3 ms·

Some data: American drives 3T miles each year. There are 6M accidents each year. That comes out to be 1 accident per 0.5M miles. Waymo's disenagement rate was

Some data: American drives 3T miles each year. There are 6M accidents each year. That comes out to be 1 accident per 0.5M miles. Waymo's disenagement rate was 1 per 12K miles. While disengagement and accidents are not the same, one can argue that for acceptable L5 self-driving, driver is not expected to be engaged and therefore we need these rates to be same. If Waymo continues to improve disengagement rate by 50% each year, it would still take at least 10 years before it becomes acceptable. So assuming funding keeps pouring, advancements continues, L5 self-driving can be expected around 2032. Carmack might lose just by couple of years.

jowday 5y ago
A level 4 car that’s dependent on HD maps, remote support, and constrained ODDs is not a level 5 car. A very safe level 4 car wouldn’t fulfill Carmack’s side of this bet.
tialaramex 5y ago
The "driver" in a Waymo is software, passengers can't become engaged and might not even be legal to drive a motor car. So if these rates were the same then somehow at least one of: * Waymo's driver is implausibly safe, never having any situations that cause disengagement except that other people hit it (resulting in an accident per your statistics) * Now we're expecting Waymo to press on after a road accident, because hey, I'm just a little dinged, lets get this passenger to their destination? That sounds both more dangerous (what if there is undiagnosed damage) and potentially illegal (doesn't the US have "must report" laws?) One disengagement per 12k miles seems comparable to public transport. Once a week I'd get on the train, it goes to Town uneventfully and back again, but twice in five years it didn't. Once it was stuck behind a train that hit a person (most likely a suicide) so we sat in the half-dark for almost an hour (recovering even a body, much less an injured person requires the electricity is turned off, and the train wants to reserve its battery power for emergencies) and once there was some kind of actual fault. I was in no danger either time but obviously my journey was much delayed, that seems comparable to a disengagement. In a lot of cases a human can remotely diagnose the problem and either tell the Waymo everything is fine now (e.g. the fire truck parked at an angle across the highway has moved, simply drive around it now) or correct its understanding (e.g. that does look like a school bus stopped in a residential area, but it is actually an advertisement for an upcoming Speed prequel, so, it's fine to just drive past it) and then your journey continues, interrupted but no worse than that. Maybe you get a refund from Waymo.

jowday 5y ago

A level 4 car that’s dependent on HD maps, remote support, and constrained ODDs is not a level 5 car. A very safe level 4 car wouldn’t fulfill Carmack’s side of this bet.

tialaramex 5y ago

The "driver" in a Waymo is software, passengers can't become engaged and might not even be legal to drive a motor car. So if these rates were the same then somehow at least one of: * Waymo's driver is implausibly safe, never having any situations that cause disengagement except that other people hit it (resulting in an accident per your statistics) * Now we're expecting Waymo to press on after a road accident, because hey, I'm just a little dinged, lets get this passenger to their destination? That sounds both more dangerous (what if there is undiagnosed damage) and potentially illegal (doesn't the US have "must report" laws?) One disengagement per 12k miles seems comparable to public transport. Once a week I'd get on the train, it goes to Town uneventfully and back again, but twice in five years it didn't. Once it was stuck behind a train that hit a person (most likely a suicide) so we sat in the half-dark for almost an hour (recovering even a body, much less an injured person requires the electricity is turned off, and the train wants to reserve its battery power for emergencies) and once there was some kind of actual fault. I was in no danger either time but obviously my journey was much delayed, that seems comparable to a disengagement. In a lot of cases a human can remotely diagnose the problem and either tell the Waymo everything is fine now (e.g. the fire truck parked at an angle across the highway has moved, simply drive around it now) or correct its understanding (e.g. that does look like a school bus stopped in a residential area, but it is actually an advertisement for an upcoming Speed prequel, so, it's fine to just drive past it) and then your journey continues, interrupted but no worse than that. Maybe you get a refund from Waymo.