You can install our site as a web app on your iOS device by utilizing the Add to Home Screen feature in Safari. Please see this thread for more details on this.
Note: This feature may not be available in some browsers.
Yes, completely and utterly useless for that. It's a general discussion thread. The prior stock trading thread was eliminated because there was too much overlap with the general thread.This thread is useless for following intraday stock moves.
There's literally only one way to do it: higher total productionNow I'm really confused. If the first ship doesn't arrive until almost halfway through the quarter in the hottest EU market for Model 3s, how does Tesla avoid the problem of end of quarter fire drill to make delivery numbers?
In other words, can someone explain from a mechanical standpoint how Tesla solves the smoothing out of the waves without completely tanking deliveries for at least one quarter?
That's what I thought, but I've been wondering if the next US expiration is really going to be a big deal. I would expect it to follow the same model as all the other expirations, but it's so SOON after the last expiration that it might not.I can't imagine that the company will keep shipping to Europe and China at the expense of US production this quarter, given the tax credit reduction in the US after 6/30.
Loss adjustment expenses, the cost of processing claims including legal expenses, is about 10% of premium. Being able to source onboard video and other data should help reduce adjustment expenses and avoid futile litigation.As @jhm pointed it out in this fantastic post, Tesla will be doing car insurance correctly if they do not chase the lowest rate offers on the market, i.e. if they just match a fair insurance rate and pocket the difference (if any), and offer "triple play" (car, financing, insurance) convenience and retain good customers long term.
I.e. they should not use any data advantage they have (360° video capture of the large majority of insurance events, fine-grained, GPS tagged customer behavioral history, etc.) to undercut competitors - they should use it to isolate loyal, low cost customer base they feel confident underwriting, and build a robust, high margin revenue stream.
Personally I'm really curious what effect Tesla's "data advantage" is going to have on underwriting costs: having legally extremely powerful video recording of insurance events will make both sides of underwriting less expensive:
In particular spurious claims of injury (fraud, or borderline fraud) should be easier to defend against with video evidence of the actual collision on record.
- Events where Tesla owners are at fault: when faced with video evidence they might pay out of pocket instead of having an at-fault claim on their insurance record. (Also, people often genuinely misremember traffic accidents - and simply having it all on record will inject a lot more reason into the process.)
- Events where Tesla owners were not at fault: both litigation and collection of damages should be a lot cheaper with good video evidence, both against insured and against uninsured parties.
There's also secondary revenue opportunities, like getting Tesla insured vehicles fixed in Tesla body shops. (Once they have free service capacity to spend on it ...)
Fair question from an infamous TSLA short:
What was the $70M for “Marketing, Promotional and Advertising Costs” in 2018 spent on?
https://ir.tesla.com/node/19496/html
Since it was not spent on traditional advertising, can we come up with a comparison of such “not really advertising” expenditures with that of traditional OEM vehicle manufacturers?
Yes, completely and utterly useless for that. It's a general discussion thread. The prior stock trading thread was eliminated because there was too much overlap with the general thread.
Domesticating the house cat was way harder, we have been working at that for 10k years and still aren't quite there.I used to think autonomous driving is super difficult - by far the most difficult project in human history. After I spent more time thinking about the details, I realize it may not be that bad.
There are essentially 2 approaches:
A. The car is super smart, understand what's around and know what's the right thing to do (General AI for driving).
B. The car can see lanes, know the basic driving rules, know speed limit for every part of road (either through mapping or reading road signs), slow down when other cars are moving slowly, keep going if the driving path is clear. Understand how to keep safe distance, understand traffic lights. Slowly drive around if there is an object in it's path. All these tasks are relatively simple.
Human drivers have a little bit of both A and B. We naturally have A, the way we learn driving is B. That's how I learned driving, I never learned the millions of conner cases. Basically I am a vision + rule-based driving system using some if-else logic, never had accidents.
I think both A and B can work in the end, but A is much more difficult to implement. So I think the likely scenario is someone will get a decent system out using approach B.
When I think about Tesla's long term valuation, FSD is the key, nothing else matters. If another company gets FSD first, they will attract tons of money and catch up on EV, battery, charger, ride sharing, etc.
Based on Tesla's Autonomy Day presentation, I guess they are doing approach A (I could be wrong). I hope they keep their mind open, maybe start a separate program using A+B approach. Wasting a billion dollars on this would be well worth it. Tesla can't afford to lose the FSD race.
Yeah, hopefully, offroad vehicles have some tank grade armor there.The (relatively small) titanium plate is there to protect against high-speed but low mass road debris hitting the battery pack in the wrong place at the wrong time - like the lose pieces of foam hit Space Shuttle Columbia.
Driving into an ice lake and hitting the car against the rocks on the shore is hitting the titanium and aluminum plates not with small road debris, but they are hitting an immovable object with a 2+ tons car. They are absolutely not designed to withstand that.
What frustrates me is that Waymo is just robotaxi company that will either sell FSD to someone else, or have to pay money for a fleet and yet... I wish $TSLA were valued like a robotaxi company
$TSLA market cap ~$44B
Waymo worth $175B
Waymo is worth $100 billion more than previous estimates, Morgan Stanley says (GOOGL) | Markets Insider
Wow Tesla is doomed
Many many many patients are issued that would never stand up to litigation. The US patent office is quite happy to patent most anything and let the merits of it be fought out in court if it comes to that.
And then right back down... weird
After hours SP goes up 8$??
Like it or not, Musk has now positioned TSLA as a robotaxi play.
I used to think autonomous driving is super difficult - by far the most difficult project in human history. After I spent more time thinking about the details, I realize it may not be that bad.
There are essentially 2 approaches:
A. The car is super smart, understand what's around and know what's the right thing to do (General AI for driving).
B. The car can see lanes, know the basic driving rules, know speed limit for every part of road (either through mapping or reading road signs), slow down when other cars are moving slowly, keep going if the driving path is clear. Understand how to keep safe distance, understand traffic lights. Slowly drive around if there is an object in it's path. All these tasks are relatively simple.
Human drivers have a little bit of both A and B. We naturally have A, the way we learn driving is B. That's how I learned driving, I never learned the millions of conner cases. Basically I am a vision + rule-based driving system using some if-else logic, never had accidents.
I think both A and B can work in the end, but A is much more difficult to implement. So I think the likely scenario is someone will get a decent system out using approach B.
When I think about Tesla's long term valuation, FSD is the key, nothing else matters. If another company gets FSD first, they will attract tons of money and catch up on EV, battery, charger, ride sharing, etc.
Based on Tesla's Autonomy Day presentation, I guess they are doing approach A (I could be wrong). I hope they keep their mind open, maybe start a separate program using A+B approach. Wasting a billion dollars on this would be well worth it. Tesla can't afford to lose the FSD race.
Except for the UK, we now have complete april registration stats for Europe: 4196 cars registered (Tesla Europe Registration Stats).
If the UK has more than 265 registrations, this is the best non-third-month-of-the-quarter result ever, beating even the worst last-month-of-the-quarter of last year.
Is driving so difficult? I’m not so sure. I’ve been driving over 60 years with severe vision problems and slower than average reflexes. In fact, I was legally blind for two years of that time. During that time, I have had very few accidents with zero in the last 23 years. Of those I had, only one was chargeable to me. Frankly, my S is already a better driver than me. It sees better and reacts faster. Knowing my limitations and compensating for them has worked for me, and I think can work for the computer driving my next car.I used to think autonomous driving is super difficult - by far the most difficult project in human history. After I spent more time thinking about the details, I realize it may not be that bad.
There are essentially 2 approaches:
A. The car is super smart, understand what's around and know what's the right thing to do (General AI for driving).
B. The car can see lanes, know the basic driving rules, know speed limit for every part of road (either through mapping or reading road signs), slow down when other cars are moving slowly, keep going if the driving path is clear. Understand how to keep safe distance, understand traffic lights. Slowly drive around if there is an object in it's path. All these tasks are relatively simple.
Human drivers have a little bit of both A and B. We naturally have A, the way we learn driving is B. That's how I learned driving, I never learned the millions of conner cases. Basically I am a vision + rule-based driving system using some if-else logic, never had accidents.
I think both A and B can work in the end, but A is much more difficult to implement. So I think the likely scenario is someone will get a decent system out using approach B.
When I think about Tesla's long term valuation, FSD is the key, nothing else matters. If another company gets FSD first, they will attract tons of money and catch up on EV, battery, charger, ride sharing, etc.
Based on Tesla's Autonomy Day presentation, I guess they are doing approach A (I could be wrong). I hope they keep their mind open, maybe start a separate program using A+B approach. Wasting a billion dollars on this would be well worth it. Tesla can't afford to lose the FSD race.
I'm talking about mainly software. Mobileye has hundreds. Tesla has 0. Even for hardware, that's not even autonomous cars related at all and as Tesla said themselves, neither of those will ever be approved (probably because of prior art or being too broad).
In acouple hours i will post the NN models in the Q4 2017 Eyeq4 versus Tesla's Feb 2019 AP.
That would be the kindergarten variety.
Real developers follow a procedure by which documentation, release information and unit tests(*) are all updated along with the actual code.
(*) Some believe that the unit tests should be written by a separate group of coders who only rely on the documentation (since this helps to test also the documentation), in which case the non-test coder gets to write more detailed documentation (instead of just writing "see unit test code for examples").
Is driving so difficult? I’m not so sure. I’ve been driving over 60 years with severe vision problems and slower than average reflexes. In fact, I was legally blind for two years of that time. During that time, I have had very few accidents with zero in the last 23 years. Of those I had, only one was chargeable to me. Frankly, my S is already a better driver than me. It sees better and reacts faster. Knowing my limitations and compensating for them has worked for me, and I think can work for the computer driving my next car.