4 ms·

No, it's a car manufacturer telling you that they do not want you handing over control of a vehicle to ridesharing programs that they cannot be sure are compati

No, it's a car manufacturer telling you that they do not want you handing over control of a vehicle to ridesharing programs that they cannot be sure are compatible with their self-driving systems. Ridesharing programs will literally be giving control inputs to the car, and Tesla has determined that those inputs should come either from licensed drivers or from their own ridesharing program. Because they're not dumb and they know that if a malfunctioning Uber or Lyft app causes a Tesla to get into an accident, they'll probably be held liable anyways.

snowwrestler 10y ago
Liability is a crucial area of the law that needs to evolve in response to software. Liability with respect to the operation of physical products is well understood. There are private standards that insurance companies and courts are familiar with (for example: Underwriter Labs, NIST, NSF, ANSI). If your product passes the relevant standard, and a customer causes harm with the product, then you've got a great argument that your company should not be held liable for that harm. Software licenses almost always just try to disclaim any and all liability entirely. "THIS SOFTWARE IS PROVIDED AS-IS AND IS NOT FIT FOR ANY PURPOSE"--anyone who has read a software license has probably seen language like that. Then on top of that, software is increasingly networked with centralized services. Tesla's self-driving technology, as described so far, will be dependent on data made available over the network--like whether the radar return at a given location is a known billboard. So liability will attach to not just the car, not just the software, but also the real-time service run by Tesla. Companies use liability as a major reason to restrict the rights of consumers. We need industry and legal frameworks to allocate liability for software and services like we have for physical goods.
- snrplfth 10y ago
  Sure, ideally, I would like to see Tesla open up its self-driving systems to new inputs and modifications, and just disclaim all liability for the consequences. I think it would promote more innovation and novel uses for the software. However, Tesla is aware that they (and others developing automobile autonomy systems) are under scrutiny, and that, in the eyes of the public, the media, and regulators, they simply cannot divest themselves of liability. This is especially true because of the hazardous nature of cars and the novelty of the technology. They've already been publically embarrassed by having one of their cars hacked into. I do not think they have any appetite for opening up a novel flagship product to unapproved mucking-about, and I cannot honestly blame them. They are covering their asses.
qb45 10y ago
Do you actually know something or is it just speculation? It looks like they are attempting to prohibit driving people for money, not connecting the car to unauthorized software. "Revenue purposes" is the exact words they used.
- snrplfth 10y ago
  It is what they appear to be referring to, given the fact that they are saying "using a self-driving Tesla...for revenue purposes will only be permissible on the Tesla Network". So you can drive people for money, or you can let it self-drive on the Tesla Network, but it can't connect to other, non-Tesla-Network software. Now, it could, maybe, also mean that a self-driving Tesla would be prohibited from being leased out for a night without using software to mediate the transaction. But I think this pretty clearly refers to hooking up a self-driving Tesla to an Uber or Lyft style app. Edit: and they've certainly seen the trouble that Uber and Lyft have gotten into insurance-wise with regards to carrying passengers as a "public accomodation". It's quite possible that they want to handle this centrally for insurance purposes. Which are bound to be a challenge with regards to self-driving cars.
EdHominem 10y ago
That's not true though. The app isn't going to crank the steering wheel around, it's going to input a destination. Their lies don't become any more compelling when you repeat them.
- snrplfth 10y ago
  Yes, by control inputs I meant destinations and probably routes as well, depending on how the self-driving program works. Maybe instead of asserting that I'm "repeating lies" you could ask whether I mean what you think I mean.
- jboles 10y ago
  The destination is not a control input?
  - EdHominem 10y ago
    Not meaningfully, no. We're talking about the user damaging the car. The car already has to check for bad user input. A hypothetical Uber app that controlled the self-driving car would just be passing the user's destination through - to be verified as if they entered it directly. It wouldn't be controlling the steering...

snowwrestler 10y ago

Liability is a crucial area of the law that needs to evolve in response to software. Liability with respect to the operation of physical products is well understood. There are private standards that insurance companies and courts are familiar with (for example: Underwriter Labs, NIST, NSF, ANSI). If your product passes the relevant standard, and a customer causes harm with the product, then you've got a great argument that your company should not be held liable for that harm. Software licenses almost always just try to disclaim any and all liability entirely. "THIS SOFTWARE IS PROVIDED AS-IS AND IS NOT FIT FOR ANY PURPOSE"--anyone who has read a software license has probably seen language like that. Then on top of that, software is increasingly networked with centralized services. Tesla's self-driving technology, as described so far, will be dependent on data made available over the network--like whether the radar return at a given location is a known billboard. So liability will attach to not just the car, not just the software, but also the real-time service run by Tesla. Companies use liability as a major reason to restrict the rights of consumers. We need industry and legal frameworks to allocate liability for software and services like we have for physical goods.

snrplfth 10y ago
Sure, ideally, I would like to see Tesla open up its self-driving systems to new inputs and modifications, and just disclaim all liability for the consequences. I think it would promote more innovation and novel uses for the software. However, Tesla is aware that they (and others developing automobile autonomy systems) are under scrutiny, and that, in the eyes of the public, the media, and regulators, they simply cannot divest themselves of liability. This is especially true because of the hazardous nature of cars and the novelty of the technology. They've already been publically embarrassed by having one of their cars hacked into. I do not think they have any appetite for opening up a novel flagship product to unapproved mucking-about, and I cannot honestly blame them. They are covering their asses.

qb45 10y ago

Do you actually know something or is it just speculation? It looks like they are attempting to prohibit driving people for money, not connecting the car to unauthorized software. "Revenue purposes" is the exact words they used.

snrplfth 10y ago
It is what they appear to be referring to, given the fact that they are saying "using a self-driving Tesla...for revenue purposes will only be permissible on the Tesla Network". So you can drive people for money, or you can let it self-drive on the Tesla Network, but it can't connect to other, non-Tesla-Network software. Now, it could, maybe, also mean that a self-driving Tesla would be prohibited from being leased out for a night without using software to mediate the transaction. But I think this pretty clearly refers to hooking up a self-driving Tesla to an Uber or Lyft style app. Edit: and they've certainly seen the trouble that Uber and Lyft have gotten into insurance-wise with regards to carrying passengers as a "public accomodation". It's quite possible that they want to handle this centrally for insurance purposes. Which are bound to be a challenge with regards to self-driving cars.

EdHominem 10y ago

That's not true though. The app isn't going to crank the steering wheel around, it's going to input a destination. Their lies don't become any more compelling when you repeat them.

snrplfth 10y ago
Yes, by control inputs I meant destinations and probably routes as well, depending on how the self-driving program works. Maybe instead of asserting that I'm "repeating lies" you could ask whether I mean what you think I mean.
jboles 10y ago
The destination is not a control input?
- EdHominem 10y ago
  Not meaningfully, no. We're talking about the user damaging the car. The car already has to check for bad user input. A hypothetical Uber app that controlled the self-driving car would just be passing the user's destination through - to be verified as if they entered it directly. It wouldn't be controlling the steering...