Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

[LN1_Casey]

What if it hit the toddler in order to block a trolley from careening into a whole group of of toddlers?

I think if there is a trolley full of toddlers running free of any adult help, I think there is more concern here than summoning into another free ranging one.

Specifically, where the duck did I just park my car where there are trolleys of toddlers??

 

[Fact Checking]

Interesting! I missed that. Weird that Enhanced Summon is excluded. Was his quote just for driving with someone behind the wheel?

This was said at the very end of the Autonomy Day Q&A session, at timestamp 13431 seconds (3h:43m:51s):

Q: "If I add my Tesla to the Robo taxi Network who is liable for an accident, is it Tesla or is it me, if the vehicle has an accident and harms ..."

Elon [interrupting the question]: "probably Tesla, probably Tesla ... but the right thing to is to make sure there are very very few accidents"​

And yes, if Tesla runs the Tesla network and runs the FSD software then assuming liability while the FSD software is operating and is carrying passengers without the owner being even there is pretty much the only possible solution - perhaps underwritten by some bigger re-insurance company - but self-insuring by Tesla is possible too.

This insurance service will basically be financed from the revenue sharing construct if say Tesla keeps 30% of robotaxi revenue.

(Of course I assume the owner is liable for the car to be maintained and roadworthy, etc.)

 

[KarenRei]

"The needs of the many?" I don’t remember that in the Three Laws.

Trolley problem - Wikipedia

Of course, there's endless variations of them - something that the internet took to with gusto, trying to create the most ridiculous trolley problems:

The Trolley Problem | Know Your Meme

E.g.:

 

[KSilver2000]

What if it hit the toddler in order to block a trolley from careening into a whole group of of toddlers?

Can’t tell sarcasm level of this post.
In this country, you will be held liable for that one toddler it did hit. Even Good Samaritan law can’t help you with that. You may or may not sleep better knowing you sacrificed one toddler.

 

[ZeApelido]

Right, but higher complexity does not necessarily mean higher difficulty. In programming you can sometimes have a very complex task resolved by some simple code. Especially when dealing with a topic where a person's knowledge is deficient (and for autonomous vehicles its pretty safe to say that everyone here is knowledge deficient) it is important to be clear what the constraints really are.

A good gymnast makes what they do look amazing, but easy and effortless. If someone hasn't solved the higher speed issues (which are different from low speed issues) then even if they have solved the low speed issues they cannot accurately compare them.

I'm surprised I that I need to spell this out, but in a parking lot you don't have to solve for vehicles doing 100mph, roll overs, etc. On an interstate you still have to solve for pedestrians, stopped cars pulling out suddenly, etc. While a grocery cart may not be likely, the equivalent happens. Pedestrians are not supposed to be crossing interstates, but it happens. On a trip earlier this year an accident I saw was the result of a semi pulling into traffic from the shoulder and being rear ended by a pickup trick.

While the probability of these occurrences is a lot lower per mile or minute than in a parking lot you still have to solve for them. And instead of having seconds to respond you may have milliseconds. And not only is more difficult to detect and avoid in time, but the probability of severe injury or death is far higher. No amount of hand waving about "inversely proportional to speed" makes these problems go away.

#1 - These events do occur much less often at higher speed as you say. Being say 10x better than humans will not be affected as much by these very sparse events (pedestrian on highway), than getting right the much more frequent interactions at lower speeds (and also the more common incidents seen at higher speeds). Sure they "have to" solve it, but not as strictly off the bat as in the more concentrated cases.

#2 - No doubt faster speeds means less time for the computer to react, less frames to process / estimate with certainty, poorer resolution. Performance of the car reacting to a deer suddenly wondering out on the highway should be worse than on surface streets. But both cases do require achieving a vision system that can detect all objects, track their movements, and estimate their intent. If Tesla can prove they can do this well at low speeds, then they a great starting point for improving it a high speeds.

 

[Fact Checking]

In this country, you will be held liable for that one toddler it did hit. Even Good Samaritan law can’t help you with that.

Yet airbags killed 175 people in the U.S. alone in the 1990-2000 decade, most of whom were children, and the auto industry didn't collapse from the resulting litigation, because airbags also saved over 6,000 people's lives in the same time period:

If we may believe Wikipedia (which is of course never, but let's):
From 1990 to 2000, the United States National Highway Traffic Safety Administration identified 175 fatalities caused by air bags. Most of these (104) have been children. About 3.3 million air bag deployments have occurred during that interval, and the agency estimates more than 6,377 lives saved and countless injuries prevented.[113][121]

Deaths or injuries by airbags are rare, so the companies deal with it on a case by case basis. In some cases it could go to Court, in others there could be an agreement.

So I think the Tesla Dash-cam feature is important in proactively collecting a very robust set of video evidence for the many scenarios where FSD technologies save Tesla owners and other people from traffic accidents and harm.

 

[KSilver2000]

Tesla has never stated that it will accept liability. On the contrary, it has already explicitly stated that it will not accept liability for enhanced summon (Elon Must tweet : Yes, but owner will have to accept tiny risk of damage. Those are very hard even for careful human drivers.) So the question is more : what is the liability/risk balance for individual users. For example, me personally, I would never use Enhanced Summon without clear line of sight on the car. Tiny risk for damage, ok smaller liability. But a tiny risk to hit and kill a toddler standing near the car is not something I can take on. It would have to be astronomically small for me to accept that, not just tiny.

Tesla will never assume liability for anything related to use of enhanced/autopilot/FSD. To cover this would be very overwhelming for them.

 

[Fact Checking]

Tesla will never assume liability for anything related to use of enhanced/autopilot/FSD. To cover this would be very overwhelming for them.

That's contrary to what Elon said on the Autonomy Day:

This was said at the very end of the Autonomy Day Q&A session, at timestamp 13431 seconds (3h:43m:51s):

Q: "If I add my Tesla to the Robo taxi Network who is liable for an accident, is it Tesla or is it me, if the vehicle has an accident and harms ..."

Elon [interrupting the question]: "probably Tesla, probably Tesla ... but the right thing to is to make sure there are very very few accidents"​

And yes, if Tesla runs the Tesla network and runs the FSD software then assuming liability while the FSD software is operating and is carrying passengers without the owner being even there is pretty much the only possible solution - perhaps underwritten by some bigger re-insurance company - but self-insuring by Tesla is possible too.

This insurance service will basically be financed from the revenue sharing construct if say Tesla keeps 30% of robotaxi revenue.

(Of course I assume the owner is liable for the car to be maintained and roadworthy, etc.)

 

[KSilver2000]

Yet airbags killed 175 people in the U.S. alone in the 1990-2000 decade, most of whom were children, and the auto industry didn't collapse from the resulting litigation, because airbags also saved over 6,000 people's lives in the same time period:

Okay...? And so?

What does that have to do with what I said?

 

[KSilver2000]

That's contrary to what Elon said on the Autonomy Day:

First, I didn’t mention anything about “robotaxi”. Don’t know why some people keep twisting what I say. But, I’ll play along.

Probably != Yes

Okay, he didn’t say “never” or “will”.
Once he understands the true cost and risk, I doubt they would though. They have yet to do so for anything.

 

[Green Pete]

#1 - These events do occur much less often at higher speed as you say. Being say 10x better than humans will not be affected as much by these very sparse events (pedestrian on highway), than getting right the much more frequent interactions at lower speeds (and also the more common incidents seen at higher speeds). Sure they "have to" solve it, but not as strictly off the bat as in the more concentrated cases.

#2 - No doubt faster speeds means less time for the computer to react, less frames to process / estimate with certainty, poorer resolution. Performance of the car reacting to a deer suddenly wondering out on the highway should be worse than on surface streets. But both cases do require achieving a vision system that can detect all objects, track their movements, and estimate their intent. If Tesla can prove they can do this well at low speeds, then they a great starting point for improving it a high speeds.

OR to put another way high speed with required driver attention to develop the basics (as far as I am aware AP would not dodge a tire in the road) => use that systems intelligence to develop low speed situations, like parking lots which have lower consequences for hitting something but a far higher frequency => apply what is learned at low speeds to high speeds and now AP on highways can avoid potholes and junk on the road autonomously.

Then you just need to read traffic signals.

 

[ZeApelido]

Everyone needs to focus. Like, what is going to be the next Tesla killer? Are we underestimating the efficiency of electric motorcycles?

Do you feel threatened now?

 

[Fact Checking]

First, I didn’t mention anything about “robotaxi”.

You did say "FSD", which is what I replied to:

Tesla will never assume liability for anything related to use of enhanced/autopilot/FSD. To cover this would be very overwhelming for them.

"Robotaxi" is perhaps the most important subset of FSD functionality under a number of bull theses such as ARK's.

If you wanted to exclude "Robotaxi" from FSD functionality from your "Tesla will never assume liability" claim, then please do so explicitly.

 

[T@oo]

Ladies and Gentlement, we have green. Out of pure coincident, I‘m sitting on a green chair.

 

[Mars ☰mperor]

green

 

[elasalle]

If TSLA ticker was displayed on TMC, all red/green messages would disappear

 

[kengchang]

Green yes?

 

[capster]

To me this idea of sleeping in the car or no humans in the car are imprecise ways of looking at the problem of FSD.
...

Agreed. Almost everybody is using "sleeping in a car" to be an imprecise short hand for some pinnacle of autonomous driving. That is not especially useful.

However, sleeping in the car is useful if you break it down into actual use cases and consider these.

For example, sleeping in the car would be most valuable (to me) to go from one city to another via the freeway.

Now, let's try to simplify the problem for that use case, like so:

Let's say I first have to drive to a rest stop on a freeway. Then, I can lay down in the back of my Model 3 and sleep while the car takes me to the rest stop on the freeway nearest my destination. Once there, the car wakes me up and I drive off.

We could even add in a restriction on hours of no driver behind the wheel on the freeway travel to begin with. There are usually few cars and mostly professional truck drivers on the freeway in the wee hours.

My experience with the Model 3 on the freeway with Navigate on Autopilot but without the new chip has been phenomenal.

With the new chip? I suspect the 'rest-stop-to-rest-stop sleep in the car' use case should be solvable (by Tesla and especially in the wee hours) to a reasonable number of nines by now or in the near future.

This use case alone is a huge value driver for Tesla. Yes, eventually I hope to clamber in the back in my garage and wake up at my destination, but that functionality shouldn't be the rate limiting factor for the sleep use case as I outlined it above.

This yet another reason why I consider each TSLA share a golden ticket.

 

[urarthur]

Here's the latest TSLA options open interest for tomorrow's expiry, in $10 price steps, trimmed at $300:

 PUT $010:  3,695 #                   ,   CALL $010:      2                 
 PUT $020:  2,374                     ,   CALL $020:      0                 
 PUT $030:  1,846                     ,   CALL $030:      0                 
 PUT $040:  3,264 #                   ,   CALL $040:      0                 
 PUT $050: 58,792 ###################+,   CALL $050:      1                 
 PUT $060:  2,565 #                   ,   CALL $060:      0                 
 PUT $070:  2,704 #                   ,   CALL $070:      0                 
 PUT $080:  4,971 #                   ,   CALL $080:      0                 
 PUT $090:  1,820                     ,   CALL $090:     35                 
 PUT $100: 47,881 ################    ,   CALL $100:     81                 
 PUT $110:  2,570 #                   ,   CALL $110:     40                 
 PUT $120:  2,509 #                   ,   CALL $120:      6                 
 PUT $130:  6,248 ##                  ,   CALL $130:     41                 
 PUT $140:  9,747 ###                 ,   CALL $140:    255                 
 PUT $150: 11,748 ####                ,   CALL $150:    126                 
 PUT $160: 11,517 ####                ,   CALL $160:    311                 
 PUT $170:  7,235 ##                  ,   CALL $170:    343                 
 PUT $180: 12,596 ####                ,   CALL $180:  2,473                 
 PUT $190: 11,253 ###                 ,   CALL $190:  1,667                 
 PUT $200: 27,591 #########           ,   CALL $200:  5,739 ##             
 PUT $210: 35,092 ###########         ,   CALL $210:  3,308 #               
 PUT $220: 20,011 ######              ,   CALL $220: 18,190 ######         
 PUT $230: 46,687 ###############     ,   CALL $230:  6,905 ##             
 PUT $240: 46,413 ###############     ,   CALL $240: 32,733 ###########     
 PUT $250: 45,444 ###############     ,   CALL $250: 31,023 ##########     
 PUT $260:  7,222 ##                  ,   CALL $260: 33,838 ###########     
 PUT $270: 15,953 #####               ,   CALL $270: 24,038 ########       
 PUT $280:    907                     ,   CALL $280:  8,284 ##             
 PUT $290:    373                     ,   CALL $290:  5,659 ##             
 PUT $300:    132                     ,   CALL $300:  8,187 ##             
 
2019/Jul/19:  PUTs:   451,305 ; CALLs:   207,422

Observations:

• Huge open interest of 658,000 options (65 million TSLA shares-equivalent position ...), with a 3:1 put/call ratio. The delta-hedging effect must be significant.
• Bearish (PUT) consensus is in the $230-$250 range.
• Bullish (CALL) consensus is in the $240-$270 range.
• Market maker equilibrium (losses to puts equals losses to calls) is around $250.
• If the "max pain" hypothesis holds true tomorrow (it might not: fundamental news is usually stronger than market-maker forces), then market makers would want the price to be in the $250 range, to deny all but the $270 PUT profits, while paying out much less to the call side.
• If there's a bear raid on any negative news they'd want to push the price below $240.
• If the price holds up a huge number of PUTs (420,000 of them) will expire worthless, including over 130,000 bankwuptcy bets at $100 and lower.

I really don't see how this could have significant effect on SP. Looks like 100.000 call contracts will end in the money, at ~$10 on average, that's only worth $100 mln. Daily average dollar volume in the stock is 11 mln shares x $250 = $2.75 bln. So only 5% of the daily $ volume in the stock will be handed over to call buyers.

I have a hard time believing ppl could significantly manipulate SP with calls/puts given these dollar volumes are relatively low compared to daily volume in the stock.

If market is liquid and volume is on average, option market players IMO really dont stand a chance in manipulating the SP to their advantage.

 

[Foghat]

How California can build 3.5 million new homes - Alfred Twu - Medium

It might not come as a surprise to many of us that Tesla Energy is setting up to for significant revenue stream boost come 2020 and beyond.

Might a large scale Solar roof roll out (along with powerwalls) with home builders in California be just the beginning?

What about the multi-year-in-development Tesla grid services being launched with it as well?

Again, the most exciting product launch line up of any public company out there. Next stop, energy industry.