Tesla Q2 2019 Letter • Hacker News Board

Re: Tesla Q2 2019 Letter

#171

Post by repolfx » Thu, Jul 25, 2019, 10:15 AM UTC

abledon wrote: post #94

kolbe wrote: post #59

Earlier quoted context omitted.

Hence why many investors don't mind that Tesla isn't generating wonderful cash flow. If there's a reasonable expectation that this data collection will lead to full autonomy, then Tesla should price their cars at whatever level is necessary to move their maximum production into consumers' hands, even if it were a loss (which it's not). And even if they don't use the data to train, they can use it as an argument for h…

Do you think they keep all that data on premise or use AWS/Azure/GCP? edit: probably AWS [1] [1] https://www.theinquirer.net/inquirer/news/3027077/hackers-hi...

For compute, during the autonomy day video they mention they use their own clusters, because they've designed their own "full self driving computer" with a custom set of neural network accelerator ASICs. So to do test runs on their own driving computers, of course, they need to build their own machine clusters.

Custom built clusters can be much cheaper than the cloud anyway.

For storage, they say they actually pull data on demand from the fleet. They have NNs that can detect similar-looking stuff to input samples, so if they want more videos of construction sites they just ask the fleet to send them more videos of construction sites. Actual storage of all video isn't required except for their test suites.

Re: Tesla Q2 2019 Letter

#172

Post by repolfx » Thu, Jul 25, 2019, 10:18 AM UTC

Klathmon wrote: post #26

boldslogan wrote: post #16

Earlier quoted context omitted.

neat. I would say im optomistic on this actual self driving stuff. But coming from almost no machine learning knowledge, doesn't this make tesla way ahead of google/waymo and others? Because they can simply check when they failed and tweak. I'm just curious if it does make them way ahead, and if so how much more ahead? Is this driving a car vs horse wagon levels?

I think the real answer is nobody knows. It's obviously got some benefits, and I think most people in ML would agree that real world data from the "average person" (as in "not a Dev or paid to do this") is worlds better than simulations. Still, "ahead" and "behind" isn't really a thing before you reach the finish line here. Many think Tesla will never reach full self driving or won't reach it without additional senso…

Tesla's approach means there isn't a finishing line. Just a long stream of over-the-air updates that incrementally enable more and more complex scenarios. They could be at it years. One day people will cross a psychological threshold where manual intervention is needed rarely enough that we talk about it as fully self-driving rather than limited driving assistance.

Re: Tesla Q2 2019 Letter

#173

Post by toomuchtodo » Thu, Jul 25, 2019, 1:34 PM UTC

navigatesol wrote: post #81

toomuchtodo wrote: post #75

Earlier quoted context omitted.

Is that Youtube video fake? The ~90k vehicles Tesla sold this quarter? "Fake news"? I have no excuses, just proof of hundreds of thousands of Tesla vehicles on the road, and billions of dollars worth of manufacturing capacity running. Come back in a year we shall.

> The ~90k vehicles Tesla sold this quarter And made less revenue than last quarter. Brilliant. Is the 1 billion dollars lost this year after saying profits forever fake? The lawsuits from whistleblowers? Months long waits for parts? The miniscule growth after slashing prices greatly? The securities violations? Tesla Solar selling the least it ever has? I could care less about some factory shell in China. They told y…

> I could care less about some factory shell in China.

Images of equipment installed GF3 in China.

https://i.redd.it/lzgtskpyubc31.jpg

Re: Tesla Q2 2019 Letter

#174

Post by mavhc » Thu, Jul 25, 2019, 1:56 PM UTC

BoorishBears wrote: post #163

snuze wrote: post #116

Earlier quoted context omitted.

As someone in the industry (with domain knowledge), I can tell you that you are oversimplifying the problem and ignoring Tesla’s advantage and leadership in this space. Other manufacturers have to account for legacy systems/designs, it is not as trivial as you make it sound.

I even state that I'm oversimplifying the problem. Other manufacturers do have to account for legacy systems, you don't need to be in the industry to understand the relationships manufacturers have with companies like Bosch and know they're not in full control of their destiny. But their advantage is not a "let's ignore dumpster fire financials quarter of quarter to earn them a multi billion dollar market cap" big. T…

Everyone else not doing updates is the symptom of the larger issue, everyone else is a slow moving hardware company that outsources their software. Tesla is a software company first. And software beats hardware whenever you're doing something complex.

Plus lack of legacy self competition, no one else wants to cannibalise their other sales, and no one else has really invested in batteries. Most other car companies can't make enough EV cars to fulfill demand. They're all out of stock until next year because they make so few of them.

Re: Tesla Q2 2019 Letter

#175

Post by intopieces » Thu, Jul 25, 2019, 2:23 PM UTC

Fins wrote: post #159

kinghajj wrote: post #99

Earlier quoted context omitted.

The stalks and buttons on the Model 3 can access all of those same features, as well as adjust the follow distance for adaptive cruise control. Climate settings are through the screen, via a fan symbol button in the center. That pops up a temperature selection slider with buttons to increase/decrease the setpoint, and (de)sync the driver/passenger set points. Press the button again, and you get a full screen with mor…

Sounds like just the thing one should be doing in a moving vehicle! /s

Your hypothesis is that this model (screen instead of knobs) is more dangerous. Do you have any data to back this up?

Re: Tesla Q2 2019 Letter

#176

Post by slg » Thu, Jul 25, 2019, 3:34 PM UTC

new_realist wrote: post #144

slg wrote: post #77

Earlier quoted context omitted.

Your comment directly conflicts with what was stated in this letter: >During the quarter, a majority of orders continued to be for a long-range battery option and the Model 3 average selling price (ASP) was stable at approximately $50,000. At the same time, manufacturing costs continued to decline.

Yes, the ASP was stable... during the quarter. In 2018 the Model 3 ASP was above $57K.

It is a bit misleading to use an ASP number from 2018 as a measure of demand since they weren't delivering the entire range of vehicles at that tine. Tesla started out focusing on the cars with a high price and margin which artificially increased the ASP over where it would naturally settle. Therefore a decrease from that initial high is not any type of indicator of demand.

Re: Tesla Q2 2019 Letter

#177

Post by Fins » Thu, Jul 25, 2019, 3:36 PM UTC

intopieces wrote: post #175

Fins wrote: post #159

Earlier quoted context omitted.

Sounds like just the thing one should be doing in a moving vehicle! /s

Your hypothesis is that this model (screen instead of knobs) is more dangerous. Do you have any data to back this up?

Are you seriously suggesting that something that does require you to pay attention to a screen entire time you're using it (and touchscreen in a moving vehicle is not the easiest thing to operate even when you're not the one driving) is less distracting that something that is operable entirely by touch?

I'll stick with my experience trying to operate both. ANd I'll stick with cars that have real controls. Not that there were much reliable data yet, what with Tesla being pretty much the only ones who made something like that. Their crash statistics aren't all that impressive, BTW, but it is hard to discern which ones are pure accidents, and which ones are tied to controls, or AutoPilot, or general stupidity.

It will be interesting to see crash statistics of e-Tron vs. regular Q series, though, that would give some apples to apples comparison options.But then that assumes that Audi can actually sell a meaningful number of them...

Re: Tesla Q2 2019 Letter

#178

Post by ReptileMan » Thu, Jul 25, 2019, 5:14 PM UTC

mavhc wrote: post #174

BoorishBears wrote: post #163

Earlier quoted context omitted.

I even state that I'm oversimplifying the problem. Other manufacturers do have to account for legacy systems, you don't need to be in the industry to understand the relationships manufacturers have with companies like Bosch and know they're not in full control of their destiny. But their advantage is not a "let's ignore dumpster fire financials quarter of quarter to earn them a multi billion dollar market cap" big. T…

Everyone else not doing updates is the symptom of the larger issue, everyone else is a slow moving hardware company that outsources their software. Tesla is a software company first. And software beats hardware whenever you're doing something complex. Plus lack of legacy self competition, no one else wants to cannibalise their other sales, and no one else has really invested in batteries. Most other car companies can…

This is a glorified bumper car. Why does it needs updates at all? It should be KISS... instead they complicate for the sake of complication.

Re: Tesla Q2 2019 Letter

#179

Post by daryas » Thu, Jul 25, 2019, 8:19 PM UTC

BoorishBears wrote: post #105

jtchang wrote: post #68

Being able to test your code out in the real world and dark launch a feature gives Tesla a ridiculous advantage. No other company has the infrastructure to be able to do that. Plus they can even tell if the decision the neural net made was wrong because I assume even Tesla knows you got into an accident. And this over thousands of cars. The speed at which they are collecting data is unprecedented.

I'm so tired of this unapologetic Tesla hand holding. > No other company has the infrastructure to be able to do that. Because they don't want it. They don't want AP regressions we've already seen happen. They don't want engineers testing code out on real people's lives. You really think other companies just don't know how to send OTA updates to an ECU or other car components? I could literally put that together with…

> They don't want engineers testing code out on real people's lives.

When Tesla pushes and runs code in the shadow mode, it is run in parallel to the "production" version of the software. The actual behavior of a car, even with the test code in, still relies on decisions made by the "production" version of the software[1]. The purpose of such testing is to compare decisions made by the updated code vs those made by the old version, and use this info for fine-tuning the next gen algorithms.

[1] https://youtu.be/Ucp0TTmvqOE?t=6679

Re: Tesla Q2 2019 Letter

#180

Post by BoorishBears » Thu, Jul 25, 2019, 8:19 PM UTC

icanhackit wrote: post #168

BoorishBears wrote: post #167

Earlier quoted context omitted.

I hate it when people say "keep shifting goal posts" in the most random contexts to sound pithy. I said no OEM wants their engineers to have a capability and you're fixating on the uses of the capability. My comment is saying it doesn't matter if they wanted to use it for something as trivial as ECU updates, They. Do. Not. Want. It. If you want a case study in why, AP regressions are a fine one. They don't want the l…

> My comment is saying it doesn't matter if they wanted to use it for something as trivial as ECU updates What I'm saying is that you've completely missed the OP's point: The advantage Tesla has in the autopilot space is that they are able to have hundreds of thousands of real-world test units for their autopilot systems by running the neural net in parallel with the driver inputs (i.e. shadow mode) versus companies…

I haven't completely missed any point, you just seem to be insistent on missing mine.

My point is the way that Tesla has gained this, using consumer vehicles as testbeds, not just for collection, but for actual running code in charge of managing peoples lives is utter nonsense and traditional automakers are choosing to stay away from it.

I mean, this isn't even a hypothetical, we've literally seen it happen, AP regressions where a lane transition your car was taking fine one day suddenly sends your car aimed at a concrete barrier!

You can call them slow, or whatever you want, but thank god the real auto manufacturers are not so flippant about the value of human life.