> Imagine the conversation New York legislators could have had if every robotaxi ride generated a cryptographically signed, tamper-evident record of the vehicle's behavior — independently verifiable by any regulator, auditor, or safety agency.
This is putting the cart before the horse. The problem with autonomous vehicles is that there's no standardized third party test suite to validate and certify ahead of time and periodically that the vehicles are capable of behaving correctly in all road conditions even with their sensors degraded as will happen over time.
This kind of validation should be done by the state or a neutral third party like UL in the US before a single vehicle is allowed on public roads.
Without it they're just unleashing multi-ton weapons on the streets with no guarantee they'll perform adequately. The problems with this were evident in SF when the blackout happened and the Waymos clogged up the roadways. If that happens during a real emergency and blocks the roadways it could result in thousands of deaths.
Autonomous vehicles are a bad idea anyway for the reasons in this video[1].
The cart is before the horse in some instances, but each state and each city will approach risk a little differently. Cities like SF and Austin are willing to take bigger risk without data, other states and cities will want the data from the risk taking cities to see.
I agree though, these products should be certified ahead of time by industry standards. That is what PhyComp will help accomplish.
The article somewhat strangely proposes a technology solution to a problem that they claim isn't a technology problem. Disclosure and reporting regulations can be enacted that don't need a technology element other than what vehicles already have installed.
I had aasumed the plan was killed for cynical political motivations, the same as apparantly caused the hangup and Hochul's near killing of NYC congestion pricing. Its really hard to gather evidence about what is the motivation for things though...
They likely are, but say I have a fleet of Waymos and Teslas. I’ll need two separate systems for telemetry, safety, and compliance.
There’s also the problem of independent evidence when incidents happen. Insurers, OEM, operators, and possibly the public should all have the same view of data and can trust its presented honestly and in its entirety. Phyware systems prevent tamper and Phyware itself has no stakes in the game, all the data is provided as captured.
The specific dishonesty is the sentence: "None of them answer the question legislators actually care about: what is this vehicle doing, moment to moment, on our roads?" Legislators did not care about that question. They cared about the Transport Workers Union and the New York Taxi Workers Alliance telling them they'd face primary challenges. No amount of cryptographically signed telemetry data changes that calculus.
Save yourself time reading it, it's just a shameless plug for some SaaS service.
The only reason why Waymo was denied access to New York is they haven't given a bribe big enough that the city government requested. That's it. As soon as Waymo gives they bribe they requested they'll be in New York.
Plausible take, but what about when a humanoid violates its rule and causes massive damage in a warehouse? Who pays? Who provides the evidence of what happened?
> Imagine the conversation New York legislators could have had if every robotaxi ride generated a cryptographically signed, tamper-evident record of the vehicle's behavior — independently verifiable by any regulator, auditor, or safety agency.
This is putting the cart before the horse. The problem with autonomous vehicles is that there's no standardized third party test suite to validate and certify ahead of time and periodically that the vehicles are capable of behaving correctly in all road conditions even with their sensors degraded as will happen over time.
This kind of validation should be done by the state or a neutral third party like UL in the US before a single vehicle is allowed on public roads.
Without it they're just unleashing multi-ton weapons on the streets with no guarantee they'll perform adequately. The problems with this were evident in SF when the blackout happened and the Waymos clogged up the roadways. If that happens during a real emergency and blocks the roadways it could result in thousands of deaths.
Autonomous vehicles are a bad idea anyway for the reasons in this video[1].
1. https://www.youtube.com/watch?v=040ejWnFkj0
The cart is before the horse in some instances, but each state and each city will approach risk a little differently. Cities like SF and Austin are willing to take bigger risk without data, other states and cities will want the data from the risk taking cities to see.
I agree though, these products should be certified ahead of time by industry standards. That is what PhyComp will help accomplish.
Thanks, that's a good video.
The article somewhat strangely proposes a technology solution to a problem that they claim isn't a technology problem. Disclosure and reporting regulations can be enacted that don't need a technology element other than what vehicles already have installed.
Yup. If someone vomits in a vehicle, it's not the lack of a blockchain detailing the vomit event which is the problem.
That's the actual trust issue.
I had aasumed the plan was killed for cynical political motivations, the same as apparantly caused the hangup and Hochul's near killing of NYC congestion pricing. Its really hard to gather evidence about what is the motivation for things though...
> Its really hard to gather evidence about what is the motivation for things though...
It can be, but it can also be hard to know where you should look.
https://tylervigen.com/the-mystery-of-the-bloomfield-bridge
[dead]
> PhyWare is a data platform for autonomous systems, and we're building it specifically to close this gap.
Why isn't Google/Waymo building it. Sounds like they are much better equipped for it.
They likely are, but say I have a fleet of Waymos and Teslas. I’ll need two separate systems for telemetry, safety, and compliance.
There’s also the problem of independent evidence when incidents happen. Insurers, OEM, operators, and possibly the public should all have the same view of data and can trust its presented honestly and in its entirety. Phyware systems prevent tamper and Phyware itself has no stakes in the game, all the data is provided as captured.
The specific dishonesty is the sentence: "None of them answer the question legislators actually care about: what is this vehicle doing, moment to moment, on our roads?" Legislators did not care about that question. They cared about the Transport Workers Union and the New York Taxi Workers Alliance telling them they'd face primary challenges. No amount of cryptographically signed telemetry data changes that calculus.
New York City was not included in the proposal likely for the reason you mentioned, only cities with less than a million people
Unnecessary solution to imaginary problem. I'd rather have this required for human drivers.
Long story short: Liability. We need a clear cut way of placing blame in the event something goes wrong and it can’t be the NYS.
Save yourself time reading it, it's just a shameless plug for some SaaS service.
The only reason why Waymo was denied access to New York is they haven't given a bribe big enough that the city government requested. That's it. As soon as Waymo gives they bribe they requested they'll be in New York.
Plausible take, but what about when a humanoid violates its rule and causes massive damage in a warehouse? Who pays? Who provides the evidence of what happened?
Who does those things today?
Agreed, this isn’t necessarily the problem of today.