Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Identify car crash editorial anti-patterns using NLP

visionzeroreporting.com

Post reply on HN
Print view View original

201–210 of 213 posts

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#201

Post by burnished »

speedybird wrote: post #144
burnished wrote: post #136

Earlier quoted context omitted.

Its more the realization that you aren't safe, and that your safety is not only not a priority but somehow actively repugnant, all due to you being on a bike. If you are a kid driving isn't an option. If you don't feel safe on the roads as a pedestrian or bicyclist then your options are rather limited. This was my own experience, and if I had been less reckless I would have gotten less exercise. As it was, telling so…

Color me unconvinced. Kids know that skateboarding isn't safe but I see plenty of kids skateboarding, many of them without helmets! Generally speaking, I think kids are insensitive to risk and think they're invincible. When weighing the risk of cracked skull against looking uncool with PPE, they often forgo the PPE. I think perception of physical danger is low on the list of reasons why kids don't get as much exercis…

>>and if I had been less reckless

Yes, you see kids behaving recklessly. You don't see the less reckless kids behaving that way. It is really not a surprise, right? Some kids, I'm sure, don't skateboard or bike because on one hand there is a social pressure to not wear a helmet but on the other hand they don't want to get their skull cracked.

'Perception of danger' was probably my primary criteria as a child for selecting activities. Why do you think children don't think about this? Do you remember being a child?

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#202

Post by speedybird »

burnished wrote: post #201
speedybird wrote: post #144

Earlier quoted context omitted.

Color me unconvinced. Kids know that skateboarding isn't safe but I see plenty of kids skateboarding, many of them without helmets! Generally speaking, I think kids are insensitive to risk and think they're invincible. When weighing the risk of cracked skull against looking uncool with PPE, they often forgo the PPE. I think perception of physical danger is low on the list of reasons why kids don't get as much exercis…

>>and if I had been less reckless Yes, you see kids behaving recklessly. You don't see the less reckless kids behaving that way. It is really not a surprise, right? Some kids, I'm sure, don't skateboard or bike because on one hand there is a social pressure to not wear a helmet but on the other hand they don't want to get their skull cracked. 'Perception of danger' was probably my primary criteria as a child for sele…

> Do you remember being a child?

Heh, I remember laying down on my skateboard and riding it down hills headfirst without a helmet when I was a kid. Like most children, I was invincible.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#203

Post by Karrot_Kream »

ryantgtg wrote: post #198
Karrot_Kream wrote: post #90

Earlier quoted context omitted.

The US is so car dependent that it's very politically fraught to make it harder to get a driver's license. Not allowing a person to drive in much of the US is a matter of equity; people that can't drive can't make it to their jobs, can't buy their groceries, etc. What that means of course is that the US usually just knowingly increases the risk of non-drivers because of its brittle dependence on cars, and thus disinc…

The lack of a license doesn’t necessarily prevent people from driving. Making it harder, or more severe punishments for poor driving, would not only result in fewer drivers on the road but also more “undocumented” drivers.

Indeed and this is yet another aspect of the car dependency trap. It's so hard to move around in the built environment without a car that individuals will rightfully ignore licensing requirements to tend to their own affairs. Once you force all of your infrastructure to be dependent on a single mode of transport, this mode becomes essential to functioning of the system. This is why it's so difficult to encourage multimodal transport in the US.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#204

Post by ryantgtg »

tshaddox wrote: post #27
dymk wrote: post #9

This is neat, but something bugs me about the framing of the intended goal versus how you propose to get there. If the intent is to raise public awareness and to put pressure on leadership to make roads generally safer, why do almost all the corrections follow a pattern of shifting blame from the individual pedestrian onto the individual driver? Many of the suggested fixes are just their own form of counterfactual -…

Couldn’t part of the larger “why?” be that individual drivers are not sufficiently subject to liability for causing accidents with pedestrians, cyclists, etc.? If it is the case that it’s almost impossible for a driver to get in any trouble for causing such an accident (except in the most egregious cases of reckless driving) then it doesn’t make sense to avoid talking about that.

I can see value in not consistently talking about “btw killing someone with your car is basically a get out of jail card”. Most people probably don’t realize this.

(Or am I misunderstanding your last sentence?)

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#205

Post by Karrot_Kream »

dionidium wrote: post #193
throwaway0a5e wrote: post #190

Earlier quoted context omitted.

>Automated enforcement tools like red light cameras and speeding sensors run into intense political pushback in the US Of course they do. The laws they enforce with machine level reliability do not reflect the reality of how society behaves. Nobody actually cares whether or not people obey the rarely relevant number on the sign so long as people travel a safe and reasonable speed.

The typical driver has no idea what a safe speed feels like. They routinely travel beyond the safe speed for a given environment. Charitably, they are unaware of the connection between their actions and the increased risk to pedestrians and cyclists. Uncharitably, they just don't care. For example, did you know that the risk of severe injury for a pedestrian struck by a moving vehicle is about 25% at 23 MPH but jumps…

Just to add to this, though I'm obviously in agreement with you, is that most engines (gas or electric) used in US cars barely distinguish between 23 MPH and 39 MPH. You could probably just push your foot down a tad bit harder on the gas pedal and make up the speed differential. That's fine; many previous limitations on speed were based around actual deficiencies in automotive development (heavy car bodies, lack of synchros when shifting, heavy clutches, etc) but now that consumer vehicles can so effortlessly go from "0-60 mph" (the common metric used to judge acceleration of cars), it's harder than ever to educate folks on the effects of their speed.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#206

Post by burnished »

speedybird wrote: post #202
burnished wrote: post #201

Earlier quoted context omitted.

>>and if I had been less reckless Yes, you see kids behaving recklessly. You don't see the less reckless kids behaving that way. It is really not a surprise, right? Some kids, I'm sure, don't skateboard or bike because on one hand there is a social pressure to not wear a helmet but on the other hand they don't want to get their skull cracked. 'Perception of danger' was probably my primary criteria as a child for sele…

> Do you remember being a child? Heh, I remember laying down on my skateboard and riding it down hills headfirst without a helmet when I was a kid. Like most children, I was invincible.

>Like most children, I was invincible.

No, like most children you knew. The quiet or more risk averse ones weren't out tearing shit up like we were.

Besides, lets get back to brass tacks here. Currently your argument is that bike safety isn't important because kids feel invulnerable, which is sort of untenable. Road safety is important, and to get back to TFA, I think its important that when describing collisions between bikes and cars some of the onus is placed on the person piloting the car. I find it suspicious that there is a concerted lack of transparency in this regard.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#207

Post by psd1 »

dionidium wrote: post #195
throwaway0a5e wrote: post #130

Earlier quoted context omitted.

Oh please. This is the kind of comical rhetoric that gets road safety advocates ridiculed. In practice speeding or at least the instance that resulted in getting ticketed is usually inconsequential. Between highways signed well under the normal traffic speeds, revenue enforcement and fishing stops the overwhelming majority of speeding tickets are trivial. People engaging in the kind of speeding most people can agree…

Pushing back on a banal, obvious truism like, " we know speeding is a significant cause of and contributor to death, destruction, and injury in automobile crashes " is the kind of comical rhetoric that should get drivers ridiculed. It really shows just how disconnected the typical beliefs about driving are from reality.

Trying to be charitable to both arguments:

- speed is obviously a factor, but there's a wealth of other factors. 90 on a clear highway with little traffic, and the driver is alert? Fairly safe. 90 while weaving between lanes, leaving little reaction time? Not safe. - drivers seriously overestimate their competence. Humans undervalue outlier events. This is the root of the stat about accidents occurring within a mile of home.

That's why we have generally-lax enforcement of motorway limits but "strict" enforcement of urban limits in the UK. I'm not asserting that we've got it right here, but it looks more sensible in my eyes.

Caveat: urban centres are increasingly adopting 20mph. I think there's a tacit understanding that people will drive at 27. When it was 30, they'd do 34, so that's a big increase in survivability.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#208

Post by ryantgtg »

rsj_hn wrote: post #199
ryantgtg wrote: post #198

Earlier quoted context omitted.

The lack of a license doesn’t necessarily prevent people from driving. Making it harder, or more severe punishments for poor driving, would not only result in fewer drivers on the road but also more “undocumented” drivers.

> The lack of a license doesn’t necessarily prevent people from driving. OK, but you need to first decide whether the state should require a license to drive in the first place. And if it requires a license, whether this license should be required to identify you or not. Because you are going to have a hard time with a licensing system in which the license cannot identify the holder. Maybe it shouldn't. That's a usef…

I’m not sure I 100% follow. By “undocumented” I mean that their license has been revoked (and thus they have no insurance), or they never got one. They lack the document to legally drive.

> If you require it, then you should enforce it, and you should not refuse to enforce it because you think that enforcing it does more harm than good.

There’s a third situation: that enforcement is difficult. The last time I was pulled over was 19 years ago. If that situation had resulted in my license being revoked, I could have continued driving without my license for the next 19 years with no consequences (maybe registering my car under a family member’s name). But regardless, how would this be enforced on a wide scale? The authorities don’t have the resources to check that every person behind a wheel is carrying a license. So, these checks generally occur after an infraction.

My point is simply that revoking a license doesn’t necessarily take those people off the road.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#209

Post by dash2 »

what_is_orcas wrote: post #120
dash2 wrote: post #60

Earlier quoted context omitted.

Of course if there is evidence that the driver deliberately tried to kill someone, nobody will call it an accident. I would imagine that the vast majority of car deaths are indeed accidental, not murders. This seems to contrast with the situation of gun deaths. No?

There's a few really busy streets in my neighborhood where people regularly go 10-20 MPH over the speed limit. Not surprisingly, there are quite a few hit-and-runs on those streets. While it's probably the case that the folks who are speeding are just being selfish ("I want to minimize my time on the road") but they are doing so at the cost of safety. The equivalent w/r/t guns might be that someone was firing their g…

Speeding is bad, but it is obviously much less dangerous than firing a gun at random in public.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#210

Post by gault8121 »

chiefofgxbxl wrote: post #168
gault8121 wrote: post #87

This is an interesting tool! It would be cool if you could take say 20 articles from Gothamist, Streetsblog, NYT, NY Post, Pix11, etc. and see how they all ranked in this system. That could be a really interesting blog post. -- Which tools are you using for these NLP judgments? Are you using SpaCy at all? Are there any models you've built here, or is this all rules on top of an NLP model? I'm working on NLP models fo…

Thanks! Comparison between news outlets is on the radar, once I improve the tool based on feedback from this Show HN. My backend uses spaCy (NER/NLP), textacy [1] for extracting subject-verb-object triples, and coreferee [2] for coreference resolution. I did build a custom spaCy model for which I manually annotated over 700 crash-related news articles in Label Studio [3]. This custom model identifies counterfactuals,…

Awesome, thanks for following up! We also use coreference resolution - we specifically found Allen NLP's coreference to be the most accurate in an analysis of a few different offerings: http://docs.allennlp.org/v0.9.0/api/allennlp.models.corefere...
Post reply on HN