My neighborhood is shown in this video multiple times. Number of times I have seen an AV act dangerously? Zero. Number of times an SUV has hit me while I was legally in a crosswalk lifting me off my feet and throwing me through the air? more than zero. Number of times I have seen an AV parked in the box on a red light? Zero. Number of times I have seen a human parked in the box on a red light? All the time, every frackin day.
Are AVs perfect? Nope but they might already be safer drivers than the median driver in SF. There have been 310 pedestrian deaths in SF over the past 10 years. Number of deaths the NHTSA has attributed to full self driving vehicles? Zero.
Given there's over a million registered vehicles in SF, and only ~300 fully autonomous vehicles in SF, if there's only 310 pedestrian deaths over the past 10 years, with all else being equal, you'd definitely expect the vast majority of those deaths to be from normal vehicles, given that there's you know, over a million of them, and only ~300 AVs.
As AVs are .0299% of the vehicles on the road, all else being equal, they'd account for .00929 deaths per year. You can't really point at the lack of deaths at this point as proof of anything.
According to the DMV of CA in 2023 there were 453,180 including motorcycles and trailers. Most vehicles are used very little or not at all where as autonomous vehicles are used most of the time. Also, the NHTSA numbers are for all AVs everywhere in the US so your astonishingly numbers precise are way off.
Still, as a pedestrian, I am pretty happy the AVs are not parking in the crosswalks all the time.
You need to look at per mile driven rather than just number of vehicles because the AVs are much higher utilization. The states confirm that per mile driven, AVs are much safer.
Also anyone that has walked near one can easily tell. The AVs actually yield to pedestrians, especially if you step near the crosswalk unlike many SF drivers.
Not to discredit your numbers, but the video mentions deaths and injuries in the US attributed to self-driving - it's a real thing. Do companies have a long way to go before these things are ready for the streets? Absolutely.
Just don't export that kind of road behaviour to other countries, please.
That's another important point he makes: these self-driving cars will probably not be retrained for every single country, so you risk exporting the dangerous behaviour of American drivers to other countries.
A better solution is less cars, more other modes of transportation.
And the other side of the coin is if you are a driver who isn’t a brainless idiot, who hasn’t ever been in an accident, self driving cars are worse than your average performance even if they are better than the median american idiot.
I had an autonomous vehicle almost hit me. It didn’t understand to yield to pedestrians in the crosswalk when turning at a light.
I walked into their office and reported it (there’s no other mechanism that I know of). They shrugged it off.
Autonomous vehicles can be done safely, but nobody and nothing is perfect so without robust error correcting and reporting systems they’ll never be safe.
Despite big promises in the title, the thesis is just
- self-driving cars aren't perfect yet (trots out the usual double-standard headlines which get zero attention when humans do them all the time)
- our cities were already destroyed by poorly-designed infrastructure, and self-driving cars don't suddenly fix that (something which was never promised or expected?)
- whenever you mention reducing millions of human-caused traffic fatalities, use a sarcastic tone of voice
Disappointing. I had much higher hopes going in.
He says it himself: self-driving cars are just cheaper taxis.
> He says it himself: self-driving cars are just cheaper taxis.
Which can lead to more traffic (@21:07). Which can encourage more resources to be moved from people to vehicles (@32:37). More traffic can increase pollution (@38:05), both particle and noise, even with EVs because tire wear, road wear, and brake wear (even with re-gen) generally increase with higher vehicle weights/masses.
He then contrasts London (Ontario) and Utrecht (@43:28), and how even given their similar beginnings and similar populations, the decisions each made on how to design things made one more human-habitable and the other one less.
More traffic = more transportation services provided, so your first point isn't inherently bad.
I agree we should invest in more walkable infrastructure (your second point), but that doesn't happen for many reasons, and AVs can also assist that process by eg removing parking spaces and making buses cheaper.
The remaining noise and pollution concerns call for building robotaxis that are quieter and lighter-weight, which seems to be what's actually occurring.
I admit his shallow dismissal of mass traffic fatalities as "marketing" and "corporate profits" immediately destroyed all credibility in my eyes. Millions of grieving parents aren't helped if some video 'dunks on' tech companies with irrelevant non-argument quips, what actually helps is (controversial opinion) to stop the killing. AVs are the only realistic path to that world.
Do you prefer the world dominated by EVs that is portrayed towards the end of the video or Utrecht? As can be seen in the video, Utrecht used to be car-dominated up until the 80s, but later chose to reject the influences from car companies and rather focus on enabling more mixed forms of transportation.
> what actually helps is (controversial opinion) to stop the killing. AVs are the only realistic path to that world.
You will have to explain why, because there's absolutely no evidence for that claim, and plenty of evidence for better solutions. As shown in Utrecht, right there in that video.
Cars are inefficient and dangerous, and it's not clear that AVs are going to be significantly better, whereas it's pretty clear that bikes and trams are.
Only in utterly narrow-minded USA-centric view of the world... I'm fine with US being terrible place to function without a car but please don't spread that cancer abroad.
> More traffic = more transportation services provided, so the first point isn't inherently bad.
And as he points out (@29:56), more traffic would mean a push for more roads, which is inefficient and wasteful: it takes over a dozen lanes of highway to move less people per hour than a signal lane-width of subway. E.g., (@32:11) more car traffic caused the elimination of the trolley/tram/streetcar on the Brooklyn Bridge so that another car lane could be added which decreased the services provided:
> If the job of the Brooklyn Bridge is to move people between the two boroughs, the reallocation of space from transit to cars has been disastrous. In 1902, one year before the photograph was taken, the Brooklyn Bridge moved roughly 341,000 people a day across all its modes, according to the Federal Highway Administration. It hit its peak capacity a few years later, with 426,000 people using it each day in 1907.
> Today, 125,000 motor vehicles cross the Brooklyn Bridge each day [PDF], as do roughly 4,000 pedestrians and 2,600 cyclists. For the bridge to carry as many people as it did at its peak, each of those cars would need to carry more than three people, but they do not. In 1989, when the city counted around 132,000 motor vehicles crossing, the FHWA estimated that 178,000 people crossed the bridge daily.
Every mode of transportation needs space to operate, and by focusing on cars for private transportation it has decreased route capacity compared to every other mode, even walking:
> The noise and pollution concerns call for building robotaxis that are quieter and lighter-weight, which seems like what's actually occurring.
Noise is a function of speed, so any vehicle travelling >50 kph will cause noise: is there a desire to restrict speeds? If we're going with EVs for our future, the mass of the batteries for a given range can hardly be reduced.
> I admit his shallow dismissal of mass traffic fatalities as "marketing" and "corporate profits" immediately destroyed all credibility in my eyes.
And his other point (@13:02) that if American road designs were similar to (e.g.) Swedish road designs there would be a 80% drop in fatalities with current technologies? No (possibly) pie-in-the-sky tech needed. Perhaps there could be benefits to autonomous vehicles (AVs), but mass traffic fatalities can be reduced now if the actual desire is there, e.g., Hoboken, New Jersey, has had zero traffic deaths over the last seven years:
AVs can be buses, so the point about "private transportation" is orthogonal to AVs.
It went unmentioned that noise also correlates with drag, and upcoming mass market AVs are far more aerodynamic than existing cars. Quantity of noise matters here, not just "makes [non-zero] noise."
If we want to restrict speed, AVs can reliably achieve that (vs speed enforcement of human drivers). Computers never get impatient! Again AV buses are left out in your analysis.
Restructuring all American cities to be like Swedish cities is far more "pie-in-the-sky," sorry to break it to you.
"Will destroy cities" remains vastly overblown given the actual concerns being voiced here. And again, zero credit for ultimately preventing millions of fatalities annually (which is literally "cities" worth of people being "destroyed"). Where is your concern for those people?
> AVs can be buses too, so your point about "private transportation" is orthogonal to the AV issue.
AV public transit mentioned @30:58. More general point is continuing (over-)emphasis on cars is inefficient and wasteful compared to other options, regardless of whether the cars are human- or computer-driven.
> You neglect to mention that noise is correlated with drag, and upcoming mass AVs are far more aerodynamically efficient than existing cars. Quantity matters here, not just "makes noise.
Even now, with perhaps less aerodynamically efficient cars, tire noise is the largest source, so if aerodynamic efficiency goes up then tire noise will become a larger proportion:
> continuing (over-)emphasis on cars is inefficient and wasteful compared to other options
Again I don't disagree, but if it were really so easy we would have done it decades ago. It's not like this is a new idea.
Quite ironic to call AVs "pie-in-the-sky," then propose a gargantuan fixed infrastructure overhaul (which has been consistently mired in inaction for decades) instead.
> tire noise will become a larger proportion.
Who cares about "proportion?" Absolute quantity is what matters here.
Anyway I think this has played out. Good discussion, cheers
The video points out that the Netherlands only began this "gargantuan fixed infrastructure overhaul" in the '90s, and until then Dutch cities were car-centric and looked pretty much just like North American ones, so it shouldn't really be that hard.
> Again I don't disagree, but if it were really so easy we would have done it decades ago. It's not like this is a new idea.
It was done decades ago… like in the Netherlands. As he explains in the video (@43:28), Utrecht and "Fake London" (London, ON, CA, where he grew up) have similar populations (300K vs 400K), and at one point were about equally car-focused. But one city deemphasized car use and the other continue to emphasize it.
Similarly Amsterdam switched focus and now is a poster child for active transportation:
In contrast, many US cities are enacting policies and street designs that make things car centric (discussed @34:16) and hostile towards humans. Contrast that with designs that don't hinder humans as much:
> Who cares about "proportion?" Absolute quantity is what matters here.
If there is X dB of noise, and (say) 80% of it is produced tire-road interaction, there is a floor that it cannot go below. And if "More traffic = more transportation services provided" is desired as you stated above, the proportion reduced by aerodynamic improvements will be eaten by by more vehicles on the road, for no net gain.
And you'll still have non-noise pollution like tire and brake dust.
> Quite ironic to call AVs "pie-in-the-sky," then propose a gargantuan fixed infrastructure overhaul (which has been consistently mired in inaction for decades) instead.
From a business perspective, why on Earth would you price it dramatically cheaper? Especially when you have to recoup R&D costs, need to show investors a profit, customers have already shown a willingness to pay market price, and maintenance on the cars and the in-car customer service agents isn't free?
And we see that today in the app - Waymo is often more expensive that Lyft or Uber.
In the long run, if there is competition, it should drive down costs. If the sensors / compute hardware / software costs more than the labor of a human driver then self-driving vehicles won't be viable, but the bet these companies are making is that it will, and in the long run, those costs will be amortized. And if the costs of offering a robotaxi service are lower than a human-taxi service, and if AVs are not a monopoly, then the price will eventually end up cheaper as well.
> He says it himself: self-driving cars are just cheaper taxis.
That's not what he's saying, that's what he's refuting.
When taxis are driven by different people competing against each other, taxis are cheap. When they're controlled by a single tech company that controls the entire market, they'll cost as much as the tech company can get away with.
But he makes a lot of other good points too; that it will mean even more cars on the road.
I too normally like this channel but a few minutes in and it started to sound whiny. Encouraging “coning” was pretty low. It is starting to become pretty obvious that the many benefits of progress in this area overwhelm minor grievances.
This dismissal doesn't do justice to the content of the video. There are a lot of other points made, including:
- Self driving cars are not a pipe-dream, the tech is developing and will be in every city in the future, this can't be ignored
- There is a real danger that they will destroy the fabric of our cities
- They will increase, not decrease, the number of cars on the road and the use of road space
- As cars shift electric, gas taxes will be less effective at "charging for usage" of road space, and other means will be necessary, or else AVs will treat the cost of occupying road space as zero, and circle endlessly.
- Despite hype that AVs won't need parking space, they will, because people will still want to own private cars and need to park them, and robotaxis will compete for response times and so will need to stay available close to high-demand areas
- As AV companies gain adoption and become more available, city design may warp such that AVs may even become required to use roads or get around at all. There's a strong potential for the AV industry to lobby for changes to road infrastructure to favor AVs, including banning human-driven vehicles and increasing speed limits or removing them altogether. This may even go to the point of fencing off streets from pedestrians, because otherwise it would be too easy for people to step in front of vehicles (forcing them to brake). This is not unprecedented based on how quickly the auto industry reshaped laws, standards, and infrastructure to favor automobiles in the 20th century.
- AV companies are likely to spread American driving practices and road culture to other countries, because AV tech is mostly developed in the US.
- AVs may make it more difficult for cities and countries to update road rules, because of the burden it would put on AV companies to keep up with rule changes, and AV companies might successfully lobby against this. Cities will come under pressure to become more uniformly like Phoenix AZ, in order to be more AV-friendly, rather than AV companies putting in the work to localize their tech to work in every city.
- There is a risk that AVs will displace other forms of public transit which are better for the environment, for pedestrians, for land use, and for the surrounding areas
- By reducing the perceived cost of spending time in traffic, AVs will encourage people to commute longer distances, resulting in more traffic, infrastructure costs, and emissions overall
- The AV lobby has suggested that AVs help mobility-impaired people, but public transit and walkable streets would help these people better than AVs would
It's a long video, but I do think there are a lot of points that are worth deeply thinking about, especially for anyone here who might be involved in developing this technology.
>- There is a real danger that they will destroy the fabric of our cities
This is just not true, I far prefer to bike next or in front of waymos on shared roadways because they keep proper distance and are predictable. Far from destroy the fabric, they are like benevolent sea turtles, providing a cover of safety for more vulnerable road users.
>They will increase, not decrease, the number of cars on the road and the use of road space
Possibly true but let's see some evidence.
>As cars shift electric, gas taxes will be less effective
Didn't SF just pass a tax on AVs? Easy fix.
>- There is a risk that AVs will displace other forms of public transit which are better for the environment, for pedestrians, for land use, and for the surrounding areas
Waiting for a train sucks, people want to go point to point.
>- By reducing the perceived cost of spending time in traffic, AVs will encourage people to commute longer distances, resulting in more traffic, infrastructure costs, and emissions overall
Same argument can be made for public transit / trains. It's not a good argument.
> they are like benevolent sea turtles, providing a cover of safety for more vulnerable road users.
Hopefully they stay that way! But right now they are in their testing phases and the #1 priority of AV companies is to demonstrate their safety and obtain trust and social license to operate. The argument made in the video is somewhat more dystopian, of a future where AVs are entrenched and society becomes dependent on them, and they have secured enough of a foothold to start raising speed limits, removing stoplights, banning non-AVs from roads, etc. It's only one possible future, of course.
But it's a common trajectory, where initially a service is introduced very benevolently, and only once it's secured traction do they start to turn the screws. I'm sure we can all think of some examples...
> Possibly true but let's see some evidence.
The details are in the video; you can watch it if you'd like, and refute its claims with counterarguments.
>- By reducing the perceived cost of spending time in traffic, AVs will encourage people to commute longer distances, resulting in more traffic, infrastructure costs, and emissions overall
> Same argument can be made for public transit / trains. It's not a good argument.
The same is actually true of trains, except that the cost to society is much less, because of how much more efficient trains are. No matter how many people commute by train, you never need ever-widening train-highways to alleviate the traffic jams from too many people riding the train. It's instead a positive feedback loop of increased ridership leading to higher train frequency, increased service quality, better speed, better network connections, and a stronger and more walkable area around and between train stations. It's exactly the opposite of cars, but not because the argument is wrong. It certainly does lead to more suburban development (and lower costs of real housing), but this becomes an upside rather than a downside.
I don't buy the raising speed limits thing because the tech is actually much easier to regulate and enforce than human drivers.
I guess the example is Uber/Lyft, but remember that those went to a vote and the vote passed. And if non-AVs are banned from cities, wouldn't our cities be much more bikeable and safe? How many pedestrians are killed again by human drivers?
Trains are not magical, they also suffer from congestion issues, in fact moreso than roadways because they generally cannot pass each other. If a city were pure AVs, you could absolutely have train-size AVs for key arteries/events and still enable much more flexible point-to-point travel.
AVs can and will have pooled ridership in the near future.
Trains can pass each other! I see it happen all the time in Japan.
It's easy to enforce speed limits on AVs, but also easy for AV proponents to argue that speed limits are obsolete and unnecessary when cars have millisecond reaction times, a legacy from back when sluggish humans piloted vehicles around. It's hard to say what the future may hold.
It's crazy that people are so pro-train. Imagine if they were invented now, the idea of building train tracks throughout cities would be laughed at. Trains cannot turn to avoid someone and their stopping distance is awful making them incredibly unsafe unless operated on tracks that completely exclude humans.
If trains were an AV they'd never be approved. The only thing they have going for them is passenger density but that is now easily achieved with battery electric vehicles.
The smart thing to do is actually pave over existing tracks and allow high-density AVs exclusive use to those thoroughfares. You could achieve much higher utilization and have routes that go more direct since they could exit the thoroughfare to load and unload rather than block it.
Trains are vital to a point that the government instantly shut down a train strike over paid sick days. They are still the best cost value for transporting cargo long distances.
Trains with current car-enabled cities would be laughed at because they are car enabled cities. Yes, you would be laughed at for using the wrong tool for the job. That's why these arguments also involve reshaping the job itself to be more efficient. It's still a popular idea for Californians to go from Los Angeles to San Fransisco three times as fast than any car ride could hope for, so it's not like people don't see the immediate benefits of how trains can bridge together cities.
>The smart thing to do is actually pave over existing tracks and allow high-density AVs exclusive use to those thoroughfares.
ignoring the details the video already counters against this: This would be a political landmine on multiple fronts. For the exact same reasons the high speed rail project above is taking decades instead of maybe 1 decade.
It's rare that I encounter such a fascinatingly different viewpoint to my own that I'm staggered to consider how it could have been arrived at.
Trains don't need to turn to avoid people because their motion is entirely predictable. You can stand one meter away from a passing train with no fear, because you know exactly what its trajectory will be -- it's built into the track under your feet. This is why people are comfortable walking next to trams in crowded areas, whereas they have to give cars a wide berth -- you don't know if they're about to steer into you. For this as well as several other reasons, trains are orders-of-magnitude safer than road vehicles. Especially when weighted by passenger-miles, because then we're comparing the collision frequency of one train against that of hundreds of cars (or even a thousand -- one shinkansen train comfortably carries about 1300 riders).
And yes, because the passenger density is so high, noise level is low, and the traffic is so intermittent and predictable, this does allow the construction of both separated-grade tracks (whether elevated or subway), and also protected crossings for at-grade trains. The safety level is so high that people rarely get hit by a train by accident -- more often suicide or crime is involved. And better safety features, like barriers at stations, can go along way to reducing these remaining risks. It's hard to imagine these kind of safety features being feasible for AVs.
Battery electric vehicles do not and will not come close to the passenger density of trains. Trains also afford higher comfort (it will probably be a long time before robovans provide washrooms with running water!).
By moving between fixed stations, investments in fixed infrastructure can be made with confidence to accommodate the high throughput of people. The routes naturally become direct not because trains stop at every single block, but because travel destinations naturally move close to the stations rather than sprawling out. And because people stepping off a train don't bring a car with them, these areas are intrinsically walkable and pleasant, and don't need to be cris-crossed with high-traffic roads that consume much of the available space.
If you ever have a chance to visit Japan or the Netherlands, let me know if you update your ideas about trains.
I haven't been to the Netherlands yet, it's on my list, but I was just in Japan a few weeks ago.
The thing that surprised me most about Tokyo was actually how generally crowded the trains were. It seems obvious that they should be running near twice as many trains in general and especially during rush hour. The tracks were still mostly empty compared to streets. While everyone was super considerate, it was still relatively low comfort traveling with a kid because of the uncertainty around getting a seat.
Why couldn't we build barriers along AV roadways and code them to drive an exact path? It's actually much easier to code an AV to act like a train than this freeform driving that they are doing. They could go into train/convoy mode to utilize the train infrastructure then back into car mode to take a side street.
You can easily put bathrooms and running water on AVs if it makes sense.
I do agree about having the nice walkable areas near stations, but was somewhat surprised that car right-of-way was still super dominant in Japan. Like the Shibuya Crossing, you have thousands of people all scrambling across then cramming onto the sidewalks while vehicles get 70% of the right-of-way timeslice. I'd actually flip that so that peds are the primary street users and AVs must yield.
Did you watch the entire video? His point that once mass adoption happens, the risk is that you won't be able to easily go without them in many places.
Ever heard of Dallas, Houston, where you literally can't walk 500m from where you are, because you're meant to drive 3km to get there, instead.
> This is just not true, I far prefer to bike next or in front of waymos on shared roadways because they keep proper distance and are predictable.
Safe biking is already being done. Instead of having a waymo, having good street design and more people using the roads in things other then cars makes it far more save.
If its a low speed road, driving with human on it is pretty save. If its a high speed road, you shouldn't share it with a car anyway.
> Possibly true but let's see some evidence.
Even if you can't prove its more, its unlikely to be significantly less. And even say 20% less, would still be far to much. Utrecht has a much lower number of car trips compared to Fake London without AVs.
So at best, you are still far worse then simply adopting better urban design.
> Waiting for a train sucks, people want to go point to point.
Waiting in traffic jams sucks as well.
People care about what is faster and cheaper. At scale, on aggregate a proper transportation system is more efficient, faster and cheaper overall for society. If people actually had to carry the true cost of their 'point to point system' they wouldn't want it.
The US with their amazing 'point to point' system has hilarious amounts of traffic jams and cities with incredibly long commute times. Even while being far less dense, far less efficient and owning car is more expensive then a city wide public transport ticket in a comparably sized city Europe.
> Same argument can be made for public transit / trains. It's not a good argument.
Same dumb argument people make about induced demand. You want to encourage public transport because, its environmentally friendly, highly space efficient, on aggregate highly cost efficient and leads to urban design that is incredibly for living quality.
Encouraging people to use the least efficient means of production is bad, encouraging people to use the most efficient means of transportation is good.
Talk about not actually listening. You are scratching the surface ignoring everything else in the video.
> zero attention when humans do them all the time
Literally his whole channel is about the problems with humans. This one video is about AVs. That it doesn't get more headlines in general is part of the problem.
The more important point that you ignore, is that we already know how to make things much saver.
> and self-driving cars don't suddenly fix that (something which was never promised or expected?)
Except the companies and all the fans who constantly claim that self driving will be better then all public transport and solve future urban problems.
To say that nobody makes such claims is just a strange ignoring of reality.
And you again, ignore the deeper point that self-driving cars reenforce these problems and potentially make things worse.
> - whenever you mention reducing millions of human-caused traffic fatalities, use a sarcastic tone of voice
Not sure what your point here is. The point of the video, partly is that we can already reduce the traffic deaths by far cheaper means then AV and nobody cares about that.
So all these claims about the amazing benefits (that hasn't been delivered) is mostly just hiding that what these companies really want is profit and the politicians want campaign benefits from these companies, rather then improve safety in the city.
Again, the point being that AV as currently sold and advertised re-enforce existing problems, rather then solve them.
> He says it himself: self-driving cars are just cheaper taxis.
He didn't say that. He said, 'at best they are cheaper taxis', but at worst the have many non-beneficial impacts on society.
And more important, there are simply better solutions.
Are AVs perfect? Nope but they might already be safer drivers than the median driver in SF. There have been 310 pedestrian deaths in SF over the past 10 years. Number of deaths the NHTSA has attributed to full self driving vehicles? Zero.
reply