"Ford CEO: Why OTA Updates Take Forever For Legacy Automakers" - Article

[Tooney]

"The problem is that the software is all written by 150 different companies and they don't talk to each other," Farley said. "So even though it says Ford on the front, I actually have to go to Bosch to get permission to change their seat control software. Even if I had a high-speed modem in the vehicle and I had the ability to write their software, it's actually their IP. We call it the 'loose confederation of software providers.' One-hundred-fifty completely different software programming languages. All the structure of the software is different. It's millions [of lines] of code."

"That's why at Ford we've decided in the second-generation [EV] product to completely insource electric architecture," Farley continued. "To do that you need to write all the software yourself—but just remember car companies haven't written software like this, ever. They've never written software. So we're literally writing the software to operate the vehicle for the first time ever."
https://www.thedrive.com/news/ford-...-automakers-take-forever-to-issue-ota-updates

 

[tigerbalm]

Tesla doing this from the start – being one of their true innovations – and an early moat that competitors are only starting their efforts to bridge.

When the history books are written – Tesla's software architecture will probably have given them a near 20 year lead in this area.

vw.OS being VW's attempt at this – and though they got an earlier start than Ford – indications are that it is still pretty badly off-course. Car's like the new Porsche Macan EV were originally supposed to launch with it – but they'll have to launch with latest version's of PCM – or risk being delayed for years.

With Porsche's lightweight separation from VW (via their IPO) – and their recent cosying up to Apple(*) – it's possible that they'll never go with the new VW software architecture. Though, as long as VW themselves remain largest shareholder – Porsche's public owners will never have the influence to set direction.

* – they've also made overtures to Microsoft in times past.

 

[f1eng]

This is a double edged sword IMO.

One the one hand having one bit of software entirely written in house allows all the ease of OTA updates and probably massively simpler code and maybe a single controller.

On the other whilst writing software to use a touch screen to control stuff is widely used and so programmers capable of doing it are probably freely available, the car related stuff which has decades of expertise and IP protection built in will be very hard to do well, both from the knowledge and experienced programmer perspective.

Producing state of the art ABS without Bosch or similar IP, getting superb power steering feel and suspension dynamics control without Porsche or similar IP and so forth is on a completely different level and not all of it will be easy to license/purchase.

 

[whitex]

What all that the Ford CEO said boils down to is complete vertical software integration, which is what Elon said he was after from the very beginning. It has its advantages, a lot around ability to patch and innovate.

There is yet another difference that the Ford CEO has not yet internalized, possibly because that is the next step after vertical software integration, or in-housing all the software. That second factor is development and test strategy. Classic automotive manufacturers test the crap out of things, because they know it takes a lot of time and a ton of money to patch things. With an almost instant OTA you can afford another approach, using Elon's own words "Ship it now, fix it later via OTA, pay for hardware fix only if we absolutely have to". While it has its bad sides, especially from a customer point of view, it has it's good sides too. After a decade of driving Teslas, I was a little tired of Tesla essentially using customers to test their software - literally, they do minimal in-house testing, then deploy to a small number of cars, see if any of them break, then deploy to more, etc, etc. Then came the Chevy Bolt battery fires fiasco and it got me thinking, would I rather drive a car which is less likely to have bugs (more thorough testing), but any bugs take a year or longer to fix, or would I rather have a car which has a higher chance of having bugs but those bugs can be fixed a week later. When Bolt batteries started catching on fire, Chevrolet just told owners to not park it near anything the owners don't want to catch on fire, life their residences, and the fix took over a year to come. When Tesla Model S'es started to catch on fire, after just 2 of them Tesla sent out OTA updates to mitigate the damage, 2-3 week later? Yes, the update ended up limiting maximum state of charge for some of the older cars (it was actually based on in-car measurement which estimated some electrode foreign material buildup), there was no world wide issue (to this day some places had "No Chevy Bolts" or "No EV's" parking policy in apartment buildings, caused by the Bolt).

Worth noting that Tesla at some point realized that there exists some middle ground between treating customers as a test fleet for their latest software, and keeping customers happy. They added an option, so each customer can choose to be running bleeding edge software - be the first to get it, or run stable releases, ones which thousands of customers already used and their didn't report major issues.

So I'm still on the fence, would I rather drive a car with software that has been tested less, but which can be updated on a day or two notice, or would I rather drive a car which is better tested, but any bugs which are found are not going to get resolved for a year or longer, as it will take that long to test the next release. Imagine being a MY 2022 Taycan customer and having the alarm going off randomly at night, would you be at peace not getting a fix for that for year, because that is how long it takes to build and test a new release, or would you rather they used the Tesla approach and disabled the internal sensor via OTA until they can get you a new one? Yes that patch might add new bugs, but they will likely still be fixed way before a year has passed.

PS> Full disclosure, over a decade, I did receive 2 Tesla OTA updates which were very detrimental to me. One almost caused a serious accident, and one actually did result in a damage to my car. Both times the problem was fixed within a week, I just happened to be in the bleeding edge group (back then there was no choice of bleeding edge vs. stable releases).

 

[mikezhang31]

That makes a lot of sense, it is incredibly how messy a typical company's software stack is. Billing system, CRM, the controller inside the actual product itself, are all from different sources and it takes forever to make an incremental update. The convoluted bureaucratic mess is also what keeps half of the population employed, ha

 

[tigerbalm]

Producing state of the art ABS without Bosch or similar IP, getting superb power steering feel and suspension dynamics control without Porsche or similar IP and so forth is on a completely different level and not all of it will be easy to license/purchase.

Do we have any sense of how much of the "chassis" control systems (RWS, PDCC, 4D Chassis control, braking, steering, etc) is in-house Porsche? That software always feels flawless to me and I assume whatever is being used – has a long heritage and has been battle-hardened for generations.

Is the "Porsche feel" in-house developed?

 

[f1eng]

Do we have any sense of how much of the "chassis" control systems (RWS, PDCC, 4D Chassis control, braking, steering, etc) is in-house Porsche? That software always feels flawless to me and I assume whatever is being used – has a long heritage and has been battle-hardened for generations.

Is the "Porsche feel" in-house developed?

I believe so from my contact.