity in the outcome. One real-life example that comes to my mind is whether we should have done COVID vaccine challenge trials. I always thought they were a good idea. Yes, it would have put people at risk, but it could have potentially saved many more lives! My father is a doctor, and in fact is an infectious disease specialist. He's been very busy this past year, working primarily on COVID research. When I talked to him about the idea of a challenge trial, he was completely against the idea—so much so that he got a bit upset, and we had to stop talking about it. That doesn't happen often. The idea of purposefully giving participants a deadly disease—even with their consent—seemed fundamentally wrong to my father, regardless of how many other lives could have been saved in the process.
  - davidgay 6y ago
    https://en.wikipedia.org/wiki/Primum_non_nocere https://en.wikipedia.org/wiki/Primum_non_nocere would seem to be the ethical principle at play here .
    - Wowfunhappy 6y ago
      Right, but that's the essence of the trolly problem, isn't it? Whether you can take one life to save multiple lives.
      - MrPatan 6y ago
        You're right, and for exactly the same reason: If somebody dies in the human challenge trial it's very specifically your fault. The people that die due to a delay in the vaccine because of not using human challenge trials? Nobody knows them, they don't count, and they can't sue you.
  - jerf 6y ago
    "One real-life example that comes to my mind is whether we should have done COVID vaccine challenge trials." You don't know what the result of the trial will be, already taking us out of the domain of the trolley problem. If you knew it was going to kill everyone who had it but you'd get this particular bit of info that was guaranteed to have this result for the general population, then we'd be much closer, but there are multiple false assumptions to get there... and we still haven't discussed all the different ways we could run the trial because it's not a matter of "Either we do the trial this exact way, or we don't"; do we do a double-blind? Base it on multi-armed bandit math? Do we focus on certain demographics? And so on and so on. Not a binary choice between these two guaranteed results; a huge range of possibilities, unknown outcomes, even things that are hard to measure outcomes.

jerf 6y ago

"Perhaps a more fitting real-life analogy to the Trolley Problem would be as follows." A non-trivial part of why the trolley problem bothers so many people is that it can't be salvaged because it is intrinsically not anything like real life. In real life you are never presented with such a binary problem. Everything is a mix of arbitrarily unbounded priorities, predictive ambiguity, and even ambiguity in the outcome. My personal "resolution" to the problem is, yes, obviously, you save as many lives as possible even if that means actively changing the outcome... but that's essentially irrelevant, because you will never in your life be faced with such a stark, mathematically-rigid choice. There's always other options... not always relevant ones or desirable ones, but always something. Even when you construct the problem as close to a real life one as possible, you end up with something more like "You just realized you're driving on ice and going way too fast. In front of you are two pedestrians crossing in front of you, one on each side of the road. You have about .3 seconds to consider this problem. Your System 1 instincts think you might be able to weave between them but there's a chance you'll start spinning and hit both. You could also try to jank to the right and go up on the curb, but you might crash yourself into that tree pretty hard and still hit one of the pedestrians. Regardless of what you choose System 2's gonna be second guessing you pretty hard for the rest of your life." You don't get "Choose A and this will happen, choose B and this will happen", it's a big mess. It's always a big mess. Even if you literally encountered the trolley problem in real life, one must consider whether your understanding is accurate, or if the person who has explained it to you is lying or has ulterior motives, and so on. (And that's ignoring "this is clearly the trolley problem and I must be on TV or in a psychology experiment"; let's pretend it's in a world that isn't obsessed with this particular problem.) Now, that is arguably the point of the trolley problem, to get down to that level and ask ethics problems. But it's dozens of layers of abstraction away from anything relevant to the real world. It's a mistake to think otherwise and stress about it as much as the gestalt seems to.

Wowfunhappy 6y ago
> In real life you are never presented with such a binary problem. Everything is a mix of arbitrarily unbounded priorities, predictive ambiguity, and even ambiguity in the outcome. One real-life example that comes to my mind is whether we should have done COVID vaccine challenge trials. I always thought they were a good idea. Yes, it would have put people at risk, but it could have potentially saved many more lives! My father is a doctor, and in fact is an infectious disease specialist. He's been very busy this past year, working primarily on COVID research. When I talked to him about the idea of a challenge trial, he was completely against the idea—so much so that he got a bit upset, and we had to stop talking about it. That doesn't happen often. The idea of purposefully giving participants a deadly disease—even with their consent—seemed fundamentally wrong to my father, regardless of how many other lives could have been saved in the process.
- davidgay 6y ago
  https://en.wikipedia.org/wiki/Primum_non_nocere https://en.wikipedia.org/wiki/Primum_non_nocere would seem to be the ethical principle at play here .
  - Wowfunhappy 6y ago
    Right, but that's the essence of the trolly problem, isn't it? Whether you can take one life to save multiple lives.
    - MrPatan 6y ago
      You're right, and for exactly the same reason: If somebody dies in the human challenge trial it's very specifically your fault. The people that die due to a delay in the vaccine because of not using human challenge trials? Nobody knows them, they don't count, and they can't sue you.
- jerf 6y ago
  "One real-life example that comes to my mind is whether we should have done COVID vaccine challenge trials." You don't know what the result of the trial will be, already taking us out of the domain of the trolley problem. If you knew it was going to kill everyone who had it but you'd get this particular bit of info that was guaranteed to have this result for the general population, then we'd be much closer, but there are multiple false assumptions to get there... and we still haven't discussed all the different ways we could run the trial because it's not a matter of "Either we do the trial this exact way, or we don't"; do we do a double-blind? Base it on multi-armed bandit math? Do we focus on certain demographics? And so on and so on. Not a binary choice between these two guaranteed results; a huge range of possibilities, unknown outcomes, even things that are hard to measure outcomes.