Let's assume that Tesla is going all in on Robo taxies (I think a reasonable move given their core automotive business is not sound from a shareholder perspective). Then the problem they're facing is less technological than regulatory. If you were to assume that their self-driving is perfect, which it's not, they're meaningfully behind Waymo in an important way. Waymo have put the work in to establish themselves with regulators, gone slow in the roll out, responded to pressure and criticism and built trust. Musk seems to imagine that he can just put a 2 tonne death trap on the street and go "Tada!" and people will just accept it. They won't.
The core problem is they'll need to go through years of hurdles to get regulatory approval, building up through the same steps that Waymo are going through. It's not a technological problem, it's a regulatory and political problem.
Waymos advantage is also cooperating as a means to help shape the regulations that competitors will later have to meet.
From a public acceptance view, I think after the first wave gets deployed by any company the companies that come after it will have significantly more leeway and the public will more easily accept them.
That might be true, but it also destroys the economics. A massive factor in this is that to justify the valuation of Tesla you need to beleive not just that Tesla are going to solve self-driving (again) but they're going to do so and have a monopoly for some period of time. If Waymo really do solve it and roll out at scale, then all that Tesla is going to offer from an economic point of view is a race to the bottom. They'll entirely lose pricing power, which defeats the purpose of the pivot in the first place.
Waymo’s advantage is also that it is actually operating a robotaxi service, which isn't just making self-driving cars or having regulations in place, while Tesla is not.
Sometimes it is the second mover in a market that wins for this very reason.
For example, say Waymo does all of the hard work of establishing regulations and making all the big mistakes. Tesla then comes in later and is able to benefit from Waymo's public lessons. Because Tesla can set a direction more clearly, they might leapfrog Waymo.
I Personally prefer the Tesla go to market strategy. They are tackling a much harder problem (machine vision) and if it ends up working, it will be done far more elegantly (a few cameras rather than tons of sensors) with far more significant implications to our day to day life (the tech can be extrapolated to tons of automation, like garbage roadside cleaning, side walk garbage collection, robots, etc).
On regulation, I believe regulators are prone to bending to the will of local constituents. Uber was very effective at galvanizing voters when local markets were threatening legislation via the app. I imagine the many millions of people that will own a Tesla in 2 years will have no problem clicking a “message your local congressman” button on the screen to get this approved in most markets. It won’t matter if NYC, Portland, San Fran and LA ban it.
Of course, the data will have to fully support the safety of Tesla FSD. I’m curious to watch it play out.
Relying on vision might be good for the advancement of computer vision, but is it good for automated driving?
It seems to be an unnecessary anthropomorphization to me. One of the greatest weaknesses of human drivers is the limited information available to act on. And there’s no possible scenario in which computer vision overcomes information unavailability issues. Fog or obstructions will always defeat it.
> Uber was very effective at galvanizing voters when local markets were threatening legislation via the app
By “very effective”, you mean just throwing around millions [1] of dollars to spam the local population to “galvanize” them to vote for/against certain laws?
Didn’t work in Austin, TX [2].
You are overestimating the power of what FSD can do. The unfortunate mistake most TSLA drivers do in the first place.
TSLA is playing catch up and burning massive amounts of cash. Laying off people left and right. The market and interest for TSLA is shrinking.
I find it funny to think back to the fervor of comments on HN a year ago about Tesla doubling down on vision only and how this would prove to be dangerous and ineffective. A year has gone by and not only have we not seen the catastrophe that was thought to be inevitable, but Autopilot has gotten leaps and bounds better. I use it every day and over the past few months it's gotten so good that there are few scenarios where I ever intervene, and when I do, it's because it's always because it's being a little too slow/hesitant for me at an intersection. I'm not a Musk fan one bit, but I see this as evidence that the popular narrative has diverged wildly from the reality of Tesla's product.
It's wild to me that people argue automated driving is great based on individual personal experience.
If I were to argue about manual driving from my personal driving, I'd be way off. I haven't had an accident in decades. Not even any close calls. If I extrapolating my accident rate out to the world, I would assume that human-driven cars are a perfectly safe technology.
In the real world, all accidents are a long-tail phenomenon. They happen pretty rarely, often under conditions that are unusual or unexpected. So somebody saying "in these last few months I hardly ever have to intervene anymore to not die", a standard that would cause a driving instructor to fail a teen driver, not only doesn't give me confidence that this technology is a-ok for road use, but suggests to me they haven't really thought about what is required for safety at global scale.
It's perfectly reasonable for me to make a determination and offer my first hand experience about Autopilot's capabilities relative to that of my own and that of other drivers I see on the road.
Accidents might be long-tail phenomenon but driving habits are observable. Autopilot has a consistent, safe, alert demeanor on the road. It reacts to vehicles I as a driver can't as quickly see, changing situations that it would take me an extra half second or more to react to.
Compare that to the average human driver on the road - a third or more of whom you can expect to see holding a phone - and I think it's a reasonable conclusion to say it's already at least marginally better. And if we are talking about accidents, you can't deny that people are generally terrible at choosing the best possible reaction to an emergency situation that happens in the space of a few seconds.
It's reasonable for you to offer your firsthand experience. It's unreasonable for you to think that you can offer a determination broader than that experience.
I have seen this "huge improvement" comment about FSD just about in every discussion about FSD ever, since the first version. It keeps getting huge improvements, and it's still not done. At this point I expect version 579 of FSD will still be a "huge improvement" over 578, but just not 100% yet, and will soon be finished.
Each of Tesla's "huge improvements" seem to approximately halve the number of interventions. You can go from one intervention per minute to one per two minutes to one per four and so on. But you'll never hit zero.
I’m not on the same boat, NTHSA has found Tesla’s Autopilot linked to increased accidents with “safety issues with Tesla Autopilot contributed to at least 467 collisions and 14 deaths from January 2018 through August 2023”. They have done a recalls and ar still investigating.
> Tesla doubling down on vision only and how this would prove to be dangerous and ineffective. A year has gone by and not only have we not seen the catastrophe that was thought to be inevitable, but Autopilot has gotten leaps and bounds better.
Maybe it’s better in terms of customer satisfaction or reliability, but in terms of the SAR J3016 standard, it hasn’t improved at all.
Exactly. The technical argument about capabilities is largely settled at this point. Vision won, LiDAR didn't "lose", but it's clearly not any better. Yet every time the subject comes up we rehash the same arguments from 2020/21 as if the evidence of the last ~billion miles (or whatever) of road testing doesn't exist. The crashes didn't show up. The bodies aren't there. The disaster didn't happen. The product is "safe" for whatever reasonable definition you want to pick.
And to be clear, my sense is the same: FSD microdriving at this point is better than mine, interventions are about planning and intersection recognition. It'll be slow to take turns sometimes, it'll be in the wrong lane to prepare for a turn, stuff like that.
Vision didn't "win" in any way, shape, or form. The only products on the road that offer actual self driving don't rely on vision only, they have vision + lidar + other sensors.
Teslas are not "self driving", they don't allow you in any way to rely on the car alone, for any length of time, to drive itself: it's the drivers who are preventing accidents, not Tesla's FSD. When drivers of these vehicles lose attention, accidents do happen and people have died.
But they didn't hold up. There are literally two orders of magnitude more vehicle-miles driven under FSD now than there were then, and there's absolutely no data showing a safety effect above the noise floor. Your argument is non-falsifiable nonsense, basically. You can continue making it forever, but that doesn't make it right.
There are exactly 0 miles of FSD driven by the public without constant attention from a driver keeping their hand on the wheel. There are disengagements of FSD in virtually every trip, where the driver feels that they need to intervene or the car would do something wrong. The safety record of FSD is basically due to the human drivers, not the AI. Tesla themselves don't have any confidence that the car can drive without constant human attention.
This is in contrast to Waymo and Mercedes, where the car actually drives itself, using both vision and other sensors, and the driver can actually disengage.
Again, non-falsifiable so I can't argue. Good trick. My car still drives me around every day and no one is crashing, so I guess everyone wins. I have an amazing supercar and you don't need to feel wrong about it.
Do you take your hands off the wheel? Do you take out a book and read it? If not, do you ever intervene on these trips?
What I'm saying is the official line from Tesla: they don't even claim to have achieved level 3, nevermind the level 4 driving you seem to think they have.
I repeat: I cannot falsify your contention. We will both go to our graves "not knowing" whether or not the driver is "really" the magic that makes the car drive itself. But only one of us is actually right.
Of course it's falsifible. I don't own a Tesla, and I wouldn't trust it even if I did, but I could check for myself by setting it to self-drive and then not paying any attention whatsoever. If I get in an accident, I was right.
More realistically, if Tesla believed their car is safe without human intervention, they could, you know, put them out on the streets without drivers, like Waymo did. Or at least tell their customers that it's actually safe to not pay full attention and that Tesla would accept liability if accidents happen, like Mercedes did. Or something in between. Since they are not doing any of these things, I think your opinion that is just obviously wrong.
This is disingenuous. And the reason is you clearly refuse to take testimony from those of us who do this every day. I'll tell you right now that it's been well over a year since I had any kind of safety-related disengagement. So, you'll change your mind?
Of course not. You're going to say that that's not enough data, and then go and cherry pick a single accident or testimony to the contrary to invalidate the data that you just said was a valid test.
I've been seeing such comments for 4 years or more. Mercedes and Waymo actually have full self driving on the roads today, with regulatory approval and proven efficiency. Tesla has a nothing burger that disengage if you take your hands off the wheel.
Because the Mercedes solution works, today, with legal guarantees, under quite precisely specified conditions. It allows you, in those specified and advertised conditions, to play a game or read a book or take a nap while driving.
In contrast, Tesla only has promises. In practice, you can't do anything with FSD that you couldn't do without it, because you have to pay exactly as much attention when it's engaged as when it's disengaged, and you can't take your hands off the wheel or it full-on disengages.
Mercedes has an excellent base to build up from. Tesla has next to nothing.
not to mention much more scalable than waymos approach, both in terms of rollout, but more importantly in terms of data collection - the gold of the AI era.
Weren't we supposed to have Tesla robotaxis making everyone $10k a year back in 2021? I will be shocked if there is an actual product that is actually usable within 6 month of August 8.
That $10K a year figure never made any sense. Let's say it was true -- you could hire out your Tesla for 10 hours per day while you weren't using it. Sure it's making you $10K/year, but you're putting 100,000 miles per year on your car and it wears out after 3 years rather than after 20 years. Having a way to monetize depreciation may be useful in some cases, but it certainly isn't making you money.
I believe that million-mile electric cars are quite possible, but 2021 Tesla's certainly weren't that.
That guy who put thousands of working Internet satellites in orbit, is currently the only way for U.S. to get humans to and from ISS, essentially created a non-existent electric car market from nothing, and built the heaviest self-landing rocket in history, among so many other things.
That charlatan? If that's how you define charlatan, I wish we could all accomplish that much.
(what other accomplishments did I miss; tbh I don't even like electric cars from a practical or even environmental perspective, and yet even I can admit that what he did was simply incredible.)
Hard to know if you're being sarcastic or not. I'm not a fan of how they handle FSD marketing, but they are very good at automating car manufactory. I believe they are more or less the only company that can mass-produce EVs at that scale _for a profit_.
The traditional car brands are losing money on EV's and/or can't scale the production up enough. BYD is probably the only other manufacturer that can compete with Tesla on scale and cost.
Anyone who owns a Tesla and has used the new V12.3 understands why Elon is so bullish about the Robo-Taxi's. He's said they've internally seen a 5x-10x reduction in disengagements on 12.4 (already trained, out soon), and that 12.5 is even better. Their GPU stack by end of the year will also be 8x the size it was earlier this year. Combine all this together, and the "feedback loop" with millions of people using FSD, it's looking like the "vision only" strategy will work.
5x to 10x seems too vague to be the result of objective benchmarking.,. those sound more like aspirational numbers from a guy who's been promising FSD every year since forever.
Some car markers will die, others will rise (BYD), hopefully most will pivot successfully and survive, perhaps by buying some of the tech from Tesla and others.
> Waymo declined to tell me whether—or how often—remote operators intervened during my ride
Or just the entire ride.
Declining to provide the information on how much humans were driving to a journalist writing an article about the experience just makes it look like Waymo decided that there would be zero errors.
Meanwhile, Tesla just said go for it, and then the author wrote a review disparaging Tesla, who gave "full" access without human help. Instead of Waymo, who declined to say whether humans drove the car.
Waymo could offer a Mechanical Turk to play chess with and this author wouldn't even blink.
This is an information-free article with a click-bait title.
"While no one was behind the wheel during my rides, Waymo has remote operators that sometimes provide guidance to its vehicles (Waymo declined to tell me whether—or how often—remote operators intervened during my rides). And while Tesla’s FSD works on all road types, Waymo’s taxis avoid freeways."
I mean, probably? That's a two-mode failure, so the generally correct engineering is to (1) not make things worse and (2) not worry about cost, because it's vanishingly rare. In this case, it's a "disabled vehicle in the road", which is not a new thing at all, and you rely on the existing infrastructure (tow trucks!) to handle it.
People seem to think you're joking, but I think you're spot on.
Putting it in the best possible light, Tesla represented a bet on developing both a new carmaker and the dominant car of the future: electric and fully automated. That was a big gamble, and it required gobs of money. Existing carmakers could try to-self fund that, Tesla couldn't. So Musk, marketer extraordinaire, built up a Jobsian reality distortion field around Tesla. He made his cost of capital incredibly low by hyping the stock to the moon.
If the bet pays off, Musk will historically be seen as a genius. But if it doesn't, he'll be seen as a fraud who milked credulous investors for a project that was never going to work out. And I think that's true even if he ends up with a modestly successful niche carmaker. Many others would see that as a huge accomplishment, but it'll be so far short of what investors thought they were paying for that even visions of an automated future won't make up for their large losses.
I know this was wit, but Tesla has and always will be an energy company.
The cars have obviously done well for them, but their solar and charging are the real innovations. I’ve been waiting for the day there’s pressure to split the car business away from Tesla
I'd expect a commodity like charging to be a very tough sell in the stock market. Turns out charging is actually not easy, and having boots on the ground in lots of places where people actually want to charge makes a formidable moat (booths on the ground, if we stretch the term just a tiny bit). But I don't buy into any claims that this was a planned outcome from the start. It's not failure when you do something well and stumble into something better while doing the thing you set out for. Not everything is Bezos launching "anything that has a GTIN and fits in a box" with the subset where the number happens to be ISBN.
Snark aside this is pretty true. Tesla's valuation by the market is completely detached from reality. Something about market can remain irrational but if/when people start believing that Tesla is a mature company whose stock price has to ride on realized quarterly numbers instead of future growth/ambition then they're gonna have a lot of trouble maintaining their price.
>whose stock price has to ride on realized quarterly numbers instead of future growth/ambition then they're gonna have a lot of trouble maintaining their price.
Sure, that's also true of every tech company. Tesla's P/E ratio is around 47, which isn't great, but also isn't terrible.
A publicly traded company is whatever those with capital (or those managing capital) and the highest grade of optimism about the company want it to be. Obviously when it comes to Tesla, there's some very strong optimism in the market. Seen from that perspective, I'd say that Tesla remains refreshingly far on the car side of the "product is car vs product is stock" spectrum.
But I agree that taking on the robotaxi market does move them further towards the stock end. In financial terms, deep integration is awesome until it isn't. Eventually most, perhaps all, layers will rely too much on the rest of the stack propping them up and then it all falls down. Mid-term investors will happily bet on abandoning ship right before that happens.
This was such a poorly written opinion piece that I wouldn't be surprised if someone with a large GOOG-TSLA options spread commissioned it as a hit piece. It just repeats the fact that Waymo uses remote operators over and over, completely ignoring the fact that Tesla uses a local operator - the driver that takes control when it disengages. Its the same class of data being collected either way. The rest of the article read like ChatGPT generated fluff to make it longer, without real substance. It does make excellent engagement-bait, I'll give it that.
Honestly at this point we could have just had standardized infrastructure for self-driving vehicles to guide them along the road.
We had government programs doing exactly this until republicans gutted them.
Would have been more stable, cost-efficient, safer, and it'd actually exist today.
The reality is that there are fundamental, unsolvable problems with the current "self-driving" stack. From the cameras to signal processing.
Everyone involved in this space has been repeating the same thing since basically the early 2010's.
Just stop pouring needless amounts of money into these projects that have offered very little fruit. Build the standardized, simpler infra with the money and fund trains while we're at it.
This just doesn't work. You can make standardized infra, ignoring the insane cost it would take, but still the cars need to handle unusual situations even within that standardized infra. If you want autonomous vehicles on the road, they need to handle situations just like humans can. Reality is messy.
Even in factories, within intense control, autonomous robots can struggle and it takes a lot of effort to keep them functioning.
>but still the cars need to handle unusual situations even within that standardized infra.
So? You're missing the point. It'd still be miles and miles ahead of where we are today.
AND it'd still cost less money than we've spent today AND it'd keep money in the right places AND it'd provide plenty of useful jobs for people that need it.
This wasn't some hypothetical either. We had working standards and working prototypes for all of this. Congress was actually gearing up to make this a reality back around 2010. But Republicans.
It's what we in the business call a "no brainer."
Saying "it just doesn't work" when it CLEARLY did is wild.
I completely disagree. You're basically saying we should create something like an automotive/train hybrid system across the country hitting all communities in the country. That would take trillions. Like actually trillions, and it still wouldn't be good as having autonomous vehicles, because once we have autonomous vehicles, we also have autonomous aircraft and ships and all kinds of great things.
* be completely autonomous from day 1, and progressively increase the number of situations you can drive through;
* or drive through every legal situation from day 1, and increase the % of them handled autonomously.
I believe the 2nd approach, Tesla's, has one key advantage: it collects data about freak situations much faster and more exhaustively. Given how data has become the key resource in AI, that's probably a very strategic asset they've accumulated here.
Also, Waymo's joker (remote operation by humans when the software bails out) is totally replicable by Tesla robotaxis.
I was in SF which and happened to be driving next to an unoccupied Waymo.
While considering other foot and vehicle traffic, I swerved into the Waymo's direction to test the reaction. The car swerved a bit, honked, and slowed. I then cut the vehicle off knowing it would submit to my position and continued to drive to my destination.
The fastest way to get adoption is to isolate the lanes between driverless and human driver vehicles. Unexpected threats and bad actors like myself would not be able to cross that line, limiting the impact of foot traffic and outside interference.
> While no one was behind the wheel during my rides, Waymo has remote operators that sometimes provide guidance to its vehicles (Waymo declined to tell me whether—or how often—remote operators intervened during my rides)
Have they announce any numbers like that? The only thing I can recall hearing was having 72 operators for 50 vehicles, but that was back at initial launch.
If I were in their shoes, especially given the Cruise debacle, I'd have continuous remote human monitoring of every vehicle. They're already burning billions on this; if having one minimum-wage worker per vehicle just hovering over a big red stop button prevents a gruesome accident and a multi-month pause, it'd be money well spent.
That’s why people many people think Tesla is behind on the path of self driving. It’s good enough when someone sitting in the drivers seat and paying attention, but that’s a long way from self driving.
How could someone pay continuous attention to 10 different cars, ready to intervene in less than a second at any time for any of them? Because that is the standard that is required to drive a Tesla "FSD" vehicle.
Search for trolley maps of your area, probably late 1800s early 1900s. I think a lot about how Uber and Lyft have all the data to support where more public transit would have the biggest benefit but clearly ignore it.
I didn't realize how good question that was until I started looking into the state of self-driving in China to come up with a witty reply. It does seem they have many contenders.
Pedestrians and motorcyclists are supposed to play chicken, one supposes. The self-driving problem requires general intelligence and a theory of mind, both unsolved problems.
I'm a pedestrian in SF, and I definitely trust the automated cars less. I pretty rapidly slot a given driver into one mental model or another. But I have no idea what the automated cars are up to, and so have to give them a very wide berth.
They said cyclists for a reason though. As a pedestrian, if you're confident the human sees you, it's very easy to make sure that you get through the interaction safely. That's not true at all for cyclists, where humans will see you and still drive dangerously around you.
Sorry, I'm not following. I don't bike as much lately, but I was an ardent cyclist here for more than a decade. If I won't trust a vehicle as a pedestrian, where speeds are lower and I'm more maneuverable, I'm going to trust it even less as a cyclist in traffic. So if bryanlarsen thought the cyclist comparison to be unfavorable to Waymo, then I'm agreeing. But I think he thought it would be favorable, and I'd disagree on both counts.
What's your evidence for that? As I said, I've seen them do things that make me give them a wide berth. And as I explained, that berth would only get wider on a bike, when I'm more vulnerable.
>Pedestrians and motorcyclists are supposed to play chicken, one supposes. The self-driving problem requires general intelligence and a theory of mind, both unsolved problems.
Self driving only has to be better than people drivers, statistically, to be a benefit to society. This probably isn't a very high bar.
No, comparing against a general population average is way too low, because automated cars won't randomly replace drivers. They need to be at least as good as the sort of good driver that self-driving cars would replace. Taxi drivers, for example, are much better drivers than average people.
Further, I think a variety of factors, including politics and liability structures, mean self-driving cars will have to be radically better than the average human driver. For example, each car-caused death is seen as the responsibility of some human. But every Waymo death will be seen as the responsibility of Waymo. Responsibility for the current 100 deaths per day is diffuse, and so accepted. But if it's 40 deaths per day for Waymo and 60 for Musk, nobody's going to shrug at that.
Are you saying if auto driving kills 5% fewer people we shouldn't use it because it's not good enough? If so, that sounds like a case of great is the enemy of good.
That's the other issue: the already low baseline. Pedestrian fatalities in the US are unacceptably high already, and let's not talk about the standards of driving instruction and road safety enforcement.
The core problem is they'll need to go through years of hurdles to get regulatory approval, building up through the same steps that Waymo are going through. It's not a technological problem, it's a regulatory and political problem.