Elon Musk has been very explicit in promising a robotaxi launch in Austin in June with unsupervised full self-driving (FSD). We'll give him some leeway on the timing and say this counts as a YES if it happens by the end of August.
So far Tesla seems to be testing this with employees and with supervised FSD and doubling down on the public Austin launch.
FAQ
1. Does it have to be a public launch?
Yes, but we won't quibble about waitlists. As long as even 10 members of the public have used the service by the end of August, that's a YES. Also if there's a waitlist, anyone has to be able to get on it and there has to be intent to scale up. In other words, Tesla robotaxis have to be actually becoming a thing, with summer 2025 as when it started.
If it's invite-only and Tesla is hand-picking people, that's not a public launch. If it's viral-style invites with exponential growth from the start, that's likely to be within the spirit of a public launch.
2. What if there's a human backup driver in the driver's seat?
This importantly does not count. That's supervised FSD.
3. But what if the backup driver never actually intervenes?
Compare to Waymo, which goes millions of miles between [injury-causing] incidents. If there's a backup driver we're going to presume that it's because interventions are still needed, even if rarely. But see FAQ 7 for a gray area here.
4. What if it's only available for certain fixed routes?
That would resolve NO. It has to be available on unrestricted public roads [restrictions like no highways is ok] and you have to be able to choose an arbitrary destination. I.e., it has to count as a taxi service.
5. What if it's only available in a certain neighborhood?
This we'll allow. It just has to be a big enough neighborhood that it makes sense to use a taxi. Basically anything that isn't a drastic restriction of the environment.
6. What if they drop the robotaxi part but roll out unsupervised FSD to Tesla owners?
This is unlikely but if this were level 4+ autonomy where you could send your car by itself to pick up a friend, we'd call that a YES per the spirit of the question.
7. What about level 3 autonomy?
Level 3 means you don't have to actively supervise the driving (like you can read a book in the driver's seat) as long as you're available to immediately take over when the car beeps at you. We'll discuss in the comments how to handle this case but I'm leaning NO because another take on the spirit of the question is whether Tesla will catch up to Waymo, technologically if not in scale at first.
8. What about tele-operation?
The short answer is that that's not level 4 autonomy so that would resolve NO for this market. This is a common misconception about Waymo's phone-a-human feature. It's not remotely (ha) like a human with a VR headset steering and braking. If that ever happened it would count as a disengagement and have to be reported. See Waymo's blog post with examples and screencaps of the cars needing remote assistance.
To get technical about the boundary between a remote human giving guidance to the car vs remotely operating it, grep "remote assistance" in Waymo's advice letter filed with the California Public Utilities Commission last month. Excerpt:
The Waymo AV [autonomous vehicle] sometimes reaches out to Waymo Remote Assistance for additional information to contextualize its environment. The Waymo Remote Assistance team supports the Waymo AV with information and suggestions [...] Assistance is designed to be provided quickly - in a mater of seconds - to help get the Waymo AV on its way with minimal delay. For a majority of requests that the Waymo AV makes during everyday driving, the Waymo AV is able to proceed driving autonomously on its own. In very limited circumstances such as to facilitate movement of the AV out of a freeway lane onto an adjacent shoulder, if possible, our Event Response agents are able to remotely move the Waymo AV under strict parameters, including at a very low speed over a very short distance.
Tentatively, Tesla needs to meet the bar for autonomy that Waymo has set. But if there are edge cases where Tesla is close enough in spirit, we can debate that in the comments.
Ask more clarifying questions! I'll be super transparent about my thinking and will make sure the resolution is fair if I have a conflict of interest due to my position in this market.
[Ignore any auto-generated clarifications below this line. I'll add to the FAQ as needed.]
Update 2025-05-23 (PST) (AI summary of creator comment): Regarding the assessment of autonomy for a YES resolution:
The creator has stated that beyond no actual disengagements, there must also be no counterfactual disengagements.
This means the system must demonstrate true unsupervised capability, where it genuinely does not require active supervision. Simply avoiding interventions due to limited scope, highly controlled conditions, or fortunate circumstances during a trial period may not be sufficient if the system would have foreseeably failed in more common, challenging scenarios. [Ok, I like the way the AI put this so I haven't deleted this auto-clarification. Do holler if you have issues with it! --dreev]
1,000People are also trading
Bloomberg: Tesla Targets June 12 Launch of Robotaxi Service in Austin
"In preparation for the launch, the company this week operated a test vehicle on public roads in Austin with no one in the driver’s seat for the first time, according to the person. A Tesla engineer was riding in a passenger seat of a Model Y SUV, which drove autonomously with no remote operation, the person said."
https://archive.is/a1bGh
Oh man, this is shaping up to be so contentious. I think I'm actually most nervous, in terms of resolving this market fairly, about something that feels spiritually like a YES but that fails on a technicality.
Like FAQ1 says it has to be open to the public, at least to get on a waitlist. If it's invite-only at first but you can invite anyone, does that count? I'm genuinely not sure yet. Do chime in, @traders, if you have an opinion. Maybe if it's viral-style invites where anyone who gets in can invite anyone else, that's spiritually a public waitlist. If Elon is handpicking everyone, not so much. If it's something in between, we'll debate it. And we may have to wait a while to resolve. See again FAQ1 about steadily scaling up from this launch, which we'll have to judge in hindsight.
Or what if it's geo-fenced almost-but-not-quite to the point of being fixed routes? And the tele-operators -- so much potential gray area! If they never intervene during the trial period, how will we know if they were actively monitoring and prepared to disengage the AI at driving speed?
(I believe that's a critical part of the definition of autonomy, that the car doesn't need active supervision. Otherwise you could take a car with zero ability to perceive pedestrians and drive it around -- with supervision -- and pretend that as long as no pedestrians ventured into its path, that it was fully autonomous. There were no disengagements, after all. My claim is there need to be no actual or counterfactual disengagements.)
I think all the ambiguity will be almost intentional. My take 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. Which may be perfectly fine and prudent, if they're honest about it, but it could make resolving this market hell.
PS, I'm removing this line from the market description:
So I think it would be a huge scandal if Tesla used tele-operation and it isn't likely.
I learned about a German company, Vay, operating in a couple cities in Germany, plus Las Vegas in the US, that does actual VR-head-style tele-operation, with some big caveats:
They don't tele-operate with passengers in the car, they just use it to get rental cars to customers who then drive the cars normally
They limit the speed during tele-operation to 26mph (42 km/h)
My understanding is that the network delay is comparable to drunk driving and they've calculated that up to that speed that's not too much of a risk, for Las Vegas at least. I'm not sure what this might mean for Tesla's robotaxi tele-operators but I'm inclined to stick to the bar that Waymo has set (see FAQ8). Especially since Elon Musk mocked Waymo for years about things like this.
@dreev regarding geofencing, this is the current area that Waymo operates in Austin https://x.com/WholeMarsBlog/status/1925990174678196684
@MarkosGiannopoulos Thanks! Here's a better link: https://waymo.com/waymo-on-uber/#modal-uber-austin
To be clear, Tesla doesn't need to match that for a YES resolution here. I'm thinking they do need to match Waymo in terms of degree of autonomy, but if anyone thinks that's overly strict, do speak up.
@dreev if you are unsure but you're leaning towards a specific resolution based on the evidence, you could resolve to a %
@SimoneRomeo Good call. That may well turn out to be what's fairest. How to deal with my conflict of interest may be tricky though.
But first we can see if Tesla's launch really does land in a gray area. The FAQ here has the gray area squeezed down pretty far. If you can think of more hypotheticals / clarifying questions, that would be great.
@dreev it is very clear to me that they are going to be utilizing teleoperators to supervise every moment. The state of their software is not good enough to do otherwise. That said, because I think they will try to obscure this fact, I expect what will happen with the wait list is anyone will be able to join it but only people who that Tesla trusts to be enthusiastic will have a chance of getting off this waitlist. Again, I do not expect they will advertise that some of the people on the wait list are never going to get off of it, but that will certainly be the case. With only 10 to l20 cars running, they will easily be able to fill the queue with just Tesla Fanboys, and keep off anyone who might be a critic from ever getting off of the wait list.
Regarding point 2, in an interview today, Musk talked again about having no one in the driver's seat.
https://x.com/SawyerMerritt/status/1924899076614537727
I'm not sure how much to trust the following but it's crowdsourced data on Tesla's FSD. It seems to be saying that even the latest FSD still needs critical disengagements every 451 miles on average.
I think Musk said that they've gotten that to a critical disengagement every 10k miles for the unreleased version of FSD. But even that wouldn't be enough for unsupervised FSD.
PS: This is not me clarifying the market. If there are clarifications needed on what the threshold for safely deploying unsupervised FSD is, please ask.
@WrongoPhD Oh man, so much gray area possible here. But I think we've committed to a somewhat bright line in FAQ1. Anyone has to be able to at least get on the wait-list. If you have to find someone to invite you... I mean ultimately we may have to wait to resolve this to see if it was legitimately a slow ramp-up vs a way to be pretending to be launching.
"It has to be available on unrestricted public roads and you have to be able to choose an arbitrary destination. I.e., it has to count as a taxi service." <-- What if it's available in some areas but blacklisted in certain areas, e.g. no highway driving or not on dirt roads etc?
@KJW_01294 Good question, and let me know if you have strong opinions before we make the answer official. My thinking so far is that we want to pick a threshold that allows for "reasonable" restrictions. No dirt roads, sure, that's not much of a restriction and we can presume it would be for mundane reasons, not really about limitations on the car's ability to drive autonomously. Highways is a tougher call. Highways are actually easier for the AI but the consequences of a mistake are greater.
One idea for where to draw this line: If you can still get to any Uber/Lyftable destination within the region (see FAQ5) and the transit time is within 20% of a Waymo, it counts.
If no one has objections or suggested refinements to that, I'll add it to the FAQ.
@MarkosGiannopoulos You're right, and looking at my market description again, it's clear that highway driving shouldn't be necessary for a YES resolution here.
In a related market about private cars with self-driving, I suggested that the spirit of the question allows for restrictions based on weather or daylight or highways but that, as long as I'm flexible about when I drive and am not picky about the route, I can read a book 90% of the time I'm behind the wheel.
Presumably we need a higher bar than that for something we can call an unsupervised robotaxi though.
I said we'd discuss in the comments any gray areas with level 3 autonomy. Since "level 4" is in the title, I'm now thinking we should draw a hard line here. Are people ok with that? I'm worried about something that's close enough to level 4 that it feels spiritually like a YES, especially if true level 4 follows soon after. But I suppose we've already given a fair bit of leeway on the timing. So probably best to keep things simple with a bright line at level 4. (Review of the autonomy levels.)
This isn't official until I update the FAQ though. Speak now or forever hold your peace!
@dreev That sounds fair and unlikely to happen anyway. I do beg you not to resolve right away if robotaxis launch until we get real information on how teleoperators are being used. I am very much expecting them to launch on time but using teleoperators as safety drivers, which I'm glad you agree isn't level 4.
@WrongoPhD Sounds good. Now I'm wondering, given network latency, is a remote human operator worse than drunk driving?
I'm any case, we agree Waymo is at level 4, right? And Tesla will get some benefit of the doubt if they seem to have hit parity with Waymo?
@dreev I definitely agree that Waymo is level 4. For me, the key difference is Waymo knows when they need to ask for help versus a teleoperator watching all the time always ready to instantly take over, which is what I anticipate for Tesla.
I do understand at some point you'll have to resolve this market giving Tesla the benefit of the doubt, but I'm betting that if we wait a few months after launch, the fact that they don't have a serious L4 contender win be revealed somehow, either through an accident or a flimsy excuse to halt the robotaxi launch.
I think the robotaxi launch in Texas is going to be the equivalent of Theranos successfully being able to run an ANA blood test despite having otherwise garbage technology.
@WrongoPhD Bold prediction! And we should totally have additional markets. I'm not opposed to leaving this market open longer (not changing the the Aug 31 deadline for the launch itself, of course) until we're sure Tesla hasn't cheated. Another market on that would be great, actually. We could resolve this market as soon as the "did Tesla cheat?" market has sufficiently low probability.
I don't think we'll know immediately if teleoperators are being used. Unlike you, I expect they WILL be used but the extent they're being used will be hidden. I fully expect Tesla will try to act like their teleoperators are just like Waymos when in fact they'll have a person watching a car constantly taking on the role of a safety driver just behind closed doors. My personal prediction is Tesla won't reveal much about teleoperators until they skapegoat an accident on one.
@WrongoPhD I don't think I'm at quite that level of cynicism but we should absolutely have a separate market about that if you're game to create it?
@WrongoPhD I agree this is the most likely scenario. The financial incentive for Musk to launch robotaxi services without in-car safety drivers before the technology is at a safety level comparable to Waymo is immense.
@dreev Have you seen this article? At the very least, there is discussion of training the current in-car safety drivers to work remotely.
It's hard to come up with good resolution criteria, though, for information being concealed. Even waymo, which has been much more transparent, has not released their intervention rate as far as I am aware.
@WrongoPhD Thanks! Archived copy of that article:
It says that the backup drivers are making "critical interventions" every so many days, which Tesla is trying to spin as being... almost good enough for unsupervised FSD? Maybe we also want a market for if they pull an Uber with this.