Similar market with laxity on the timing: https://manifold.markets/dreev/will-tesla-count-as-a-waymo-competi

What about remote safety drivers/operators?

@tobiasscheuer doesn’t count. Unlikely to qualify as SAE 4.

From Wikipedia link above:

Cooperative system

A remote driver is a driver that operates a vehicle at a distance, using a video and data connection.[63]

According to SAE J3016,

Some driving automation systems may indeed be autonomous if they perform all of their functions independently and self-sufficiently, but if they depend on communication and/or cooperation with outside entities, they should be considered cooperative rather than autonomous.

@BlueDragon I agree with you if we were talking about full time remote operators. But I'm assuming they'll have remote safety operators which jump in to help if the car gets stuck, similar to Waymo, Zoox and just about every other autonomous driving company. As far as I know that still qualifies as SAE 4 and most companies are doing or aiming for that combination

@tobiasscheuer 👍 if it’s comparable to Waymo / Zoox and generally in the league of SAE 4 I say we resolve yes, right?

@BlueDragon sae 4 isn't full self driving though.

@TiredCliche … feel free to read the title, market description with particular attention to resolution criteria and then restate your point, I’ll be happy to respond when it makes sense :).

@BlueDragon My objection is Waymo and Zoox rely on remote safety drivers. Therefore, if it's comparable to those, by your standards, it should resolve NO.

@TiredCliche this market is about robotaxis and has always specified SAE 4 as the appropriate level of self-driving.

@BlueDragon I can't tell whether SAE 4 allows control by remote safety drivers. Waymo and Zoox are not autonomous. https://www.nytimes.com/interactive/2024/09/03/technology/zoox-self-driving-cars-remote-control.html?smid=nytcore-android-share

@TiredCliche as far as I know SAE lvl 4 applies to highly automated vehicles which need remote human support in edge cases. Which applies to waymo and zoox and probably later Tesla FSD with a remote safety driver. SAE levels apply mainly to operational area restrictions and only level five is fully autonomous with no help ever

@BlueDragon that's a coherent definition for me and fine resolution, but I'll sell my NO since I was thinking about no remote help since Musk was always boasting

@tobiasscheuer fair enough! But you may be overlooking the more likely scenario: Robotaxis do not roll out in June, due to the overwhelming crush of responsibility Musk has taken on for the US federal government.

@BlueDragon yes, a further delay is definitely possible. But looking at the Tesla robot event where everything was faked, I wouldn't be surprised if they do it for the cars as well. If the cars are driving with FSD say 80% of the time vs Waymos 99%, it still kinda counts as Level 4 but FSD is still way worse. I'm just looking for markets where I can squeeze people who are too confident in FSD :)

Edit: also you may want to edit the AI generated resolution criteria

@tobiasscheuer It's not edge cases though, is the problem. If the humans are stepping in every couple miles that doesn't seem autonomous

@TiredCliche I’m going to discourage you or anyone from making this market a referendum on whether semi-autonomous robotaxis that get guidance from employees remotely should count as autonomous. This market is not about adjudicating what autonomous means, it is about adjudicating whether Musk’s Robotaxis can live up to the hype. Thus just as Zoox and Waymo are “considered SAE 4” if Robotaxis are “considered SAE 4” that will suffice for that portion of the resolution criteria to resolve.

I take your point, with companies like Waymo and Zoox and Cruise before it folded, employees were intervening a lot. The NYT article you posted posits that in urban environments, unexpected situations are the norm, and with the ratio reported for Cruise of 1.5 employees for every vehicle can’t be profitable and the claims of automation are highly dubious. But strictly in terms of SAE classes, these vehicles are still considered SAE 4 or semi-autonomous by the industry.

I will pay attention to the media commentary in June. If, when Robotaxis come out, they are offering guidance in line with this industry practice, they will likely still be considered SAE 4. However if the remote operators are doing more than offering updated routes in real time, such that they are actually driving or operating the steering and real time maneuvering remotely, I can see consensus moving towards “these Robotaxis can’t be considered SAE 4”, in which case that resolution criteria will not be met.

I respect you and your argument, please be advised that I likely won’t respond to it again. For this market to resolve fairly I need your help refocusing on the core question, which is whether Robotaxis will roll out in Austin in June as planned.

@BlueDragon Thank you for the detailed explanation.

@BlueDragon I have a sad meta-prediction, namely that the soi-disant launch is going to be exquisitely ambiguous and commenters here are going to try to bully you about what's "obviously" the case. (Please yell at me if I get anywhere near that line! I do have strong opinions about all this and have actually staked my own credibility on Tesla not pulling this off in this timeframe.)

Examples of potentially agonizing ambiguities: Does being invite-only count as open to the public? Is the geo-fenced region so restrictive as to essentially count as fixed routes? Do the remote tele-operators count as safety drivers? What if they don't happen to intervene during the trial but they need to be ready to intervene? And on and on.

I think all the ambiguity will be almost intentional. My own object-level prediction is that Tesla is making impressive progress but is not close to Waymo, and so they're aiming for the most restricted possible thing that they can still tout as technically a launch.

I do think it's important to understand where Waymo is at, since I see a lot of confusion about the way Waymo uses remote humans. I tried to clarify this in FAQ8 of my similar market. Short version: Waymos don't use remote operators the way some people are imagining. When the car is confused, which happens plenty, it stops and then asks a human what to do. The car is always autonomous when moving. (Only exception is the rare case where it needs to get off the road onto the shoulder for safety and is too confused to know how, then a remote operator can nudge it a couple car lengths at walking speed to accomplish that.)