Is Waymo better than a human?

Subscribe using your favorite podcast service:


note: this is a machine generated transcript and may not be completely accurate. This is provided for convience and should not be used for attribution.

Anthony: You’re listening to There auto be a law, the center for auto safety podcast with executive director, Michael Brooks, chief engineer, Fred Perkins, and hosted by me, Anthony seminar for over 50 years. The center for auto safety has worked to make cars safer. The

story I wanted to capture. That’s why

Michael: I wanted to start. I’ve got it on the webcam. If you want to see that. Wait, you have

Anthony: to have a webcam like on your dashboard? I

Michael: put a dash cam in just to see if I can, first of all, I put it in because after you watch a large number of videos of crashes and the aftermath, you realize at some point that a lot of people don’t tell the truth about what happened before the accident.

Although that’s a shock. A lot. A lot of Kyles out there , right? But that was a post crash questionnaire. But this is the before. A lot of people claim that they were not at fault when they clearly were post crash. So a dash camera is one great way to resolve any uncertainty in that area.


Anthony: with that, listeners, welcome to a happy holidays episode of the Center for Road Safety Podcast, There Ought to Be a Law, not to be confused with the 1950s Looney Tunes cartoon called There Ought to Be a Law. Yeah, everyone survived this is good, Michael was not sideswiped by a car in, on Mobile Bay, which sounds lovely, by the way.

Is it lovely? It

Michael: is a, it is a nice drive, usually, it’s probably A little too much traffic than we’d like, you’re from New York, you don’t care. Being sideswiped on a interstate where there’s a drop into a bay though is always a little concerning. Oh yeah, I don’t want to do that. But we avoided all contact.

I don’t even know if the person even noticed.

Anthony: Were they texting?

Michael: It’s certainly possible. I saw a lot of people doing way too much with their phones while also trying to control a multi ton vehicle.

Anthony: Maybe if if it was a Waymo, you’d be safer, because that’s how we’re going to start

Michael: off this week.

I was going over 50 miles an hour, so

Anthony: No, it’s not a level three. Wait, Waymos can go over, can Waymos go over 50 miles per hour? No, they’re restricted,

Michael: aren’t they? They can go, I believe in the latest report, they say they’re going up to 50 miles per hour in a lot of the areas they operate.

Anthony: Well, Waymo’s now driven over 7 million miles.

They claim only three injuries. They’ve driven 5. 3 million driverless miles in Phoenix, 1. 8 in San Francisco. And again, they, we think this sounds like a lot, but this article we have in Ars Technica. Ends with this nice little tidbit. All the data so far suggests that Waymo vehicles are making roads in San Francisco and Phoenix at least a little bit safer, and Waymo’s case gets stronger with every million miles it completes.

But it’s going to be another couple of years, if not longer, before we can be confident about whether Waymo vehicles are helping to reduce the risk of fatal crashes. First of all, a small round of applause to Waymo for actually sharing data and opening things up as opposed to all of their competitors who are just a bag of dicks.

Waymo’s been forthright. They’re still playing that game. Hey, 7 million miles, we’re safe as can be. Whereas you guys have both pointed out to me add a few hundred million miles and then we’ll talk. But what do you guys think? Is this you guys ready to ditch your cars and just rely on Waymo?

Fred: Let’s see. If you look at the NHTSA Data, it says that there is roughly one fatal accident for every hundred million miles, and that translates into one serious injury about every ten million miles. WAMO says three injuries per seven million miles, so they’re only one order of magnitude away from the overall statistics, or safety of conventionally driven vehicles.

They’re good there. Kudos to them for sharing some information. But, it’s always the devil’s in the details, right? We’ve got to look at where those occurred. What are the circumstances for similar human driven accidents or collisions? Is there a systematic flaw? Is there really something going on here?

If, because, to say that There is one fatal human accident for roughly every 100 million miles. Means that you’ve put together you, me, Michael, a lot of little old ladies along with all the drunks and drug users and really aggressive drivers. That’s not an appropriate way to look at the statistics overall.

So you really need to parse this information. With a lot of additional information about when and how those injuries occurred. I’m not saying Waymo is trying to, pull the wool over anybody’s eyes, but you really need to look at the details of how and when and where these incidents occurred before you can say that.

Somehow there’s overall safer than the average human driver.

Michael: Yeah, and I would encourage, anyone who’s interested in looking into this subject of, comparing autonomous vehicles to human drivers to look at the paper and the research that Google released accompanying this.

So they were taking a specific look at injury crashes because they haven’t had any fatality crashes yet. So they were basically comparing. alL the police reported crashes in the areas they operate, Los Angeles and in Arizona, San Francisco to the police reported crashes in those areas. In some respects, I believe they were trying to narrow down as much as they could, but, ultimately, as Fred alluded to, you’re comparing, your best offering to, the worst drivers in many cases. So just simply comparing this vehicle’s performance or the system’s conformance to the human crash rate doesn’t really get us to where we’re saying that this vehicle is as good or better than, a train human who’s not speeding, not drunk, yada, yada.

It’s still a little unclear just how much safer the Waymo would be than me or than Fred or than Anthony or than, the guy off the street. So it’s, versus how good it is compared to, teenagers stealing Hyundai’s people running from the police and people who just came out of a bar.

So there’s. And then part of that really isn’t due to the fact that Waymo is trying not to show us that it’s because those data aren’t really available. It’s really difficult to parse this stuff out. And when you look at their paper, they go through a lot of that and they make a lot of really good explanations and, open up how difficult it is to arrive at these decisions.

And, even when you’ve gotten the numbers in front of you, you still have that. underlying question of, how safe is safe enough before we can actually pop these things out onto roads across America, expand our services. So there’s still a lot of considerations that come even after you have gotten your numbers lined up.

And to

Fred: geek out even a little more, I know you guys find it unusual for me to geek out, but there’s something called granularity of data. That is important in this case. There’s very few samples, happily, of injuries that involve the Waymo vehicles, and good for them. Let me put it in perspective, okay? If you take out a penny from your pocket, and you flip it, and it lands on the table, it’s likely to be either heads or tails, right?

So if you flip it, and you, let’s say it comes up heads. You take it out of your pocket, and you flip it again. And it comes up heads again. You could come to the conclusion that every time you flip that pen, that penny, it’s gonna come up heads, right? So you test the hypothesis and you flip it again, it comes up heads again.

Oh, now you got great data that says three times in a row, I flip this coin, it’s gonna call up heads. You’re ready to go to Atlantic City and, put your money down because you’ve got a trend going right. That’s how math works. The problem is that if you flip that penny a million times, you’re going to find that the average is very close to 50%.

And so that’s how confidence, statistical confidence builds up. So the problem that Waymo’s got is that the numbers of vehicles involved in this, numbers of incidents involved in this. Is very small compared to the overall population. So it’s going to be a long time before they accumulate enough information, the flips of the coin, if you will, to say that what, what really is the number that’s associated with the long term use of these vehicles.

It’s a statistic, it’s not a slam on Waymo. It’s just a statistical inference they might make based on these data that is really not justified based on the small number of samples that they’ve got at hand.

Anthony: Okay, so we need a lot more data and also what I thought was a glaring error in Waymo’s safety data was there was no mention whatsoever and we heard about it here from a guest a couple weeks ago that Waymo may try to steal your girlfriend.

whY isn’t that in their reports? Again,

Fred: statistical sampling is, there’s just not that many instances.

Anthony: That we know about, that have been police reported at least. Yeah. We’ll see. Okay. mOving on to another car that is is surprising that it’s out there is the Chevy Blazer EV.

Now, Chevy made a very cool, small EV. And It’s not that small, but no, the original, not the Blazer, the Bolt. Oh okay. The Bolt it was, small, like every reviewer, all these super car geeks were like, this thing’s awesome, I would love to have it for myself. Me, as a consumer, I’m looking and I’m like, how long does this thing take to charge?

Because it was ridiculous. But, apparently as a commuter car, it was great. So GM did what GM does, and they’re like, hey, people like this, let’s get rid of it. Let’s replace it with this thing called the Blazer EV, which is You know, it’s a SUV style thing, and it’s not good. Edmunds is basically saying it’s had a, it’s had 23 fault codes on its diagnostic test since they’ve had for, had it for two months.

A writer at Inside EVs whose week long test ended after 28 hours. The vehicle’s infotainment system went blank while he was driving, and then an attempt to charge the battery failed, producing a service vehicle soon. Error message. That’s never a service vehicle soon.

Michael: Oh, just basically right now that vehicle qualified.

Edmonds could take that vehicle back to the GM dealer. It qualifies under the California limit law where they receive delivery of it right and get their refund. ThAt’s quick, but it’s, in California, there are a lot of ways to qualify for the women law. It would be, one safety issue, which I think if you read that list of faults, it’s pretty clear the vehicle has a safety issue that’s uncorrected.

And hasn’t been able to be repaired for two weeks, which is, as long as that vehicle’s been in the dealership uncorrected. And it just looks like a, Edmund said it was the, the longest list of faults they’d ever seen on a vehicle. Either, either at the, at this point, you’re thinking either GM, gave them a bad vehicle or, the Evie blazers, a hot mess.

They suspended production of the Evie blazer as well. So that probably tells you that there’s a lot they need to work on. It looked like a lot centered around the infotainment system, but all of those faults were, or many of those faults were related to. Communications with, engine and other safety related parts of the car.

Anthony: How does its backup camera perform? Backup camera, okay. I

Michael: didn’t see a fault on the backup camera, but the infotainment screen didn’t come up as Edmunds noted, during large portions of its drives even right after the vehicle started. So I’m assuming that would include the backup camera, which is, one of our leading causes of a recall.

Anthony: Chevrolet Vice President Scott Bell said, We are aware that a limited number of customers have experienced software related quality issues with their Blazer EV. Customer satisfaction is our priority. And as such, we will take a brief pause on new deliveries.

Fred: Having it as a priority doesn’t mean it’s a top priority,

Anthony: does it?

We’re just going to take a brief pause. Okay, everyone think about it. Okay, everyone go into your safe space here. And breathe. And let’s go on to sell a larger, stupider car.

Michael: Yeah. What are you going to do in an America where 80 percent of your buyers are looking for an SUV and not a sedan?

Are you just going to ignore that? Are you going to fill that market? At

Anthony: least, make one that works. At least like normally when they supply vehicles to journalists, like they go over them with a toothbrush and a fine tooth comb. And this one we’re like, Hey, it’s 28 default messages.

That’s better than last week. Last week it was 37. So it was pretty good. Get it out there. What does Edmonds know? Oh, again,

Fred: granularity. GM would say this is just a bad penny flip. In the future, everything will be better.

Anthony: Maybe not for GM. Toyota. This is a surprising one.

So Toyota has a small division called Daihatsu and Toyota, they make pretty good cars, right? Fair to say, they make cars of really good reliability.

Michael: Yeah, I believe they were number one in quality in consumer reports

Anthony: recently. Yeah, I think they’ve been that way for decades. They do well in crash tests.

They’re well done, but they have this small division called Daihatsu that makes small cars and trucks and primarily in Japan. And they’ve been submitting bogus safety tests. Daihatsu’s probe found 174 new cases of irregularities in safety tests and other procedures in 25 test categories on top of problems reported earlier.

They have reported improper testing on door linings, proper problems in side collision testing, data falsification, and unauthorized, and use of unauthorized testing procedures. Speaking to reporters last week, Daihatsu president, not gonna try to pronounce his name, acknowledged the cheating on safety testing and procedures, saying it was tantamount to neglect of safety certificates.

He attributed the problems to pressure on workers to meet ambitious demands for tight development deadlines. Hey. We got to hit those production numbers. People are going to die. Have a nice day.

Michael: Yeah, that’s a, it’s a huge safety scandal. I don’t believe, and from everything I can tell, there aren’t any vehicles that have been sold in the United States that this applies to.

Mainly because of something I was just talking about. Daihatsu typically makes really small pickups and cars, and nobody in America wants to buy those right now. But what they’re, what they’ve done a couple of cases is, not just, yes, falsifying side impact crash test results. It looked like and measurements they were taking.

I think one, one of the reports I read mentioned that they basically took the results from one side of the vehicle and flopped them onto the other side and reversed it to make it look like they had tested that side. So there’s clearly some type of. Money saving operation going on in Daihatsu in Japan to try to, limit the number of tests they’re having to conduct, basically limit their research budget.

They’re not doing their due diligence on safety, period. And, if they had done that in the United States, where we’re typically looking at recalls and such, it would be a huge issue and NHTSA would be all over them because they would basically be violating the NHTSA certification regulations, you can’t, even though you’re signing a form that says we’ve certified everything, if in fact you haven’t certified everything, then you’re subject to some pretty severe penalties for lying effectively, which is what DIHATS is doing in this case.

They have, issued. Giant apologies and covered their website with them and produced a report on the matter. And, this is going to be something that impacts a lot of vehicles in Japan. But since dsu, I believe the last they did manufacture cars in America from, I believe they, they sold two models in America from 1998 to nine or 1988 to 1992.

One of them was called the charade. Which is probably appropriate but those did not catch on in, in the United States at all. And they left town in 92. So this would not apply to those vehicles. It looks like this is the more modern, small trucks, small SUVs and the types of vehicles, a few Toyota models that are sold primarily in Japan, Indonesia and the Asian market.

There’s there may

Fred: be a cultural issue at work here. I have some experience working with Japanese manufacturing and Toyota bought Daihatsu. So it’s not an organic part of Toyota. And the, in Japanese culture, it’s typically very, I don’t know how to say this exactly, but it’s very much a trust basis. So if you give your supplier a requirement.

You assume that what is delivered will be in conformance with those requirements. They very rarely will go into a supplier and do an audit of the supplier to find out exactly what’s happening and how that’s done. It’s all assumed to be on the basis of the honor of the supplier and, that yes, of course, they’re going to do exactly what’s required.

And no, they don’t have to have a step by step procedure that is going to provide the kind of quality you’re looking for, because I’m trusting the provider to do what they said they’re going to do. I’m not sure how this fits in to that, but there may be a cultural problem at the heart of this, that the house who was struggling and somebody there Just let the ball drop, and Toyota, for cultural reasons, never had the presence in the supplier to really detect that was happening until it became apparent in the public use of these vehicles.

So it is a different culture over there, and when they do purchase a supplier, from the American perspective, you’ve got to be very careful about what’s coming out of that supply chain, because there may be hidden defects that are pushed behind that. wall of trust that the suppliers typically have with their customers.

Michael: That’s interesting to me that wall of trust still remains, eight years after we saw what happened with Dakota. And all of the fallout over there. But, just generally from a safety perspective, I would say that you should never trust your suppliers and you should be testing and retesting everything you get from them.

And we see so many recalls because of Suppliers that were outside of specifications or didn’t follow proper procedures were cutting corners to make a buck. And, I think it’s the trust system doesn’t really work, as you move into, a software controlled vehicle where you need to be validating every step that every, of every, type of software component that a supplier is putting in a vehicle, it seems like a really bad system to use in the future.

Fred: No argument for me.

Anthony: Yeah. Trust, but verify, and good luck figuring out a way to verify this AI software. Have a nice day. Yeah. Speaking about having a nice day, Consumer Reports, those bunch of liberals. Tesla had a recall. Yes, it was a recall. It was a recall because NHTSA called it a recall.

And Tesla even says, hey, we had a recall of all their vehicles. And so Consumer Reports says, yeah, that’s not enough. Kelly Funkhauser, the non profit’s organization’s associate director of vehicle technology and person with possibly the coolest name in this industry, tells TechCrunch she discovered it’s still possible to cover the cabin camera while using autopilot, meaning drivers can neutralize one of the two main ways the car monitors if they are paying attention to the road.

Oh my god. What’s more, Funkhauser said she did not notice any difference when activating, are using Autopilot’s flagship feature, Autosteer, outside of the controlled access highways where Tesla says the software is designed to be used. Wasn’t this recall to be like, hey, I’ll, part of the recall say, hey, we’re, Essentially setting up an operating design domain.

I pay attention. No. God damn it. Okay. It wasn’t

Michael: good. They said it all. They alluded to putting additional restrictions. They said they were going to put additional restrictions on the engagement of auto steer on Non on basically on interstates limit controlled access highways.

They only want it used there So if you’re off of one, they want to limit the engagement of the feature We think the best way to do that is by geofencing Hey, your vehicle is clearly based on GPS not on a controlled access highway. You can’t turn it on Tesla thinks their little camera and their AI can figure all this out without a map so they put a system into place where the vehicle is supposed to recognize where it is.

Somehow. We’re still not even sure how just like you just said, ask the well, it’s not giving us any answers. So they’ve put this system into place. They apparently have done a software update and, Kelly hops in the car to test it out. And, if you can disable your in car driver monitoring system, the same way you disable the camera on your laptop with a post it note or by covering it, it’s not working.

You’ve got to, figure a way around that problem, Tesla. You’ve got to be able to monitor with the camera. You can’t just cover it up and let drivers do whatever they

Anthony: want. Yeah, doesn’t everyone else’s camera, interior camera, doesn’t it do eye tracking to see that you’re actually Paying attention to the road.

You can’t

Michael: operate like the Ford and the Ford’s Blue Cruise and the GM Super Cruise if you’re blocking the camera monitoring system. It’s not going to let you because they’re not stupid. And I don’t understand why Tesla can’t just get with the program and geofence the feature. It would solve everyone, it would make everyone’s life easier.

You don’t

Fred: understand why? Come on, Michael. They’re doing what they’re doing because they can. There’s NHTSA that says you have to do things a certain way.

Michael: And this is not really forcing them here, right? They’ve been negotiating with them and negotiating with them. We’ve seen this go on and this investigation drag out and it’s Come on, Nitzo, when are you going to jump on top of Tesla the way you’re jumping on top of, ARC over there airbags where there are far fewer incidents involved and far more airbags than there are Tesla’s in the room.

There’s clearly a big issue here and you guys are not pushing them hard enough to do it. Whether maybe that’s because Tesla has great lawyers and they’re paying them billions of dollars. To support this case, but it’s pretty clear that if you come in and with a recall, you say, you’re going to do something you better do it.

Or, or is it going to be another 3 or 4 years of negotiation with more crashes and more people died more injuries before that’s actually makes Tesla do something about this. It’s just been a long, disappointing saga for us to watch. And at the end of it, you see a recall the updates already out and the update isn’t even really doing what it’s says it’s going to do.

It’s, hugely disappointing. And I hope NITSO follows up here and conducts a recall query, an audit query of this recall and goes back and, It forces Tesla, if necessary, to put in a system that actually works to prevent drivers from disengaging and to ensure that when drivers are operating this technology that they’re paying attention.

Anthony: Paying attention and driving? Come on! What’s wrong with you? Oh, the Mississippi’s not strong in him anymore. Let’s continue with some Tesla. I think we’ve talked about something like this before. Where there’s an article we have in from CNBC. Where it’s it’s titled, Tesla blamed drivers for failures of parts it long knew were defective.

This is one of Tesla’s key moves is, Hey, your car’s messed up. It’s your fault. The article it talks about someone here who’s who bought a car with only 115 miles on the odometer it suddenly lost steering control as he made a slow turn into the neighborhood. The vehicle’s front right suspension had collapsed and parts of the car loudly scraped on the road as it came to a stop.

The complex repair required nearly 40 hours of labor to rebuild the suspension and replace the steering column, among other fixes, according to a detailed repair estimate. The cost? More than 14, 000. 1, 000. Tesla refused to cover the repairs, blaming the accident on prior suspension damage. Now, in 114 miles, 115 miles I’m sorry, how much suspension damage can you do?

If you work hard at it,

Fred: you can do a lot, but yeah,

Michael: that’s a lesson. Yeah. Yeah. Come on, Anthony. But

Anthony: come on. This is ridiculous. This is, this was somebody who like really excited to buy their first electric car. Oh my God. He’s a ton of family savings. We were quote over the moon.

This is bought by an electronics engineer. So you know, they’re not going, Hey, let’s go off roading or maybe they were, he was driving with his wife and his three year old. So I don’t, he could be.

Fred: You know how engineers are, they never read the instruction manual. They just assume that if they get into a car, they know how to drive the car.

Have you ever rented a car and read the instruction manual? I think not. No,

Anthony: it’s not my car. I like to bring it back and they go, wait, what’s that? And I go, ha. American Express is paying for the insurance damage. Are you sure? I’m like, I’m not paying for it. Sorry. That’s my personal

Michael: experience.

This suspension issue is something we’ve heard most suspension steering problems that we’ve heard, most often from consumers over and over again for many years now about Tesla’s. There was even a NHTSA’s complaint database got invaded by some Australian guy who was filing complaints for vehicles he found on the internet in junkyards and polluted the NHTSA complaint data on this for a while about three or four years ago.

So it’s been a kind of a An important issue for Tesla to address and the way they’ve gone about it, like some of the examples in this article doesn’t appear to be the greatest customer service and seems to be a heavy handed approach to owners who are experiencing, a common problem.

It would be really interesting to see Nitsa do an investigation here. figure out, how many repairs Tesla has made across its fleet on the suspension issues and also look into whether they, what kind of changes they’ve made in the designs over the years as because this appears to be affecting a large percentage of the fleet of Tesla’s that’s out there and whether it’s build quality or design issues still somewhat unclear.

Anthony: Yeah, Tesla has blamed frequent failures of several parts on the Tesla owners themselves, alleging they abused the cars. This car was a lovely car when it left our factory, and you’re just bad people who managed to scrape together, what, 50, 000 to start, and then, you’re Abusing it. I

Michael: mean, that’s that’s basically the starting point when you’re arguing over some kind of defect with most dealers.

Anyway, it’s, they are invariably, looking for evidence of misuse. It’s the first thing that comes up when your engine goes bad. Yep. Let’s see a list of all of your oil changes, so it’s common approach for a dealer to say you need to show, we’re going to blame this on you, and you have to make an affirmative showing that you haven’t abused the vehicle, which is probably not the best approach for someone who’s only had the car for 115 miles, but Yeah, that’s

Anthony: less than 24 hours.

Yeah. Yeah, the from the article, the automaker told him the suspension collapse was caused by the separation of a lower control arm from the steering knuckle, which connects to the wheel assembly. The owner expected Tesla to cover the damage, and Tesla’s nah, suck it. Which is And that

Michael: sounds like a manufacturing problem.

Anthony: Does that qualify

Michael: as a lemon? Yeah, that would be one severe safety issue right there. Hey! Was it

Anthony: California? They might have been Massachusetts, I think this was. Massachusetts

Michael: might have that way of Oh, no, this is, sorry, Cambridge, England. There’s a sprinkling of states that allow you This was Cambridge, England, this one.

They don’t have a lemon law in England, or a First Amendment. Boo,

Anthony: England. Okay? But hey, if you live in another state and you’re like, wait, my Tesla’s falling apart, you should go to autosafety. org and click on lemon laws. You read all about your lemon laws. And then when you’re like, I’ve had enough lemon, click on donate because it’s the end of the year and now’s the time to do it.


Fred: Michael, how many states have lemon laws and what was the Center for Auto Safety’s participation in getting that done?

Michael: So every state and the District of Columbia now have lemon laws. The effectiveness of those laws varies greatly. We see some states with great lemon laws that are really helpful for consumers like California.

Versus states, I believe Illinois comes to mind who just don’t have the types of protections. We’re looking for, basically a lemon long. We’d love to see if you have a safety issue that the manufacturer can’t identify, or that requires your vehicle to sit like the Edmonds blazer in the shop for.

Two weeks to a month and they still have no idea what’s going on. You need to be getting a new car. It’s, it, there’s no reason for someone to put, a huge down payment in and be subject to a multi hundred dollar, a monthly fee for a vehicle that’s sitting at the dealer, unable to be repaired.

So that was the intent of the lemon laws. And we, the first lemon law, I believe, was passed in the eighties and. The center was heavily involved in promoting lemon laws across all 50 states as one by one, the states began passing lemon laws to protect consumers. Because up until that point, you literally had to rely on, some of wide variety of state warranty laws, if they existed or not, that your protections were really You didn’t really have a lot of protections when it came to buying a new car and you had multiple problems with it.

Limit laws typically, the typical way you’ll see that people qualify for them is they’ll have the same problem come up 3, 2, 3, 4 times and the dealer is attempting to repair it and each time they fail and the problem continues to reoccur. But then you have a smattering of states where there are about.

I would say probably 15 to 20 states. That’s just right off the top of my head. I could be wrong that have a safety issue. So if you have a safety problem in the vehicle, then it is, and it can’t be repaired. Then it’s an automatic lemon because you shouldn’t be, allowing the consumer to drive a vehicle off your lot with a safety issue.

So a common sense law. That’s something I think more states should put into play. But that’s sad. Depending on where you live and where you buy the vehicle, in some cases, or where it’s registered, you could have a great limit law in your state, or you could have one that’s completely ineffective.

And despite the fact that there is 1 in every state, there’s still a lot of work to be done to make sure that all of them are fair. So hey, if you’re looking

Anthony: for a reason, wait, why should I donate to you guys? Lemon Laws. You’re welcome. Okay? Another reason, Lemon Chiffon Cake. Anyone who donates from now to the end of the year, I will eat Lemon Chiffon Cake.

I’m not really sure what it is, it sounds delicious. And hey, how about this for a third reason to donate? Have you met Fred Perkins? Have you? I have it in person, physically, but I see him every week. How about the Tao of Fred? How about we do that? And he’ll deep dive us on ARC submission from automaker techs and all sorts of qualifications around that.

How does that sound? You’ve now entered

Fred: the Tao of Fred. That sounds good. I think that the people who do know me have already checked out, but that’s okay. We’ll catch the rest of these people. ARC one of our favorite company. A lot of other companies responded to Nitsa concerning the a RC airbag inflator problems that have been discussed earlier.

The problem with those a RC airbag inflators is that they occasionally explode, and a RC is trying to determine through, oh, blowing a lot of smoke, that it’s really not a problem and nothing to see here. Folks. Why don’t you just move on? So NHTSA has collected comments from not only ARC, but a lot of other companies.

And what, the heart of this is actually discussed on page 11. I’m going to read this, it’ll put you to sleep, but it’s not long. So it says, the relevant LAT for NHTSA’s investigation involves ballistic testing, typically at various temperatures, i. e. hot, ambient, cold. These tasks can be performed in various amounts, at various frequencies, or at different frequencies across periods of time depending on customer requirements set as appropriate for the particular implementation in the end vehicle and other factors present during a particular period of production.

For example, a new model launch may involve more testing at startup than stable production. According to information in the confidential file, these tests were performed at ARC since production began and were deemed successful. It’s worth noting that where issues were identified in LAT, the process at ARC involved yada.

What is LAT? LA, I’m glad you asked. I was just circling back to that. LAT stands for Lot Acceptance Test. Now, when components are developed for industry and military use, they go through a two stage process. Essentially there is something called a qualification. Where you test the device in a lot of challenging circumstances to make sure that it actually will do its job over the long term.

Lot acceptance test is done with every group of these devices that you manufacture to make sure that they’re consistent with the design that you put forward and tested in the qualification test. So the qualification test is a higher level test. It’s more severe. Loud acceptance test is done more frequently, is less severe.

So there’s a lot of landmines in this statement that I just read. For example, it says that the testing is done in various frequencies or different frequencies across periods of time, depending on customer requirements. Who would the customer setting these requirements and why is Anitza making sure that the customer requirements are consistent with personal safety?

That’s not happening, right? There is no NHTSA requirement that says all of these customer requirements must meet at least this minimum standard for safety. So whatever the hell a customer says they want to do is what these customers are testing to. Or what these

Anthony: suppliers are testing to. The customers are the vehicle OEM, so the Ford, the Honda, the Toyota.

Okay. buy an airbag like what level of detail they are they asking for they’re kind

Fred: of nobody knows

Michael: all we know is that there were about 40 different levels of detail that were asked for so amongst this population of arc airbags of 50 million airbags there are about 40 different types of inflators, even though the inflators fall into two main branches or two main designs, the designs differ from that level down.

And, there are about 40 different implementations. So the, these folks say that you can’t really conclude that a safety defect exists in the subject inflators because there’s so many different types here. I’m not sure if that argument flies or not. Oh, no,

Fred: that’s pure bullshit.

Anthony: I guess from a consumer perspective, I’m thinking of like an airbag as being like, Oh, I need an ethernet cable.

It’s just I just, I need this long. That’s what I need. Great. Done. I don’t think about it. ethernet

Fred: cable were filled with energetic material, they needed to Blow up every time you use the internet. Yeah, that would be similar, but that’d be different. ,

Anthony: Like what I’m trying to say is, but

Fred: typically those cables don’t blow up.

Anthony: True. What I’m thinking is it’s just it’s just something like, oh yeah, I need this part. Like it’s, I need it. Yeah. I’m possibly thinking about it way too simplistically. Yeah. Obviously I am, but I’m like, Hey, I need a driver’s side airbag that fits in the steering column. Hey, it’d be funny if when it pops out a little jack in the box face comes with it too.

No, we can’t do that. So you’re just

Michael: assuming safe functionality, in a

Anthony: way. Of course! It’s an airbag. I’m not assuming it’s, I’m not expecting it to be a jack in the box to pop up my face with a knife and a gun. I’m expecting it to be, yeah, it’s, it seems self explanatory, but apparently I am naive.

Fred: Oh, you are. And it’s a charming part of you. We like that, but again, parsing this information, depending on customer requirements. What is the customer requiring? Okay, that’s different for different different for different applications. And then it says, according to information in the confidential file, the tests were performed since production began and redeemed.

A lot of acceptance tests were performed. Since production began and were deemed successful. What are the criteria you’re using for a successful lot acceptance test? If, and then it goes on to say on page 12, according to A. R. C. Beyond their inline process controls and quality checks, in the case of a lot acceptance test abnormality, the production lot is held for further investigation, and the lot is potentially scrapped, depending on the results of the investigation.

Boy, a lot of weasel words in that one, huh? Yeah. We’ve built some, and we found one that’s really junky, and yeah, maybe we’ll scrap them, maybe we won’t. So let me contrast that with. The standard that’s used for industrial and military use. Now there’s a standard called MIL STD 322.

This is freely available on the internet and as a helpful hint to NISA, you can use something called the internet and a search engine to find it. 1. 1. General, the standard has a twofold purpose, establishing tests and procedures where electrically initiated explosive components, like airbag inflators, must pass, as well as characterizing electro explosive device EED designs.

Is that so hard? Why can’t NIST do the same damn thing? Because we’re talking about people dying here.

Anthony: I assume they did.

Fred: They did not. You’re charming. Okay, 1. 2. Application. This standard applies to electrically initiated electro explosive components prior to completion of development. Again, what’s, oh, what is hard to understand about this?

Before you put these things out on the street, check them to make sure that, they’re done. So what are the criteria that is used by this military standard? It says the reliability of functioning for the samples shall be at least 0. 99 with an 85 percent confidence level when tested at a predetermined all fire input.

The criterion for passing the test is that no more than one failure occurs or no more than one sample fails to meet the output requirement when testing 385 samples or the 230 consecutive samples function without a failure. They give the opening to say one of them didn’t work, but they tell you what you have to do to pass if one of them doesn’t work.

Anthony: And they have a specific number that you have to

Fred: test. There’s a specific number, and that’s based upon the statistics of what it takes to get to that level of reliability with the 85 percent confidence level. Now, to remind people, what that confidence level means is that if you were to conduct this test many times, you would find that 85 percent of the time in that test series, you ended up with 99 percent reliability.

Okay, so it’s like flippin that penny again, okay? If you don’t get the result, you, if you flip it a million times, you’re going to come up with a 50 percent reliability of head versus tails. If you flip it 10 times, you’re not going to have that. You’ll have some other number because, there’s the granularity of the data, right?

As we said, if you flip it three times and come up with heads every time, you’re going to say, I got a hundred percent chance of getting heads. But that’s going to fall apart as you get to more and more samples. So that’s what that confidence number is all about. How well have you really done the test?

How do you know that in the future, if you did the test over, you would get the same kind of results? So what are the, again, what are the criteria that they use as part of their lot acceptance testing and qualification testing? They are high temperature storage, functional tests at extreme temperatures, sensitivity to static electricity.

Jolt, which is a sudden acceleration, a 12 meter drop, transportation vibration, thermal shock, temperature and humidity, and they cite the standards used for that, a high frequency vibration, and waterproofness. Now we’ve, we know that airbag failures in the automotive industry are accelerated by their exposure to humidity.

And thermal cycles, right? So at least, for the last 25 years, according to this middle standard, the three 22 people have known about this, people know how to build these things. The manufacturers who are building these inflators and all their various sizes and shapes have been working in the military environment for years and years.

They know all about this. This is not a, this is not news to them. So why is

Michael: that standard real quick with the GM Hummers? The Hummer or the a MC Hummers that are built for military applications be required to meet an airbag standard, a military airbag standard versus,

Fred: I don’t know the answer to that, but

Michael: I, I assume they could be contracted out, but I just wonder if that comes into play

Fred: at all.

Yeah. I’d have to go back to military two and a specific circumstance, so I don’t know the answer to that. If it were a standard military. application. Yes, it would. I don’t know if the military includes passenger

Michael: airbags, right? I know they’re probably not required because it could be an exception.

Yeah, over a certain weight. But it’s interesting to wonder.

Fred: So I don’t know the answer. But my point is only that This information is out there, and it’s again, MIL STD 322. Look it up, read it, use that as the basis for criteria that are required to preserve the safety of the 300 million Americans who are driving these damn cars.

This is not rocket science, folks. This is quality control. We know how to do Qualification tests on these devices, we know how to do lot acceptance tests, we know how to do all this stuff. We just need the government to stand up and say, you can no longer put explosive lethal junk in the airbags that are three feet away from people’s faces.

You have to use these criteria to make sure that they’re safe. And these criteria exist, they’re well understood by the industry. I’m just mystified and chagrin that the government has not stepped up to, if not this standard, then at least put some standard in place that says this is what’s required as the minimum standard for safety of these explosive lethal devices that surround everybody in the car.

Michael: They’re also very life saving devices at the same time. So there’s a lot of tension going on there. So I would expect that of the 50 million inflators that ARC built, there had been far more lives saved and injury prevented by them than have. been damaged as a result of the defect, which is, I think that’s absolutely right.

And what NHTSA is really going after there, I think is a safe airbag. We’ve had a history with airbags where we know they’re saving more lives than not. And we know they’re making a profound impact, but, we’ve had the problems with. Airbag inflate or the airbags hurting children, small women in the eighties, nineties, where we had to have some, get some sensors into the cars to detect people, to make sure that airbags aren’t deploying, and hurting people.

And now we’re having somewhat a kind of a rehash of that with these electric explosive problems with Takata and with ARC. And so I think NHTSA just wants. To have an Anthony airbag, I’d call it one that you don’t have to worry about its safety at all. It just fits in your steering wheel and goes.

Yes. The

Anthony: Anthony airbag brought to you by gloob. Do you want to crash? Do you want to live? Oh boy. Oh boy. Get the Anthony airbag available for purchase online.

Fred: Hey, remember earlier in this episode, we talked about Daihatsu and the quality problems because perhaps Toyota never looked deeply into what the Daihatsu is doing.

That’s a very similar problem. Nissan needs to look into these airbag manufacturers, set standards for them, and then force the manufacturers to subscribe to those standards. It’s a very simple step. It’ll take them about a week and a half to do the engineering on this, and then however long it takes to get the process done, but, boy, is this a huge gap.

Anthony: We’re going to have a, for our listeners at NHTSA, we’re going to have a link to MilSpec322.

322. I was looking at it right there. All right. I found a link to it. That’d be great. So you donate. Then you go drink a little coffee and then you. Then you read Mil Spec Standard 322. Oh boy, oh

Michael: boy. You may need some eggnog

Anthony: with that one. Are you suggesting people drink and drive?


Michael: I’m not. No, they’re reading military. Oh. They shouldn’t be driving at the same time either, for sure.

Anthony: Even in my Mercedes Level 3 driving car? No, we shouldn’t be doing that either. But here, we’re going to change We’re gonna change tone a little bit, and talk about what kind of cars people like to drive drunken.

Yeah! That’s right, they collect data on this. There’s an article we have a link to from Jalopnik, cut titled, DUI offenders drive these brands the most. Okay, so I don’t know if you guys look. Yeah, I

Michael: mean, I have one take on this, that whole article. And it is, are you being

Anthony: upset? Cause your vehicle’s number

Michael: one?

No, my vehicle was actually number 10. You’re reading the article backwards. Number one is, just something that you would expect. It is BMW drivers and they were almost double. Literally, almost double the rate of their nearest competitors. So BMW people, we already know that most of you are assholes.

Don’t be drunk assholes too. Except for you, my next door neighbor. You’re great.

Anthony: Yeah. BMW with a staggering 3. 13 DUIs per driver. Oh my god. Not per driver. That’s what it says there. With a staggering per driver. That’s their phrasing. They’re far more likely to be driving under the influence compared to the general population.

Last year, a highway trooper in Florida had to put her cruiser in the path of a BMW 335i hurtling towards a 10k 10k foot race. Ah. But hey, number 10 on the list was Volkswagen. Surprise. Let’s see hey Subaru doesn’t, oh,

Fred: Subaru, let me just jump in for a second because I want to report that I one time saw a BMW on an interstate highway, it was in Maryland, actually the interstate highway between Washington and Baltimore.

I saw a BMW being driven at the speed limit. It was a remarkable

Michael: thing.

Fred: And it was about 10 years ago, it was it really sticks in my mind. In fact, the other thing that sticks in my mind about that, we were talking about coin flips earlier, is that I was once in a friendly restaurant, West Housatonic Street, Pittsfield, Massachusetts, with a friend from high school named Kim Lovejoy.

And we flipped a coin to see who was going to leave a tip and the coin landed on its edge, a penny. It was a remarkable thing. I never expect to see that again, similar to the fact that I never again expect to see a BMW driving at the speed limit.

Anthony: These are great coin flip stories. I don’t think I’ve seen a coin in a decade, but I’m glad that you lived inside of a Twilight Zone episode briefly.

You can flip your

Fred: credit card. It’ll work out the same way.

Anthony: Yeah, so Subaru makes the list. They surprisingly a third among Ameri Their, their language in the article is very poorly written. But I will repeat their confusing phrasing. Subaru is surprisingly third among a sea of American automakers, with 1.

45 DUIs per 1, 000 drivers. Subaru is not an American automaker. But hey, my my vehicle of choice does not make the list, they just haven’t caught me yet. Ha, eggnog in the trunk, straw to my mouth, I don’t know what I’m saying. Hey, you guys wanna do some recalls? Let’s do some recalls.

There are some big strategies. There are some, and okay, look, I’m gonna start, I’m gonna This one here personally affected me. I was, I read about it when it came out and I was like, Oh my God, is this me? This is Toyota. Toyota recalls certain 2020 to 2022 Toyota and Lexus vehicles. And this is an airbag related issue.

It’s the sensor in the front passenger seat could have been improperly manufactured, causing a short circuit. So it would not operate. And so this morning I was like, I have a 2020 Toyota. Oh my God, my model’s among the list. And I did my VIN check. I didn’t use the VIN, I used the. License plate, and I don’t have a recall.

Michael: Hey, guess what, Anthony? They don’t update that for a few days. Yeah, you’re going to have to go back. See what they do first is they announce the recall because they’re required to announce that within a certain amount of time and determining there’s a defect. But that information and the VIN numbers that it applied to don’t get uploaded into the NHTSA database for at least a couple of weeks.

I’m not sure the exact timing of that. But I searched on Toyota’s database. Toyota announced, you may want to search again in a few weeks,

Anthony: just to be sure. They list, I might be okay, because it lists the Corolla, and I have the Corolla Hatchback, and that is not listed,

Michael: so maybe I’m okay. You may be right.

Yeah. That one is I’m trying to figure out, this is a pretty large one, so it’s about a million vehicles. They actually just, Dump the five, seven, three on the Nitsa’s website this morning. That was Toyota put that out in a news release around Christmas. I guess to beat the official announcement from Nitsa.

But basically. Just like I was talking about with the airbags, there’s an occupant classification system sensor that detects, how big the person in the front seat is, how big the person is, and where they’re positioned in relation to the airbag and how much force the airbag can deploy with and.

In these vehicles, the sensors got a cracked capacitor and is inoperable in many cases and won’t be able to report to the vehicle the occupant size for airbag deployment. So you could see a situation where an airbag is deploying is. aggressively in response to a crash when there’s a small female or even a child in the front seat.

And that’s not something we want. And that’s why the Occupant Classification System Sensor needs to work.

Anthony: All right. So get that checked out next week to see if you’re in the recall. Is that what you’re suggesting? I

Michael: would say, when you hear about a recall, I would suggest, waiting at least a couple of weeks to a month before verifying whether your VIN is In the recall just to be sure, because there is a delay between the announcement of the recall and the time where that system’s entered into the VIN check system.


Anthony: that’s my second complaint against NHTSA this episode. Come on, that should be faster, okay? Making me wait a month. Anyway, let’s go on to Honda, a rare entrant to the recall roundup. Potentially two and a half million vehicles. 2018 to 2020 Honda Accords and 2018 to 2020 Acura TLXs and Honda Civics.

And yeah, it’s a Honda Clarity FIVs and CRVs and CRV hybrids and the Honda Fit and oh my word, the list goes on and on. And I can’t even get to the description. Okay. Fuel pump issue. The fuel pump. The fuel pump impeller was improperly molded, resulting in low density impellers. Over time, the low density impeller can deform and interfere with the fuel pump body, rendering the fuel pump inoperative.

Oh that’s not good. I want my fuel pump to pump.

Michael: Yeah It’s a, basically the ultimate problem here is that if you don’t have fuel pump working, you don’t get gas to your engine and your engine can stop stall in the middle of the road causing a crash or just putting you in a situation that’s unsafe.

So I believe you’re going to since we have a problem here with a, with a. With a low density impeller, the fix is going to be installing a higher density impeller, and also it looks like they’re going to add some clearance between the impeller and the fuel pump body. So although they, they alluded to this being a manufacturing defect, it just has all of the markings of a design issue.

And. It’s, that’s a big recall for Honda. That’s definitely hurts, but it’s obviously necessary.

Fred: iT could be both a design defect and a manufacturing problem, but I, yeah, when you qualify something properly, you freeze the manufacturing process. Once you get it right, you do everything that’s required to make sure that you consistently get it right, particularly when you’re putting out high volume parts that have to last a really long time.

It’s part of the validation of this design. Might have been and should have been a testing and it’s a highly accelerated life test to make sure that this impeller would last for the life of the car design standards should be set up so that things work until the expected life of the vehicle until the expected life of whatever is being used has been exceeded.

There are ways to do these tests and clearly Honda did not do that for this particular part. This is probably once again, that’s something that was provided by a supplier. I’m guessing on that, but it seems typical of something that would be provided by a supplier. And you have to flow down the right requirements to people.

If you expect them to do the right kind of manufacturing process.

Anthony: If you got that Honda code, check it out, get that. Fixed on. All right. Last one we have here is a little company called Tesla. They don’t have recalls. What are we talking about here? They just had one of two and a half million vehicles.

Now they have one of 120, 000 vehicles. This subject population includes all model year 2021 to 2023 model S and all model X vehicles between. Produced in 2021 to 2023, certain date range for you Tesla freaks. No, mine was produced on the 17th of August. It’s okay. Yeah. Relax. Okay. The cabin door latch mechanism that does not comply with federal motor vehicle safety standard number 214 subparagraph 9.

2. 3B1 may increase the risk of injury during crash. So I’m going to crash and my side door pops open.

Michael: At least your latch at least your door on latches that’s basically that they’re violating FMVSS 214 here on latches and Tesla discovered this supposedly during crash testing in early. This is a, this is a quick recall process here.

They were doing routine validation site impact testing and their engineering department observed the cabin door. Unlatch after an impact on the other side of the vehicle and that’s not compliant with FMDS 214. I believe this is another one where the OTA’s already been released and the fix is in place.

So at this point you should be able to hop in your Tesla and go down to Piggly

Anthony: Wiggly. Oh, played, sir. Played. I love it. With that, let’s who’s this comes out with, and this is an auto news Ford generated the most U. S. recalls for the third straight year. Aw, look at that, Ford. For a third year in a row, Ford Motor Company had the most recalls of any U.

S. auto manufacturer. According to partial NHTSA data, they issued 54 recalls, affecting nearly 5. 7 million vehicles in the U. S. this year. Ah, but hey, on the bright side, that’s down 21 percent from the number of recalls it issued in 2022. So hey, Ford, quality is something. But

Michael: they’ve tried, they continue to recognize at least that they have a quality issue.

They’ve been doing that publicly for a little over a year now. But yes, they’re still leading the pack and recalls. And, from my perspective, it’s difficult to take the recall number and say, Hey, this company’s having a quality problem. I wouldn’t be able to say that if Ford hadn’t explicitly came out and said they had a quality issue.

Because. It could just be a sign that a manufacturer is taking safety really seriously and recalling a lot of vehicles. We see Mercedes sometimes do a lot of recalls with 1, 2, 8 vehicles, that type of thing. And, when you’re just counting up the number of recalls in a year, it, as far as.

Insight into a company’s quality issues. It doesn’t always give you that. It may give you a better insight into their safety culture than anything else. But any

Fred: Chevy owners will often say that Ford stands for fix or repair daily.

Anthony: Well,

Fred: I know that’s harsh, but they

Michael: also put really stupid stickers on their cars with Calvin and Hobbes.

Yeah, I’ve

Fred: seen that as well. Yeah,

Anthony: but this kind of matches data. We’re going to release next week. Start of the new year. We’ll come up with the list of the vehicles that are most complained about that. Most complaints we got in 2023. Cause we have our own complaint database and it’s like 10, 000s of complaints have come in.

And Ford tops that list multiple times and it’s

Michael: stuff that’s not just safety related, but could be that’s been going on for years. We get a lot of complaints on melting dashboards and Nissan’s for example, and it sounds like a sticky. Issue of something you wouldn’t want, but it sounds cosmetic, but in fact, it actually causes some visibility problems out of the front window.

So little things like that rise up a lot in our complaints when they happen to a lot of owners and those owners don’t feel like they’re being treated fairly. Hey,

Anthony: hey listeners thank you so much for sticking with us. This is the end of another year. I hope you have good and happy holidays. I want to let you know that, listeners like you, you’re not alone.

We had over 10, 000 downloads this month so far. We’ve had a total of well over a hundred thousand downloads and so you’re part of this fun little Autosafety nerd fun community we’ll see you next year and please, as I’ve said a thousand times on the show, please go to autosafety. org and click donate.

Thank you.

Fred: Thank you folks. Thanks for listening. Happy New Year.

Michael: Happy Holidays everyone.

Anthony: Happy New Year. Happy Holidays. Bye bye.

Michael: For more information visit

Fred: www. autosafety. org