4 ms·

> Why not? It's how everything else works. If you buy a house and then install your own electrical wiring which causes your house to burn down, you have no clai

> Why not? It's how everything else works. If you buy a house and then install your own electrical wiring which causes your house to burn down, you have no claim against the construction company and nobody blames them for it because it was your fault, not theirs. This recent attitude of "people won't understand, therefore corporations have to be paternalistic" is patronizing and factually incorrect. People do actually understand. I think you're really overestimating both the level of technical comprehension generally, and people's attitude towards "due diligence". People understand what a house is and what wires are. They are not, as of yet, that clear on how neural nets, sensor systems, and dynamically updated software come together to make a self-driving vehicle. If modders cause problems, especially with something as mobile and potentially destructive as a car, the general reaction could easily be "this is mostly the fault of the self-driving Tesla, and the fact that they didn't do due diligence in preventing misuse." It is not a very reasonable reaction, but it is a likely one - and one that could have serious consequences for Tesla as a company. >The worst outcome is that technology-ignorant bureaucrats require companies to clamp down on modding. But having companies do that to begin with is no improvement. It's not the same. A bureaucrat's rules are mandatory for everybody. Tesla has competitors who may choose a different route. > The basis of this seems to be the conceit that manufacturers can actually control their products after they've been sold. It can be true for a time, in the sense that it takes people that long to figure out how to break back into the things they own, but it happens. If you can jailbreak an iPhone then you can jailbreak your car. A self-driving Tesla is not an iPhone, because an iPhone is unlikely to run somebody over or block a freeway. Risks are relevant. Self-driving systems are also dynamically updated and dependent on networked information. It already is a service, not a fixed product. You don't own a self-driving system (at least Tesla's idea of it) any more than you own an Amazon Web Services server rack by having something hosted on it. > Which means that we have a choice. The first option [....] which makes it much more likely that people will die. What does that do for your brand? Maybe in the future, when such things are more familiar, Tesla will feel confident enough to open their products up to modders. But right now? When even the idea of self-driving cars is a challenging sell, and they're being criticized for it not being secure and consistent enough? No, not now.

AnthonyMouse 10y ago
> They are not, as of yet, that clear on how neural nets, sensor systems, and dynamically updated software come together to make a self-driving vehicle. They don't have to be, for the same reason that you don't have to understand Maxwell's equations to understand that it's possible to cause a fire with bad wiring. The concept of "the user modifications caused the problem" is not beyond the public understanding. Anyone who cares about more details than that can take the time to understand them but those are not the people you have to worry about. > It's not the same. A bureaucrat's rules are mandatory for everybody. Tesla has competitors who may choose a different route. Your arguments either apply or they don't. Their competitors are not in a different situation and there are not very many of them. They'll likely all make the same choice. > A self-driving Tesla is not an iPhone, because an iPhone is unlikely to run somebody over or block a freeway. Risks are relevant. That's the point. You don't want a black box in control of life or death situations. You want something the user can understand when they're capable of it, so that the 2% of users who are can identify and fix problems that will affect the other 98% before it turns into deaths and liability. Opening up the car could reduce their liability, because it gives them somewhere else to point the finger sometimes, and a large population of people willing to find and fix problems. If they control the car entirely then all the liability is always on them and no one can help them when they make a mistake. > Maybe in the future, when such things are more familiar, Tesla will feel confident enough to open their products up to modders. But right now? When even the idea of self-driving cars is a challenging sell, and they're being criticized for it not being secure and consistent enough? No, not now. Right now is when the precedent and expectations are being set.
snrplfth 10y ago
AnthonyMouse, I can't seem to reply to your comment directly, so I'll have to comment to mine. >They don't have to be, for the same reason that you don't have to understand Maxwell's equations to understand that it's possible to cause a fire with bad wiring. The concept of "the user modifications caused the problem" is not beyond the public understanding. Anyone who cares about more details than that can take the time to understand them but those are not the people you have to worry about. You try convincing the public of that. Every time a Tesla has caught on fire, it's been reported on in numerous news outlets. Despite the fact that thousands of gasoline cars catch on fire every year, people are worried about much rarer, milder lithium-ion battery fires. It's a new thing, so it gets way more scrutiny. People aren't always reasonable, and regulations are often made based on perceptions of danger, not actual risks. I think Tesla is not being unreasonable in thinking that "well, it's the owner's fault for messing with it!" is going to come off as anything but mealy-mouthed blame deflection. Even if they're in the right. >Your arguments either apply or they don't. Their competitors are not in a different situation. They'll all make the same choice. So whatever choice Tesla makes, all their competitors will also make? Or, in the case of Uber's in-development car, probably already have made? I think you're according them more influence than they really have. The point stands that whatever Tesla's choice on this matter, it does not apply a legal obligation on their competitors to do the same. So it is not "no improvement". >That's the point. You don't want a black box in control of life or death situations. You want something the user can understand when they're capable of it, so that the 2% of users who are can identify and fix problems that will affect the other 98% before it turns into deaths and liability. Opening up the car could reduce their liability, because it gives them somewhere else to point the finger sometimes, and a large population of people willing to find and fix problems. If they control the car entirely then all the liability is always on them and no one can help them when they make a mistake. But that's the thing - they don't want to be pointing the finger anywhere because that's almost always bad press for them. Throwing blame just looks bad. While helpful users are certainly an asset, it's mostly as a resource for identifying problems, not attempting to solve them on their own cars. Trying to aggregate, understand, vet, and integrate users' solutions is a huge and difficult undertaking, and not something that I think they want to attempt while trying to roll out a reasonably consistent product.

AnthonyMouse 10y ago

> They are not, as of yet, that clear on how neural nets, sensor systems, and dynamically updated software come together to make a self-driving vehicle. They don't have to be, for the same reason that you don't have to understand Maxwell's equations to understand that it's possible to cause a fire with bad wiring. The concept of "the user modifications caused the problem" is not beyond the public understanding. Anyone who cares about more details than that can take the time to understand them but those are not the people you have to worry about. > It's not the same. A bureaucrat's rules are mandatory for everybody. Tesla has competitors who may choose a different route. Your arguments either apply or they don't. Their competitors are not in a different situation and there are not very many of them. They'll likely all make the same choice. > A self-driving Tesla is not an iPhone, because an iPhone is unlikely to run somebody over or block a freeway. Risks are relevant. That's the point. You don't want a black box in control of life or death situations. You want something the user can understand when they're capable of it, so that the 2% of users who are can identify and fix problems that will affect the other 98% before it turns into deaths and liability. Opening up the car could reduce their liability, because it gives them somewhere else to point the finger sometimes, and a large population of people willing to find and fix problems. If they control the car entirely then all the liability is always on them and no one can help them when they make a mistake. > Maybe in the future, when such things are more familiar, Tesla will feel confident enough to open their products up to modders. But right now? When even the idea of self-driving cars is a challenging sell, and they're being criticized for it not being secure and consistent enough? No, not now. Right now is when the precedent and expectations are being set.

snrplfth 10y ago

AnthonyMouse, I can't seem to reply to your comment directly, so I'll have to comment to mine. >They don't have to be, for the same reason that you don't have to understand Maxwell's equations to understand that it's possible to cause a fire with bad wiring. The concept of "the user modifications caused the problem" is not beyond the public understanding. Anyone who cares about more details than that can take the time to understand them but those are not the people you have to worry about. You try convincing the public of that. Every time a Tesla has caught on fire, it's been reported on in numerous news outlets. Despite the fact that thousands of gasoline cars catch on fire every year, people are worried about much rarer, milder lithium-ion battery fires. It's a new thing, so it gets way more scrutiny. People aren't always reasonable, and regulations are often made based on perceptions of danger, not actual risks. I think Tesla is not being unreasonable in thinking that "well, it's the owner's fault for messing with it!" is going to come off as anything but mealy-mouthed blame deflection. Even if they're in the right. >Your arguments either apply or they don't. Their competitors are not in a different situation. They'll all make the same choice. So whatever choice Tesla makes, all their competitors will also make? Or, in the case of Uber's in-development car, probably already have made? I think you're according them more influence than they really have. The point stands that whatever Tesla's choice on this matter, it does not apply a legal obligation on their competitors to do the same. So it is not "no improvement". >That's the point. You don't want a black box in control of life or death situations. You want something the user can understand when they're capable of it, so that the 2% of users who are can identify and fix problems that will affect the other 98% before it turns into deaths and liability. Opening up the car could reduce their liability, because it gives them somewhere else to point the finger sometimes, and a large population of people willing to find and fix problems. If they control the car entirely then all the liability is always on them and no one can help them when they make a mistake. But that's the thing - they don't want to be pointing the finger anywhere because that's almost always bad press for them. Throwing blame just looks bad. While helpful users are certainly an asset, it's mostly as a resource for identifying problems, not attempting to solve them on their own cars. Trying to aggregate, understand, vet, and integrate users' solutions is a huge and difficult undertaking, and not something that I think they want to attempt while trying to roll out a reasonably consistent product.