I though so too (what the title says) when I read the Wired article referenced, but I realise now, reading the article above, that I have no idea how many "robotaxis" [1] are in San Francisco. The interesting statistic -that the Wired article didn't mention either, as far as I can tell- is how many such disruptive incidents are caused by number of robocars circulating, or by number of miles driven by them, or some other relevant ratio; and compared to ordinary cars, of course.
It's the same way we consider accidents caused by robocars. Not in terms of absolute numbers, but in terms of relative ones.
- "K Ingleside blocked near Ocean and Capitol by a non-Muni collision."
- "N Judah blocked at 8th and Irving by an automobile on the track way."
- "Non-Muni collision at 16th and Mission. Expect delays and reroutes for the 14/14R, 22, 33, 49, and 55."
Those are just human automobile-related delays in the last day. Lots of others - police activity, fire hoses, Muni vehicle problems, signal system failures.
For comparison: Vienna has a large tram system and their trams are block by other vehicles four to five times a day on average. This number is going down because of changes to street design so car drivers are less likely to park their car wrong. https://www.derstandard.at/story/2000145150263/anzahl-der-fa...
Agreed, this doesn't give a relative magnitude nor an accurate comparison to comparable services operated by humans nor against vehicles operated by (potentially impaired) humans.
I encourage anyone interested in the topic of self-driving cars (this author calls them "robocars") to read more articles from this author. He provides a very reasoned analysis of most every self-driving car topic: Tesla sensors, safety, parking issues, business models, etc.
> I founded ClariNet, the world's first internet based business, am Chairman Emeritus of the Electronic Frontier Foundation, and a fellow of the Foresight Institute. My current passion is self-driving vehicles and robots. I worked on Google's car team in its early years and am an advisor and/or investor for car OEMs and many of the top startups in robocars, sensors, delivery robots and even some flying cars. Plus AR/VR and software. I am founding faculty and computing chair for Singularity University, and I write, consult and speak on robocar technology around the globe.
(Emphasis added)
This doesn't sound like the bio of a person who gives a fair shake at AVs. This sounds like a person who is quite literally invested in the success of AVs and is using the Forbes platform to lend legitimacy to his opinions.
To be fair, it also sounds like the bio of a person who's quite knowledgeable about AVs. But I'd rather get my commentary about AVs from somebody who doesn't have a financial interest in their success.
People don't want experts, who probably have at least some stake in a topic, writing about it. And they don't want some theoretically "objective" journalist or other outside observer writing about it. I'm not sure who that leaves. (Obviously some journalists and other writers are pretty knowledgable about the topics they cover but they're still often filtering and interpreting what experts have told them.)
Maybe I should have clarified, that quote came from the bio which you get if you click on the author's name -- not from the body of the article. It does appear in a little "show more" box (at the end of the article) but the juicy conflict of interest bits are obscured by the "show more" link. That's not clear disclosure by journalistic standards.
Journalists put relevant conflict-of-interest disclosures right at the beginning of the article, with emphasis added. This guy is doing the opposite: using the Forbes site to make it look like he's a journalist employed by them, when in fact he's a consultant/investor with the very industry he's writing apologist articles for. Sure, he's not lying... he put it in the bio field. But it's still misrepresentation.
About 10 years ago, my first non-academic job was with a start up causing a major stir winning major awards and was profiled all over the place for their work in food traceability. Once I started though, I realized everything was theoretical - in reality it could be used for food. But no one actually was and none of the applications they talked about were done using these systems.
Then one day I was called into a meeting with an editor at Forbes. And the conversation began with the fee and moved into topics I knew we couldn't pull off.... Which our marketing guy said to me later was normal and that all the articles and awards (from Red Dot to Fortune to Barron's) were effectively paid placement. No one checks, no one verifies, no one blinks an eye. "This is how you build the market and our story is compelling"
They never made a breakthrough in food because it never worked. This in spite of that Forbes profile leading to a CNN mention and eventual WEF appearance... In which no one asked to verify or reproduce our data
So the author may be a good SME, and I will check him out but take anything you read in places like Forbes for what they are: likely advertorials hiding the smoke and mirrors
Was kinda hoping for fast deliveries with drones by now instead we get self driving cars. Wish we could go for a future that’s car free in cities and services by drone delivery.
Walkable and safe streets.
Not complaining, but I just don’t really care for self driving cars personally. Seems like a waste of money when we don’t even have good streets and transit.
I'm definitely on the side of walkable and safe streets, but I guess I'm interested why you think drones would be a better option than delivery vehicles, or that they would deliver on the promise of a car free future?
Drones are loud, and it can't be avoided (they need to move air to provide lift, and doing so creates vibration - I.e noise). Cars are definitely loud but not on the scale of a swarm of drones.
Drones don't have the cargo capacity of, e.g. a moderately sized van.
We don't have the traffic management policies for low air traffic, and we do have those policies already for street traffic.
Most importantly, deliveries don't account for the majority of traffic: in fact, most commercial deliveries happen at off hours for traffic. If you eliminated deliveries in cities, you may get rid of some annoying double parked Amazon vans, but you wouldn't meaningfully reduce traffic. The majority of traffic is people in personal automobiles (usually one or two per several ton SUV).
We get walkable and safe streets by changing our building codes and parking requirements to make walkable neighborhoods legal to build. And by building useful public transit, at a local and regional level. And by incentivizing personal mobility options like (e-)bikes and scooters, while disinventivizing cars. But delivery vehicles, along with work trucks and vans: I have no problem with them staying exactly as is. If we get rid of the vast majority of personal vehicle traffic, then commercial delivery vans will be no big deal, and small deliveries will happen via ebikes and cargo ebikes. No need for drones!
> Drones are loud, and it can't be avoided (they need to move air to provide lift, and doing so creates vibration - I.e noise).
I think owls disprove that theory. If it can’t be avoided to create noise for flight then how do they do it? After all they create lift, yet somehow avoid being as noisy as an equivalent size or weight drone would.
Ok, show me the owl silently towing a few cases of beer and I'll take that into account ;)
Edit: OK, on reread this is far more snarky than is deserved. Sorry!
I get what you're saying about the noise per unit weight! Clearly nature has potential ways to improve on the technology, though I have to wonder how much development is necessary until a drone could restock my corner store at the same noise level as a van does.
Overall, my larger concern, which should have been listed first, is that delivery traffic is not the primary cause of traffic on our roads. Eliminating it won't solve our transit issues in cities.
On the noise issue, I saw one company recently have the idea of keeping the main drone itself at a relatively high altitude, high enough that you can't hear it, and lowering its cargo along with an extra device for stability/maneuverability down below. Called zipline.
Of course, still far too small to be able to deliver e.g. a fridge, but can easily handle your typical food or small package deliveries.
Drones move in 3d space. Maybe a lot more investment would have yielded some cool ways to navigate in 3d space around cities. I know there are regulations around where and how drones can be used but in a city like SF it’s probably fine. Besides when have regulations stopped startups from defining something new and better?
Cars and trucks just take up way too much space. They also pollute. I like scooters but they don’t have a lot of range:time advantage. At least with drones you could deploy relay style movement so the drones could move items between multiple drones, of various types and capabilities, to get the item to the final destination. Or work in tandem with the scooter people.
> Drones move in 3d space. Maybe a lot more investment would have yielded some cool ways to navigate in 3d space around cities.
Well it turns out that human tend to enjoy the ability to open a windows and look at the sky while enjoying birds. This is hardly compatible with a cool way for delivery service to navigate in 3D space 24/7.
I once hiked on a remote part of a beautiful island where helicopters is the only viable delivery system because there is literally no road. Even with very limited and restricted helicopters delivery rotations for the small hamlets there the nuisance was noticeable.
Drone delivery only to sustain a megalopolis would likely fill the air with background sound not unlike an army of barking dogs.
>Drone delivery only to sustain a megalopolis would likely fill the air with background sound not unlike an army of barking dogs.
No, barking dogs would be far worse. Drones are just a constant buzz or whir, so they're basically white noise, though the noise gradually changes in pitch and volume as the drones move around due to distance and doppler effect. Dog barking is sharp and irregular. Sane humans find that far, far more annoying.
Having smaller deliveries by drone could be quiet and nice if they invent anti-gravity systems (or "suspensors"). But that probably won't come until after we have ornithopters...
> Drones move in 3d space. Maybe a lot more investment would have yielded some cool ways to navigate in 3d space around cities. I know there are regulations around where and how drones can be used but in a city like SF it’s probably fine.
In many iconic areas of SF it's illegal to use those video camera drones because of the disruption they cause, in terms of noise, and also disrupting beautiful scenery. And those are drones with nowhere near the capacity to carry a few hundred pounds of cargo. Do you want some quadcopter with a few cases of beer blaring their rotors past your bedroom window all night because there's "cool ways to navigate 3d space"? Personally, I don't.
> Besides when have regulations stopped startups from defining something new and better?
Maybe different strokes for different folks, but this is the exact attitude I actively dislike when it comes to city planning. "Just let the venture capitalists throw their shitty tech project into the city and let the city workers clean up the mess". It's not a good strategy for transportation planning. In fact, I'd far rather that elected officials make a weird and crazy decision with little community input (we can always reverse it and vote them out) rather than a VC funded startup trying to "disrupt" local transit in my area.
> Cars and trucks just take up way too much space. They also pollute. I like scooters but they don’t have a lot of range:time advantage.
Fully agree right up to the range/time thing. Electric mobility devices on the street, with even the shitty infrastructure SF has, are like a cheat code to transit. Traffic can't go that fast in a dense city, so scooters and the like don't need to go that fast to outdo, or at least rival them. And as for range: SF is 7x7 miles, it's a very limited problem in a city. Plus, Lyft is introducing 20+ mile range heavy ebikes for trips which may be a bit beyond the city.
> At least with drones you could deploy relay style movement so the drones could move items between multiple drones, of various types and capabilities, to get the item to the final destination. Or work in tandem with the scooter people.
And none of this solves the root cause of traffic: personal mobility. Unless we're having drones ferrying people Jetsons style, i.e. helicopter noise all over the city, then this is just a pipe dream that doesn't alleviate traffic, but ruins the cityscape and dramatically increases noise pollution.
Drones have a lot of useful roles in important deliveries (e.g. medicines) to remote areas, possibly farming, commercial photography with permits, etc. One of those useful roles is not delivering hamburgers in an urban environment to someone who doesn't want to get off their couch.
Sure, and I'm happy to see those uses get developed. :)
The original argument I'm pushing back against is that by replacing deliveries in cities, drones could deliver the car-free pedestrian/cycling utopia of our dreams.
>Was kinda hoping for fast deliveries with drones by now instead we get self driving cars. Wish we could go for a future that’s car free in cities and services by drone delivery. Walkable and safe streets.
You don't need to wait for some sci-fi future to get this, you just need to move someplace where this (or something closer to it) is the reality, which means not-America.
Here in Tokyo, the "walkable and safe streets" part is the norm. There's still cars of course, and there's no drone delivery, but you don't really need those things to have walkable and safe streets. You just need high density, and infrastructure designed more for people than for cars (so, very small roads most places, very little parking, low speed limits, sidewalks on all streets except tiny 1-lane residential streets, etc.). And of course you need really good public transit. There's still roads, of course, which is how trucks get around to move cargo and do deliveries, and also how taxis get around, but overall "walkable and safe" describes this place perfectly.
It's not just Japan either: lots of cities in Europe have this as well. Go check out Berlin, Amsterdam, Copenhagen, Munich, Prague, or many more.
In a nutshell, if you want "walkable and safe", you need to move out of the US.
Robots are not failing in a way that emulates some teenager in his 90s shitbox doing donuts in the street. They are failing in a way that emulates some middle aged middle manager in her $50k SUV who stops at a cloverleaf because the traffic is intimidating.
The latter case is far more frustrating to most people because the person or robot is inconveniencing everyone not willfully, but out of their own inability to do something the rest of the driving public can do fine whereas the teenager doing donuts can be more easily empathized with by most people in a Falling Down sort of way. And unlike a semi truck that simply can't move fast enough to not inconvenience others or an elderly person taking a long time to go up some stairs the robo-car and incompetent driver don't have what most people consider a good excuse for their behavior.
I doo feel that part of what defines what people will and won't tolerate within a disruption to their lives is spectacle. People seem to love to rail against the ordinary, while they crane their necks and gawk at the extraordinary.
Nobody stops to peer into hole full of infrastructure and machinery that's turned a 4 lane road into 1.
Everybody seems to want to stop and marvel at the upside down trailer engulfed in flames though, despite it being an equal inconvenience to their commute.
I believe that you'll find the plenty of people complaining about both of them. Speaking as one of those people. The sideshows have woken me up at night on a couple occasions.
Reminder that forbes.com is mainly a blogging/self-publishing tool like Medium. The author, Brad Templeton, is not employed or paid by Forbes.
Edit: forbes.com/sites/ is not a self-service as I thought. Though not full employees, it looks like there is some editorial vetting of Forbes "Contributors".
Publishing an article on Forbes costs a couple hundred bucks. A bunch of "contributors" sell a spot on SEO forums. I was doing a SEO optimization for my friend's website and a black hat agency sent me a price list with pretty much all known big media websites listed in the spreadsheet.
That is crazy. What a fall from grace for Forbes. It is possible to write regulations to prevent blackhat SEO? I keep thinking about it. Why don't all websites eventually rent their reputation to the highest bidder, but in a sneaky blackhat SEO way? I think many do. <sad face> The modern web feels like 50% great and 50% terrible. It is a hostile landscape to navigate!
Twenty years ago, Forbes was considered an excellent business magazine in the United States. For a while (last few years), I was "fooled" when I saw yet-another click-baity / poorly-researched Forbes piece. Then someone (here?) pointed out that Forbes has the "contributors" section with far lower editoral standards (err... asymptotically approaching zero!?). The contributor stuff is just rubbish. I think blocking it on HN is too much, but, please, stop upvoting this crap from Forbes contributors.
I do think Forbes sold its soul for clicks and I don't have any first-hand knowledge of how their contributors and other third-party content works these days.
That said, most of the trade press for example uses outside contributors--sometimes paid, often not. As I wrote elsewhere, there are various problematic aspects to the arrangement but there's usually enough editorial oversight to present blatant conflicts of interest around promoting a company, etc.
This can't be reiterated enough times. Forbes articles, in my mind, are automatically bunk now, since they accept minimally vetted blog posts from their "contributors" who are just looking for SEO and a boost from the brand name.
Not that Wired is some bastion of perfect journalism, but it's funny how this blog post levels criticism at Wired with little additional data, and simply says "I think they're wrong, because I think AVs will be better one day, so it's worth letting them screw up all over our cities for a few years".
Forbes pays contributors, who do so by invitation by a Forbes editor who likes their writing. Forbes also allows companies to pay to put in stories, but they are clearly marked as such.
It's amazing what people will write by assumption without knowing.
And yes I believe in the value of the technology. But also in the truth which is why I was very harsh on cruise in the story earlier this week.
> The easiest way is to know what you're talking about, write often and follow the best practices of news craft (i.e., pick up the phone and call people, be clear and don't make shit up). Our channel editors and staff writers are always on the watch for people they trust, would read and would want their posts adjacent to. It is an evolving vetting process. I've turned people down who are mostly issue advocates or run non-profits. This kind of mouthpiece stuff is what made HuffPo unreadable. We've parted ways with contributors for writing shoddy, libelous posts. We've taken people down for flogging their own newsletter, investment advisory business, PR agency, whatever. We have economists, financial advisors, lawyers, environmentalists, lockpickers, liberals, libertarians, China watchers, NY tech scene watchers, people covering only celebrity tweets. We like freelance journalists, who only have themselves to flog. I would personally rather sign someone up who wrote short every day than someone who wrote 2,000 word posts three times a month.
(Bruce Upbin)
So I don't think it's accurate to describe Forbes as a self-publishing tool.
> Hmm, if they're not paying him, why does he just put links to these articles on his own blog, instead of copying the whole article
Because Forbes also still just about has enough cachet for people to want to associate their subject matter expertise with it - far more of a motivation than contributor fees for someone with Templeton's background and earning power - and probably also has rules around duplicate content
It's closer to Huffington Post than Medium, but I think the point being made is that there's zero editorial oversight in between the author drafting the post and it appearing on Forbes' website, just the possibility that someone at Forbes might remove "shoddy, libellous posts" at a later stage. I don't think Forbes would have any issue if Brad thought cheap sensationalism was a better way of promoting himself to the Forbes audience (and I doubt he'd quit writing if he wasn't potentially able to earn as much as $50 per blog!)
I don't think "zero editorial oversight in between the author drafting the post and it appearing on Forbes' website" is equivalent to "mainly a blogging/self-publishing tool like Medium", given the large amount of editorial oversight evidently involved at other stages of the process. And, though I agree that Forbes's cachet is more of a motivation than contributor fees, "The author, Brad Templeton, is not employed or paid by Forbes" is evidently a flat-out lie.
Based on what I knew from about five years ago, "it's complicated."
At least at the time, brands could buy a slot. I wrote for a company a couple of times. I think the pieces were very lightly edited.
There were/are also "contributors" who were mostly individuals who didn't have gross obvious conflicts of interest in the topics they wrote about. I think they either wrote for free or were paid some nominal sum. Again, I think there was light editing.
The latter is also what you saw with the CNET Blog Network which I did write for over a number of years at a time when everyone was branding and blogging including journalists and analysts. (I declined to accept money for various reasons. Other people were paid a modest amount.)
The problem with these things is that there are always going to be conflicts of interest and biases to various degrees. Some organizations try very hard to weed out direct examples of the former at least outside of explicit op-ed pages. But it's hard to manage. To give a current example, maybe you're bullish on generative AI because you work or consult for a company doing it no matter what reservations you have. Or maybe you work for the company because you're bullish on it. Or back in the day Microsoft's a client of your's because you really like what they're doing and they engaged you and passed on the Unix expert who didn't think much of what Microsoft was doing.
CNET got increasingly uncomfortable with the whole model and dropped it.
>Hmm, if they're not paying him, why does he just put links to these articles on his own blog (ideas.4brad.com), instead of copying the whole article?
Probably because of some combination of his deal with Forbes doesn't allow him to do so and it's often better to have a post/article in one place where it gets the most traffic. (There are other reasons why you might post in multiple places if you can, but SEO is probably not one of them.)
It's a little more opaque and less open than I thought. Not as self-service as I thought. Here's one description from https://joshsteimle.com/writing/become-forbes-writer.html : "Contributors at Forbes are unpaid writers, domain experts with day jobs, as opposed to staff writers who are full time employees of Forbes"
Brad is a robocar supporter ever since he worked with Waymo. He's fair, but he has little patience for anything but full speed ahead rhetoric, he's one of the true believers that think "humans are such bad drivers that virtually nothing robocars do can possibly be as bad as letting humans drive even a year longer than absolutely necessary". So it would be rare to get a critical viewpoint from him.
> according to city transportation authorities, who collected the data from social media reports, 911 calls, and other sources, because companies aren’t required to report all the breakdowns.”
Seems like making companies report all break downs will allow to reason about progress they are making better.
During the FTX fallout, I saw many articles share here -- some from major and others from mid-sized media websites. Across the board, they were mostly worse that any reporting from Financial Times (ft.com) and their blog FT AlphaVille (ftalphaville.ft.com). FTA "guerilla analysis" is so good that it breaks major scandals 2-5 times per year. (See Wirecard in Germany.) So, no, not all big media websites. Plus, The Guardian (UK) does some excellent reporting in the United States on touchy subjects that domestic media does want to touch. (NZZ Neue Zürcher Zeitung is legendary in German language media for the same type of coverage in Germany and Austria.)
I hope so, but for some reason, I doubt that small cars will catch on soon, whether driverless or regular.
They make a lot of sense economically and ecologically. Despite that, companies have been building things like Mircolino and Isetta for a while and no one seems to be buying.
The post you’re replying to is talking specifically about driverless (probably electric) taxis, which will likely be a pretty different market and form factor than anything which exists today.
Big cars are seen as a status symbol even if one is being driven in them. Even modern electric cars are big. I’m not sure going driverless will change much.
Seems impossible to tell without comparable data on humans. I'd assume 99%+ of all miles are still by humans? Unless you have that base rate who knows.
In 2022, human drivers in San Francisco killed 20 pedestrians and another ~25 people in vehicles (probably motorcycles and scooters). I think one cyclist.
If the human to software driven mile ratio is greater than 50, then the expected number of fatalities for software driven miles would be <1 right now. So the data is insufficient.
Is "robotaxis" the word we're going with here? Because that sounds more like it ought to be the name for a subfield of robotics relating to optimizing the physical arrangement or orientation of robots...
Chemotaxis is the behavior of a small organism to follow a chemical gradient to find a higher concentration. Likewise, thermotaxis follows temperature. It took me several rereads to realize that robotaxis wasn't some kind of behavior that followed a "robot gradient".
In India an "auto" is a three-wheeler autorickshaw, commonly known in English by its Thai name tuktuk. The "auto" here means it's motorized, as opposed to the original human-powered vehicles, this being the literal meaning of the Japanese jinrikisha (人力車).
And I'm happy to entertain the calls of "where's the data" in response to anecdotes. But the thing is, this article just takes the same information reported by Wired and says "it's really not that bad [presumably compared to human drivers], and even if it is that bad, I don't care because I hope AVs will be better than humans someday(TM)".
What really matters here is how disruptive the AVs are in terms of minutes of public transit delay per vehicle mile traveled (VMT), compared to human driven vehicles. Of course human drivers cause the majority of incidents on SF roads: they're the vast majority of VMTs, and they are the ones permitted to drive during the daytime that makes up the bulk of Muni service. Until there's data on real SF streets saying that AVs are causing less delays per VMT, then IMO, the criticism can't be ignored. Especially since we've been lead on by AV companies insisting that autonomous technology would be safer (and presumably less disruptive) than humans. We need to take their assertions with a grain of salt, given that they're trying to extract profits on our streets.
And plus, there's a whole area of the original wired article that is much harder to contradict. Dealing with the disruptions caused by AVs is so much more demoralizing than dealing with a human. A human realizes they've made a mistake, moves their car to the side, and awkwardly waves people by. An AV throws on its hazards and waits in the middle of an intersection until its "handlers" can be bothered to intervene. It's a faceless and inhumane experience that will never be quicker than a human driver making an innocent mistake. No amount of data can change that.
SF Muni has reported on at least a dozen instances where a transit vehicle has been blocked from proceeding by a driverless vehicle. That’s only “surprisingly little” if one had expected widespread chaos.
It only takes watching the widespread chaos of Ubers murderous testing program and Cruise's brilliant performance during the last wind storm to come to expect widespread chaos from the chaos monkey techbros who decide what algorithms risk our lives on the public streets.
(I am anti-robocar because I am a professional software developer. I know what the sausage is made of.)
Cities need to become denser to accommodate the growing population and reduce footprint. Efficient public transportation (trams/subway) are no longer regulatory or financially feasible enough to build (how is the second avenue line in Manhattan going?). Robotaxis and ad-hoc buses are a potential solution.
Autonomous cars are a localized regulation, so can be adopted gradually. The cities that will innovate will reap the benefits and those that will still dedicate their human resources to driving will lag.
> Efficient public transportation (trams/subway) are no longer regulatory or financially feasible enough to build (how is the second avenue line in Manhattan going?).
Uh are you sure about that? I don't think self driving cars are that more efficient than just regular human drivers. Giving up and accepting mediocrity (but fully automated!) doesn't seem like a very compelling argument
I think you need to compare it to overall number of incidents. I personally have seen many more blocks from double parked Amazon delivery trucks than driverless vehicles.
This analysis fails pretty dramatically on the size (46 square miles of San Fransisco vs. ~3.5 million square miles of the US) population (810k of San Fran vs. ~330 million of the US) autonomous cars percentage (not doing that lookup) and scale of incident (blocking public transit vs. lethal crash).
Agreed but the point is humans are terrible drivers and we need to solve that. We basically accepted a covid level death toll on the roads each year. It will be tough to do when seemingly every major media company has set the bar for robotaxis at “perfect from day 1” when humans themselves are absolutely atrocious at driving
> We basically accepted a covid level death toll on the roads each year.
That's not even close to true. COVID killed at least an order of magnitude more people than road deaths (and that's not having an end date for COVID, just averaging everything over the past 3 years).
Humans seem fine at driving
For every 100,000,000 miles traversed, about one person dies. The latest Tesla FSD total distance numbers I saw were about 35,000,000 miles in total (but that number is 10 months old).
Humans are involved in accidents once every 19,000,000 miles. In December 2021, FSD was involved in accidents once every 4,400,000 miles
Human drivers do dangerous, stupid, and illegal things all the time. I was in an Uber today that drove around bollards to go down a blocked off street. Also today, my wife was driving and witnessed a road rage incident where two cars were over taking each other aggressively in the city and throwing bottles out their windows at each other. A few months ago I walked out of my house to find a drunk driver getting booked for driving over the curb and into a public park.
Self driving cars, on the other hand, are always courteous in my neighborhood, stop at stop signs, and never speed.
Not only that, the expectation for safety in automated devices is much higher. The benchmark shouldn't be "as safe as cars driven by Californians"; it should be "as safe as elevators in California, per mile."
The median driver is a risk to his own life as much as the life of other motorists. The average robotaxi is not only an imperfect driver, but also has no concept of its own mortality, which is not the case for almost all the worst human drivers. The benchmark for "safer" is a lot higher than you think.
None of those had a photo of the tunnel entrance… why is there a rail tunnel that’s easily accessible to cars? Here in DC, the Metro is completely segregated from auto traffic. Is the SF rail running on streets?
Oh, well I can totally see how a car would end up there if it was dark, stormy, or the driver was drunk/high. We gets cars on the local bike path all the time - usually a drunkard, but sometimes just somebody who thought it was a driveway.
It's the same way we consider accidents caused by robocars. Not in terms of absolute numbers, but in terms of relative ones.
________________
[1] jameshart I'm looking at you :P