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.
As of April 2025, Tesla seems to be testing this with employees and with supervised FSD and doubling down on the public Austin launch.
PS: A big monkey wrench no one anticipated when we created this market is how to treat the passenger-seat safety monitors. See FAQ9 for how we're trying to handle that in a principled way. Tesla is very polarizing and I know it's "obvious" to one side that safety monitors = "supervised" and that it's equally obvious to the other side that the driver's seat being empty is what matters. I can't emphasize enough how not obvious any of this is. At least so far, speaking now in August 2025.
FAQ
1. Does it have to be a public launch?
Yes, but we won't quibble about waitlists. As long as even 10 non-handpicked 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.
A potential litmus test is whether serious journalists and Tesla haters end up able to try the service.
UPDATE: We're deeming this to be satisfied.
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.
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. This would be tantalizingly close and a very big deal but is ultimately a NO. My reason to be picky about this is that a big part of 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.
9. What about human safety monitors in the passenger seat?
Oh geez, it's like Elon Musk is trolling us to maximize the ambiguity of these market resolutions. Tentatively (we'll keep discussing in the comments) my verdict on this question depends on whether the human safety monitor has to be eyes-on-the-road the whole time with their finger on a kill switch or emergency brake. If so, I believe that's still level 2 autonomy. Or sub-4 in any case.
See also FAQ3 for why this matters even if a kill switch is never actually used. We need there not only to be no actual disengagements but no counterfactual disengagements. Like imagine that these robotaxis would totally mow down a kid who ran into the road. That would mean a safety monitor with an emergency brake is necessary, even if no kids happen to jump in front of any robotaxis before this market closes. Waymo, per the definition of level 4 autonomy, does not have that kind of supervised self-driving.
10. Will we ultimately trust Tesla if it reports it's genuinely level 4?
I want to avoid this since I don't think Tesla has exactly earned our trust on this. I believe the truth will come out if we wait long enough, so that's what I'll be inclined to do. If the truth seems impossible for us to ascertain, we can consider resolve-to-PROB.
11. Will we trust government certification that it's level 4?
Yes, I think this is the right standard. Elon Musk said on 2025-07-09 that Tesla was waiting on regulatory approval for robotaxis in California and expected to launch in the Bay Area "in a month or two". I'm not sure what such approval implies about autonomy level but I expect it to be evidence in favor. (And if it starts to look like Musk was bullshitting, that would be evidence against.)
12. What if it's still ambiguous on August 31?
Then we'll extend the market close. The deadline for Tesla to meet the criteria for a launch is August 31 regardless. We just may need more time to determine, in retrospect, whether it counted by then. I suspect that with enough hindsight the ambiguity will resolve. Note in particular FAQ1 which says that Tesla robotaxis have to be becoming a thing (what "a thing" is is TBD but something about ubiquity and availability) with summer 2025 as when it started. Basically, we may need to look back on summer 2025 and decide whether that was a controlled demo, done before they actually had level 4 autonomy, or whether they had it and just were scaling up slowing and cautiously at first.
13. If safety monitors are still present, say, a year later, is there any way for this to resolve YES?
No, that's well past the point of presuming that Tesla had not achieved level 4 autonomy in summer 2025.
14. What if they ditch the safety monitors after August 31st but tele-operation is still a question mark?
We'll also need transparency about tele-operation and disengagements. If that doesn't happen by June 22, 2026 (a year after the robotaxi launch) then that too is a presumed NO.
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 2025-11-01 (PST) (AI summary of creator comment): The creator is [tentatively] proposing a new necessary condition for YES resolution: the graph of driver-out miles (miles without a safety driver in the driver's seat) should go roughly exponential in the year following the initial launch. If the graph is flat or going down (as it may have done in October 2025), that would be a sufficient condition for NO resolution.
Update 2025-12-10 (PST) (AI summary of creator comment): The creator has indicated that Elon Musk's November 6th, 2025 statement ("Now that we believe we have full self-driving / autonomy solved, or within a few months of having unsupervised autonomy solved... We're on the cusp of that") appears to be an admission that the cars weren't level 4 in August 2025. The creator is open to counterarguments but views this as evidence against YES resolution.
Update 2025-12-10 (PST) (AI summary of creator comment): The creator clarified that presence of safety monitors alone is not dispositive for determining if the service meets level 4 autonomy. What matters is whether the safety monitor is necessary for safety (e.g., having their finger on a kill switch).
Additionally, if Tesla doesn't remove safety monitors until deploying a markedly bigger AI model, that would be evidence the previous AI model was not level 4 autonomous.
Update 2026-01-31 (PST) (AI summary of creator comment): The creator clarified that passenger-seat emergency stop buttons should be evaluated based on their function:
If the button is a real-time "hit the brakes we're gonna crash!" intervention button, this would indicate supervision that could rule out level 4 autonomy
If the button is a "stop requested as soon as safely possible" button (where the car remains in control until safely stopped), this would not rule out level 4 autonomy
This distinction applies to both Waymo (the benchmark) and Tesla. The creator emphasized that mere presence of a safety monitor doesn't rule out level 4 - what matters is whether there is supervision with the ability to intervene in real time.
Update 2026-02-01 (PST) (AI summary of creator comment): The creator has proposed a concrete scenario for June 22, 2026 (the one-year deadline from FAQ14) that would result in NO resolution:
(a) Longer zero-intervention streaks but not to the point that unsupervised FSD is safer than humans
(b) More unsupervised robotaxi rides but not at a scale where tele-operation becomes implausible
(c) Continued lack of transparency on disengagements
(d) Creative new milestones that seem like watersheds but turn out to be closer to controlled demos
Conversely, if Tesla demonstrates a clear step change in autonomy before June 22, 2026 (such as declaring victory, opening up about disengagements, and shooting past Waymo), there would still be a debate about whether Tesla was at level 4 on August 31, 2025, but it would be more reasonable to give Tesla the benefit of the doubt on questions about tele-operation and kill switches.
Update 2026-02-02 (PST) (AI summary of creator comment): The creator has clarified terminology and concepts around supervision and disengagement:
Supervision refers to a human in the loop in real time, watching the road and able to intervene.
Real-time disengagement is when a human supervisor intervenes to control the car in some way - a gap in the car's autonomy. If the car stops on its own and asks for help or needs rescuing, those might count as other kinds of disengagement but not a real-time disengagement.
Evidence threshold: Human drivers have fatalities roughly once per 100 million miles, or non-fatal crashes every half million miles. A supervised self-driving car needs to go hundreds of thousands of miles between real-time disengagements before we have much evidence it's human-level safe.
With less than 100k robotaxi miles, seeing zero real-time disengagements would still be fairly weak evidence that the robotaxis would crash less than humans when unsupervised.
For miles with an empty driver's seat, we need to know:
If safety monitors had the ability to intervene with a passenger-side kill switch
If that kill switch was real-time (like an emergency brake) or just a request for the car to autonomously come to a stop as quickly as possible
If the robotaxis have been remotely supervised (using the definition of supervision from FAQ8)
Update 2026-02-02 (PST) (AI summary of creator comment): The creator has analyzed data suggesting Tesla robotaxis may have markedly worse safety than human drivers, even with supervision. If this analysis is fair, the creator indicates that Tesla's safety record could be too far below human-level to count as level 4 autonomy, regardless of questions about kill switches or remote supervision.
The creator notes that human-level safety has been assumed as a lower bound for level 4 autonomy throughout this market. A safety record significantly worse than human drivers would not meet the level 4 standard, even if other technical criteria were satisfied.
The creator acknowledges a possible Tesla-optimist interpretation: that Musk "jumped the gun" in summer 2025 but may have achieved unsupervised FSD later (possibly January 2026). However, this would still result in NO resolution for this market, since the criteria must be met by August 31, 2025.
1,000People are also trading
Fully unsupervised rides are now also available at night in Austin. From other comments, they are still using passenger-monitors for rides including highways.
Abhimanyu Yadav (@WorldlyReviewer) on X
Apparently, there is a new guy daily tracking availability
Abhimanyu Yadav (@WorldlyReviewer) on X
@MarkosGiannopoulos Oh, last I had heard the passenger-seat monitor moved to the driver's seat for highway rides. Do you know when that changed?
As for this new tracker, any reason to think it's more accurate than robotaxitracker.com?
@dreev Not sure about the new tracker; it seems to be a single person tracking cars every day, doing rides and checking waiting times throughout the day (!) :)
He has tracked 50 different cars in Austin in August https://austin-robotaxi-tracker.netlify.app/ (which is similar to the current number of 67 cars in https://robotaxitracker.com/?provider=tesla&area=austin for the last 30 days)
I will have to look into the highway situation; I seem to recall some regulation issue.
Not a reg issue according to Tesla Robotaxi, also confirmation that the safety monitors were still around. Also, because they weren't required by reg, is this not additional evidence that this market should've already resolved NO? Driver's seat = capacity to intervene immediately. Pls Resolve NO thx. 😂
@MarySmith There is no requirement in the market that rides need to not have someone in the driver's seat in both city and highway rides.
@MarkosGiannopoulos @MarySmith
It is possible to have a level 4 system but put someone there to observe out of caution. However does that apply to this case? If the system really was adequate to be considered a level 4 in June or August 2026, then we would expect to see continuing growth of the system because it was good enough and the monitors being removed within a few months.
Seems like there were several incidents of poor driving that prompted official inquiries in ~July 2025. I think the robotaxis had a robotaxi command and control shell with v14.0.x driving software loaded into that. There were lots of v14.1.x driving software releases in ~October and 14.2.x and 14.3.x from Nov. We first knew of monitor less vehicle 18 Jan 2026 and by 22 Jan 2026 they were giving rides to the public. It is possible from comments made at earning calls that there were monitor-less tests right at the end of 2025. V14.2.x onwards may have been quite a bit better and been the proper fix for several known issues rather than working around problems in earlier versions.
Now we have a significant ~62.5% retrenchment of VMT and it seems they are waiting for v15 before starting a significant expansion of service.
I think this all paints a picture of the software not being ready at June or August 2025. Maybe it reached the level of being good enough to be level 4 somewhere in October 2025 to January 2025 or maybe you could consider it is still not good enough yet because it isn't being expanded and might only become good enough when v15.0.0 or v15.0.2 or whatever is released.
As you can tell, I maintain you can't really launch a [level 4] system if the system isn't up to the level required at that time.
We ideally want a launch event after the system is up to the level required. Maybe when v15 is adequately tested Tesla will have a launch event after which they really start rapidly growing expansion without major retrenchment events. That would be helpful and might give us a more plausible launch date of a system that seems up to the standard required. If they don't then we seem left with launch possibilities as
a) June 2025 launch but it seems like the system wasn't really ready then.
b) 22 Jan 2026 launch of properly monitor-less robotaxi rides. Tesla didn't make a huge fuss of this as a launch event but it seems the nearest thing to a launch event after the system was good enough. It is quite possible history won't remember this date preferring to quote the June 2025 date.
c) v15 released for public use date (assuming that seems good enough to scale deployment) but if there isn't any fuss made of it as a launch date maybe it doesn't seem like the relevant launch date.
If history's conclusion is that it was available to the public in limited ways from June 2025 but the rapid expansion didn't really start until Nov 2026 [or some date like that] then I think it would be reasonable to conclude the system wasn't really ready at Aug 2025 and 22 Jan 2026 or v15 release date is when it is properly launched. Might take a while before we know how history will view it.
But now we are here in Aug 2026, may as well wait at least until we have a graph of VMT in the year after August 2025 before making an effort to persuade creator he needs to grow a backbone and reach a decision.
@MarySmith markos is right about highways. See faq items 4 and 5.
In terms of restrictions on operating domain specifically the robotaxis are actually pretty far beyond the minimum this market required.
@ChristopherRandles well said. This matches my thinking pretty well. Lots of ways for a NO to end up unambiguous, hence my inclination to keep waiting. But I'm listening if the consensus is that continued waiting is just too annoying. I do think that so far there's still enough ambiguity that I'd probably want to resolve to a (low) probability if forced to resolve now
Unless I missed something, this resolves no?
Ashok Elluswamy distinguished the summer-2025 phase, when cars had passenger monitors, from the “first fully unsupervised Robotaxis” around the end of 2025. Is that close to an admission that the August system was not then considered fully unsupervised?
Also, Tesla still has not provided comparable transparency about passenger-monitor disengagements AFAIK? FAQ14 required transparency about both teleoperation *and* disengagements by June 22, 2026. It is past that date.
@MarySmith This page has a very long discussion on the issue of monitors on the passenger seat. Which Ashok quote are you referring to exactly?
@MarkosGiannopoulos On the July 22nd 2026 call:
"Second, I’d like to discuss scaling. We started the Robotaxi program roughly a year ago in Austin. We had safety monitors in the passenger seats of the cars back then."
A year ago would be, ~ July 22nd 2025, that they had safety monitors.
"Around the end of last year, we had the first fully unsupervised Robotaxis in Austin."
End of last year is what, December, November, maybeee October? July 22nd is already almost August.
Tesla has not provided transparency about passenger-monitor disengagements and it's past June 22, 2026. I'm not sure why this market hasn't resolved NO yet 🤔
Market creator says right in FAQ 14 "We'll also need transparency about tele-operation and disengagements. If that doesn't happen by June 22, 2026 (a year after the robotaxi launch) then that too is a presumed NO."
@MarySmith If you ask your favourite LLM to summarise this page, you will see that my argument since last summer has been unchanged: the market has been a Yes already since last year. Tesla operated paid rides with the cars driving on their own; the passenger-seat monitors were an expected safety precaution (similar to what Waymo had done). Since then, Tesla has shown a much faster pace of progress than Waymo (when they started), by removing the operators from the cars within 6 months (Waymo: 3 years) and operating a year later in 6 cities (Waymo: 9 years). It will be 7 cities soon, with Vegas being added.
@MarySmith Those are good arguments. My own attempt to steelman the YES position (or, rather, the let's-wait-to-be-sure position) for your two points:
Tesla is playing marketing games with the definition of "fully unsupervised" in order to keep churning out exciting-sounding sound bites. In particular they didn't want to admit in their Q2 earnings call that they've scaled back the program so they just ignored all the passenger-seat-safety-monitor miles. Now they can claim "double-digit growth" in really-for-real-this-time autonomous miles. It's super dumb. More in my recent AGI Friday. The point is, last summer's robotaxis may have counted as level 4 despite the passenger-seat supervision.
I no longer believe that if there was a physical switch it could've been used for real-time intervention at driving speed. Which means it wouldn't have been disqualifying for level 4 autonomy.
Of course there are strong counterarguments to both of those. The scaling back of the robotaxi program is itself part of the NO argument. Unfortunately we didn't pin down criteria for scaling. Just that for YES it has to eventually go mainstream and, looking back, we'll agree that summer 2025 was when it started.
For point 2 the counterargument is the technicality about transparency, as you say. What I'm nervous about is that in retrospect we may decide we had enough transparency all along -- that real-time kill switches were never plausible. Like Tesla is only failing to deny that because it would be a protesteth-too-much thing. So what I'm hoping is that when we finally do have full transparency it will be clear in retrospect how reasonable it was to have suspected Tesla of wiring up a physical kill switch that was more of an emergency brake than a signal to the car like "you are screwing up here, please autonomously come to a stop". The latter would be compatible with level-4, I think, despite being a far cry from Waymo.
To be clear, I think those are high bars and burden of proof is on YES at this point. I want to honor the spirit of the question as well as honor the technicality I stipulated. So unless we somehow decide that in retrospect it never actually made sense to have suspected that that passenger-side button was something like a brake, we should resolve NO. And similarly with scaling up, we'd have to zoom out see a nice exponential, with the current scaling back looking like mere wiggles in retrospect. And even then I'm uncertain if the YES case is strong enough.
You all have been amazing in asking hard questions and helping to gather evidence (and not just to support your own side!). Keep resisting the temptation (I need this reminder myself) to preface anything with "obviously" or otherwise try to cow/browbeat/jawbone those who see the situation differently. I think the reality here is extremely ambiguous.
PS: I think we should factor out the points @MarkosGiannopoulos is making about how fast Tesla is scaling vs Waymo. It feels like apples-and-oranges in a few ways. Waymo had to bootstrap training miles before end-to-end neural nets, for one thing. Plus Tesla announced (falsely) that they had prototype self-driving in 2016, making that probably a fairer time to start the clock if we did want to compare prototype-to-product speeds. But, again, I don't think we can meaningfully compare that and don't think it matters for this market.
PPS: And number of cities is a metric Tesla is currently gaming the crap out of. It's so tricky to form a fair impression of where Tesla is really at on this, which is another reason I prefer to keep waiting. So much will be clearer in retrospect!
@dreev I've sold my stake, because at this point the market feels more like predicting what your resolution conditions will be than any matter of facts on the ground, and I don't participate in such markets.
We'll also need transparency about tele-operation and disengagements. If that doesn't happen by June 22, 2026 (a year after the robotaxi launch) then that too is a presumed NO.
What is your criteria for such transparency? The date has come and gone, without anything that you've pointed out as meeting the bar; now you say it might become clear in retrospect and they just didn't bother saying so right now, but... I don't think that's compatible with the definition of "transparency".
2. What if there's a human backup driver in the driver's seat?
This importantly does not count. That's supervised FSD.
Arguably this was a bad criterion in the first place, because per SAE's definition, level 4 is a matter of "is the vehicle driving itself, such that the human doesn't need to monitor the road", regardless of where any such human is located. That is, it's entirely possible to have level 4 with a human in the driver's seat, so long as they are not expected to pay attention. Admittedly, it's implausible that Tesla would put a non-passenger human in the vehicle for any reason other than monitoring the road, such that the car is not operating at level 4 even if capable of it, so "non-passenger ready to take control means it's not level 4" is probably a reasonable assumption in practice... which brings us to:
9. What about human safety monitors in the passenger seat?
Oh geez, it's like Elon Musk is trolling us to maximize the ambiguity of these market resolutions. Tentatively (we'll keep discussing in the comments) my verdict on this question depends on whether the human safety monitor has to be eyes-on-the-road the whole time with their finger on a kill switch or emergency brake.
Like I said, it seems deeply unlikely that Tesla would put supervisors in the car if they weren't expected to supervise the driving (I guess they could be expected to supervise the passengers instead? But that's not what it looks like at all). Combine that with their own admission that unsupervised rides didn't launch until well after August, and... well, I don't really see it mattering whether they had emergency brake buttons, the question that determines the driving automation level is "were they responsible for proactively overriding the car's actions (level 2), could ignore the road until the car announced it didn't know what to do and then had to take over (level 3), or not responsible for controlling the car at all (level 4)?" Given that the market defined a supervised FSD as disqualifying, and both you and Tesla used the same term, the case for Yes seems to only be... that the supervisors weren't actually supposed to intervene, and were just there to reassure and/or confuse passengers. plus pad casualty counts if something went wrong? I don't think that's plausible, but I guess it's not impossible.
@SeekingEternity Yeah, this keeps happening to Musk-adjacent markets. Ask a seemingly simple question like whether Tesla will launch robotaxis and then reality plays out in the most ambiguous possible way.
What is your criteria for such transparency?
Tesla sent a letter to a congressperson detailing the conditions in which they employ tele-operation. That counted as transparent (and the tele-operation turns out to be limited enough to still count as level 4).
It's actually possible that there's something similar for kill switches and we've missed it. So that's another reason I'm hoping to wait this out longer.
Arguably this was a bad criterion in the first place
Oh my, yes, so many of my criteria are bad in retrospect. But my stipulation that the driver's seat be empty I think was fine even though, as you say, it would technically be possible to have level 4 autonomy with a human in the driver's seat. Oh, you go on to say as much.
seems deeply unlikely that Tesla would put supervisors in the car if ...
My latest belief (my beliefs have been all over the place over the course of this market!) is that the safety monitors were there for a bunch of reasons, some embarrassing for Tesla, but none of them quite disqualifying for level 4 autonomy. For example, Teslas are still so, so bad at parking. And sometimes (not too often) they're too timid to finish making a turn. Or they can get stuck in infinite loops because they have no memory. All sorts of reasons it's helpful to have a human available to rescue the car. As long as the car autonomously stops and explicitly relinquishes control to a human, that's still level 4.
Oh yeah, and we've learned that the tele-operators are pretty terrible at controlling the car, so that's another reason to prefer a human on the scene. But again, it's all still level 4 as long as the handoff from autonomous to manual isn't at driving speed or human-initiated.
I'm not sure about this though. If the safety monitor was (back in summer 2025) pressing a button every few blocks to tell the car to pull over and relinquish control to a human because the car was screwing up in some way, that starts to sound like level 4 by the barest technicality. Or maybe that's disqualifying if a human is watching in real time and providing any input, even if it's not steering/braking input. I guess this is yet another question mark to resolve. Sigh.
@SeekingEternity "their own admission that unsupervised rides didn't launch until well after August" - If we are going to resolve the market in a pedantic way, we should be accurate. Ashok spoke of "fully unsupervised Robotaxis" (not simply "unsupervised"). If you could ask Ashok if the robotaxis were on Level 4 in June 2025, he would most certainly say they were.
"it seems deeply unlikely that Tesla would put supervisors in the car if they weren't expected to supervise the driving" - Supervising the car (without taking control) is not against SAE Level 4
@MarkosGiannopoulos I have quibbles here!
I'm not sure how Tesla would answer (in an official capacity, like in an earnings statement) the question of whether they were at level 4 in summer 2025. Maybe 70% chance they'd say it counted? This would be hugely helpful (not to say dispositive) to find out.
I know what you mean here but since people are perennially confused on this point I'll reemphasize it: when total mileage is as low as it is for Tesla robotaxis, counterfactual interventions matter. The canonical example is a kid running into the road. If you need a supervisor to take control if that happens then you can't claim it's unsupervised just because it hasn't happened to happen so far. Or consider the zero-disengagement cross-country trip: lack of disengagements might just mean that nothing too unusual happened on that particular trip.
So we need either assurance that the supervisor couldn't intervene in real time if they wanted to, or enough miles to be assured that enough corner cases cropped up. Even now we don't quite have that, looking only at the robotaxis. Combined with other evidence from privately owned Teslas, I'm personally convinced that human-level safety has in fact been exceeded. But for just the robotaxis, the error bars are pretty wide:
(That's Waymo in blue like 10x safer.)
PS: More evidence that the robotaxi program is quasi-paused right now despite officially operating in 6 cities:
That's 35 weekly active vehicles at the moment -- 7 "fully unsupervised" and 5 fully supervised, leaving 22 with a passenger-seat safety monitor.
@dreev Ashok covered the issue you raise about the number of cars and cities
- "The reason we have been expanding across different cities instead of just doubling down on a single city is that we want to make sure that our stack is very general one. It is a general one. We just want to, like, you know, both prove to ourselves and to other folks that it is working across a lot of different cities without too much effort per city."
- "That’s why it’s hard for others to comprehend in terms of, like, miles versus vehicles. Since these vehicles operating the Robotaxi fleet drive basically continuously as opposed to human drivers who use vehicles for maybe a couple hours a day or something like that. These vehicles are in mostly continuous operation, which means that even for a few vehicles, you can get a lot of miles out of them."
Regarding your impression of a pause in the program, GPT disagrees: "Tesla’s first unsupervised Robotaxi mileage ramp appears roughly 10–12 times faster than Waymo’s earliest comparable driverless ramp." https://chatgpt.com/share/6a725923-ec40-83eb-b457-2958cfdef634
@MarkosGiannopoulos Valuable context. Hard to be sure how much is rationalization and spin, but we can look at vehicle miles traveled (VMT) directly and see this:
I continue to dispute the meaningfulness of comparisons to Waymo's VMT numbers from years ago but if we zoom out as far back as I can find good numbers for we see this for Waymo:
I know we haven't pinned it down well enough and I'm not saying we should wait this long to resolve but I'm treating it as one of the in-principle resolution criteria that Tesla's graph should eventually have the rough shape Waymo's does.
Tesla has scaled back at least somewhat and that should be an update towards NO. One of the multiple hurdles for YES is for this retrenchment to look like meaningless wiggles on the left side of the hockey stick when we later zoom out.
@dreev if what you're waiting for to resolve this NO includes years to see "this retrenchment to look like meaningless wiggles on the left side of the hockey stick" , that's kind of a bummer 😅. I'd be willing to bet that Robotaxi is an eventual success, and that this retrenchment is meangingless wiggles, eventually, given current trends.
@MarySmith If you think that, but still think NO is correct, I take that seriously. My thinking is that there is still a conceivable way the future could play out where a YES resolution would feel most true to the spirit of the question. I don't think that's likely! And I'm still up for the debate that should rightly ensue in that possible world, about letter vs spirit and such. So what I'm mostly hoping, in some sense, is for the actual world to deviate enough from that conceivable world that we can resolve NO without requiring the invidious debate. 😅-emoji, indeed.
Btw, I'm assuming that with Manifold's loan feature it isn't a significant hardship to have money tied up in this market. I get that it's still kind of annoying! But let me know if it's worse than I'm presuming. That should be a factor in how long we're willing to wait on resolution here.
@dreev
I have sold out my position, well reduced to 1 mana payout on no to keep an eye on what happens. Happy to cash out at a profit and avoid risk on such an interpretation based claim. Cashing out earlier would have resulted in a loss, so thanks to Hacking for cashing out in other direction to allow me to do this. Anyhow ...
>"Tesla has scaled back at least somewhat and that should be an update towards NO. One of the multiple hurdles for YES is for this retrenchment to look like meaningless wiggles on the left side of the hockey stick when we later zoom out."
A scaling back but with an explanation that they are concentrating on expanding to other cities to prove the generality of the system could seem like part of an expansion just in a different way than VMT or it could seem like the system still needs to be proved out before properly launching it.
If Tesla driving system eventually becomes the the dominant system then inevitably the curve looks exponential-ish and the numbers now won't look like wiggles it will look more like it is still firmly on zero before the exponential growth starts. It is also possible that Tesla launches or even has already launched the system, becomes a contending system but someone else system becomes dominant and the VMT graph shows an growth phase and then levels out or declines.
I have always been a little concerned that 'looking exponential' is not the right test and perhaps this VMT drop and above paragraph should help clarify thoughts on whether the VMT graph looks exponential (and only meaningless wriggles) is the right test. Should it be more a question of when is the starting point of the exponential growth, before or after 31 Aug 2025.
A drop from ~400k VMT in March to ~150k VMT in June, 3 months later seems like a significant % drop of ~62.5%. There is a couple of 3 month declines on the Waymo graph but nothing like 62.5%.
So it seems sensible to ask you:
1. How committed are you to exponential like growth curve and meaningless wriggles as the test as opposed to something more like looking for when the exponential growth starts.
2. Is a 62.5% drop in VMT over 3 months a breach of 'looks exponential'? Is it sufficient enough drop that it is becoming more likely that the start point of exponential growth is after this 62.5% drop? If you consider 62.5% drop 3 months to be too short or not steep enough, then how long or steep does the drop have to be to say looks exponential is breached and/or the growth started later than the deadline for this market? (We are getting close to a year after 31 Aug 2025 and is seems VMT does (and will) not look exponential for that 1 year.)
3. Does an expand to other cities represent some development in the growth of the system that you might accept that as adequate explanation for such a large percentage dip in VMT?
Your version sounds potentially better. At this point we should just match what the FAQ above says as best we can so we're not throwing curve balls at anyone. Of course the FAQ just doesn't say much beyond the vague "it has to go mainstream eventually and summer 2025 has to be when it started".
Hard questions! These questions seem like they'll be much easier to answer with more hindsight. I feel like fundamentally the whole problem, just in terms of spirit of the question, is that Tesla has done something that's like a perfect superposition between launch and demo.
That does sound fair, I think?