4 ms·

Regulating safety and noise are not a buzzkill. They are essential in preventing corporations, like most regulation, from killing people. Tesla's autopilot is

by maweki 5y ago

Regulating safety and noise are not a buzzkill. They are essential in preventing corporations, like most regulation, from killing people. Tesla's autopilot is the best example that the hurdles are not regulatory in nature, no matter what musk is trying to tell us. As of now, these machines will kill people on the regular. Even in situations that a human could handle. And that's the bar we're setting and I'm glad we do so. Edit: I know that people are killing people on the regular in situations that a machine could handle. And introducing this stuff as safety systems (like autonomous emergency breaking) is not hindered by regulation.

xorfish 5y ago
The main way you reduce traffic fatalities, is by desgining streets and roads in a way that increases safety for all users. One of the most effective ways is to get as many people to use something other than cars for transportation. You do that by making cycling, walking and public transport faster and saver.
- maweki 5y ago
  This side of the equation is public policy though. But I agree, that this is the best mechanism we have.
  - xorfish 5y ago
    It will also become a economic necessity. Car dependence is not sustainable. Infrastructure costs need to be covered by the income they generate. That is not the case with car dependent cities. Here is an excellent playlist about how car dependence is so expensive: https://www.youtube.com/playlist?list=PLJp5q-R0lZ0_FCUbeVWK6OGLN69ehUTVa https://www.youtube.com/playlist?list=PLJp5q-R0lZ0_FCUbeVWK6...
- andrepd 5y ago
  Absolutely. If you want cars to go 30kph in a residential zone, you cannot make a straight, four-lane street and plonk a round sign with a red border saying "30". You need to make a narrow street with bumps and/or curves to make it physically impossible for cars to drive at unsafe speeds (and in the process make most of that space walkable/cyclable and safe for people. YT channel suggestion: "Not just bikes"
  - afandian 5y ago
    As a pedestrian I am terrified by streets with occlusions and random obstructions in the name of traffic calming. There's a road in Oxford with build-out with a tree in the middle of it. You can't see cars coming and they can't see me. Same goes for removing lines from the road. I'd rather cars knew where they stand and focus on what's important like looking out for other road users. I've always thought it was a perverse conclusion that the way to make someone operate heavy machinery more safely is to increase cognitive load of the people operating them. Why not have a the straight roads you describe, with rigid automatic enforcement and penalities that take dangerous drivers off the road? And then put in place infrastructure to make sure that not driving a car is a viable option, so that stricter enforcement has public support. I see this option as a self-reinforcing route to reducing car use.
    - cge 5y ago
      On the other side of this: there's a residential street I cross rather frequently in California which is a straight road with good visibility, and this being California, both unsignalled crossings and a 35mph speed limit. Everyone drives at or above 35mph, and almost no one yields for pedestrians at cross walks despite all the yield to pedestrian signs that are clearly visible. I've seen people wait for several minutes afraid to cross; I've had one driver turning right onto the road run into me while I was crossing at a crosswalk, despite their stopping and presumably checking the road before turning; and I've frequently seen cars have to swerve or heavily brake in order to avoid running into the cars that have yielded to let pedestrians cross, despite those cars being quite clearly stopped and visible. In all my experience with high-visibility roads, even though you'd expect people to be better able to pay attention to everything on road with good visibility and clear markings, they actually seem much worse in those conditions. And given the ways in which they are worse, it's not clear how automatic enforcement would work; speed limits and traffic signals can be enforced, and perhaps crosswalks, with cameras, but often regulations seem focused on banning, rather than encouraging, this sort of enforcement.
      - afandian 5y ago
        Crosswalks with traffic lights and speed cameras seems like such a self-evident solution to this. That doesn't seem any more authoritarian or heavy-handed than literally digging up the roads to make them more miserable to use. (Also, turn-on-red is such a weird idea. Especially as, from the sound of it, the rules vary from locale to locale.)
netjiro 5y ago
Regulatory compliance is, in many areas, a major cost of development. Sometimes the largest. Much of the regulation is in form, not function. Comply with box ticking and you pass, regardless of whether the product is safe in reality or not. If regulation was focused on function it would be less expensive and yield better results for society. > ... is not hindered by regulation. It is effectively stymied by added total cost.
- maweki 5y ago
  I wonder which regulation you're referring to. Most ticked boxes I can think of are indeed of the "does not kill a certain kind of person"-kind, or of the "can be used by a certain type of person"-kind. And of course, regulation can not cover every case. And if it turns out to be a problem, we can cover the additional cases. This way (in Europe at least) we gradually require cars to be safer for passengers and the environment.
  - TeMPOraL 5y ago
    I think GP is right that there's plenty of "form, not function" regulation - of the form, "must be no longer than X / no heavier than Y", etc. But I think GP is wrong in considering this to be a problem. Rules may be expressed in a non-functional way, but that's often just a way to embody functional rules in a heuristic that doesn't require expertise to apply. For a computing example, you may have a processing algorithm that has a tendency to explode in computational complexity on some datasets, because of complicated algorithmic reasons. Instead of explaining the exact problem to users, most of whom will not be able to understand it, you could determine that the failure mode is very unlikely to happen for smaller datasets, and just say "the input should be no larger than 10MB in size". Other reason for "form, not function" is coordination - there may be many good ways to do something, but all of them require everyone to do the same thing. So it makes sense to pick a reasonable default and tell everyone to stick to it.
    - andrepd 5y ago
      You have still not given a concrete example of what you mean.

by maweki 5y ago

xorfish 5y ago

The main way you reduce traffic fatalities, is by desgining streets and roads in a way that increases safety for all users. One of the most effective ways is to get as many people to use something other than cars for transportation. You do that by making cycling, walking and public transport faster and saver.

maweki 5y ago
This side of the equation is public policy though. But I agree, that this is the best mechanism we have.
- xorfish 5y ago
  It will also become a economic necessity. Car dependence is not sustainable. Infrastructure costs need to be covered by the income they generate. That is not the case with car dependent cities. Here is an excellent playlist about how car dependence is so expensive: https://www.youtube.com/playlist?list=PLJp5q-R0lZ0_FCUbeVWK6OGLN69ehUTVa https://www.youtube.com/playlist?list=PLJp5q-R0lZ0_FCUbeVWK6...
andrepd 5y ago
Absolutely. If you want cars to go 30kph in a residential zone, you cannot make a straight, four-lane street and plonk a round sign with a red border saying "30". You need to make a narrow street with bumps and/or curves to make it physically impossible for cars to drive at unsafe speeds (and in the process make most of that space walkable/cyclable and safe for people. YT channel suggestion: "Not just bikes"
- afandian 5y ago
  As a pedestrian I am terrified by streets with occlusions and random obstructions in the name of traffic calming. There's a road in Oxford with build-out with a tree in the middle of it. You can't see cars coming and they can't see me. Same goes for removing lines from the road. I'd rather cars knew where they stand and focus on what's important like looking out for other road users. I've always thought it was a perverse conclusion that the way to make someone operate heavy machinery more safely is to increase cognitive load of the people operating them. Why not have a the straight roads you describe, with rigid automatic enforcement and penalities that take dangerous drivers off the road? And then put in place infrastructure to make sure that not driving a car is a viable option, so that stricter enforcement has public support. I see this option as a self-reinforcing route to reducing car use.
  - cge 5y ago
    On the other side of this: there's a residential street I cross rather frequently in California which is a straight road with good visibility, and this being California, both unsignalled crossings and a 35mph speed limit. Everyone drives at or above 35mph, and almost no one yields for pedestrians at cross walks despite all the yield to pedestrian signs that are clearly visible. I've seen people wait for several minutes afraid to cross; I've had one driver turning right onto the road run into me while I was crossing at a crosswalk, despite their stopping and presumably checking the road before turning; and I've frequently seen cars have to swerve or heavily brake in order to avoid running into the cars that have yielded to let pedestrians cross, despite those cars being quite clearly stopped and visible. In all my experience with high-visibility roads, even though you'd expect people to be better able to pay attention to everything on road with good visibility and clear markings, they actually seem much worse in those conditions. And given the ways in which they are worse, it's not clear how automatic enforcement would work; speed limits and traffic signals can be enforced, and perhaps crosswalks, with cameras, but often regulations seem focused on banning, rather than encouraging, this sort of enforcement.
    - afandian 5y ago
      Crosswalks with traffic lights and speed cameras seems like such a self-evident solution to this. That doesn't seem any more authoritarian or heavy-handed than literally digging up the roads to make them more miserable to use. (Also, turn-on-red is such a weird idea. Especially as, from the sound of it, the rules vary from locale to locale.)

netjiro 5y ago

Regulatory compliance is, in many areas, a major cost of development. Sometimes the largest. Much of the regulation is in form, not function. Comply with box ticking and you pass, regardless of whether the product is safe in reality or not. If regulation was focused on function it would be less expensive and yield better results for society. > ... is not hindered by regulation. It is effectively stymied by added total cost.

maweki 5y ago
I wonder which regulation you're referring to. Most ticked boxes I can think of are indeed of the "does not kill a certain kind of person"-kind, or of the "can be used by a certain type of person"-kind. And of course, regulation can not cover every case. And if it turns out to be a problem, we can cover the additional cases. This way (in Europe at least) we gradually require cars to be safer for passengers and the environment.
- TeMPOraL 5y ago
  I think GP is right that there's plenty of "form, not function" regulation - of the form, "must be no longer than X / no heavier than Y", etc. But I think GP is wrong in considering this to be a problem. Rules may be expressed in a non-functional way, but that's often just a way to embody functional rules in a heuristic that doesn't require expertise to apply. For a computing example, you may have a processing algorithm that has a tendency to explode in computational complexity on some datasets, because of complicated algorithmic reasons. Instead of explaining the exact problem to users, most of whom will not be able to understand it, you could determine that the failure mode is very unlikely to happen for smaller datasets, and just say "the input should be no larger than 10MB in size". Other reason for "form, not function" is coordination - there may be many good ways to do something, but all of them require everyone to do the same thing. So it makes sense to pick a reasonable default and tell everyone to stick to it.
  - andrepd 5y ago
    You have still not given a concrete example of what you mean.