C’mon, you can’t be serious here. The fundamental metric of a program’s success is schedule. And again, Tesla is maybe 6 months late on their schedule. We can even say 8 months if you really want to be pessimistic about their crunch week being unsustainable.
Defense/aerospace programs are often delayed by many years–decades, even, depending on how you want to count things. And of course many billions over budget. It’s Tesla that looks like a smoothly oiled machine in comparison.
I went back and looked at Tesla’s original predictions of ramp rates. Had they performed to that level, there would be somewhere around 200,000 to 250,000 Model 3s delivered by now. Instead, there are somewhere around 15% of that number. That’s terrible. I mean, awful.
When defense programs go bad, it is overwhelmingly issues that delay the start of production: but that isn’t the case with the Model 3. It’s been in production for closing in on a year, and the problem IS production. That’s the difference. And I am serious. (And I’m glossing over the “feature not a bug” issue of the M3 development apparently continuing such that there isn’t really a fully stable design at this point.)
I continue to be amazed that Tesla can’t use its zillions on dollars to purchase established expertise in production. Outside of battery production (which itself should be solvable via massive parallelization), there’s very little involved in building Model 3’s that’s very different from building Corollas. Poach some talent from Toyota or VW or wherever, give them adequate authority, and let Elon do the things that Elon actually does well. I’d love to see Tesla succeed in the long run, but it really does seem like they just refuse to solve their production problems by incorporating the existing solutions. And mass-producing cars is very much a solved problem.
That’s absurd. Your entire argument hinges on semantics.
Suppose that 12 months ago, Tesla said “Production is delayed by a year. Pre-production starts now with limited cars going out to the public.” No change at all in production ramp or anything tangible.
According to what you’re saying, that would have satisfied you, because they’d have come out the door running and everything before that doesn’t count because it’s just pre-production and troubles there are expected. It’s just a one-year delay to the start of production and that’s not a big deal.
As an aside, I have no idea where you’re getting your 200-250k numbers from. I think those are well above Tesla’s most optimistic projections.
Funny how you were saying “SpaceX is not Tesla” over and over but use the comparison when it fits your needs…
Gwynne Shotwell is indeed a gift to SpaceX. But it’s not like she was the CEO of Boeing or whatever. Her entire work history as a manager was a 4 year stint as a director at a little podunk aerospace company. There was no evidence from her resume that she’d be fantastic at running a big successful company like the present-day SpaceX, let alone all the intermediate stages.
Tom Mueller–VP of propulsion at SpaceX–is another great example. Musk didn’t just hire the top guy at–well, wherever, because there’s hardly any propulsion research in the US. But there was this guy working at TRW on yet another dead-end project that probably got a few bucks thrown at it by the Air Force or someone. More importantly, this guy liked going out to the desert in his spare time firing rockets he built in his garage.
I really wonder where you think Musk can find this off-the-shelf talent that is apparently widely available. All of Musk’s success stories involving talent have been “diamonds in the rough” situations; people who were talented but dramatically underutilized. It’s not like there’s been zero attempt at poaching higher-up talent; it’s just that these people haven’t been nearly as successful.
That would be an embarrassment in and of itself, but I think more understandably. If somebody has a piece of technology that doesn’t work right, don’t put it into production. But if the technology generally works, but you can’t build it, that usually should be chalked up to incompetence – especially when tens and tens of millions of cars are built each year in modern, efficient factories that would very likely be able to do final assembly on Teslas in a similarly efficient manner.
Based on ramping from nothing last summer to 5,000 per week by the end of 2017, per Tesla’s May 2017 letter to shareholders, I swagged about 50,000 cars built in 2017. Some Tesla fans estimated it could mean as many as 83,000 based on the same projection.
Then, with 2018 production supposed to ramp from 5,000 to 10,000 “sometime in 2018,” per the same letter, I estimated that the year would have a smooth ramp to 10,000 per week by week 52, which is both charitable and conservative. That yields about 150,000 cars that would be built by this point. So that’s 200,000, per a totally smooth production ramp using the most charitable Tesla estimations of 5,000/wk by week 52/2017, and 10,000/wk by week 52/2018. But since I’m guessing Elon was aiming higher than that bare minimum, I swagged that he may have meant a sharper ramp, putting the total production closer to 250,000.
Again, those two datapoints are what Tesla told its shareholders. May I ask, at the time Tesla was putting those numbers forth, did you think those were realistic promises to make?
Where did Marv Vander Weg and Tim Buzza come from? McDonalds?
McDonnell, McDonalds, same diff.
I’d never heard of these guys so they obviously aren’t at the same level as Shotwell, Mueller, Straubel, etc.
Buzza came in early on, clearly bored out of his mind at M-D and willing to take a shot at a startup. So a bit like Mueller in that regard (who was in a similar situation at TRW). Weg appears to have been hired solely on his experience in EELV contract negotiations, and probably entirely because he knows the Air Force people in charge. Useful but not exactly at a “designed the first new US liquid-fueled rocket engine in years” level.
Some Tesla fans are idiots. Musk has said over and over that the ramp would be exponential (or rather an S-curve, with the initial part being exponential). For whatever reason, people don’t like to hear that. Maybe they’re just bad at math.
The ramp therefore depends on the time constant. Their 17Q2 shareholder letter states an expectation of 1500 cars in Q3 and 5k/wk by the end. Doing a little curvefitting, this corresponds to a weekly growth rate of 23%. That would mean ~22k cars in 2017.
It’s neither charitable nor conservative. Tesla was clear that the 10k/wk number required considerable improvements to their factory infrastructure.
A more charitable interpretation would be that their 5k line would stay constant until mid-2018 (i.e., today) while they prepare some things (reconfiguring the factory, buying new land/buildings, etc.), and then another exponential ramp of just that additional 5k over the next 6 months. That would put us at 152k today and 304k by the end of the year.
I dunno. I don’t work there. But I’ve worked on plenty of projects that we thought would take 6 months and instead took 12. It happens.
How many of your projects involved delays in mass production?
ETA: as opposed to delays in development, I mean.
cough already addressed that here. Tesla’s reputation as a workplace is well known throughout the industry. Nobody with that kind of experience needs (or wants) to work in such a place.
that you can’t understand the difference between why I was saying that then, and why I’m making the comparison now, is very telling.
There’s not really a distinction we make (and really, making hard distinctions between design and production sounds like dinosaur thinking).
To invent an example, suppose a chip comes back with a timing defect. We wanted to clock it at X Mhz, but it only works at 80%*X Mhz.
We could “spin” the chip to fix the timing. That would add around 6 months to production, depending on the breaks.
We could sell the chip as something slower and ship it now, but if the competition is faster, then it’s pointless.
We could, depending, work around it in software. If the specific function can be implemented in a different fashion (usually more slowly, but it’s not always important), then maybe the timing doesn’t matter.
Finally, we could “bin” the chips. There’s always process variation, and even with a timing bug there will be some golden chips that can hit full clocks. So we ship small numbers of chips “on time” (even though they’re hardly available) while we fix the original problem.
Most of this stuff is invisible from the outside. Sure, product delays can’t be hidden and people notice if something is always out of stock because hardly any product is being shipped. But all of the chaos and ugly decisions are under the covers.
Let’s see–you said:
That is a very broad claim that you made–essentially, that we can’t make any points of comparison between the two companies because they are so different. Surely, this also applies to their management structure and hiring methods.
For the record, I actually agree with the point about public vs. private. In fact that may be the single greatest distinction between the two companies. We would not be hearing even 1/10 as much noise about Tesla if they were private. All of the hot garbage coming from the shorts, the financial “analysts”, Jim Cramer and his ilk, etc. would completely disappear. But that boat has sailed.
Yeah, new engine. Heh.
So, why can dinosaurs build cars extremely efficiently, and a car company that thinks its a software company can’t?
Musk has noted some irony in that (ground launch) rockets are the one thing that will never, ever be electric. Not to mentioned that they guzzle fossil fuels at a rate that would make a city-scale power plant blush.
That said, the future is methane fuels and that can be synthesized via electric power (though it’s expensive at the moment).
You’re the one that made that hard distinction. My understanding is that part of the Japanese revolution in manufacturing is tightening the feedback loop between design and production. The dinosaur way was that when the design was done, that was that and it’s up to production to sort out any problems. I don’t know to what degree this philosophy has been embraced across the board but it’s clearly the right direction and not unique to Tesla.
This was probably a bigger deal with SpaceX vs. the incumbents than it is with Tesla vs. the rest. Though Tesla’s ability to fix (some classes of) “production” problems in software certainly is a step forward.
But again, we aren’t talking about fundamental design problems with the Model 3. They have designed a car that doesn’t seem to be anything special in terms of final assembly. The problem is that Tesla made very big errors in embracing bad ideas on building cars, and wasted money and time in trying to build the “alien dreadnought” and is paying for it now.
It seems perfectly clear that if they had set out to build a car using the incredibly efficient methods that are adopted across the embassy, they would be far, far ahead of where they are today. That would surely mean more revenue, more happy customers, more people out of dirty gas transportation, and surely a stock price of a bazillion dollars a share.
But what is surely happening, as has been said a lot, is for some reason Tesla is embracing what is proven to be terrible management methods to deal with their past mistakes. For example, the reports of pulling people out of their jobs to chip in on final assembly. Give me a break. Imagine if a football team decided, “We gotta win the game! For the last three minutes of the game, we don’t need linemen - everyone is a wide receiver!” Madness.
I suppose you think this is a “gotcha” or something, but sorry, no.
in any other manufacturing company (like, say, every other car company) the design and construction of the assembly line is done by VO/Manufacturing Engineering, day-to-day operation is handled by plant management, and maintenance and modifications are done by plant management and VOME together. That’s their job. That’s what’s called “delegation of responsibility,” which means you hire people who know how to do those jobs to, well, do those jobs. 'cos honestly, when a company Tesla’s size has the C-E-fucking-O personally going around and fiddling with stuff on an assembly line, that’s all sorts of screwed up. That’s doing shit that people 6 or 7 levels under him should be doing.
I brought up Shotwell merely as an example of Elon delegating the operation of one of his companies to someone who knows what the hell she’s doing (and I like how your response to me bringing her up is to piss all over her background,) so he doesn’t need to stick his ass in it every day. If he could do that at SpaceX, why can’t he do it at Tesla?
That’s not at all clear to me.
Sure, given the knowledge we have now, if we could pick and choose industry-standard practices for some stuff and Tesla “special sauce” for the rest, then they’d be in a better position. It’s not obvious that the knowledge of which way to pick was available 24ish months ago when they started building out their production systems.
By most accounts, the Gigafactory is reasonably close to the “alien dreadnaught” dream. It hit a few rough spots but pack production hasn’t been a bottleneck for a while now. It is very likely that Tesla has the lowest pack production costs in the industry right now.
While most EV makers have outsourced some of all of their drivetrain (battery+motor) manufacturing (LG being a common partner), Tesla has kept all their expertise in-house. This has almost certainly been a net win but meant they created a lot of work for themselves.
So when Tesla decided that they should (say) automate seat installation instead of using human+mechanical assist, I’m not sure why it should have been obvious that this was the wrong choice–especially since it wasn’t completely outside industry practice, just uncommon. Let’s not pretend that there’s some single unified way to auto production in the industry instead of a spectrum. The Japanese lean more heavily on the manual side; the Germans (IIRC) lean more toward automation. Tesla pushed heavily toward automation.
Even if you ignore all the 20/20 stuff, it’s still not obvious that going automation-heavy was a mistake. If viewed solely from the criterion of getting Model 3 to high production levels ASAP, then sure. If you include the value from learning how and how not to do certain things–then maybe it was worth going through all that pain.