@aashiq Is it your intention that waymo would resolve no currently, because I think it currently might?

This, from perplexity:

Can you say more? I don’t see why the screenshot shows that Waymo would resolve No. do you have stats on Waymo rides?

• Rides that require a human to be present or to take control will not count as driverless.
  
  This means waymo would currently fail this, right?

@NathanpmYoung what I've heard is that Waymo allows humans to tell the cars what to do at a high level (E.g. Reroute, or go through a construction zone) but they can't actually take control.

@SeekingEternity Agreed. And it would be quite a monkey wrench in these self-driving car markets if Waymo were declared to not count because it's SAE level 4 and not level 5. Other traders can chime in if I'm wrong but I think the universal assumption for this and related markets is "driverless ride-hailing à la Waymo".

Okay but that is not what it says:

"Driverless rides must be completely autonomous with no human available as a backup."

IIUC waymo rides have people in an HQ who can take control.

>"and the service is still operational"

I got impression they will do a series of 'test runs' (Tesla may well use a different term to test run) each probably wider than the last test run. e.g. starting: just in Austin, with careful monitoring, just tesla owned vehicles and just tesla employees taking rides and then later test runs gradually reducing these restrictions. This may make interpreting the 'still operational' requirement difficult:

If the total rides are more than 10,000 but there are gaps between some of the tests then do only rides after the last gap to end of year count towards the 10,000? In this case, if it is known at the year end that this test run will come to an end in early 2026 and there will then be a small gap before the next test run, does this affect the 'still operational' to end of year requirement?

Or is it something other than timing gaps? e.g. if each test run is strictly more comprehensive than the last, does that count as still operational and a small gap does not matter? i.e 'still operational' is largely only a restriction if Tesla "pulls the plug" completely / a months long gap. So can add all (/most) test runs with small gaps to see if 10,000 rides reached (subject to other restrictions like excluding tesla employee test rides).

Or maybe test runs that can be identified as different test runs are of a different service regardless of whether overlapping periods or whether strictly more comprehensive. So only if one of these test runs reaches 10000, and it is still operating at year end, will it count.

@aashiq how are you planning on interpreting your 'still operational' requirement as Tesla expands the tests by reducing restrictions? I don't know whether there will be gaps between test runs or if it will be more like seamless expansion to more situations. However I think knowing how you will interpret may change the expected probabilities.

Okay, what if I park my Tesla at the end of my street, then do this:

1) Let one of my neighbors "hail" me.
2) I open the doors for him, he gets in the back seat.
3) I use the "Actually Smart Summon" feature to bring the car to me, at the other end of the street.
4) When the car stops next to me, my neighbor thanks me, and pays me $1 for the ride.

There was hailing involved. There was no human driving involved. I only pressed a button to unlock doors, then held down a button to activate Actually Smart Summon.

Would that count? Arguably, that was "driverless ride hailing." Heck, it was even, "paid driverless ride hailing."

@StCredZero perhaps it would be useful to clarify that the rides have to be “bona fide”. As in the rider wanted to legitimately get from point A to point B, not just test the service at a trivial distance. And not because the rider is a Tesla employed tester.

@aashiq A little more needs to be specified here. If the neighbor is elderly, and the walk from their door to the mailboxes is "significant" to them, then this is not just to test the service, and it's arguably not a "trivial" distance. It's a point A to point B journey of genuine utility.

I think you're better off just setting a short arbitrary distance. I think 400m is the max range of Actually Smart Summon, so that would close off that loophole.

@StCredZero sure I’m happy to set a range of 400m. But I don’t know if we will actually get stats that have an accurate range on them. In practice we will get some number like “rides” from Tesla, which will probably exclude shenanigans like what you are saying by construction

Also note that I am requiring at least 10k rides so one-off tricks like this shouldn’t matter much, only a systematic usage of smart sunmon

@aashiq I'm an avid user of Actually Smart Summon. I'm pretty sure 400m is the max distance.

@StCredZero yup, agreed. Of course, there’s some chance they realize an even smarter summon with a bigger distance sometime this year right?

@StCredZero ah i see the confusion. my point about the accuracy of the range was not about smart summon. i am saying that Tesla may not tell us how long the actual rides were. So we won't be able to tell how many are < 400m

@aashiq Doesn't smart summons require your finger be on a dead man's switch? That still seems like you're the driver even if you're controlling the car from the outside like an RC car. A good way to prove that you're the driver is that you'd be the one responsible if the car ran into someone, not Tesla.

@WrongoPhD I think it really depends how the running over happened. I think there exist smart summon failure modes where Tesla would lose a lawsuit.

@WrongoPhD

> Doesn't smart summons require your finger be on a dead man's switch? That still seems like you're the driver

Is holding a dead-man's switch really "driving?" It's more "supervising." A student driver with an instructor in the passenger seat would still be the one "driving."

@StCredZero @aashiq According to Tesla, you are required to have control over the car at all times. This doesn't sound like Tesla thinks they have any liability or that they are the ones in control of the car. It's also noteworthy they are explicitly still calling you the driver

@WrongoPhD I agree that’s what they think. I still think there are cases where they would lose a lawsuit, such as if you have a model S that suddenly kicks into ludicrous mode with no user input or something

@WrongoPhD but yeah in general I agree that’s smart summon does not count for my market

Can you define "driverless"? Does it count as driverless if a human is required to be available to drive?

@HankyUSA no it doesn’t count