2017 Investor Roundtable:General Discussion

[myusername]

Fuzzy pattern matching is the human behavior that machine learning is really trying to mimic.

Early autonomous cars - think DARPA challenge - weren't doing it that way. It was much more like you think, teaching a car to drive by writing the rules of the road into computer code. What Tesla is doing is much closer to Dad teaching a 16 year old to drive by showing him.

It's a big part of why Tesla's can autonomously drive in many conditions that more primitive autonomous vehicles like Google's car cannot.

Machine learning works. It's a problem of collecting a large enough data set for it to be able to infer how to deal with new situations, and Tesla has certainly got the data part down.

look at all the leaps and speculation you guys are making here to justify your points... then you collectively declare @neroden incorrect and somehow that reenforces your opinions... (i'm picking on you here racer, but commenting on the general points by others in the conversation).

do you guys really think you know what you're talking about when you say GPS is a tool for automated driving?... do you understand the latencies of radio waves in the distances of geostationary orbit?... do you guys think you can actually work through the details of Tesla's implementation here on a message board in an educated way?

@larmor: what does this even mean?... "Parts of driving by humans is better for now, but feed that data into multiple supercomputers...."

it sounds like dialog from NCIS?

@racer26: "It's a big part of why Tesla's can autonomously drive in many conditions that more primitive autonomous vehicles like Google's car cannot."

racer -- is this your justification of why Tesla will achieve what Google hasn't?... because if you think Google is not capable of accomplishing what Tesla is capable of... then you are so incredibly wrong... AND MORE IMPORTANTLY... your statement is actually just wrong... Google has actually been using machine learning approaches in their implementation for years.

I won't speak for @neroden... but I'll just restate my general opinion on the subject...

much of "Automated Cars" is a sales pitch... and many of you are now convinced it's the near future... I mean duh... AI has now made our cars sentient beings... soon there will be coffee shops made only for cars so they can go have philosophical conversations about sharing a planet with humans.

guys... fully autonomous cars is hard... autopilot in planes is super easy compared to this... we are nowhere near the TV fictional reality that you think we are... and many of the timelines Elon has put on this and other subjects are getting closer... will the goal post just get moved again?... and will once again nobody give a &*$# and just continue to declare Tesla the never arriving future?

 

[myusername]

RSI: 81.77 -- a multi year high

the only time it's ever been this high was in May of 2013 to $80-$100... where it then began to steadily climb and did not retrace... since then... all other RSI peaks like this retraced anywhere from 50% to 150%... this has to be at least a short term top.

 

[Ovulation]

Well a lot of people putting billions of dollar into the self driving problem ... They vote with their wallets on self driving cars and have done some work (a lot of) in that area, so maybe they have a better judgment than we have.

BTW. If self driving cars are impossible it's not bad for Tesla, they still will sell a lot of electric cars, otherwise if self driving is possible, it's crucial to make the early investment and not get disrupted.

 

[dakh]

Visual pattern matching is basically the only one that matters for driving. If it's better than humans by your own admission, then we've got the building blocks we need.

I do not challenge your view that driving as a problem is *hard*. I work in R&D for a technology firm as a programmer and product designer, and spent much of my youth building robots for the FIRST Robotics Competition. Trust me when I say that I understand how difficult it is to get a complex electromechanical system to autonomously respond to the world around it, even in very narrowly scoped ways.

All I'm saying is that there is no technical reason a computer with the capabilities that autopilot has, couldn't do a better job than a human provided it knows how.

I believe that eventually it will. You seem to believe the problem domain is so large that it can never be fully solved. On this we disagree.

Actually you two agree, you just for some reason empathize specific aspects of the story. The rock bottom here is getting these dangerous two+ ton hulks of metal from point A to point B without running into things and not letting other 2+ ton hulks of metal run into you.

Narrow AI will not be able to match humans in handling complicated corner cases any time soon. But it pays attention 100% of the time, sees though things and got a hair trigger on the brake pedal. So it'll solve the A to B problem better than humans, eventually. Better being more efficient and safer. With an added benefit of not having to spend a lot of resources on training the new "drivers" and dealing with their rookie mistakes. Just look up insurance rates for teenagers.

 

[dakh]

guys... fully autonomous cars is hard... autopilot in planes is super easy compared to this... we are nowhere near the TV fictional reality that you think we are... and many of the timelines Elon has put on this and other subjects are getting closer... will the goal post just get moved again?... and will once again nobody give a &*$# and just continue to declare Tesla the never arriving future?

You think lip-reading is hard?

Elon can run his mouth on when they're gonna deliver stuff all day long and that's expected it's not gonna quite be that. But!

Name me another team that has as good of a shot at achieving full autonomy first.

Name me another company that can monetize that as efficiently as Tesla. Yeah, Uber might be a contender. But at the end of the day they've got a pretty weak hand, in my opinion. They get a mention because they got SOME hand, unlike the rest of 'em that don't have any.

 

[dakh]

Visual pattern matching is basically the only one that matters for driving. If it's better than humans by your own admission, then we've got the building blocks we need.

I do not challenge your view that driving as a problem is *hard*. I work in R&D for a technology firm as a programmer and product designer, and spent much of my youth building robots for the FIRST Robotics Competition. Trust me when I say that I understand how difficult it is to get a complex electromechanical system to autonomously respond to the world around it, even in very narrowly scoped ways.

All I'm saying is that there is no technical reason a computer with the capabilities that autopilot has, couldn't do a better job than a human provided it knows how.

I believe that eventually it will. You seem to believe the problem domain is so large that it can never be fully solved. On this we disagree.

We're talking about timelines of commercially available products here. "Never" simply means "not in my investment timeframe" in that context.

And yes there's a technical reason, at least for now. AI isn't trained on general life on Earth knowledge, it just gets to know how to drive vehicles. It has a lot less context than a non-artificial neural network that's exposed to life beyond the road, and some driving decisions do require that context.

 

[myusername]

You think lip-reading is hard?

Elon can run his mouth on when they're gonna deliver stuff all day long and that's expected it's not gonna quite be that. But!

Name me another team that has as good of a shot at achieving full autonomy first.

Name me another company that can monetize that as efficiently as Tesla. Yeah, Uber might be a contender. But at the end of the day they've got a pretty weak hand, in my opinion. They get a mention because they got SOME hand, unlike the rest of 'em that don't have any.

I don't think any team is close... but I think basically ALL auto companies have teams working hard on it... and what I absolutely do know is Tesla is the one company that's publicized it the most... put the future of the company on the line for it... and that's it... GM doesn't constantly come out with declarations of timelines... but they absolutely are spending hundreds of millions of dollars on it... along with Toyota and VW and... and... and...

so... again, my opinion is... I don't think Tesla is in the lead... I just think they are the loudest about it and taking the most risk on the roads... the stock is highly publicized... and that combined with human's constant interest in "futuristic" concepts... Tesla is getting a lot of attention.

 

[schonelucht]

View attachment 211848

I foresee a lot of word parsing on this thread come 3 to 6 months about what constitutes 'noticeable'. What would your yardstick be?

 

[vgrinshpun]

I foresee a lot of word parsing on this thread come 3 to 6 months about what constitutes 'noticeable'. What would your yardstick be?

FSD officially staring to be designated as capable to handle (with the driver supervision of course) surface streets, as oppose to highways for EAP.

What about yours?

 

[schonelucht]

I don't think any team is close... but I think basically ALL auto companies have teams working hard on it... and what I absolutely do know is Tesla is the one company that's publicized it the most... put the future of the company on the line for it... and that's it... GM doesn't constantly come out with declarations of timelines... but they absolutely are spending hundreds of millions of dollars on it... along with Toyota and VW and... and... and...

so... again, my opinion is... I don't think Tesla is in the lead... I just think they are the loudest about it and taking the most risk on the roads... the stock is highly publicized... and that combined with human's constant interest in "futuristic" concepts... Tesla is getting a lot of attention.

I think so too. Mercedes basically had the same autonomous drive promo video that Tesla released in November back in 2012, it's still on youtube. Volvo's committed to a solution that is 100% safe while Elon is claiming 90% reduction only in fatality. Basically every single major car company is working on it and there is may be something where they can claim a lead. Fact remains, no team delivered yet but Tesla's certainly received the most media attention so far due in part to having the most aggressive time line of them all. Unfortunately sometimes Tesla timelines suggest progress that isn't quit there yet (see the whole 'expected to reach parity by 2016' episode)

Interestingly enough, the configurator still claims that parity is expected by end of 2016. Yet it is end of January and we are not at parity. At some point you can't claim honest optimism anymore and you must admit that keeping that misleading text there throughout various other changes to the configurator must be a conscious decision.

 

[Hogfighter]

TSLA is now up 38% in only two months from the Spiegel Bottom. Tesla will pass Ford in market cap at $304 if Ford keeps its current price. On that day I think it's fair to assume Mark locks himself alone in his work/live apartment and blasts 'Black Hole Sun' on repeat for 18 straight hours.

I love that phrase, 'Spiegel Bottom'. Almost as good as 'Brodered'.

 

[schonelucht]

FSD officially staring to be designated as capable to handle (with the driver supervision of course) surface streets, as oppose to highways for EAP.

What about yours?

I think that's more 'enhanced autopilot feature' than an 'autonomous feature' in my mind. For me the essence of self-driving means that the final responsibility is not anymore in the hands of the driver. I would suggest : the ability to handle complete park/summon scenario in a multi-car car park/garage with the driver getting out of the car. If regulatory approval is a problem, I am fine if it's only on private grounds. And obviously I am not talking about a demo video but about firmware that is actually in customer's cars.

 

[vgrinshpun]

I think that's more 'enhanced autopilot feature' than an 'autonomous feature' in my mind. For me the essence of self-driving means that the final responsibility is not anymore in the hands of the driver. I would suggest : the ability to handle complete park/summon scenario in a multi-car car park/garage with the driver getting out of the car. If regulatory approval is a problem, I am fine if it's only on private grounds. And obviously I am not talking about a demo video but about firmware that is actually in customer's cars.

We see it in different terms. For me FSD is not autonomous, it requires presence of the driver that pays attention and supervises situation. IMO we will not see EAP being officially sanctioned on surface streets (see Tesla description below), this would be main distinction between the two.

Also note that Tesla puts self parking as part of EAP, not FSD.

EAP:
The significantly increased sensor information is processed by a computer that is over 40 times more powerful than before. Your Tesla will match speed to traffic conditions, keep within a lane, automatically change lanes without requiring driver input, transition from one freeway to another, exit the freeway when your destination is near, self-park when near a parking spot and be summoned to and from your garage. That said, Enhanced Autopilot should still be considered a driver's assistance feature with the driver responsible for remaining in control of the car at all times.

FSD:
This doubles the number of active cameras from four to eight, enabling full self-driving in almost all circumstances, at what we believe will be a probability of safety at least twice as good as the average human driver. The system is designed to be able to conduct short and long distance trips with no action required by the person in the driver’s seat. For Superchargers that have automatic charge connection enabled, you will not even need to plug in your vehicle.

 

[sdtslafan]

I don't think any team is close... but I think basically ALL auto companies have teams working hard on it... and what I absolutely do know is Tesla is the one company that's publicized it the most... put the future of the company on the line for it... and that's it... GM doesn't constantly come out with declarations of timelines... but they absolutely are spending hundreds of millions of dollars on it... along with Toyota and VW and... and... and...

so... again, my opinion is... I don't think Tesla is in the lead... I just think they are the loudest about it and taking the most risk on the roads... the stock is highly publicized... and that combined with human's constant interest in "futuristic" concepts... Tesla is getting a lot of attention.

Have you seen the uncut self driving video on the Tesla website? I don't see any other car companies providing such a demonstration. They are so modest right?

 

[vgrinshpun]

I think so too. Mercedes basically had the same autonomous drive promo video that Tesla released in November back in 2012, it's still on youtube. Volvo's committed to a solution that is 100% safe while Elon is claiming 90% reduction only in fatality. Basically every single major car company is working on it and there is may be something where they can claim a lead. Fact remains, no team delivered yet but Tesla's certainly received the most media attention so far due in part to having the most aggressive time line of them all. Unfortunately sometimes Tesla timelines suggest progress that isn't quit there yet (see the whole 'expected to reach parity by 2016' episode)

Interestingly enough, the configurator still claims that parity is expected by end of 2016. Yet it is end of January and we are not at parity. At some point you can't claim honest optimism anymore and you must admit that keeping that misleading text there throughout various other changes to the configurator must be a conscious decision.

BTW, some time ago I recall you mentioning that you exited your TSLA position. Could I ask if you are back in, at least partially?

 

[Irishjugg]

It will NEVER have the level of context-sensitivity that a human is capable of acquiring from a human's years of experience. It will always be an idiot savant.

You know the more I think about this the more I think it's irrelevant. Is there any data out there that would speak to how much of this context is truly required vs. only required because of humans slow processing and very limited set of input data?

Basically with the much richer input data from 360 cameras and radar and such paired with evaluating thousands of times per second I have a strong feeling that this completely removes the need for these subtle non visual cues that humans use to compensate for being comparatively Ill suited to the driving task.

 

[schonelucht]

Also note that Tesla puts self parking as part of EAP, not FSD.

EAP:
The significantly increased sensor information is processed by a computer that is over 40 times more powerful than before. Your Tesla will match speed to traffic conditions, keep within a lane, automatically change lanes without requiring driver input, transition from one freeway to another, exit the freeway when your destination is near, self-park when near a parking spot and be summoned to and from your garage. That said, Enhanced Autopilot should still be considered a driver's assistance feature with the driver responsible for remaining in control of the car at all times.

That is about automatic parallel parking ('near a parking spot') and the summon that we know from AP1 : basically move forward to line up in garage. I would EAP in this context perform better in terms of following curves in your driveway than AP1 etc.

Compare with the relevant quote from the self-driving option

When you arrive at your destination, simply step out at the entrace and your car will enter park seek mode, automatically search for a spot and park itself.

Again the line between EAP and FSD is control : with EAP the driver is in charge of selecting the parking spot. With FSD the car takes over and doesn't need driver input anymore.

Your highlighted text out of the same FSD says the same thing : it will be able to drive with no action (therefore no supervision) from the person in the driver's seat.

 

[schonelucht]

BTW, some time ago I recall you mentioning that you exited your TSLA position. Could I ask if you are back in, at least partially?

Yes, I held no Tesla for a very long time. I went back in after Q3 which basically blew away any of my concerns. Suffered through the dip and here we are. I am considering getting out before the annual report but have not yet pulled that trigger.

 

[vgrinshpun]

That is about automatic parallel parking ('near a parking spot') and the summon that we know from AP1 : basically move forward to line up in garage. I would EAP in this context perform better in terms of following curves in your driveway than AP1 etc.

Compare with the relevant quote from the self-driving option



Again the line between EAP and FSD is control : with EAP the driver is in charge of selecting the parking spot. With FSD the car takes over and doesn't need driver input anymore.

Your highlighted text out of the same FSD says the same thing : it will be able to drive with no action (therefore no supervision) from the person in the driver's seat.

No action is NOT the same as no supervision

 

[sundaymorning]

Yes, all the dinosaur autos are on the same level of autonomy as Tesla, they're just better at hiding their achievements until official rollout. Dinosaur autos are also on the same battery level as Tesla, in fact, they all have GFs hidden underground somewhere bc you they're all working hard to bring 10 models of 300mp charge by 2020. C'mon shorts, get real and it's time to pay up. This rise is long overdue.