Hacker News Reader

Comment by apeace | Hacker News Reader

I'm really shocked at the trend in comments here.

Yes, we get it. Fatalities will happen sometimes, they are unavoidable sometimes, and what matters in the long run is if we can achieve a significant overall reduction in fatalities.

But my god, a person was just killed by a computer. Can't we have some compassion and humility?

Let's, as a community, set the standard for how we will react to these events. Let's make sure Uber releases detailed data on what happened, whether they were at fault or not. Let's hold the media accountable for their reporting. Let's mourn the loss of life and think about how we can solve these problems.

But for christ's sake, please stop posting the same thing everybody already knows, which gets posted on HN whenever a self-driving article comes out.

Can't we do better?

---

EDIT: I clarified what I meant here [1]

Wasn't trying to say this thread should be all about mourning. This is HN so we should talk about technology, even when a person died. I'm pointing out one specific argument that gets repeated over and over in place of a substantive discussion, and I think we can do better.

[1] https://news.ycombinator.com/item?id=16621589

View on HN

What trend in the comments is bothering you? It's a little unclear from your comment.

Fatalities will happen sometimes, they are unavoidable sometimes, and what matters in the long run is if we can achieve a significant overall reduction in fatalities.

I haven't even read any of the comments on this story yet, and this type of pro-self-driving-car response is what I expect to be in the majority. That and anything anti-Uber :)

The trend I'm seeing is to immediately defend self-driving cars based on "but if you look at it statistically I'm willing to bet this is better than human drivers".

It is as if HN users expect the news headline to be: "an event occurs which is statistically speaking non-anomalous".

If one doesn't feel we have enough data to draw conclusions yet, then sure, fine. But a person just died. This is a newsworthy story and it deserves our best attention and discussion, not repetition of something obvious ad nauseam.

The root problem is that that there isn't enough specific new information to slow down the discussion and make it substantial. Lacking that, people recycle the generic points. Meanwhile the story is a dramatic one, so the energy to discuss it is high. That's a double whammy: lots of energy driving the gears, but no grist to slow them down. This produces reams of commodity comments. I'm sure the story being at #1 amplifies the effect, too, so it's a triple whammy. But it doesn't make sense for the story not to be there.

When the sensational-to-informational ratio is so large, we always get this.

> The root problem is that that there isn't enough specific new information to slow down the discussion and make it substantial. Lacking that, people recycle the generic points.

Is there a term for this tendency? Curious to read the research on it.

It would be great to have just the right term for it, because then there would be a better chance of the community absorbing the idea, and that kind of feedback loop is the only thing that can alter behavior. So far I don't know of one. I use the distinctions 'specific vs. generic' and 'reflective vs. reflexive' in moderation discussions a lot, but those don't quite cut it.

Alan Kay pointed out that Kahnemann's slow-system vs. fast-system also applies here.

I hereby propose the Dang Double Whammy Effect.

"Bikeshedding" should cover it if there isn't a more specific term. If not, then bikeshedding will.

I still don't understand what specifically you're objecting to (at least I think I don't). As best I can tell, you don't disagree with the points about statistics but think the discussion is somehow "without compassion", which stops it from being "our best discussion".

Do I understand you correctly that you'd be okay with the comments if the were just prefixed with some "thoughts-and-prayers" genuflection? And if so, why not say it directly?

I think dang described it best here [1]

It's not the lack of compassion, it's the lack of substance.

And I suppose it strikes me as lacking compassion and/or humility to come and use HN as a platform to recycle the same obvious idea, when the story is about a death. It would be different if it were about a tire blowing out.

Everyone knows how these statistics work on a macro level. That's not worth commenting here on HN, and especially not on this story.

I just think we could do better.

[1] https://news.ycombinator.com/item?id=16621716

I read dang's comment before posting. It still didn't help -- I can't come up with a mental model that generalizes the criticism into actionable rules for improved discussion (except as noted below).

>Everyone knows how these statistics work on a macro level. That's not worth commenting here on HN, and especially not on this story.

If statistical comparisons were obvious, and people didn't overreact with unjustifiable countermeasures, then indeed it wouldn't be important to make the point. But the world we live in is one where nuclear deaths are scarier than heart disease deaths, and shark attacks more than auto collisions.

I think you're committing the fallacy of "my social circle doesn't make that mistake, therefore it's not worth pointing out."

I can agree that users should be more careful about making the common point when others have made it, though -- I had a reply on this thread about "Uber flouted the $150 license" that I deleted when I saw others made a more informed response to the same point.

>> not repetition of something obvious ad nauseam.

The fact that Uber's self-driving car program has a worse deaths-per-miles-driven metric than human drivers is anything but obvious to me.

After an incident like this, it's natural to question whether anything needs to be done policy-wise. And, I don't see any reason why that discussion can't take place here in the comments section.

It seems like you want HN users to use the comments section to express condolences? Or, to just have an empty comments section?

I find what might help those that quote statistics is to self-reflect and ask the question, how would you react if your father/mother/sister/son/daughter/etc was the victim?

Would you care that statistically self-driving cars have killed less people than human drivers?

Yes. I'm in favour of safer drivers, robot or human.

Well, just because people react badly when confronted with personal tragedy doesn't mean you have to do the same when it's not applied to you.

For example it's clearly excusable to be angry / depressed / looking for someone to blame when a personal tragedy happens to you. But that doesn't mean you should replicate that behaviour when a tragedy happens to someone else. We should distance ourselves from subjects where we're emotionally compromised, not guide our decision by the example of emotionally compromised people.

I think that's a strange thought experiment, since it's basically requesting that you think emotionally and as irrationally as possible to draw conclusions.

It's requesting that you remember that a human life was just lost, and to not reduce this woman's life to a rounding error.

The only strange thing in this thread is the emotional detachment many are showing and one could probably argue that's the exact kind of thinking that will lead to more cases like this.

> how would you react if your father/mother/sister/son/daughter/etc was the victim?

How you would react when emotionally distraught shouldn’t be the model for how you act all the time.

This. And I think this is why these stats end up posted here when stuff like this happens. Because the emotionally distraught look at this situation and they're like BAN SELF DRIVING CARS! and the people posting stats are here to remind them that they're still safer than regular cars and accidents happen but we can't impede progress because of one accident here and there.

Could you please not use allcaps for emphasis in HN comments?

This is in the site guidelines: https://news.ycombinator.com/newsguidelines.html.

I didn't realize that, it's been a while since I read those. Thank you for the information!

> Would you care that statistically self-driving cars have killed less people than human drivers?

I don't think there are enough data points right now to make such claim, but please correct me if I am wrong.

Would you care that seatbelts are statistically likely to save lives if your loved ones died drowning in their car due to them?

What do we do differently based on the answer to that question?

> …statistically self-driving cars have killed less people than human drivers?

I realise that the point you're making is about emotional turmoil and not really statistics, but it looks like so far Uber aren't doing so well in terms of driving safety:

https://news.ycombinator.com/item?id=16620736

So yeah, if it turned out that Uber were sending cars out into the world without sufficient engineering or testing* and they killed someone (especially someone I know) then I would be understandably furious.

* In order to get to market sooner? I don't trust Uber and I wouldn't put this past them. Nothing I have seen of the Uber self-driving car program makes me want to be anywhere near one on a road.

> how would you react if your father/mother/sister/son/daughter/etc was the victim?

Shit, of course. But I'd like to believe that I'll realize that banning self-driving cars is not going to make things any better, just like I'd never propose banning cars in general. I'm not sure how it would help here to imagine it's my daughter; it's not as if we imagine it being our daughter in every traffic accident we hear of, since that would be crippling given the sheer number of them...

All deaths are a tragedy to someone. A statistic is an aggregate account of untold suffering and misery. To be truly compassionate, we must learn to see the humanity in statistics. We must recognise that our emotional reaction to the death of a loved one and our indifference to the death of a stranger is not a sign of compassion, but egoism. As enlightened citizens, it is our duty to use all our ingenuity and effort to reduce the sum total of suffering.

This death is a tragedy. There were 37,461 such tragedies in 2016. Every single one of them matters equally. It is only by studying and analysing them in aggregate that we can prevent future tragedies. Seat belts, air bags, crumple zones, anti-lock brakes, median dividers and rumble strips are tremendous acts of compassion. Statisticians, scientists and engineers have made an incalculable contribution to human welfare.

The news isn't about the fact that a human died, unless I'm somehow missing the constant flood of ~50 news items a day about other deadly car accidents. The news is very much about self-driving cars and our relation to them. That's what people react to. The human death is tragic, but not really the topic.

Lately, I'm quite disappointed with the tech community's approach to the society.

Lot's of people are talking as if human beings are a resource which exploitation should be secured for the businesses and technical curiosities.

No, people, humans are what matters and it just happens that there are business opportunities and scientific curiosities. No, it not O.K. to kill few people to speed up the development of self-driving vehicles. No, it is not O.K. to have total surveillance so that we improve AI.

It’s not just you, it’s a lot of the country that’s rapidly going from concerned to pissed. If they read more threads like this, they’d probably already be coming for us with torches and bad intentions. They’re worried that tech is full of disconnected, vaguely autistic (and they don’t know what that means beyond iRobot), too-smart-for-their-own-good people who care more about making money than poeple. In my experience that’s incredibly unfair, but it only takes a small percentage of loud and callous voices to make us all look bad.

We need to start cleaning our own house, or it’s going to be cleaned for us by people with agendas and little understanding of what they’re cleaning!

Do better than what? This is a technology news forum, the relevant question is how this affects the technology. People are going to talk about the potential long term trends in self-driving technology. What exactly were you expecting?

I think the relevant question must be how the technology is going to affect lives.

Okay, and is that not what is being discussed here? I'm just really confused as to what the OP is complaining about.

I think what bothers me is that the whole story gets dismissed by one idea: that this event doesn't matter, it's only the overall statistics that matter.

That this event doesn't necessarily reflect a larger trend is not incorrect--it's just that it's obvious. That self-driving cars could potentially reduce fatalities in the future is not only obvious, it's also somewhat off-topic when the story is that a person just died.

So my question is, when a computer kills a person, can't we do better? Better than stating the obvious, and better than being dismissive.

"The relevant question is how this affects the technology"

Is arguably the whole problem with our industry in a nutshell. The more relevant and largely ignored question should be "how this affects human beings".

This just seems like a rhetorical trick to me. Obviously we're talking about humans, it's just through the lens of technology. The website has to be about something in particular besides "aren't humans great??".

Actually I think rhetoric matters enormously - how you frame a question affects how you come to think about a situation.

If you ask a question "through the lense of technology" ("Can we solve this hard computer-vision problem"?) you arrive at different answers to those you would get if you ask the question through (say) the lense of societal benefits or humanity's relationship to technology ("Is it ok for a badly programmed computer to accidentally kill someone"?).

I say this not to criticise the enormous benefit this technology could bring, but rather to provoke us to think about the implications of the technologies that we are creating outside the confines of an IDE.

I'm far from being an Uber-supporter here but I think you're overreacting. As others have pointed out, this story is only relevant to HN (and to the national media at large) because it involves Uber and AV, not because it involved a pedestrian dying. It makes sense that the discussion centers around policy and systems since few people here can claim to have even known the victim. Nor does participating in this discussion preclude anyone from going to the victim's Facebook page/funeral and expressing condolences.

What is that supposed to mean? If the only way it directly impacts a forum is because of the company doing the killing, the killing is irrelevant?

Uh, no. Sorry, at a loss to understand where you see "killing is irrelevant" in my comment.

edit: NM, I can see where you'd see that, even though I argue that both being concerned about the death, and wanting to defend Uber, aren't mutually exclusive. Instead of saying "this story is only relevant to HN", I should have said, "this story is only noticed by HN and the national media". Which is true, as far as I can tell. I can recall few nationwide stories about other pedestrian deaths, even though nearly 6,000 occur every year.

A person was not killed by a computer. There was a person behind the wheel responsible for taking control of the car in a situation like this. If this is a computer killing someone, then every instance of a person being killed while cruise control was engaged also counts.

Yeah, someone put that poor person there despite all warnings that humans would perform terribly as standby drivers once the AI becomes good enough to not require interventions all the time.

> But my god, a person was just killed by a computer. Can't we have some compassion and humility?

You have to recognize that your culture and your moral standards are not universal and you are in no position to force them on everyone else. It is also foolish to expect everyone across the globe to react in a certain way to certain things.

On the other hand, death topics is the side of HN that is heavily biased towards what mods feel about it and their culture. They even ban accounts and give out warnings if you happen to disagree with them and react differently. So there is not much sense to even discuss this.

You are seeing the Trolley Problem Dilemma[https://en.wikipedia.org/wiki/Trolley_problem] in action. Majority of respondents when asked that problem would choose to let a few people die to save a larger number of people. Emphasis being on "let a few people die" and not "kill a few people", machines (i.e. lever in that problem and autonomous car in this scenario) somehow evoke a more logical and less emotional response, I guess machines create an emotional distance between the people who would die and the respondent. If the same respondents would somehow feel more responsible for the death, i.e. the machine is removed from the scenario the response would be much different.

Not implying that the lack of empathy is in anyway acceptable, but people should know that when machines are involved in such scenarios it is much easier for them to distance themselves emotionally, this is not new, weapons are a prime example of this. Maybe by knowing this dilemma people can react better to such tragedies.

I hope people show more empathy knowing this.

I think the argument needs to be made for ETHICAL development of mission critical software. The Aerospace industry has been writing software that has peoples lives in their hands for years. NASA has been doing the same.

We need to see accountability for any death caused by a car being operated by a computer. We need to see integrity in testing and metrics that show that these are actually safer for humanity. One more death than expected due to software vs human is too many.

I don't remember where I saw it but the process for creating mission critical software at NASA is insane. They analyze everything. It's a completely different and slower process than the one we use for building websites. SpaceX has probably increased the productivity of said systems, but I'd like to see how they develop mission critical software to keep things safe.

The point of self driving cars is to reduce death, if you can monetize that great, but if you can't hit that baseline in some sort of testable way, then you shouldn't be able to put your car on the road.

I understand your point. But people are capable of recognizing the tragic loss of a life while also trying to understand both the cause of it and the implications.

The victim's life had meaning. To herself, to her friends and loved ones, and all of those affected by her untimely loss. We don't know her story, nor the stories of the over 6,000 pedestrians (a number that's been increasing) killed annually in the US alone as of 2016.[0] Nor the 37,461 killed in US motor vehicle crashes in the same year.[1]

We talk about the technology, because it was involved in this woman's death and because the technology has the potential to drastically reduce the overall number of such accidents. Understanding why she died, and how similar accidents can be prevented in the future, can give society back something out of an accident that only took. If we can learn from her death, anything at all, it behooves us to do so.

Yes, it's cold. It's extremely cold. But it's something doable. Even if these discussions offer little impact on the actual development of self-driving cars, they can still impact the political debate that's only now just starting. For better or worse, there will be those who point to this woman's death as an argument against the technology. Recognizing that her death is not simply one death, but an example of a type of incident that claims thousands each year without self-driving car involvement, is critically important. At times, statistics can dehumanize and cause us to overlook the very real pain depicted as a small part of a number. But at the same time, they also permit us to recognize problems and measures ways in which we can mitigate and even solve them.

0. https://www.npr.org/2017/03/30/522085503/2016-saw-a-record-i...

1. https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/...

What's noteworthy of this news event is the involvement of an autonomous vehicle.

People are being killed in accidents constantly. We aren't going to be in a constant state of mourning over it.

This is implicitly a discussion about autonomous vehicles, and the very currently relevant and heated topic of wether we want these things on our streets.

40100 deaths in 2017. Averaging 110 per day.

According to this[0], I reckon there are 16 cars (order of magnitude 10s at least) in the pilot down in Az. In almost a month, one has died.

There are about 260M[1] cars in the us in 2015, so perhaps more today. That means there are ~4e-7 deaths per car per day. On the other hand, we have 1e-3 deaths per autonomous vehicles per day if we naively use the death of the person in this story. That means so far, autonomous vehicles have a rate of killing people a ~2500 times that of normal cars.

Of course, it's one data point, but out the gate, it's not looking so strong.

[0] https://www.theverge.com/2017/2/21/14687346/uber-self-drivin...

[1] https://en.wikipedia.org/wiki/Passenger_vehicles_in_the_Unit...

What bothers me is that people are jumping at all sorts of narratives and conclusions when the article clearly pointed out that the incident is still under investigation.

re: your edit. I agree with the sentiment, but I didn't quite get your comment because before you posted it, the most upvoted comment/discussion on this story was about Uber failing to meet safety standards [0]. People have been posting about "the greater good" but seems far from a consensus.

[0] https://news.ycombinator.com/item?id=16620042

1.75 people die per second. Calling for compassion and mourning of one stranger over the others seems fake. Personally I'm pleased over the circling of wagons, the very real danger is emotional kneejerks to this will set back the entire self-driving industry by at least 5 years. The trend of comments is a helpful loyalty signal even if it's not very substantive.

Reminds me about this comment a few weeks ago (which made an impression on me) :

>'How low has the SW development bar gone, if "it's okay" now means "at least it's not directly killing people"?' https://news.ycombinator.com/item?id=16541235

We now see the standard has slipped to: "Is the death of one person relatively so bad? People die every day."

Next a large crowd of people will be killed and the standard will become: "It's not so bad compared to creating skynet or starting thermonuclear war."

I'm not sure what comes after that.

"It's not so bad compared to the alternative"

I mean, we two are still capable of discussing it, at least!

Interesting, isn't it?

But if you think about it, that's not a pattern anyone should be unfamiliar with. History is full of it.

I don't know, I really have no strong opinion here. But I think stuff like this is truly fascinating. There seems no end to where rationalization can go.

Welcome to the ambivalence of reality...

I don't think the people arguing for a different standard are saying that it is ok for one person to die because people die every day. They're asking for observers to look at self-driving cars in comparison to human-driven cars. Because people dying is such a bad thing that we want to prevent, the relevant question should always be, "will this change make it more or less likely that somebody will die." And that is why it is relevant to compare pedestrian deaths from self-driving cars to pedestrian deaths from human-driven cars, but not relevant to compare it to skynet or thermonuclear war.

I don't know, however, how that comparison turns out. From the other comments, it looks like Uber's self-driving cars might be less safe than human-driven cars--in which case you and I would end up on the same side of the argument.

Sure, but in the oculus rift comparison someone could have easily died because the VR surgery equipment failed for a reason so stupid as a certificate expiring during surgery.

You can make the same argument that people die in non-VR surgery all the time so it is no such a big deal.

You're absolutely right that designers, regulators and, in some cases, courts should take every death (and accident/near miss) seriously. But it seems to me like there are two questions here. The first question is, "should we have self-driving cars?" and the second is "how can we make those cars as safe as possible?" When people say, "compare it to a human driver," they are suggesting a way to answer the first question, not the second. And that does not mean that they think the second question is "no big deal."

It's also worth noting that the quote he presented from the Oculus Rift thread was criticizing someone for pointing out that it is less concerning if surgical training equipment crashes than if surgical equipment crashes, not for saying that it is no big deal if someone dies in surgery. I didn't see anyone in that thread saying it would be ok for VR surgery equipment to fail.

While you're admonishing people for their comments and employing stealth implicature like, "let's make sure Uber releases detailed data", do you think that maybe you should include a prominent notice disclosing your (and your former colleagues') ties to the industry that Uber is operating in—most importantly as a competitor?

Yikes - bringing someone's personal details into threads as ammunition in an argument is offside. It crosses into personal attack, so please don't.

https://news.ycombinator.com/newsguidelines.html

That's not at all what happened here (least of all because I'm not the person who was trying to advance an argument).

I'm not sure I understand, but the high bit here is that bringing in someone else's personal details crosses into personal attack and isn't ok.

> Can't we do better?

no. there's a reason our community has the reputation of being heartless. it's because it's true.

It's not at all true. People here care very much. It would be weird if a large population sample (4M a month or so) had a different human nature than humans in general.

What is true is that large internet forums suffer from systemic effects that compound into problems that no one has yet figured out how to deal with.

> large internet forums

what in the world did you think i meant?

you can try and pretend all you want, it ain't my fault if the mirror ain't a pretty thing to behold.

>Fatalities will happen sometimes

How many self-driving cars are there on the roads right now? And what's fatality rate? I' quite sure it's higher now than compared to human-only driver.

> they are unavoidable sometimes

Were autonomous cars not designed to prevent such situations?

>Let's, as a community, set the standard for how we will react to these events

Yes and let's punish non-conformists!

> which gets posted on HN whenever a self-driving article comes out.

I honestly have no idea what that comment is.

Probably not since Google drove millions of KM without a single fatality.

As of late 2017, Google/Waymo had recorded just 4 million miles of real-world self-driving in its entire history: https://techcrunch.com/2017/11/27/waymo-racks-up-4-million-s...

NHTSA data measures number of fatalities by the hundreds of millions of miles driven -- the latest rate being 1.18 fatalities per 100 million miles: https://www.nhtsa.gov/press-releases/usdot-releases-2016-fat...

Yes, absolutely, let's ostracize people who react to this situation (a machine killing some random person in a fatal beta test) with anything other than compassion and humility.

Those are not people I want in my engineering community; just impossible to trust they'd make the right calls and have the right priorities in developing cool new stuff.

Reply on news.ycombinator.com