It’s a terrible time to be in the software department of a major automaker, as some of these organizations continue to demonstrate an inability to do the basics. The basics, in this case, are incredibly, almost overwhelmingly complex. More than 1,000 of GM’s software engineers just got the terrible notice that their services will no longer be needed. It’s possible they won’t be the only ones.
The car market in the United States has been defined by increasing inventory, which is not quite matched by increasing buyers. That’s finally starting to turn around, but the 2025 models are coming! The 2025 models are coming!
Unsurprisingly, this coincided with a drop in profits for dealers through the first half of 2024.
Full Disclosure: I am Baja Blasted after Monterey Car Week and am writing this from a hotel bed in Los Angeles I’m scarcely able to lift myself out of. Let’s Dump!
‘Legacy Automakers’ And The Software Problem
The term ‘Legacy Automakers’ feels like something Tesla stans made up to describe companies like Volkswagen and General Motors. It was meant to imply that these organizations lacked the high-tech, Silicon Valley-style agility of startup car companies (even though these startup car companies were initially made up of engineers largely from ‘Legacy Automakers.’)
When it comes to software, however, the stans maybe have a point. The approach needed to build new cars (shove a lot of engineers into the problem) doesn’t seem to work when it comes to coding software, whereas the Silicon Valley approach of leaner teams seems to be more effective.
Volkswagen learned this the hard way, ultimately agreeing to spend about $5 billion with its competitor Rivian in a bid to gain access to that company’s software (in spite of having its own giant software arm, CARIAD). When it happened I wrote:
Volkswagen, like most automakers, realized a few years ago that what Tesla was doing was shifting the desirability of cars purely from their stats/comfort/style (hardware) to their added electronic abilities (software). This is how we get the industry term Software-Defined Vehicles or SDV.
At the time, VW did what a big company is going to do and created a company called Cariad (originally Car.Software, which, lol). The company staffed it up with thousands of engineers to build a software platform for all Škodas and SEATs and Porsches and Bentleys and everything else.
Believe it or not, the move-slowly-and-don’t-break-things approach of VW didn’t work.
I mention all of this because what happened to Volkswagen seems to have infected General Motors as well. From CNBC:
General Motors is laying off more than 1,000 salaried employees globally in its software and services division following a review to streamline the unit’s operations, CNBC has learned.
The layoffs, including roughly 600 jobs at GM’s tech campus near Detroit, come less than six months after leadership changes overseeing the operations, including former Apple executive Mike Abbott leaving the automaker due to health reasons.
“As we build GM’s future, we must simplify for speed and excellence, make bold choices, and prioritize the investments that will have the greatest impact,” a GM spokesman said in an emailed statement. “As a result, we’re reducing certain teams within the Software and Services organization. We are grateful to those who helped establish a strong foundation that positions GM to lead moving forward.”
That is a lot of layoffs.
The departure of Mike Abbott for health reasons was a big surprise, though it’s not clear how related this was. GM eventually replaced Abbott with two more ex-Apple folks this summer: Baris Cetinok and Dave Richardson. Both of those leaders are based out of GM’s Mountain View Technical Center in California and most of the layoffs seem to be outside California (though the Warren Tech Center is way way bigger than the CA operation, so it might be just proportional).
It’s bigger than geography, however. GM’s software and services division had grand ambitions and, to date, has successfully accomplished very few of them. The company earned a lot of skepticism when it said it was eschewing CarPlay and Android Auto in some of its new vehicles, a move followed up by a series of massive software-induced blinders including delaying its Silverado over software issues. If that wasn’t bad enough, the Chevy Blazer EV won MotorTrend’s “SUV of the Year” and then sales of the Blazer EV and most of its EVs had to be stopped over more software problems.
This is somewhat existential for GM as it tries to double its revenue in an increasingly complex market. Two of the company’s big strategies to get there were: Cruise and subscription software (like OnStar). With Cruise, GM seems to have let the Silicon Valley team do what it wanted, which ended in disaster. The approach to software/subscriptions seems to have been the opposite, with a neverending series of Apple and Microsoft execs coming into a more traditional culture and failing.
There has to be some happy medium between the two and I’m guessing this is GM’s approach, but it’s yet another warning to big automakers that this is way harder than it looks.
Have We Have Reached Peak Inventory?
Here’s a fun graphic:
Product shortages related to the pandemic (and the short-sighted-yet-profitable decision by automakers to cut semiconductors) are over, with carmakers suddenly finding themselves with a little more supply than demand can bear. Will lower interest rates help juice demand to eat some of this inventory? Possibly.
In the interim, it looks like we’re finally reaching a little balance. This graphic comes from S&P Global Mobility and you can see advertised new cars are starting to retreat a little bit. We’re still way up year-over-year, but inventory couldn’t keep going up forever.
The 2025 Model Year Cars Are Already Here
This graphic, again from S&P Global, shows another reason why we might be leveling off a bit: Production of 2024 MY cars is switching over to 2025 MY, some folks are waiting it out, factories are switching over, and all the usual summer model-year transition stuff.
Time flies like an arrow (fruit flies like a banana) and we’re almost in freakin’ 2025. When did that happen? Why is time moving so fast?
By halfway through the year, dealers managed to move most of their 2023 inventory and now the 2025 models are arriving. This means that you might start getting good deals on 2024 vehicles, just FYI.
Dealerships See A 33% Decline In Profit: Report
A new report from a couple of advisory/training firms in the industry (Presidio Group and NCM Associates) is out, and it shows that dealership profits have dropped so far this year.
What’s going on? Dealerships made a ton of money during the pandemic and are still up relative to where they were in 2018, but not all dealerships have adjusted to all this inventory. Here’s the explanation via Automotive News:
“We’re surprised that the typical dealership hasn’t adjusted its spending and variable costs as quickly and appropriately as it should have given the decline in profitability over the last two years,” George Karolis, president of Presidio Group, said in a statement. “This is a time when dealers must be nimble and adjust quickly to the industry’s changing circumstances, and many of them have not.”
Variable gross profit at the average dealership has plunged 32 percent since 2022. Meanwhile, total personnel expense fell by just 6 percent in the same period, according to the report.
This doesn’t surprise me, though there are also some basic minimums to running a car dealership in terms of staffing.
What I’m Listening To While Writing TMD
It’s Boise’s own “Built to Spill” with “Distopian Dream Girl.” This was suggested by Apple Music as part of my “Get Up Mix” and I never thought of this as a pump-y song, but it kinda is.
The Big Question
How much do you care about software in your new car?
Aside from problems with desired skill sets, lowball comp packages and antique tools used in automotive software: there’s a big financial problem in transition.
An analogy is “Why is it so hard to move automotive electric systems to 48Vdc?” So many advantages: lower current, smaller conductors, simplified packaging. The reason seems to be that legacy OEMs want commonality across different platforms and they want to use – let’s say a Body Control Module – across their entire product line. Make it dual voltage and you’ve added cost to the cheap cars, even if it’s a couple square inches of PCB you depopulate. A new OEM doesn’t have this problem.
Software is worse. In 2024 how many of the cool kids want to write functional safety software in C++? How many want to deal with CANBus message optimization? Cybersecurity in CANBus gateways? Code reviews with mechanical engineers and accountants? 6 year release cycles so CPUs and memory are always to small? Which OEM will be the first to build a language and tools group?
The silicon problem is even worse. No single OEM can get to more than a couple 100K wafer starts for advanced silicon, which is mouse nuts in the silicon world.
Software is a tough biz to be in. Companies need to “catch up” and want to develop something on their own in like 1 year. So they hire like crazy, get it done, and dump them. It happens over and over. My company is currently undergoing a technology “transformation”. Anyone in dev has to see that once we’re “transformed” we have no need for devs to be like 15% of the company.
It’s a tough road but it’s not like this hasn’t been happening for the last 10 years. If you want stability don’t chase the money and bed in with a company that has more consistent needs (ie, not hiring a boat load all at once and throwing cash on the hood) and maybe dabble in systems support, get some MS certificates or something. Then you can have hope you’ll last past thanksgiving when cost cutting becomes popular to meet year-end financial goals.
On the other hand, these engineers will most likely find another company that is ramping up for the next cycle. So they can ride the wave again.
Sounds like they were working on turning every drive into a stream of micro transactions or auto-renew subscriptions. Probably the same group who sold customer data to insurance companies.
I have limited sympathy.
It doesn’t matter what I want. The OEMs want their piece of this revenue stream. Selling cars isn’t enough. Most of their customers already carry around a device that does everything anyone could need in terms of “connectivity”. Just mirror it on the big screen.
But no, you can’t let Apple and Google get all the money. Your car manufacturer can offer their own connectivity experience! Download our apps! Pay us for 5G service you already have for your phone! We want to feed you ads too! Because everyone wants more ads!
Regarding inventory, I wonder what the breakdown is with brands. The Toyota dealership is still relatively devoid of new cars unless you want a Tundra. I’ve looked for Chevy Tahoe/Yukons….those aren’t around either. Is that graph being entirely driven by the Stellantis and Nissan lots?
They only had all of those developers to try to squeeze more profit from android automotive so it can be batched with onstar as a pricy subscription that sells your data. And that is exactly where they fall and are scaring away customers.
Other automakers are just using Android automotive as it is. Simply configuring Android automotive to connect to hardware is not very difficult and does not require a large workforce. If they can make an affordable data only plan for the infotainment then they can easily correct more revenue and appease their customers.
As one with a newer car running Android automotive, I will admit than the reliance on infotainment controls is excessive but the connected services are extremely valuable.
I’m a techie Gen X and love me some good tech but sadly haven’t met one in a car yet. It’s either Tesla-extreme (see how to pinch to adjust vents) or plain dumb like VW ID or others. So with this as options, I say give me nothing newer than 2017 so I can live a little and adjust my own damn hvac while driving and without looking at it.
Now, if you came up with something like the Spinner in Bladerunner 2024 then yes, that’s cool and useful tech. Make it so and then we talk.
Side note: is Lucid’s software better? I never drove one but it looks promising. Can A-utopians maybe write an in-depth thing about UI and software about today’s cars?
I’m gen X as well and I’m currently shopping lucid, they seem promising you can’t lose with Saudis ha
^^^THIS^^^ x1,000,000!!!
I get it for flagship brands, but the automakers’ insistence that every vehicle be a rolling showcase of super computing, capable of intergalactic travel, is beyond absurd. It’s also why the average vehicle in the US is nearly $50k. A huge segment of buyers want no part of all the superfluous and nonsensical tech gimmicks and setting adjustments that make it so time consuming and unnecessarily complicated to accomplish what was a relatively easy task a little over a decade ago–driving a car.
Ford sync 3 in my 2016 Cmax is serviceable. So is the GM infotainment in my 2018 tourx.
Neither is going to provide a new revenue stream for their OEM, but they do the things I need and aren’t tremendously frustrating.
Ford actually added CarPlay support to the CMAX at some point. I had to replace the USB hub because the original one didn’t meet apple’s specifications for power output. The new hub was only about 50 bucks and it was a five minute install. otherwise it was just a software upgrade that came for free.
In both cases, the basic functionality of infotainment is accessible and adequate. Each Has serviceable navigation.
GM saved a single hard button by making me go to infotainment to turn off or on the AC. That’s annoying, but otherwise the car is reasonably normal to use.
Of course, this ignores the fact that they sold me out to my insurance company.
They think those developers were expensive before they fired them, wait till they find out how much it’s going to cost to figure out what that software does without anybody who was involved in the project around to ask or finish documentation.
A great many people who manage software are idiots.
It’s conceivable that they’re throwing away the software These folks were working on.
Also conceivable that their software operation is big enough that every team still has people.
Maybe I’m old and out of touch (who knew that would happen at 42), but I like the fact that my 2018 GMC Canyon doesn’t need software to run anything other than the infotainment system, which if it breaks, will not brick my truck.
I don’t want something that needs smartphone levels of capability to drive. I don’t want something that needs an app for access. I want a car that I can unlock and start with a key and drive away without a retinal scan.
Anyway.
Just about everything relatively small and cheap is flying off the lots, there’s rarely more than a few in stock at any dealer.
Meanwhile I saw the area’s first post-covid reappearance of an offsite overflow parking lot, at a movie theater that hasn’t been open since before covid, full of brand-new Fords. Every one of them is an optioned-up (possibly just lightly, at least not “fleet white”) fullsize pickup or SUV.
It’s not so much too much inventory as the wrong kind of inventory, and that ties back into the “post-pandemic era will last forever” mindset of the dealers. Someone, somewhere, fails to realize how much the big and fancy market’s been saturated and that so many of those customers are on such long payment plans that they won’t be back for a very long time.
I love my Bolt. Having said that, the five year “free” access to the synchronization between my car and the MyChevrolet app (that I didn’t know was free and only for 5 years) just ended. To continue, it will be $15/month. Until I pony up, I can’t use the app for anything but scheduling service. No checking the charge percentage… no remote start… no find my car. (I already pay for the OnStar security package).
Now if the app was decent… think charge percentage + plug share + ABRP, I might pay. But the app was crap… and an EV without an app is not great.
Had you asked me last month if I would choose GM for my next EV, I would have said for sure. Now I have no brand loyalty and in fact a bit of disinclination towards GM.
They should fire them all. Given how universally terrible the interfaces are in cars today, they obviously suck at their jobs.
The miserable user interfaces are a large part of why I have no interest in buying any new cars at all. There is nothing I need to do in a car that I can’t do with the handful of knobs and buttons and two-line display that my pair of 2011 BMWs have. And 95% of the time, I can do it without taking my eyes off the road.
You can (and probably should) blame the UI designers, but it’s a bit unfair to hold the guys who write the code responsible. They’re stuck between the hardware handed to them and the crappy UI design forced down their throats whether they liked it or not. (Source: been there, done that.)
No it’s fundamentally down to bad management and bad process. The problem with is that UX at legacy OEMs is run by stylists, not designers, who have no idea how to think strategically or develop a framework for what the end user values. Instead they specialize in making flashy iconography, whether it’s exterior sketches or UI icons. That also falls flat because taste at OEMs is so far behind the zeitgeist due to the slow moving culture at OEMs and more generally in the Midwest where most of this stuff is developed (Detroit is about 10 years behind the coasts in knowing what’s cool). There are more than a few young people in the design departments who try to propose more forward-thinking approaches, but they get smacked down by clueless boomer managers.
If I were to be charitable, a capex intensive industry like automotive leaves little time to think strategically and the bloat/ unnecessary layers of management certainly doesn’t help. In reality, the lack of a user value framework and a way to validate that framework means that way too many projects are happening at once, each driven by whatever the manager in charge thinks is cool and each trying to capture a marginal part of the market. A more focused product roadmap with more time and resources (but less bloat) would have a higher chance of success.
Source: am a young person in a OEM design department.
Sounds like you are there and are doing that. Best of luck, from a hopefully not clueless boomer non-manager. (Brutal take down, by the way, but probably accurate!)
PS – every failure in business eventually boils down to bad management/bad processes. They’re two sides of the same coin. That it’s “bad” isn’t the issue, it’s what is or isn’t done to improve the situation that distinguishes good management from bad management, and yes there is way too much bad management out there.
“clueless boomer manager” was definitely unnecessarily harsh and I apologize for that, especially because these generational divisions are completely made up. I meant people who are too stuck in their ways about how things “should be done”, which definitely transcends age.
I’m 53 and work in software field support for a government agency. “Struck down by clueless boomer managers” is the status quo. Anything that can be fixed or improved needs to “follow the process”. Anything and everything new or improved I can suggest or even show proof of concept is pointed towards the process where it goes to die. So it’s not just young folks who can’t stand our antiquated processes, lack of risk or innovation, and the managers who think this is the way.
Get rid of it all and then there is no need for coders.
Well, that’s not entirely true, but there would definitely be a lot fewer. At this point embedded systems are here for the long haul, so at least SOME programmers will always be necessary.
I look for the minimum required software interference, but it encroaches on and infects more things every damn year. Whether it’s constantly changing clueless customer requests, managers, budgets, garbage engineers, or a combination of them all, software development has got to be the industry flooded with the most incompetence, or perhaps its ubiquity and pervasiveness has made their prodigious weaknesses more frustratingly apparent than any other industry that might more fightfully claim such an honor. Software should work in such a way as to be nearly unnoticeable. I find it to be anything but and it’s noticeable for the wrong reasons. It seems like they knew how to write code until sometime in the early ’00s. Vee below mentions 2006 and I wonder if that’s no small coincidence and that much of what gets done just seems to be patches of bad code written upon layers of bad code by incompetent people who change things not to improve, but to merely justify their continued employment. Perhaps the bad code is intentional job security/opportunity, but that almost seems like too much credit. To finish on something more positive, when I do find well well designed and thoughtful UX, I feel like grabbing the shoulders of any random person I can find and shouting about the miracle I’ve experienced and how perhaps humanity is redeemable after all.
The UX is a tiny, and hopefully easily replaceable, part of the software. Unfortunately, that’s all anybody ever sees.
These tech companies operate on the idea of lateral development. Anything they do is repackaging something that exists, just in a way that allows further extraction of wealth from it. It’s a once in a decade thing where they actually come up with true forward developments. The image they trade on comes from a period of rapid advancement that ended in 2006. That’s almost twenty years ago now. Another way to look at it is that we’re no better technologically than we were in 2006, we’ve only gotten better at making money off of what we had. No, large language models don’t count, they’re technology that has existed since the 1970s that made a splash because the idiots pulled a Nedry and let them out of the highly necessary cage designed to protect us from them.
Because of this inflated and undue level of recognition these tech companies disregard any laws to grab as much money as possible as quickly as possible. They don’t run leaner, they run immorally and illegally because can perform crimes faster than the law can catch up. From mass layoffs in the tens of thousands (Hello Google! Hello nVidia!) to stealing user information (remember the recent data broker leak of three hundred million people’s Social Security Numbers?) to insane forced legal agreements (Disney saying that trying Disney+ means they’re not responsible for your death in an unrelated amusement park), they’ll pull any lever they can to keep the machine running. No, that is the correct lever, Kronk. There is no wrong lever.
Ultimately the automakers’ problems aren’t the software. It’s the fact that they and other manufacturing companies are trying to compete in the new scheister’s sphere by using the tricks from back when they were the reigning scheisters. You cannot underhand the student who learned from you and your peers’ worst behaviours, because they know what your moral and legal limits are and they will train themselves to not develop any limits.
“It was meant to imply that these organizations lacked the high-tech, Silicon Valley-style agility of startup car companies”
As a sometimes-Tesla-stan, the term “Legacy Automakers” means more than that.
It means “Automakers heavily invested in legacy/ICE tech”.
And tied in with that is the implication that these automakers resisted putting any real effort into BEVs.
There was also the implication (which has been well documented) that these automakers often actively lobbied against BEVs and spread disinformation/lies/FUD about them.
And when you look at all the problems that VAG and others have had with software and producing BEVs in volumes, it’s clear that legacy automakers have lacked the tech and agility of the startup BEV makers.
Regarding the 2025 models and the statement “By halfway through the year, dealers managed to move most of their 2023 inventory and now the 2025 models are arriving”
We should also address something else… the fact that a lot of dealers still have “new” inventory from 2022, 2021 and much earlier. Here are a couple of examples:
https://www.autotrader.com/cars-for-sale/vehicle/663565366
https://www.autotrader.com/cars-for-sale/vehicle/644800406
Why is that? Because these sellers/dealers are still living in fantasy world where pandemic pricing will last forever.
That 2nd ad I linked… the fucking dealer wants full MSRP on a vehicle that has been sitting on their lot for 3 years!!!
Fuck ’em.
That thing should be marked down by at least $20,000 IF they were serious about selling it.
Worse yet, you see shit like this:
https://www.autotrader.com/cars-for-sale/vehicle/526186086
A 2017 Ford Fiesta that is still “new”… and even though the fucking thing is 7 years old, they’re only offering $2161 off the MSRP. That thing is also WAAY overpriced.
There are over 4000 pre-2023 “new” vehicles listed on Autotrader alone.
And they’re not gonna sell any time soon with these dealers being seriously detached from reality.
Dealers are a big part of the problem.
And regarding the big question… I care enough about software to avoid vehicles with known bad software. For example, I won’t buy any Fords with the shitty Microsoft-based Sat-nav system.
When I was looking at used Ford C-Max vehicles, I limited it to ones from 2016 and newer which had a newer QNX-based sat-nav system. Also by 2016, Ford fixed a defect with the transmission.
That Fiesta only needs a few more years with no additional miles before it ends up on Cars and Bids or Bring a Trailer…
If it were an ST it would likely already be there. As it is, a base model with the hellspawn eggshells-and-glass DCT, and not even a good color, yeah…no.
Past model years being still in inventory is an issue, but none of those 3 examples are very good ones here.
The Ram dealer has several 2022s, but they all appear to be commercial spec, so not really your average sale.
The Traverse does not show on the dealer’s actual website and the Fiesta’s selling dealer appears to have changed ownership entirely at some point.
Typically the manufacturers have a limit to how low a dealer can actually advertise a vehicle based on incentives and discounts, so they wouldn’t be able to advertise 40% off. Any decent dealer is trying to get rid of that kind of aged inventory, if not for their floor plan but also at a certain point there won’t be any more support for the old year with incentives, so you’re stuck taking a loss on an older model out of their own pocket to sell it for the same or less as a newer model. It’s likely the Traverse and Fiesta are stuck in some inventory feed no longer in use by the dealer or updating properly that is going to the listing site, and not actually physically in inventory any more or part of the bottom line.
That same Ram dealer does have a better example of old inventory still around, they have 17 2023 Jeeps still, for some reason. Although given the Stellantis product pipeline, who knows, maybe they aren’t motivated to move them, so they aren’t left with nothing on the lot to even try to sell.
They seem to not want to turn paper losses into actual losses.
Ah yes, the strategy of hiding and delaying a bad situation, even though it will only ever get worse.
That always turns out well. /s
Ummm, what’s a sat-nav, doc?
Is it like something you get on a $400 phone except it works much better?
Incidentally, mapping is another example of how carmakers could never hold a candle on catching up to silicon valley on the most basic of apps actually needed in a car: a damn digital map. In your dash! GM could have bought tom tom or garmin if they chose to, or pay google early on for integration, but they sat on their hands for, oh, two decades? So yea, catching up to them.
I don’t think the over the air thing is a problem, just as many cars can’t connect anymore because 3G, soon 4G and 5G will be outdated and won’t work either…..
And that’s an issue that OnStar subscribers have faced not once, but TWICE! First time was during the 2008 Analog to Digital transition, but GM had a hardware solution in place if the vehicle in question was upgradable or had a “Dual Mode” VCIM, but this wasn’t the case with the 2G/3G sunset just a couple years ago. Their solution for the 2G/3G Sunset was a halfassed smartphone app or “Trade it in for a new overpriced hyper connected truck or SUV”
“How much do you care about software in your new car?”
it depends…
– ECM engine mapping (stoichiometry!), ABS, traction control are features I want.
-entertainment, hvac, windows, etc. all i want is bluetooth to pair my phone to make/take calls, listen to music and l
ignore prior comment- posted before i was done and won’t permit me to edit.
How much do you care about software in your new car?”
it depends…
Some s/w control for features like engine mapping, ABS & traction control are great.
some s/w control for safety features like windshield wipers and lights is likely unavoidable and ok…as long as the UI is NOT the accursed touchscreen.
s/w control for windows, HVAC, mirrors, remote lock/start are ok-ish, but not necessary to me (and again, as long as UI not thru touchscreen).
ADAS – I am not interested (other than “DAS” cruise control, ABS &TC) cuz they do not improve safety but do increase maintenance costs.
s/w control of entertainment is foolish move by OEMs cuz people want AA/CP…not some half-assed kluge solutions that seem so common.
s/w intended to subscription-ize paywall-able features and monetize operator driving data – F**K NO!
this last spy-ware issue will control my next car purchase…are any OEMs listening
Thank you for saving me the trouble of typing that
I would prefer if getting newer cars that it does not have software bugs like *insert Todd Howard it just works meme here*. But also I think there is just to much software at this point in cars it has gotten ridiculous. I work in R&D for semi’s which lags behind the normal automotive world but even then there are so many different modules that have to communicate with each other if one has issues it can down a truck.
But the nice thing about modules is that you don’t have to fix the entire stack. The architecture should allow the system as a whole to function even if individual modules fail.
The test should not be “how well does it work? “ the test should be “how well does it fail?”
Margaret Hamilton figured that out in the 60s.
While in some cases you can have a module fail and it not affect other systems, certain failures can take out the CAN bus and thus everything else. Chrysler instrument clusters used to be great for causing a no start and no communication when you tried to scan them. Unplug the cluster and its included module and the car would start and the scan tool could communicate.
I have run into issues at work a can bus gets over loaded because some parameter is not set correct to record data properly and the truck goes haywire and won’t start or shift into gear. Granted this is due to being in R&D and having extra devices for data logging so it shouldn’t happen in a normal driving situation without all that data acquisition.
I never understood the reasoning behind can bus. “Everyone loves standards, that’s why there are so #%¿!ing many of them”
Well in any other industry one piece of hardware on a network taking the network down would not be acceptable. I guess when you are your own customer the vendor can screw you over.
But…
When I was at Cisco, a bug was defined as anything that was documented as something you could do with the product ie “supported” and was reported by a customer as causing a problem. If we discovered in testing that some random thing ( I wrote a random thing generator for my own use as a developer) caused a problem, it was not a bug until a user not only triggered it but could reproduce it and report it through the chain of support.
Part of the problem was they thought of themselves as a hardware company, but they were in the software business.
For various reasons I left.
I suspect that auto makers generally think of themselves as being in the automobile business not the software business, yet there they are.
The upcoming vehicle to vehicle communications protocols really need to be taken away from the manufacturers btw.
I think the big reason that CAN prevailed/was mandated is that Bosch had the profit motive and $$$ to make it happen, while SAE and ISO didn’t.
As to why do they do busses in general it is cheaper to have only one _____ sensor and let the module that it is connected to share that with the other modules that need the same data.
The problem is when one of those modules go down that has some widely used data it can affect a number of modules.
One of my cars uses the ATE MK25 ABS module which I recently had repaired. When they fail, and they will fail, they take the VSS signal with them. The instrument cluster, PCM, Power Locks and even the Radio are all looking for that VSS data. Driveablity definitely suffers with rougher shifting and weaker throttle response but it will still get you where you are going. The speedo and odo of course don’t work and neither does the auto locking and speed sensitive volume.
A device blabbering garbage can take out any network.
That said I presume canbus failures can be caused just with a bad link in the middle.
Yup as can one that shorts the bus out. A bad connection somewhere in the middle won’t necessarily take out the entire network but modules that are still online can be left wanting for data from those off-line modules.
I don’t care about software in my current car at all because I have a DIN head unit and no drive modes/traction control to deal with.
But if I had drive modes, active driver aids other than ABS, and/or an embedded infotainment, it would become extremely important. If the software locks me out of certain controls (such as steering, suspension and drivetrain modes, or undefeatable stability control) or lacks basic features (like wireless android auto), I’d think twice about buying a new car with it.
Embedded systems made software inextricable from the chassis, and now that manufacturers finally realized they need to keep up with the tech industry, they’re in too far. They could’ve gotten away with lackluster or outsourced software if they kept the head unit independent, customers would just replace or upgrade it as needed without bricking the entire car.
Worse even, cyber-security evolves hundreds of times faster than vehicle production life cycles, and when the module that talks to your phone also talks to the level 2 ADAS, the stakes are way higher than an isolated system. I still find it mind-boggling that manufacturers are willing to connect chassis controls to the internet.
This is something I don’t understand how it is legal? Like why yes my car should always be connected to the Internet so hackers could do who knows what to it while driving.
Because you know what moves even slower than vehicle production life?
Government.
“ Embedded systems made software inextricable from the chassis”
I had a Chevy Volt. I loved it, until I didn’t. Snap your fingers and that’s how fast I went from like to dislike.
Chevy had a module on the bus that in hot, humid weather, would brick the car, requiring a full system factory reset by a Volt technician.
NOT covered by warranty, according to Chevy. Say what? The hardware in the drivetrain is useless without the software, so the “drivetrain warranty” should cover everything in the drivetrain system. But even that wasn’t what moved me from like to dislike.
It was the battery cell that costs $99 but $5k in labor to replace that killed it. Sold the car for black book and won’t by a GM product again.
Meanwhile my 2 Toyota Prius’ (2011 and 2016) run just fine, and the 2016 hs Bluetooth that my phone “just take to” vs Checy’s “your device might work” approach.