The real lever to the discussion: why does it take an hour to continue, and how to shorten that? In an ideal world, the emergency break was applied, someone communicated the situation with the driver via intercom, the driver resets the brake from his cab, train proceeds as usual - a delay of a minute, a minute-a-half?
Someone made the deliberate decision to make this process take an hour instead, which is completely out of proportion. Your calculcation in that case should be even worse, because the delayed commuters will delay others as well (e.g. the clinic staff who now have to cover for the delayed collegues).
The discussion on to-leve-or-not in a split second situation, specifically in groups, is futile to me: assessments' hard, so always act on the worst-case-assessment side (one person mamed for life, child drowns), think later (and scold If necessary, e.g. boy cried wolf situations).
The QALY cost of pulling the brake is not nearly so high as calculated. You can't assume the extra commuting time is life completely lost. It's life mildly degraded. There would be a lot of variance between individuals -- some might suffer cascade effects that stretch beyond the delay, and some might somehow even benefit from the delay, but I would guess the QALY cost at no more than 10% of the time wasted, and probably much less.
Meanwhile, if the guy's hand is actually caught, it not only costs the guy, and likely causes larger delays (as jeffk points out), it also affects the bystanders who witness a guy getting hurt, and maybe feel guilt about not acting, or maybe develop fear of trains. The narrative that some guy's hand was caught in the door and no one bothered to help, is also going to be deeply harmful.
Do we know for sure that it would take an hour to continue? Your story about what would happen afterwards in your ideal world sounds plausible to me as something that might happen in the real world, up to an extra minute or two for the drive to walk over, look around, and tripple check everyone and the train are OK.
absolutely correct. A brief search on train delay studies (I only found some articles) shows a direct delay of mere minutes in such an incident in most cases (subways and regional transport). Other non-emergency-incidents where commuters insisted on an "emergency" (unattended bag, phone fell out of window, feeling anxious) let to delays of ~5-15 minutes. Some older articles (~2004) suggest that the staff had to walk the whole train to find the emergency break to reset, leading to longer delays.
So nowhere near an hour, but even brief delays of 5min can result in cascading delays later due to necessary replanning of the line usage, especially on busy rail lines.
I’m highly sympathetic to the idea that this should have been resolved faster, but I think an hour is the right ballpark for the delay to the passengers on that train in the particular actual incident under discussion. The linked Reddit thread indicates that the brake was pulled around 9:30am, and then as of 10:42am, everything is reported as moving again with lingering delays estimated to remain for up to 60 minutes thereafter (with delays varying in different portions of the line). The stopped train does get moved out of the station a bit before that second timestamp but this all seems to support the idea that the train was stuck for the better part of an hour.
Another commenter in the thread indicates that the initial response took 15 minutes, and then there was half an hour of failing to get the train doors to open and/or the train to boot up. Which implies that the train was stuck for at least 45 minutes, and leaves it ambiguous as to whether the train was able to carry passengers when it left the station; if not that would add additional delays to passenger trips, probably raising the total delay to about an hour total (if not longer; except for any passengers who were able to switch to another mode of transportation at some point in that hour).
A third commenter in the Reddit thread suggested that, in this circumstance, the MBTA was wholly unable to restore power to that train after the emergency brake cut its power. I interpret that as an indication that the original train probably wasn’t actually serving passengers when it did successfully leave the station.
Eleven years ago I was on the Red Line (not the section that's a trolley), and found myself in a variant of the classic moral dilemma. I wrote about it on Reddit at the time:
I was pretty shaken up, and didn't end up blogging about it, but it came up again at one of our effective altruism dinners. This got me thinking about ways that philosophical problems do and don't translate into real life.
In this situation the stakes were lower than in the traditional trolley problem, and the situation is closer to Scanlon's Transmitter Room (What We Owe to Each Other, quoted jokingly here). The worst case was someone's arm being badly injured, maybe permanently. On the other side was delaying the whole line for an hour, perhaps losing ~4 years of waking life in total (~25k people * 1hr / 16 / 365 = 4.3y). But were these the real stakes?
I now think his arm probably wasn't caught in the door. That's what it looked like to me, and I was very sure at the time, but people who were closer disagree:
And:
This illustrates a major way philosophical problems can fail to translate to real life: we're always acting under uncertainty, without knowing the consequences of our options. People understand this, and often "fight the hypothetical", but needing to make decisions when you really don't know whether you've read the situation correctly is a key part of real life morality. We need to make decisions that are robust to us having misinterpreted the situation.
(When Julia found herself in "drowning child" situation it was similar: the hard part was assessing the world, not whether to act.)
The other big way this differs from a textbook thought experiment is that it's not clear that the "inaction" case actually would have avoided a train delay. If I'd been right about his arm, I expect once the accident happened he would have been screaming even more, and I or someone else would have pulled the brake anyway. Which might well mean a worse delay, since now there's an injury involved too.
Which illustrates another way that intuitions built from philosophical problems can lead us the wrong way: this didn't occur to me at the time. It's hard to think very far ahead in an emergency, but morality problems often suggest a frame of mind where we have time to fully reason out the likely effects of each of our options. If I had thought this through it would have been an additional point in favor of pulling the brake. Or maybe I would have thought even further ahead: "if I do not pull the brake, I will become a famous example of a calculating (derogatory) utilitarian. I MUST NOT DO THAT."
Another thing I learned is that the suggestion Cialdini makes in Influence (quoted in Yudkowsky's Bystander Apathy) to point to one single person and ask them for help doesn't reliably work. When this happened I wasn't near the brake and I pointed to the person next to it and shouted something like "you, the brake!" They didn't, so I ran down the train and did it myself.
One philosophical idea that did transfer to real life is the Unilateralist's Curse. Perhaps the reason no one else on the train pulled the lever is that they all correctly perceived that the person was only in danger of losing their bag and not their arm. The one person (hi!) who came to a different conclusion still delayed the train for everyone. I don't think that was fully what happened here, since others were also shouting to pull the brake (though they were not independent observers). But since the "it's ok" position was harder to communicate, I think part of what happened was what we could call "expressivity bias", where an idea that was harder to express lost out, combined with it just being very hard to integrate diverse viewpoints under time pressure in a stressful situation.
As someone who had often speculated about philosophical problems, the experience of being in one was very stressful, and it's not one I hope to repeat.
Comment via: facebook, mastodon, bluesky