Hacker News Reader

Comment by toomuchtodo | Hacker News Reader

It's not volume, it's vertically integrated agility. Can't get Continental radar hardware anymore? Tesla updates their vehicle software to use vision only. Tesla can't get certain chips? They have motivated engineers (culture!) in house able to refactor the vehicle software to support a new silicon configuration.

This is a symptom of legacy automaker stagnation, believing their role is to make a profit from bolting together a bunch of external supplier parts, which works until it doesn't. Their supply chain failure is Tesla's opportunity. This is really no different than SpaceX versus Blue Origin, and how each approaches their culture and management.

EDIT (HN throttling, can't reply directly to your comment): @filereaper Musk is an engineer at his core, and sets the culture of his organizations (high expectations but respect for engineering talent). Bezos and Blue Origin are what you'd expect if McKinsey & Company (the consulting firm) were building a space exploration company. Check out upwardbound's comment below, it's spot on, specifically the section on finding talent.

@malcolmgreaves: There is a delta between what we believe makes a good engineer.

View on HN

Also probably willingness to not restrict themselves to silicon parts rated for an automotive environment. They did this for the LED display...going with "industrial" vs "automotive". Where industrial has a more narrow acceptable operating temperature range, etc. Probably works fine in a lot of cases, but they are taking on some risk.

>Where industrial has a more narrow acceptable operating temperature range

*wider

Nope. Automotive grade is typically wider than industrial.

also, they probably ignore safety and warrenty pbligations, and abandon repairability because they are still a top end manufacturer

> Musk is an engineer at his core

This is objectively false. He has great advertising and PR, but Musk is a business person through and through.

He has never been, is not, and never will be an engineer. His hostile take to engineering -- from the fact that he OKs unsafe technology and brands it as "autopilot" to half-baked, rushed assembly processes that produces terrible quality in the Model 3s -- is the most obvious evidence that he ultimately only cares about getting richer, whatever the cost.

Your objection is flat embracing.

He has Degrees in physics, and many people with such degrees end up getting positions in engineering companies and are called engineers. He was accepted at Standford PHd program for materials science. That would natural lead into an Materials engineering position in some company if completed.

Of course instead of doing that, he left and started companies, as a software engineer he was the core developer one Zip2 and early on on PayPal. If that doesn't qualify you as an engineer I don't know what does.

Literally he has been the Chief Engineer at SpaceX since its founding. And the argument that its just an empty title just don't work. They were a small company for quite a while and somebody made all the final decisions. The are multiple stories where he works directly with other engineers on the rocket as well if 'making engineering decisions' is not enough. Multiple former employees have directly contradicted these sort of 'not engineer claim'.

Just recently former lead engineer at Ford with 50 years experience in automotive, defense and aerospace was allowed to sit in on a meeting for Starship. He said he had never seen anybody CEO be that involved in detailed decisions, about manufacturing processes and design choices and so on.

So just because you don't like him or his product, doesn't make him not an engineer. He has the education many other people that call themselves engineers have, and he has the title of engineer and that title has been officially recognized by NASA and DoD.

---------------

    Elon is both the Chief Executive Officer and Chief Technology Officer of SpaceX, so of course he does more than just ‘some very technical work’. He is integrally involved in the actual design and engineering of the rocket, and at least touches every other aspect of the business (but I would say the former takes up much more of his mental real estate). Elon is an engineer at heart, and that’s where and how he works best.

    Former Head of Software Quality Assurance at SpaceX

I think Elon being a true polymath scares a lot of insecure engineers out there. They’ve gotten used to being pampered and told they’re special, and here is a goddamn billionaire business guy who could do their job also. It’s just too much to swallow.

People see what he's done as a criticism of their own failings and missed opportunities. I've never seen grapes more sour than the ones devoured by anti-Tesla/Musk folks (apparently by the electric truckload.)

This is interesting. I suppose Bill Gates and Mark Zuckerberg are less threatening since they don’t work on hardware but other polymaths exist. Certainly I believe John Carmack of Id Software and Armadillo Aerospace is one - and the latter is hardware.

Elon being unstable scares a lot of insecure engineers out there.

What is your evidence of him being unstable?

I think he's referring to his tendency to pick weird twitter fights.

I have tons of respect for what Musk has done with Tesla and SpaceX, but it's undeniable that some of his tweets should not have been sent. And while I generally appreciate weirdos, he's a bit much sometimes.

Ok, somebody being competitive on twitter means that person is mentally unstable?

I mean you want to call him a asshole fine, but doesn't mean he is unstable.

If everybody that gets on fight on twitter or is sending tweets that shouldn't be sent is considers 'unstable' then most people on twitter are unstable.

I’ve had a model 3 for 2 years, I keep hearing from people who don’t own teslas about “poor build quality” but I have never witnessed it myself. It’s certainly been a lot better than the Ford and Lexus I previously owned.

I could give you heaps of anecdotal evidence to the contrary but they ranked almost last in a survey [1] and even Musk admitted it was a problem at the company [2]

[1] https://www.greencarreports.com/news/1128625_tesla-ranks-las...

[2] https://www.cnn.com/cnn/2021/02/03/business/elon-musk-tesla-...

Maybe I am just not that picky. I remember my friend almost got a Tesla but he was complaining about the stitching in the seat, I could not care less about that. So he got a Lexus instead, now he complains about rising gas prices.

I obviously can't speak to your experiences, but this conversation is one you see quite often around any popular luxury product.

Teslas are popular, so for many it's the first expensive car they've splurged to get. This also means they don't know what that amount of money could've gotten them from another manufacturer. Whether another manufacturer would win over a given customer considering a Tesla is up to the personal taste of the customer, but most people splurge for the Tesla, and not any car at a given price point, looking at the pros and cons, then choosing the Tesla as the best option.

It's also one you'll see with Apple laptops. You'll see someone swear that a MacBook is the best computer they've ever owned. However, the MacBook is $2500 and their last many computers were $600 from Walmart. Of course the MacBook is better, the question is if it's better than everything comparable. I personally think the new MBPs with M1 Pro/Max chips will be, but was a MBA pre-M1 better than the comparable XPS? I'm less inclined to agree.

I think you have a good point here. Anecdotally, I had been waiting for the electric cars to hit the markets for over 10 years before even getting a driver’s license (yes, I’m a treehugger) when I finally decided to buy, it dawned on me that for the same amount of money as a model 3, I could get a nice BMW, Mercedes or Audi. After testing them all the Tesla didn’t make the grade and I’m the happiest BMW-hybrid owner. Opportunity cost is a very real thing when buying anything in this price range.

I don't notice stuff like that either, especially not for cars, so I disqualify myself from any discussion that relates to build quality.

Tesla's build quality issues spans the range from "the stitching and panel gaps weren't right" to "the roof fell off on the way home from the dealership"[0] and "the steering wheel pulled right off the column"[1]. It depends a lot which sample you get. People with good examples seem to love them, but the range of outcomes is shockingly big.

0 - https://www.businessinsider.com/tesla-model-y-roof-fell-off-...

1 - https://insideevs.com/news/421012/tesla-model-3-steering-whe...

>His hostile take to engineering -- from the fact that he OKs unsafe technology and brands it as "autopilot" to half-baked, rushed assembly processes that produces terrible quality in the Model 3s -- is the most obvious evidence that he ultimately only cares about getting richer, whatever the cost.

Funny, these are the attributes that I think make him a good engineer. An engineer incapable of building to a minimum viable product spec is not a good engineer.

The fact that he is also amazing at business and PR is besides the point.

He clearly has an engineering and physics mindset and the very DNA of his companies is all about engineering. He attracts excellent engineers in large numbers. I think they would disagree with your assertion.

> from the fact that he OKs unsafe technology and brands it as "autopilot" to half-baked, rushed assembly processes that produces terrible quality in the Model 3s

https://en.wikipedia.org/wiki/Engineering_disasters

Engineers don't get to put themselves up on a pedestal and claim they never do wrong, are never subject to incompetence, negligence, or other personal pressure in their lives like money.

But they do take responsibility and are open to scrutiny by their peers when they actually do wrong. In this sense, it's impossible to be a CEO making profit decisions (where loss of life is a balance sheet issue), and maintain the scruples of an engineer at the same time.

> But they do take responsibility and are open to scrutiny by their peers when they actually do wrong. In this sense, it's impossible to be a CEO making profit decisions (where loss of life is a balance sheet issue), and maintain the scruples of an engineer at the same time.

I don't know what you're saying. Engineers absolutely have been involved in everything from turning a blind eye to problems, shirking responsibility, and the normalization of deviance, all the way to deliberate coverups and crimes.

You are naive about engineers. They're people pretty much like everybody else.

Then it should be easy to come up with some concrete examples? The biggest examples in recent times where top-down profit motives trumped ethics:

Boeing 737 Max - A cost cutting measure that came from above, 100s of lives lost https://en.wikipedia.org/wiki/Boeing_737_MAX_groundings

Volkswagen diesel emissions scandal - Tax avoidance, increasing market share https://en.wikipedia.org/wiki/Volkswagen_emissions_scandal

Tesla Full-Self Driving - advertising full self driving capability when it's at best assisted and has led to multiple deaths https://www.tesladeaths.com/

And that's just the tip of the iceberg.

> all the way to deliberate cover-ups and crimes.

I don't doubt that engineers have done monstrous things, but only when the engineer is a stakeholder. In this case, they aren't your typical employee.

I fail to see where any of your examples feature engineers speaking out, taking responsibility, and opening themselves to scrutiny before the problems became known. These are good examples of engineers being complicit or worse.

> Tesla Full-Self Driving - advertising full self driving capability when it's at best assisted

Since you have a bone to pick with Tesla though, I looked up how they're advertising full self driving. From their website "While no Tesla cars are fully autonomous today and require active driver supervision,".

> and has led to multiple deaths https://www.tesladeaths.com/

This page does not appear to keep track of "Full Self Driving" deaths. Autopilot deaths are 5-10% of the deaths from Tesla wrecks, however they keep track of data. Doesn't really seem outlandish but it's obviously very hard to know -- it could be anywhere from a lot worse to a lot better than human only drivers, going by these numbers.

You do know though that no company advertises these driver assists as offering complete protection from death. I don't think that's quite in the same category as the first two examples.

> I don't doubt that engineers have done monstrous things, but only when the engineer is a stakeholder. In this case, they aren't your typical employee.

No, this is not true. As well as being unintentionally incompetent, lazy, not keeping up with best practices etc., people get emotionally invested in their company, their teams, their project, their careers and cause them to knowingly and actively participate in these things even if they may not directly (or at all) stand to gain financially.

> No, this is not true. As well as being unintentionally incompetent, lazy, not keeping up with best practices etc., people get emotionally invested in their company, their teams, their project, their careers and cause them to knowingly and actively participate in these things even if they may not directly (or at all) stand to gain financially.

Once again, my assertion is that the motivation is typically not coming bottom-up from the engineers themselves, but top-down from senior management and stakeholders. Competent engineers who work in aviation, automotive, construction, etc. are generally motivated to avoid failure conditions. I'd actually be keen to know if you have any significant examples to the contrary - undoubtedly they exist.

I'm not sure how you could fill the role of CEO, with a strong profit motive, and not be at conflict with an engineering mindset. It's possible that Elon is more scrupulous than most, and better placed to anticipate failure modes with his engineering background, but his messaging particularly on FSD has been leaned more towards market optimism, driving sales, generating investment, than the safety of passengers.

> Since you have a bone to pick with Tesla

I don't, but I don't lionize Tesla or Elon either.

Your assertion was that it never comes from engineers. If you accept that was too strong and want to adjust your claim okay, but don't make as though that's what I was arguing about.

> I don't, but I don't lionize Tesla or Elon either.

If you don't address or retract any of the seemingly incorrect claims you made about Tesla, that suggests to me that you do.

> don't make as though that's what I was arguing about

You were arguing that that engineers are often unscrupulous, I'd counter argue that actually their managers and the stakeholders are often unscrupulous and force them into complicity. My claim is that the responsibility ought to be at the top.

Edit: But I think I agree that even if peers hold each others to account in engineering, it doesn't really matter. They don't hold the power to change institutions. I know there are Engineering societies... and I think in some domains you need a licence to practice that could be revoked, but this isn't true of your typical aerospace firmware coder.

> seemingly incorrect claims you made about Tesla

Such as? Regarding FSD, the deaths stats only break out numbers for autopilot. You might argue that the drivers are responsible for their actions if they leave their hands off the steering wheel, but encouraging any kind of overconfidence or complacency is putting other drivers and pedestrians at risk - all for marketing purposes.

As for having a bone to pick, I provided 3 examples of well publicized engineering failures in recent years. You zeroed in on Tesla which is fair given the OP, but not evidence of me having a bone to pick, an axe to grind, or what have you :)

> You were arguing that that engineers are often unscrupulous

Wrong.

> Such as?

I already pointed them out several posts earlier in the thread.

> Wrong.

Maybe you're thinking of black hats? But they aren't operating as engineers. Still, from you, no examples, no evidence, but intransigence. Such a mindset would not be becoming of a scrupulous engineer, so you might be right after all - or you're less qualified on the subject than even me

> This is objectively false.

I commend the response by user panick21 and would add that you can watch Elon Musk demonstrate his acumen in this three part interview on Everyday Astronaut. You can't fake this kind of breadth and depth of knowledge.

https://www.youtube.com/watch?v=t705r8ICkRw

Excellent interview series

My favourite moment is from part one (linked above) about the Tesla production line. Discussion of Elon's principles of simplification (as applied to SpaceX) begins at 13:30 and those principles lead to three contextually relevant anecdotes about Tesla between 22:00—28:20.

This interview is quite the exclamation mark on the statement: Elon Musk is an engineer.

I'd recommend interested folks to listen to Musk on Joe Rogan's podcast. Musk gives detailed explanations on the challenges they face and how they solve them.

>It's not volume, it's vertically integrated agility.

Which is easier at low volume...

>They have motivated engineers (culture!)

I love the implication that 100+ year old companies that build race cars don't have solid engineering culture. Aren't ICE vehicles far more complicated?

> I love the implication that 100+ year old companies that build race cars don't have solid engineering culture. Aren't ICE vehicles far more complicated?

Depends on what kind of engineering. Having an old company that is good at mechanical engineering struggling when it comes to software is not surprising, it's the norm.

I'd also disregard the race cars; that is often a wildly different group with its own unique culture and specialists. Tech can and does move back and forth between the two, but if we're evaluating culture, you have to acknowledge that the racing divisions are just different teams with different cultures.

So only a million vehicles a year, eh? That's not really an easy problem. Your door handle motor manufacturer ran out of ics. Please build 20,000 new door handle motor controllers. It's not documented, it needs to work reliably.

Tesla doubled their sales because there is lots of demand (as there is for legacy 'makers) but also because they are at heart a scrappy, innovative company.

> I love the implication that 100+ year old companies that build race cars don't have solid engineering culture. Aren't ICE vehicles far more complicated?

They are complicated, but in a different way. They don't get radically re-architectured from the ground up the way software does. An ICE car today has some fantastically complicated components, but by and large they're recognisably the same components doing recognisably the same things, as you would see going back decades.

> It's not volume, it's vertically integrated agility. Can't get Continental radar hardware anymore? Tesla updates their vehicle software to use vision only. Tesla can't get certain chips? They have motivated engineers (culture!) in house able to refactor the vehicle software to support a new silicon configuration.

If these are all true, why not implement these your proposed way anyway? There has to be a tradeoff somewhere...

> It's not volume, it's vertically integrated agility. Can't get Continental radar hardware anymore? Tesla updates their vehicle software to use vision only.

You say that like it's a good thing.

Given that radar is functionally blind[1] to static objects when travelling at high speeds, I'd argue it's not just a good thing, it's a necessary thing. Tesla have demonstrated the superiority of their vision approach and validated it with 1,000 years of real-world driving, running it in shadow mode on customer vehicles.[2]

[1] Technically radar isn't blind to static objects, but since it cannot disambiguate them from tunnel entrances and highway overpasses, the data is useless. Therefore it's functionally blind.

[2] https://youtu.be/g6bOwQdCJrc?t=1647

This kind of agility has some interesting knock ons in terms of support and maintenance complexity later on and the need to maintain spare parts inventory for different vehicle configurations.

> motivated engineers (culture!) in house able to refactor the vehicle software to support a new silicon configuration

I feel like this is overstated. I say this as someone who's done a fair whack porting work, compiling for a new platform isn't really that technically challenging - at least not at the talent you'd expect in the embedded space. Typically you'd have abstractions around maths, GPU compute, UI, rendering, etc. and a retargetable RTOS/kernel. That said, it's quite likely other auto-makers shot themselves in the foot by not having these abstractions.

>Can't get Continental radar hardware anymore? Tesla updates their vehicle software to use vision only.

Doesn't that just mean that I'm getting a shittier car with corners cut out?

No. Newer versions of Tesla's software disable the radar on cars that shipped with it.[1] Vision has gotten so much better (thanks in part to training from radar data) that radar adds noise to their model of the outside world. Andrej Karpathy explained the issues in a talk this summer.[2]

1. https://twitter.com/elonmusk/status/1447588987317547014

2. https://www.youtube.com/watch?v=g6bOwQdCJrc&t=589s

But you get the car. Some manufacturers have simply shut down production while they wait for parts to become available.

But at some point it stops being the car that I ordered as they strip parts from it. Arguably it stops being the car that I ordered the moment they decide to start taking stuff out of it.

You're buying the final product, not specific components when you place your order. As long as it functions as advertised and meets warranty requirements, it is what you were sold.

At least in the case of my Model Y, I was given the option to decline my order as it was placed before the vision transition [1], so there is disclosure and the ability to opt out.

[1] https://www.tesla.com/support/transitioning-tesla-vision

Yes but that doesn’t show up in the quarterly numbers

Bit off topic w.r.t Tesla but could you elaborate more on the SpaceX vs Blue Origin point?

Do they have different methods in building their rockets and their supply chains?

Here's a comparison that Blue Origin themselves made in an internal memo in 2018: https://arstechnica.com/science/2021/10/revealed-the-secret-... The article has 3 pages, make sure to check out the whole thing.

For example:

“Cost as a design constraint and important variable is embraced by their culture, instead of being viewed as an evil metric that leads to a sub-optimal outcome," one [Blue Origin] executive wrote [about SpaceX].

"I believe we study a little too much and do too little," one [Blue Origin] executive wrote. "More test [rather than] more analysis will allow us to progress more quickly, iterate, and eventually succeed."

I would have assumed SpaceX vs Boeing (Starliner) would have been a more relevant comparison. Boeing generally tends to outsource as much as they can.

SpaceX does business with the DoD, which can give them supply-chain priority over everyone else for so-called "DPAS rated" purchase orders. [1] If I'm fulfilling a rated order, I not only have the ability, but the obligation, to inform my suppliers that I'm at the front of the line.

Tesla, presumably, isn't able to play that card.

1: https://www.bis.doc.gov/index.php/other-areas/strategic-indu...

They didn't have that for all the development of Falcon 9. SpaceX deliberately vertically integrated and moved to automotive and other suppliers because space suppliers were so bad.

I would be surprised if they have to use this process all that often, specially now that they have reduced construction rates.

It’s definitely a combination of volume and agility

Reply on news.ycombinator.com