125 replies to this topic

[Magoo]

John McElroy of Autoline gets it. 

 

 

[mariner]

I watched this and I honestly don't think it adds much. It's clear software skills aside, Tesla has had two advantages

 

- it has had just one vertically integrated plant ( like Henry Ford did at the beginning) so everybody is co located by default.

 

It's direct sales model has very large advantages at start up because not only does it save inventory but when you go toa new country the start up process is just to set up an import / distribution centre and start web adverting.

 

Whether both those advantages will survive plants now in Fremont, Berlin, Shanghai and soon Mexico is a TBD.

 

Also the direct model leaves Tesla residuals in the hands of the auction operators with no chance to manage second hand inventory and hence prices. In a leasing dominated world that's a residuals risk which the leaseco's may or may not price for.

 

I can't comment on the car's centraliesd software vs distributed CANBUS but all the integrated design, order, mfg  control ,software is basically same as SAP .which most businesses use.

Edited by mariner, 20 August 2023 - 10:05.

[Chubby_Deuce]

Did they mention the part about just constantly lying until you reach profitability?

[Magoo]

Here is Ford CEO explaining some of the problems with how electronics are done in the legacy industry. 

 

What he doesn't say, because it doesn't need to be said, is that the the solution for the legacy automakers is to follow Tesla and adopt zone processing and integration. 

 

 

 

https://www.thedrive...PEUVJszQGZxM8Q4

[Magoo]

Did they mention the part about just constantly lying until you reach profitability?

 

Meh. I've been around the Detroit auto industy my entire life and I don't think Tesla lies more than any other automakers. It would be a pretty close call for sure. 

 

I know the auto enthusiast media likes to paint Musk as a liar and they are used to telling enthusiasts what to think, but I prefer to form some judgements for myself. 

 

Besides, that's kind of a kettle-black deal. I don't think they're fully qualified to point fingers. 

[Magoo]

Meh. I've been around the Detroit auto industy my entire life and I don't think Tesla lies more than any other automakers. It would be a pretty close call for sure. 

 

 

 

 

In fact, Lying Season is officially open right now with the United Auto Workers' contract negotiations under way. 

 

Historically, the automakers are so habitually economical with the truth that it has become atmospheric. We expect them to lie and it goes in one ear and out the other. 

[Greg Locock]

"It would appear that Ford's future goal would be to ultimately collapse the software (and potentially hardware) side of electronics manufacturing back in-house."

 

Oh. They'll be recruiting coders big time then. Not happening. That ship has sailed. And as for electronic hardware!

[mariner]

I suspect, but don't fully know that the great new business model adopted by the big mfrs about 15 years ago to outsource all sub assemblies work to tier one suppliers , including R+D has back fired on them .

 

That's pretty much what Jim Farley was saying but I think he is maybe blaming multiple parts suppliers too much 

 

In a sense the OEM's are al beholden to Bosch in particular for not  just the electrical hardware but the architectures like CANBUS which Bosch introduced

 

If you are Bosch etc.  all centralised software is a huge  problem as you don't own it anymore.

 

Of course if we go back to the "bad" old days of GM etc. GM made it's own electrics ( Delco), its own fuel systems ( Rochester), its own even its own spark plugs so it would have ben well suited to integration but declining profits vs capital saw it spin that all off..

 

he route back is very expensive. It is bit like when Michelin introduced the radial; tyre - it was clearly much better but as it lasted twice as long it halved volumes so alt eh investment to switch from cross plies to radials was rewarded by a lower sales volume !

Edited by mariner, 21 August 2023 - 09:34.

[Magoo]

When the carmakers buy a packaged solution off the shelf It's convenient, like a TV dinner in the frozen foods section, but there are compromises.

 

Farley says there are 150 different processors on his cars. I was stunned by that number, must be an exaggeration, but still it's a lot whatever the actual number.  

 

These are  mainly generic, commodity chips produced far down the IC manufacturing food chain --- companies that were paralyzed by COVID, some not surviving. 

 

Hence the "microchip shortage" that even now is handicapping the legacy industry. 

[mariner]

Not quite  the same " secrets of Tesla" but handy if you are an electrical engineeriig dummy like me 

 

 

[Greg Locock]

Pushing subsystem assembly (and hence design) off to Tier 1s was a management reaction to unionized labor in the assembly factories.  It always seemed weird to lose profits and control over the design, to me. I can accept that we shouldn't be building HPAS steering racks, perhaps, yet on the other hand were quite happy building engines from steel ingots. But suspension modules? 

[Magoo]

I know a guy whose company provides engineering and manufacturing services to the Detroit 3 and others. We go back. He is very successful but I had nothing to do with it. One night at dinner I asked him why the automakers were so willing to hand over their most critical functions to contractors. "You would think they would want to keep their core competencies close to their chests, " I said. After a few moments he said, "You would think."

[desmo]

Managers want to offshore management headaches/labor so they can coast. Human nature. Management is probably where AI can really shine as a human replacement, rather than on factory floors.

[KWSN - DSM]

AI is Big Data, analyzing, find trends, it's algorithms, mathematics and statistics - There is nothing truly intelligent in Artificial Intelligence, which does not mean as a tool it will have many solid uses across many businesses and daily life for all of us. 6 years ago everything was Blockchain, it may or may not exist as a contributing tool somewhere (or everywhere), but when did a general media article/show last utter the words 'blockchain'?

[desmo]

Human intelligence is highly over-rated as well.

[KWSN - DSM]

Human intelligence is highly over-rated as well.

 

Agree, not artificial though which must count for something.

[Magoo]

I think the trend is partly 

 

Managers want to offshore management headaches/labor so they can coast. Human nature. 

 

 

I agree. The trend is driven in part by the desire for "flexibility." As in the flexibility to bring in hired hands to hop on any trends once they develop, instead of investing in innovation and looking forward. They don't have to use any judgement or take any risks. 

[MikeTekRacing]

https://youtu.be/HrNN6goQe50

 

Jim Farley explains why it's impossible to compete with Tesla on software. 

150 companies (!!!) contributing, and Ford doesn't even own the IP.

 

The short term cost obsession is a huge pain down the road. Now they are in a situation where they don't control anything. 

[desmo]

If I'm a typical C-suite executive suit less than ten years from deploying a golden parachute, I'm not doing anything with no big short-term payoff that could bring potential risk or accountability upon myself. Life is good, why mess it up?

[MikeTekRacing]

That is true and that is the standard of business. That's why disruptions always happen from outside of the business, that's where the biggest threat is. And legacy businesses are most always doomed as they are entangled into complicated business models that they don't want to change.

Kodak died when they couldn't adapt to digital

Nokia died when Apple (with 0 phone history) challenged their model.

History repeats itself.

[KWSN - DSM]

AI which really is not intelligence at all:

 

https://sports.yahoo...-155002525.html

[MikeTekRacing]

AI which really is not intelligence at all:

 

https://sports.yahoo...-155002525.html

I'd argue the bar is so low for management that AI is already better than the average manager

PS before somebody tells me I don't know what I am talking about - I have spent quite a bit of my career working with managers from first level to CEO and I am passionate about coaching people and helping managers become better. The bar is SUPER low, most managers are actually terrible 

[gruntguru]

AI is Big Data, analyzing, find trends, it's algorithms, mathematics and statistics - There is nothing truly intelligent in Artificial Intelligence. . .

Don't kid yourself - its coming.

[GreenMachine]

Don't kid yourself - its coming.

 

Whether or not is is coming is not the question.  The question is where, and in what way, will it be disruptive.

[Greg Locock]

Having mucked around with ChatGPT, my conclusion is that it is basically Markov chain on steroids. There's no logic, as such, just that if these clauses go together, then they are usually followed by this clause. OK, I get the impression some managers are like that.

[gruntguru]

Whether or not is is coming is not the question.  The question is where, and in what way, will it be disruptive.

 

I see two issues that the "she'll be right" naysayers avoid.

 1. When any technology advances to the point where it can perform a task better than humans - we gladly hand that task over. Thus AI will eventually be given complete control of essential services including power grids, water grids, policing, national defense.

 2. Easy to say humankind will maintain control over AI because humans will instill the firmware and software with overriding ethical principles. . . . or not - in the case of a large number of humans (or corporations or rogue states) that any of us could easily name.

[GreenMachine]

Yeah, robodebt anyone ...  

[KWSN - DSM]

Don't kid yourself - its coming.

 

I do not think I said anything else in my post.

[gruntguru]

Hard to interpret this any other way . . .

 

AI is Big Data, analyzing, find trends, it's algorithms, mathematics and statistics - There is nothing truly intelligent in Artificial Intelligence, which does not mean as a tool it will have many solid uses across many businesses and daily life for all of us

[KWSN - DSM]

Hard to interpret this any other way . . .

 

From the post you comment on

 

"...which does not mean as a tool it will have many solid uses across many businesses and daily life for all of us"

Edited by KWSN - DSM, 23 August 2023 - 09:49.

[Magoo]

https://youtu.be/HrNN6goQe50

 

Jim Farley explains why it's impossible to compete with Tesla on software. 

150 companies (!!!) contributing, and Ford doesn't even own the IP.

 

The short term cost obsession is a huge pain down the road. Now they are in a situation where they don't control anything. 

 

 

Indeed. And all these individually sourced modules are bigger, heavier, costlier, more complicated, and less functional than they need to be, because they are by nature a compromise, and optimized for the supplier, not the automaker. 

 

This approach is going  to cripple their efforts at self driving, too. 

[Charlieman]

I can't comment on the car's centraliesd software vs distributed CANBUS but all the integrated design, order, mfg  control ,software is basically same as SAP .which most businesses use.

The SAP sales pitch is that they have researched the best ways to run a business and that is what you get with the SAP packages. it is suggested that if your business model does not fit with SAP options, there is something wrong with your business. You can customise SAP at your own expense but you can't sell your enhancements without their approval. If your changes break when SAP modifies its software, that's your problem.

 

Many companies are happy to go along with the SAP sales pitch. Others, who prefer to run their business their own way, pick or build MIS software which suits them. Personally, I think that Tesla are right to develop their own management systems.

---

The video in the opening post omits to say how much Tesla was unburdened by legacy during its growth spate -- although that burden is creeping up. It didn't have to maintain a huge spares division because it hadn't sold lots of varieties of cars. It didn't have to worry much about legal concerns in different jurisdictions because it didn't sell widely across the world. 

 

The presenter in the video makes it seem that software updates to cars are something easy to do -- because you don't have to rip out mechanical bits. Unfortunately, every manufacturer has to send out updates to designs which have minor changes or replacement parts in numerous combinations. Whilst Tesla may be able to emulate almost all of these combinations in their test environment, they will trip up and it will be embarrassing. 

---

The anecdote about designing factories in software sounded like equine generated fertiliser to me. Don't people remember news stories about the reject rate on Tesla production lines? Or industrial relations? Manufacturers build factories to facilitate workers and processes, using the products of suppliers doing the same thing.  

---

When Tesla started to make cars, they were the only people in the business. They attracted the best brains who wanted to make EVs. They now have competition.

[MikeTekRacing]

Indeed. And all these individually sourced modules are bigger, heavier, costlier, more complicated, and less functional than they need to be, because they are by nature a compromise, and optimized for the supplier, not the automaker.

This approach is going to cripple their efforts at self driving, too.

Ford closed down their self driving efforts.
They will, most probably, license FSD.

[Greg Locock]

They shut Argo down, but big companies being big companies, that wasn't the only game in town, and parts of Argo will live on, internally. More specifically that was a joint project with VW. This added a huge  anticompetition/CoI/legislative/record keeping burden to the project. BTDT.

[Magoo]

Ford shut down the Argo self-driving unit last October, but then in March it launched another AV susidiary, Latitude, based in the same city (Pittsburgh) with 500 ex-Argo employees -- around half  the Argo workforce. 

 

Outwardly, it seems Argo was focused on level 4 applications and robotaxis, etc, while Latitude is devoted to more immediate results in level 3 on Ford production vehicles. It's said Argo never attracted the outside investors required to support long-term development. 

Edited by Magoo, 23 August 2023 - 20:15.

[Magoo]

They will, most probably, license FSD.

 

 

it seems to me carmakers will either have to license TeslaVision and FSD, or embrace generalized autonomy and retrace Tesla's steps. 

 

Either that or premap all the paved roads in the civilized world every 3-5 years .

Edited by Magoo, 23 August 2023 - 20:29.

[GreenMachine]

Back in the day, there were two or three suppliers of maps to the GPS suppliers, so your Navmans/Copilots/Garmins/etc bought one or the other and hey presto GPS driving instructions.  These days I suspect it is all Google-based.

 

Is that model likely to be replicated with self-driving, either with the digital road models, or the full FSD package?

[MikeTekRacing]

it could be, as it is a very heavy lift

[Greg Locock]

True FSD (ie all weather) will need instrumented roads. That doesn't mean every lamp post will have an antenna, but I suggest each cell will maintain a local map updated in real time as changes are spotted, fed back by the cars that drive through.

[Greg Locock]

 

 

10 microns is half a thou. So he wants to build a vehicle to the sort of tolerances that need finish grinding in an engine.

[gruntguru]

AI is Big Data, analyzing, find trends, it's algorithms, mathematics and statistics - There is nothing truly intelligent in Artificial Intelligence, which does not mean as a tool it will have many solid uses across many businesses and daily life for all of us. 6 years ago everything was Blockchain, it may or may not exist as a contributing tool somewhere (or everywhere), but when did a general media article/show last utter the words 'blockchain'?

 

From the post you comment on

"...which does not mean as a tool it will (did you mean to insert "NOT") have many solid uses across many businesses and daily life for all of us"

 

What you didn't say is "AI won't always be merely Big Data, analyzing, find trends, it's algorithms, mathematics and statistics - There is nothing truly intelligent in Artificial Intelligence".

So I added it for you.

 

I might add - have you seen the current state-of-the-art in AI? It may already be truly intelligent.

[KWSN - DSM]

What you didn't say is "AI won't always be merely Big Data, analyzing, find trends, it's algorithms, mathematics and statistics - There is nothing truly intelligent in Artificial Intelligence".

So I added it for you.

 

I might add - have you seen the current state-of-the-art in AI? It may already be truly intelligent.

 

Yes I missed a not.

[Chubby_Deuce]

10 microns is half a thou. So he wants to build a vehicle to the sort of tolerances that need finish grinding in an engine.

Sounds like the fake engineer learned a new word and felt the need to throw it around.

[Magoo]

Sounds like the fake engineer learned a new word and felt the need to throw it around.

 

Opinions vary. I've listened to Musk for a few hours in various interviews on the subject, and I think he knows auto manufacturing. 

 

He knows at least as much about it as any automaker CEO, I can say that. The interviews are all over YouTube, not hard to find. 

 

He knows exactly what a micron is and how presses work. No one should kid themselves about  that. 

[gruntguru]

 

 

10 microns is half a thou. So he wants to build a vehicle to the sort of tolerances that need finish grinding in an engine.

 

I'm sure he meant 10 mm.

Edited by gruntguru, 26 August 2023 - 04:39.

[flatlandsman]

My feeling is that much like Apple the initial designs were very good,m the success is about trends and people just wanting to be involved with Tesla, it has become a brand like Apple, a coffee table brand.  Very good, but largely successful because of a sort of feeling only a few brands can manage to achieve. it is without any kind of substance, it is just a sort of look. 

What Apple did was package something in such a way to make it simple, clever and well engineered but in reality it was nothing special and there were plenty of decent, cheaper alternatives.  

Tesla did some very good initial groundwork, the rest is up to marketing, design and that weird thing, trending

[Greg Locock]

...and Superchargers. and no dealers.

[Magoo]

I'm sure he meant 10 mm.

 

 

He means 10 microns. Industry standard for panel gaps is 4-5mm, and he's saying the Cybertruck fits have to be better than that or they will look wrong  He's noting to his people that 10 microns is the tolerance for soda cans and Legos.

 

Sure, he's crazy. Like he was crazy when he decided floor pans should be castings, or that a rocket could be landed vertically on the ground.  

[Magoo]

...and Superchargers. and no dealers.

 

I know, right? What stupid ideas. What could he possibly be  thinking. 

[Chubby_Deuce]

Industry standard for panel gaps is 4-5mm

When is Tesla planning to achieve this on their other vehicles?