Happy Monday everyone. Hope you are all looking forward to the week. Remember Bike Happy Hour is back on Rainbow Road Plaza on Wednesday. I’m eager to see how that spot feels after our winter away and hope you can join us to meet the folks behind the upcoming Filmed By Bike festival!
Here are the most notable stories that came across my desk in the past week…
Traffic stops: The Trump Administration coupled ICE officers with the Tennessee Highway Patrol for an enforcement action that targeted drivers’ immigration status. (AP News)
Tour de France fundraiser: A Beaverton man will ride all 21 stages of the Tour de France as a fundraiser for his mom’s medical condition. (Kerry Eggers)
Silver tsunami: Is Portland ready for a large wave of aging residents who might not want to — or be physically able to — drive a car into their golden years? A noted voice in transportation says cities need to prepare for a low-car future when older folks stop driving. (Streetsblog USA)
Way-mo safer?: I’m very skeptical about robotaxis, but with every year they are on the road, I feel like I need to at least be open to the data. And a major new report from Waymo shows very clearly their cars are safer for non-drivers than human-driven cars. (Mashable)
More transit funding: Transit advocates and a growing group of lawmakers in Salem say the proposed framework for a statewide transportation package doesn’t go nearly far enough. (Portland Mercury)
Transit talk: Portland-based transit consultant Jarrett Walker gives a very informative overview of the state of public transit in the U.S. in this podcast episode (transcript available too). (Volts)
Congestion pricing success: The evidence is crystal clear that charging people to drive cars into Manhattan is a resounding success. Now we just need replicate it in more cities — understanding that it’s only possible when driving alternatives are widely available and competitive. (NY Times)
Automated camera politics: Interesting dynamic around automated traffic enforcement cameras going on up in Seattle as local and statewide policies evolve in an attempt to improve traffic safety without police. (The Urbanist)
Local influence: What could help with the governance of big transportation projects? A greater understanding of how local elected officials can sway the outcomes and a statewide mandate for a quicker timeline to force key decisions. (Urban Institute)
Thanks to everyone who sent in links this week. The Monday Roundup is a community effort, so please feel free to send us any great stories you come across.
Thanks for reading.
BikePortland has served this community with independent community journalism since 2005. We rely on subscriptions from readers like you to survive. Your financial support is vital in keeping this valuable resource alive and well.
Please subscribe today to strengthen and expand our work.
I’m guessing “might now want to” was supposed to say “might NOT want to”?
The USA and Portland desperately needs to move away from charging for parking (which creates very strong incentive for city governments to remain automobile-centric) to charging for the privilege of driving in urban areas and the decommissioning of public (and private) parking. Sadly, an unholy alliance of libertarian urbanists and short-sighted non-profits scuttled Portland’s proposed congestion fee. And not surprisingly, this market urbanist-centric blog pretty much ignored this very sad state of affairs. Example #1225 of how libertarian market urbanism leads to poor health and equity outcomes in urban areas.
Charging for parking is only market urbanism insofar as the parking charges are framed as creating an efficient market in car parking. It would be hard to look at Portland’s parking districts and conclude that is the goal based on their limited hours, low prices, and poor enforcement.
I think using whatever means we can to reduce auto dependence in urban areas is great, so would be fine seeing parking costs increase and congestion pricing. But doesn’t congestion pricing also introduce a perverse incentive to the local government implementing? Like they gain more revenue the more cars that enter at peak times – how does that meaningfully differ from the parking discussion? I view congestion pricing as a much purer distillation of market-oriented solutions than parking (though this is probably because parking policy feels inherently wonky and a bit silly, while tolls elicit a different kind of reaction from the public).
Sooooo … the very essence of Schoupian parking reform.
Some “means” are proven to generate transformative reforms (e.g. whole-sale removal of parking and congestion pricing [or bans]) and some are just incrementalism (greenwashing) that unintentionally or intentionally end up reinforcing Fordism.
My point is that Portland’s parking policies are not interested in creating an efficient market in car parking based on how they work in practice, even if there is some Shoupian influence in the history of parking districts in the city.
Some reforms are better than others, but I hardly think there is any proven solution to our current automotive dominance. No other country so forcefully adopted the automobile, and no other country so thoroughly dismantled its transit systems. Solutions that worked well in one historical context may not work well in another. And I see no actual citations, proof, or discussion on why a market on car driving is good (congestion pricing) but a market on car parking is bad (Shoupian parking refom).
I prefer bans and I should have made that more clear. Of course my preferences are meaningless in a society that appears to be becoming even more automobile-centric (if that’s even possible),
I agree. Many see a conspiracy, but I see utility, convenience, and efficiency at work. Do you think that imposing bans on driving/parking/whatever is a feasible way of transforming society moving forward from where we are today?
Or, from a different perspective, perhaps individualized, motorized transportation is what people really want (look at the displacement of bikes by larger, motorized vehicles, even amongst diehard cyclists), and we should instead focus on reducing the damage done by an increasingly motorized future?
What are examples #1224 and #1226?
I’m kind of curious what example #1 was which I can only assume set them on their interesting path.
That was a Babylonian around 2000 BCE who, to the horror of all around, etched his signature into a clay tablet which claimed his and a small portion of his neighbor’s porch.
Example #1224 here: https://bikeportland.org/2025/05/12/monday-roundup-trump-and-traffic-stops-silver-tsunami-waymo-safety-and-more-394308#comment-7543790
Example #1226: “no meaningful alternative to…” https://bikeportland.org/2025/05/07/port-unveils-plan-to-smooth-traffic-to-and-from-pdx-airport-394244#comment-7543484
Until it is clear that the charge to breach the city limits with an atomizer of literal and figurative toxicity (masculine is what we’re most worried about), is equal to 100% of income above the poverty level for anyone presenting with oppressor characteristics (cis, white (if only there was lower lower case for the ‘w’), male, neuro-convergent, abled, adequately visioned, excessive musculature, well kempt, etc.), I am going to hold out hope for a better solution on congestion pricing. As all biters of hands that feed, we must unionize our allegiance to eliminate all private enterprise and make sure everyone becomes a UN approved co-op citizen of urbanity, one on top of the other, side by side, shared walls and all.
**I am now adjusting my keffiyeh, raising a clenched fist and pushing “Post Comment”, in solidarity**
“Twee” comment of the week!
/big sigh of relief
I thought you were talking about me until I saw the well visioned part.
You forgot rediculously good looking though and Sigma 3 IQ though.
Oh sorry for the inconvenience, ridiculously good looking and statistically significant above median IQ will be sent to the data entry camp in the basement of Montgomery Park brought to you by TikTok, your friendly CCP data aggregator.
I agree with your intention, but doesn’t charging people for the privilege of driving keep the incentives exactly the same as charging for parking? They won’t want to do anything that discourages driving because it’ll turn off that sweet sweet money spigot.
Edit: nevermind, Blumdrew said the same thing.
My initial look at the Waymo article per ADT technologies: I would feel way more comfortable about the reported findings if these three issues had been handled differently and were not potential red flags of doubt: (1) Waymo had waited to issue the press release ‘celebration’ after the paper was published AND widely discussed publicly in peer circles; (2) the research team had not been only “in-house”; and (3) the results of the findings (pg 1) had used standard confidence intervals and were not instead relying on wider than ideal confidence intervals (87% to 99% or 75% to 98%)*. [* “The confidence intervals of the Airbag Deployment and Any-Injury-Reported F2R struck comparisons done in the current study are still quite wide in F2R Struck crashes, so it may be too soon to draw conclusions.” Kusano et al pg 44. There may be other potential issues found upon further review.
And that’s before even getting into the likely outcome where these vehicles incentivize more and longer car trips, so we can’t compare mile-for-mile either. The more cars we have on the road, the more of all associated negatives we will have to deal with, from crashes, to pollution, to the negative feedback loop of cars creating an environment where it’s unsafe and uncomfortable to travel any other way.
There’s also the issue that even *if* if were true today that these robotaxis were a net safety improvement (which I’m highly skeptical of for so many reasons including the above) – even if that were true, there’s ultimately profit motive to erode service quality (by way of minimizing ongoing investment) which will ultimately lead to worse outcomes. Just look at any big modern tech company (Google, Amazon, etc) – as soon as they captured the market, they started making their experience worse in pursuit of profit, assuming it wouldn’t cost them anything because people lacked alternatives.
We haven’t seen that kind of profit-driven safety decline (at least not as badly) in ex: aviation thanks only to intense regulatory oversight, which we’re unlikely to see for robotaxis for a wide variety of reasons.
There’s enough of the vehicles on the roads now that we can start to evaluate whether they are encouraging longer trips. I don’t know the answer, of course, but I would be very surprised if they did.
I expect robotaxis to be much more regulated than private cars are. For one thing, complete, systemwide data will be available. I doubt they’ll be as tightly regulated as aircraft, but that’s reasonable considering they’re not as dangerous.
Private for hire services (Uber, Lyft, etc) definitely incentivize more and longer trips. Lots of before and after studies of dense urban areas have demonstrated that vmt rises precipitously after ride hailing service enters the market. It stands to reason that automated ride hailing services will have a similar and more intense effect, assuming they can lower costs compared to human operated service.
Yes, much like education, if the public option would like to continue to curate the entire service to the bottom quintile of users, then the would be users will continue to seek out private options that address the reasons they don’t ride public transit. Attempting to regulate/restrict their use and force feeding the public alternative will blow up spectacularly, which means Portland will probably try that.
Do you have a link to any of those studies showing that people travel longer distances because of Lyft and Uber?
There is a ton of easily Google able info out there that supports my assertion. But here’s a link: https://www.planetizen.com/news/2021/01/112003-report-ride-hailing-drastically-increases-vehicle-miles-traveled?amp
The article cites the work of Schaller, who is a well respected transportation researcher who has written extensively about the impacts of ride “sharing” services on user and driver choices and behavior and the cumulative impacts on cities.
And it’s not necessarily that people travel longer distances between destinations, it’s that there are more vehicles on the road, deadheading around to position themselves to try to find riders or get to locations that apps tell them are in high demand. Also people are choosing private rides that might have otherwise chosen transit in the absence of ride hailing services.
“It’s not necessarily that people travel longer distances between destinations” is very different than “definitely incentivizing more and longer trips”, the claim I was skeptical about.
But sure… deadheading increases VMT. Of course.
The deadheading is part of the trip (more than 50% of the ride-sharing related vmt is deadheading). If I need to take a three mile trip across town, the driver is going to travel 1.5 extra miles before my three mile trip even starts. If you add the deadhead miles to the actual trip miles, the ride-sharing service has just induced a longer trip (my three mile trip magically turned into a 4.5 mile trip). I wasn’t in the car for some of those miles, but i generated that extra vmt by purchasing the ride.
Okay, then let’s say that ride sharing increases VMT, even if trip lengths aren’t increasing.
That’s totally obvious and uncontroversial.
Quantity has a quality all it’s own.
I’d argue the ubiquity of land vehicles elevates the danger to be greater than that of aircraft.
The danger is fundamentally different. You don’t need a pre-flight check on your car, and it’s unlikely that hundreds will die if you get into a fender bender at some point on your journey.
Those different characteristics lead to different oversight requirements, so comparing the two doesn’t make a lot of sense.
I dunno, If drivers had to meet the same requirements as pilots to operate a car there would be a lot less dead people 🙂
heck, a lot less cars on the road too.
Man – that’s it! A panacea! A silver Bullet! The answer we’ve all been searching for!
(I kid of course, I was trying for droll in the first post, but I don’t know the emoticon for it)
My thoughts also Todd. It’s fit for news but in no way does it suggest anything other than further research needed given that confidence level etc. That is, we need research to originate outside the company that has an interest in its own findings, as well as reviewed by peers. It is commendable they put the data on crashes on their website, so others can try to duplicate the results (and more than a little scary it’s the only AV company that has).
As SD said below, there are some very troubling potential results of allowing corporations undue influence on our public space: the diffuse ownership/culpability of an “acceptable” percentage of deaths by insurance companies, the use of US highways/stroads (some of the deadliest in the world) as training grounds without the public’s consent for AI to be used across the world, the increasing influence of design on urban streets that will allow for the storage and constant circulation of AVs.
Yes, robocars tend to not speed or run red lights or commit DWIs (except Johny Cab). But the whole host of problems associated with giving our streets to corporate interests is indicative of the lessons learned when Chicago decided to privatize its parking meters. Waymo doesn’t care about quality of life, it (not a person) only cares about profit. The main takeaway I hope we all can recognize is that cars are inefficient/dangerous regardless of the driver, and walkable cities, not auto-centric hell, is what we actually need.
I haven’t seen a lot of evidence that agencies like TriMet care much about “quality of life” either. Everyone has their mission, and that’s what they tend to focus on.
Cars may be inefficient on some metrics (gas powered cars emit more than a fully loaded diesel TriMet bus, for example, though less than a nearly empty one), but are quite efficient on metrics that people care a lot about, such as travel time and cost per trip (though that cost is largely subsidized so riders don’t actually pay it directly).
Portland did this without even a peep of protest when Uber illegally started operating in Portland. To add insult to injury market urbanists actually applauded the take-over of our streets by private-capital and VC-funded TNC tech unicorns, despite the predictable and now well-documented negative externalities (e.g. decreased use of transit and reinforcement of automobility — esp for young people).
Example #1224 of how libertarian market urbanism leads to poor health and equity outcomes in urban areas.
The problem with self-driving car safety is that it can be adjusted to maximize profit by the corporation. It can be adjusted to 5% less than human drivers, 50% or 99%, but the choice will still come down to political will and power. Having waymos on the road is not going to give lawmakers a spine to set policy that we should already have in place, and potentially could make things worse if the politicians are up against mega corporations instead of a mixed group of citizens who generally want to live.
Allowing self-driving vehicles on the road at scale, is a further step in giving private control to public assets- our roads and public spaces.
I am skeptical of the idea that there’s some “safety dial” that can be adjusted in the manner you suggest; that’s just not how these systems work.
But how has this played out in the cities that now have whole fleets of self-driving cars, either in the US or abroad? Are regulators overwhelmed by “mega corporations”? Are AV companies “dialing down” safety to improve profit? Is the public losing control of the streets?
What we know so far is that they are safer than human drivers. For those of us concerned with safety, that’s something to celebrate.
lost in trivial arguments
There absolutely is a safety dial. You need a threshold where you choose to ignore something that pops up in front of the car. A ball just bounced into the street? Or was that a small animal? Should we slam on the brakes?
They will absolutely dial up the risk to smaller pedestrians to reduce the instances of false braking if they need to.
https://en.wikipedia.org/wiki/Death_of_Elaine_Herzberg
“However, night-time video shot by other motorists in the days following the crash, plus their comments, suggest that the area may have been better illuminated than the dashcam footage, viewed in isolation would suggest. This raises the possibility that Herzberg’s appearing so late in the Uber video could merely be an indication that the camera had insufficient sensitivity or was otherwise poorly calibrated for the environment and setting in which it was operating”
What happened in the Elain Herzberg case was a case of software failure to recognize an obvious person, as explained in the article you linked:
This wasn’t someone adjusting the mythical tradeoff between avoiding collisions and profit that SD was describing; it was a system that didn’t work properly, with a safety driver that wasn’t paying attention, involving a company that is no longer developing AVs.
And think of it from a pure profit standpoint: These vehicles are very expensive; you’d have to make a lot of extra profit in order to justify crashing one, even if you did have a profit/crash dial.
An issue I take with Waymo vs. “the typical driver” is that it’s not a very apt comparison. Unless things shift massively for Waymo on the cost front (it’s currently twice as expensive as an Uber/Lyft), it will only ever be a taxi competitor, and should have its safety judged relative to a professional driver.
There is no evidence to suggest that an autonomous vehicle is anywhere near the cusp of adoption in the consumer market, and trips taken via Waymo are not really displacing typical car trips, they replace typical cab trips – the market of which seemingly shifted massively with the introduction of Uber/Lyft but has stabilized since.
Right and even under those ideal circumstances where it is assumed an AV is fully autonomous, there is some question as to how frequently an intervention by actual humans occurs. That is, there appears to be a difference between how often actual drivers are required, and how many of those are reported (how “disengagements” are defined). So practically speaking a lot of AVs have a human correcting their behaviors, but it’s not at all clear how frequent and necessary those interventions happen. It’s also fairly clear that there is an incentive for corporations to reduce the number of these disengagements since that stat is most reported to the media and assumes (at least on the surface) to correlate with safety.
I agree; Waymo is much less likely to speed, get distracted, or drive drunk. The comparison with a human driver is simply unfair. But even compared to a professional driver, the technology has a good record (I assume… my neighbor’s son became a professional driver driving a small delivery truck as his summer job just after he turned 18 and had driven on his own exactly once after getting his driver’s license, and surely you read about the recent dustup with Skyline CDL School and the envelopes of cash).
That said, I agree with your second paragraph. We’re still in very early days, and it’s hard to know how this technology is going to be deployed (but it’s worth noting that just a year or two ago, the arguments on this very site were that the technology would never work and the it was just a techno-utopian fever dream).
Nonetheless, Waymo’s safety record is excellent, and it’s probably the worst now that it will ever be.
In general, if a cabbie does this there should be some public or quasi public way to prevent it from happening in the future. Like if you drive drunk as a London taxi driver, surely you get fired.
I would challenge that it’s the worst now that it will ever be. Urban driving is inherently chaotic (part of why autonomous vehicles have such a long, tortured development history) and expanding into new markets – especially with adverse weather conditions – is likely to reduce the overall safety numbers. There’s a reason Waymo only operates in broadly warm places without much inclement weather. Maybe it’s confirmation that I know far more people in SF than Phoenix, but I’ve seen a lot of incidents with Waymos in SF, and not very many in Phoenix. In the report, all of their serious safety incidents are in chaotic SF, and none are in suburban Phoenix.
Granted, most of the US looks more like suburban Phoenix than SF. But the most profitable cab ride tends to be a short urban trip. This is all to say that I’m not convinced it’s the worst it will ever be, nor do I think Waymo is going to be able to operate effectively in any place where snow and ice are endemic.
It doesn’t take too much imagination to see that one solution to the “silver tsunami” of people aging out of driving might be provided by the automated taxis that are demonstrating their excellent safety records in the real world. A good system of automated taxis might make it easier for seniors to give up their cars, which seems like a win from almost every angle (safety, efficiency, convenience, reduced parking needs, etc.)
https://finance.yahoo.com/news/5-6-billion-injection-hasnt-142440151.html
I am sure Utopia will be here at some point but not quite yet…
If it can’t work in LA, where is it going to work?
The problem in LA is that Waymos aren’t using the highway system, which is pretty much everything there. That is likely to change. It is also likely that the technology will continue to improve, as most technology does.
By the way, those “too expensive” LA Waymo rides cost about what it costs to transport someone on TriMet.
As I’ve said before, it will work, or not. And if it doesn’t, Waymo will disappear and we’ll be no worse off than we currently are. I’m not paying for it, so while I hope they succeed, I won’t pay the price of failure.
Do you have any evidence of this? TriMet reports $11/ride, I think there is a 0% chance Waymo comes anywhere near this. And bear in mind, that $11 is the entire cost of running the service – TriMet does make some of that back on fare (about 10%). Waymo looses millions of dollars a year and doesn’t really compete on price with conventional cabs. In per-mile terms, TriMet’s system costs per mile are $2.96 – much more than a per-mile trip in a cab (it’s $6.75 per mile for a cab in SF for the first mile, and $3.25 per mile after the first, plus a 65¢ per minute waiting charge – source)
“Do you have any evidence of this? ”
Nope, because I misread the article. My comment was wrong in that respect.
My larger point still stands — the vehicles are not doing well in LA because they avoid the highway, which is different than an inherent flaw in the model (unless they never use the highway, in which case they will never work in LA, nor in a great deal of other useful situations).
I spoke to my partner in crime that we hope that in 20 years there will be automated taxis (Johnny cab from 5th Element maybe?) that we can take and we won’t need to use a vehicle. We’ve seen people in their 80s and 90s that should never be allowed on the road still driving until their families forcibly took away the car keys from them. Oh the battles that were had over that are of epic legends.
Total Recall. Actually, a little too prescient with Musk’s robotaxi and Mars ambitions.
“How did I get in this taxi?!”
“The door opened. You got in. *eyeroll* ”
I should note that when he didn’t pay the 18 credits, Johnny Cab tried to run him over.
Thanks Verhoeven!
“Drive, drive!”
“Would you please repeat the destination? Please state the street and the number.”
“@#$!”
“I’m not familiar with that address.”
Automated taxis are currently very expensive, and I have yet to see compelling evidence that they will ever be even close to on par with public transportation on price. Given that retirees tend to be very price sensitive, and often on fixed income, I don’t think automated taxis really stand to be that strong of a solution (in a narrow economic sense for the price-sensitive retiree).
I can imagine a future where it’s possible and practical for a senior to ditch the car for a Waymo, but that doesn’t mean it’s a good future for everyone. We probably don’t need to rehash our weekly “investing in transit vs. automated vehicles” schtick, so I’ll leave it at that.
If they’re too expensive, people won’t use them, and they’ll fade away. Some investors will lose money, and we’ll have learned something. This isn’t really a problem, except that we’ll have to deal with the Silver Tsunami some other way.
We don’t need a single solution for every person. Some folks could rely on Waymos, some on traditional taxis, some on public transit, and some on their kids. Pretty much like today, but with an additional option.
I think the only real end option for automated taxis is replacing current ones, and even that may not be enough. If they can’t compete on price with human drivers, what’s the point of doing it? And for investors, I think a huge aspect of getting in on Waymo is probably trying to own some slice of a massively profitable future. If the company falls short of eye-watering, monopoly profits, eventually investor sentiment will turn on the company and it’ll be doomed.
The hype around Waymo isn’t to replace Uber drivers, it’s to replace all human drivers. If they cannot develop a product which eventually reaches the market, they will eventually face some kind of investor backlash. Like $45B valuation for Waymo (more than Ford, about the same as GM, more than Stellantis/Chrysler) is not fully justified on the lines of “it’ll cut labor costs for existing taxi services”, it’s a based on some more utopian vision of the future. Will failing to deliver on a promised future cause Waymo to collapse under its own weight? I don’t know, it’s a Silicon Valley thing so trying to make a rational guess about what will happen is impossible, but I’m not very optimistic about self driving cars for the consumer market.
I think we agree that fleet / taxi deployment is the most likely future for these vehicles. My hope is they will perform well enough that people will be able to give up their private vehicles and rely on these instead.
It’s basically the same dream that proponents of transit have, but with different vehicles and a different service model.
We also agree that if the technology does not ultimately work out, someone is going to lose a lot of money. I don’t consider that a problem. Interestingly, one of the conversations parallel to this is that the vehicle will work too well and people will want to travel long distances in them.
We live in interesting times.
I think even if the real issues in autonomous vehicle performance in challenging conditions are solved, and even if the regulatory bodies are willing, that the most likely outcome is that they won’t be cost effective for consumers. On the hardware end, a Waymo costs like $100k, and has a lot of operational and software needs too (I’m not sure how to estimate this cost to be honest). Maybe that cost will come down, but cars keep getting more expensive and adding a bunch more sensors and computers doesn’t really lower the cost.
There’s also another angle where the whole point of automobility is personal freedom. Will people sell that out to let a corporation drive them around? I mean probably, but that’s an angle of backlash that the industry hasn’t had to deal with yet but inevitably would if it ever threatened the traditional consumer automotive industry.
Things certainly are interesting
This could well be right. And that’s probably a good thing, if it means that shared vehicles become more common (reduced parking demand, fewer resources sitting unused, etc.)
We’ve been having this conversation across multiple threads, and while it is clear you are skeptical of the technology, I am having a hard time discerning what your position is.
We know that our current transportation system is deeply flawed, and we know that transit, walking, and biking have (so far at least) failed to address most of these problems except on an individual basis. Do you think that AV technology will simply never work under general conditions? Or that it will work, but will be worse than current conditions? Or something else?
I think the technology is likely to continue to improve, and will become widespread, using the taxi model, because it offers most (but not all) the positive attributes of driving and transit while eliminating most (but not all) of the downsides of both. No other available transportation option can offer that.
My hope is that it displaces human drivers to an extent that traffic fatalities markedly decline, and urban parking becomes less necessary. I see no reason why this couldn’t happen. I see the cost of failure mostly one of lost opportunity, so I see further development as low risk, without much downside.
Do you see it differently?
I see private ownership of a shared AV being sort of unlikely to take hold. It’s sort of antithetical to modern automotive culture, so it would face a big uphill cultural battle on top of high entry costs – not something that gives me much confidence.
I have a lot of mixed and probably slightly contradictory feelings about AVs. Mostly, I feel that AV taxis are not the most effective solution to urban transportation issues – especially where cost is a concern. I also think the machine learning approach of “just add more data to the model” will face diminishing returns for things in the real world where conditions can vary and having an actual understanding of what is going on is important for safety in those sorts of conditions. Everything I’ve read about the current economics of Waymo make me very skeptical that it will pay off in its current iteration in the long run. It would not survive without massive subsidy from Google, and it cannot compete on price with human taxi services. That is not a good sign for a service predicated on competing on price with human taxi services.
I also generally don’t think a fleet of autonomous vehicle taxis offer an upgrade from public transportation in an urban setting, but we’ve gone over that many times before.
If it’s not an upgrade, and it’s more expensive, people simply won’t use it.
I read somewhere that AVs have driven something like a hundred million miles in the US. Clearly, someone is using it, and finds it preferable to their other options.
Maybe robotaxis will go the way of 3D TV. Or they might be the next generative AI. We’ll have to wait and see.
The next generative AI? You mean the next hype-driven corporate bandwagon that makes workers’ lives more precarious and doesn’t even turn a profit?
That 100 million miles figure comes from the Autonomous Vehicle Industry Association itself. I’m sure they have no reason to lie or exaggerate those numbers. 100 million is also a tiny fraction of the approximately three trillion miles Americans drive in just a single year.
As an example of as technology that appeared suddenly and has the potential to transform society, it’s hard to think of a better example than generative AI. Perhaps you have a suggestion for a more apt comparison?
Also, Cruise learned that lying about safety results in a lost licenses to operate, which is akin to the death penalty.
No, I think the comparison is fine. In that both technologies are mostly useless hype that billionaire tech oligarchs nonetheless see as tools to help them “transform society” into their own private fiefdoms. That’s not a good thing, in case you’re curious.
Time will tell if these tools will prove to be transformative or useless hype, but it’s completely incidental to what we were talking about.
Regarding Waymo: As I recently noted in a lengthy letter to the editors of Harvard Magazine,
“Vehicular violence exemplifies the harm that comes from falsely equating new technology with progress. Yet the [March/Apr 2025 cover] article readily engages in this logical fallacy when discussing “autonomous vehicles.” Companies now dangle the unproven promise of eliminating human error to promote a detrimental business model that will increase traffic, accelerate the climate catastrophe, and potentially increase deadly collisions, perhaps in ways we have not yet encountered (just imagine when cyberattacks target AV software). Mark Fagan downplays all the dangers of this new technology by suggesting there have only been “one or two bad events” that shouldn’t “end the experiment”; but many of us do not join Fagan in celebrating the fact that “way back when” safety concerns did not impede an unending embrace of automobiles and aircraft. Perhaps if safety concerns were taken more seriously then and now, we wouldn’t have 40,000 people a year killed by vehicular violence, hundreds of thousands of others badly injured, along with the many lives lost because of unsafe aircraft production and increasing air traffic—not to mention an ever worsening climate catastrophe that is already beginning to cause horrific impacts for entire generations (including current Harvard students), as well as countless other species.”
As the article linked above notes, AVs are dramatically safer to pedestrians and cyclists than human drivers are. That suggests we might have found a way to reduce injuries and deaths without losing the attributes that make cars so attractive to so many people.
So, maybe be encouraged by progress in this area?
“dramatically safer” is the marketing line, but that doesn’t make it true. I have seen these vehicles in traffic in other cities, doing things that are unexpected, unsafe, and not conforming to traffic laws. And again, the scale for MASSIVE collisions from hacking or even just from a failure in the connectivity on which each vehicle relies is unprecedented even in our motor vehicle-death induced current reality.
Agreed. But real-world testing has shown that it is, in fact, true. 85-96% fewer injury collisions with bikes and pedestrians strikes me as pretty significant.
I underestimated how far AI would advance in my lifetime, I’ve probably underestimated AV’s as well.
Unfortunately I see the predominant lifeform on our roads being fossil fuel powered, human guided vehicles for the forseeable future.
By the time I reach my early 80’s I’d estimate 1 million more people will die on our roads, mostly needlessly.
Re Waymo: One of the ways Waymo achieves good safety is by restricting where it operates, only operating in fairly small areas where it has very thorough data on the physical environment.
It’s not impossible to expand those areas, but there’s definitely a tradeoff between geographic coverage and thoroughness for a given budget. Very thorough and up-to-date data means the system can operate with a lot of certainty about where it’s allowed to drive and where it’s allowed to drop people off. Its sensors can help prevent crashes with things and people in its path but they mostly don’t have to figure out whether the car is about to drive into the middle of a bike lane that the system doesn’t know about yet.
A popular, profitable system could use profits from its operations to fund expansion of its coverage of extremely thorough map data. There are some advantages over the current system, where individual drivers are responsible for handling ambiguous situations. When an automated system finds out about that new bike lane, for instance, it can add that to its data and then send it out to its entire fleet, whereas with human drivers each one has to figure it out on their own. We might hope that each of the drivers can handle it, but … I’ve certainly seen drivers plow straight through flex-posts, seemingly oblivious. Well… in one case it was a Tesla, so it might have been both the computer and the driver both oblivious.
I’d generally trust Waymo to be more responsible than Tesla (or, say… Uber) in building out a self-driving system, just because they’ve been more responsible so far. But their system relies on a level of detailed mapping that they might have a hard time scaling efficiently. It’s possible that what we really get is a refinement of Tesla’s “let the computer figure it out” model (possibly implemented by more focused competitors) rather than an expansion of Waymo’s. That wouldn’t have the same safety benefits.
One interesting side note here is that Waymo can continuously map the areas in which is is running vehicles, getting updates dozens or even hundreds of times a day, each time one of its vehicles passes a given point. This will give them a valuable dataset that they could sell to cities, for example, who might benefit from having real-time data about infrastructure conditions. There are also (darker) ramifications for personal privacy.
I don’t think there’s a fundamental scaling problem; you map an area once, then get updates for “free”. I mean, I once thought what Google was attempting with StreetView was logistically impossible, and yet it’s been done (and repeatedly redone) using human drivers.
One other thing: Waymo has set a bar for safety, proving what is possible; if they fall too far beneath what they’re doing today, people will (rightfully) start asking questions, part of the reason I think their vehicles are as dangerous as they’ll ever be.
The Mashable article is basically a press release for Waymo to trumpet its own report on its driverless cars. A recent independent study found crash rate comparisons between humans and self-driving cars (including Waymo) to be statistically insignificant.
The article you linked to cited a study that found that crashes involving driverless cars occurred at a similar rate to those involving professional drivers.
It also found that about half of those crashes involved human drivers rear-ending the driverless car.
The Mashable article talked about something completely different: the crash rate involving pedestrians and cyclists.
These two articles are talking about different aspects of safety and compliment, rather than contradict, one another.
My personal opinion is that if I could swap out human drivers for a machine that was involved in low level crashes with other cars at about the same rate as professional human drivers, and was 85% to 96% safer around pedestrians and cyclists, I’d do it, especially if I thought it was likely that the technology would improve over time.
Uber’s gig workers are not “professional drivers”. And people keep rear-ending driverless cars because driverless cars keep stopping for no reason in the middle of traffic. But why let facts get in the way of neoliberal techno-fetishism? “Move fast and break things“, amirite?
In fact they are. Yes, some of the rear-ending happens when cars stop unexpectedly, but that’s still on the following driver. We’re talking about urban traffic, where things happen, and drivers sometimes suddenly stop suddenly. Following drivers need to be ready for that. That said, it is obvious we’re still in the early stages, and the system needs to continuously improve.
And yes, I’d prefer to focus our discussion on facts and reasonable conjecture, which you keep trying to peel us away from to get your little jabs in.
Ah yes, “reasonable conjecture” like telling all drivers to simply change their human behaviors en masse to better accommodate the robots. Sounds completely realistic, yessir.
That’s not what conjecture means, but asking drivers to prepare for vehicles to stop in an urban environment doesn’t sound unreasonable. It’s what they should be doing anyway.
But even if you disagree, and think drivers shouldn’t allow sufficient following distance, and blame all rear-end collisions on AVs, and assume that AVs today are as good as they can possibly be, we’re still talking about minor vehicle-to-vehicle collisions that occur in frequencies we’re experiencing with human drivers, accompanied by a huge reduction in pedestrian and bike collisions.
It certainly is reasonable to examine these facts and assumptions alongside the related costs and benefits and design good policy (including the possibility of prohibiting the technology). That’s the type of political determination that we know how to make. Other cities are looking at this data and allowing deployment, but that could change as we learn more.
Here in Portland, we don’t need to make any decisions for a while, giving us time to learn from the experience of other places and give the technology more time to mature before concluding anything.
One of the stories I linked to was literally about a person who died after their self-driving car stopped while driving on the freeway, not an urban street. Another was about an eight-car pileup that resulted from a self-driving car braking suddenly on a different freeway, sending a child to the hospital. Minor collisions, indeed.
Those were Teslas. Teslas are not self-driving vehicles — they are fancy cars with a very sophisticated version of cruise-control that has been dangerously and grossly over-marketed. They simply aren’t relevant to the topic at hand.
Do you have anything constructive or insightful to add, or should we wrap this up here?
Fine, here are several examples of actual self-driving robotaxis stopping for no reason in front of fire engines responding to emergencies, forcing them to wait or find an alternative route. All very minor, very low level incidents, I’m sure.
What exactly is the point you are trying to make? That the technology can never work and should be banned, and we should stick with the status quo?
No, just that your claim that self-driving cars are only responsible for “minor” safety incidents is false. If the technology (and the companies behind it) were that beneficial, you wouldn’t have to sugarcoat it.
If you’re right that I am sugarcoating the facts (which I got from articles posted by Jonathan and you), or wrong on my assessment of the cost to deploy, the technology will fail, automated vehicles will never come to Portland, and we can carry on carrying on just as we are today.
And no, typical rideshare drivers do not require special certification beyond an ordinary driver’s license. They are no more “professional” than any other driver.
“Professional driver” is not the same as “CDL holder”. One is someone whose occupation is driving, another is someone who holds a specific license to operate certain classes of large vehicles.
But it’s utterly unimportant if you agree with me or not.
Pizza delivery drivers are not known for stellar safety records despite driving being their occupation. So “professional driver” is a meaningless benchmark in this case. But that’s not the only meaning of “professional”, which also has the sense of conforming to a certain set of technical or ethical standards. But I suspect you knew that already.
“So “professional driver” is a meaningless benchmark in this case.”
It is. Someone else was using the term, so I went with it. As I mentioned elsewhere to make this very point, a neighbor’s kid got a job driving a delivery van having ventured on his own driving exactly once.
If you find a set of technical and ethical standards for “professional drivers”, show it to your next Uber driver for a good laugh.
So by driverless cars having a similar safety record to “professional drivers”, you meant “worse than normal drivers”. Weird flex but OK
I’m regards to other cars, they have a similar record as Uber drivers, if you assume getting ready-ended is their fault. You know, just what the article said. They have a much better safety record regarding pedestrians and bicycles, according to the article Jonathan posted.
Other than trying to poke me, what argument are you trying to make?
Easy on the self-victimization there, champ. You replied to my comment, remember?