Short-Term TSLA Price Movements - 2016

[RobStark]

The "obvious thing" is to put AutoPilot 1.0 on Model 3.

And put AutoPilot 2.0 on Model S and Model X first.

In part as justification for the higher price and secondly to have a smaller and much more forgiving crowd beta test AP 2.0.

 

[Intl Professor]

Keep up the great work. With your help we can go toward homeostatic moderation.

I have now digested relevant posts on a forum for local mods which is the TMC acronym rather than my green-mod or intern-mod. Since I really, really, don't want to have the immediate power to move marginal cases, it is more important for members to report cases of concern so I can handle/reassign. The TMC coders IMHO have already built-in many homeostatic processes.

With a wag hoping to praise the coders here I cried out "Samaritan" for help. Apparently only one member thought it funny, for the rest (who found it off topic), the reference was to a bad and a good machine controlling us with human moderation in the TV series "Person of Interest." The denouement should be the reconciliation of good and evil by having moderation itself moderated by citizens. That is the superiority of the TMC architects. Also, probably why Madison counseled in Federalist 51 (shortened form from memory) "if men were angels, no government would be necessary, if angels governed us there wouldn't be a problem, in constructing a constitution for men governed by men, certain precautions are necessary."

Somewhere I read what pretends to be machine learning involves at first responses by humans before the machine can learn by itself. TMC has done one better by programming the machine to moderate even moderators who are chosen for their naivete about short-term trading.

With the caveat I may have in an earlier life what anglos might call a loose screw, or a native American maker of loose connections, how wise is the great moderator in the sky. During voir dire in a gang case I was selected for trial while an expert I wanted on the jury was dismissed. Of course I am a great local mod on this site, I was a moderator in a gang fight. Thanks, gang.

 

[electracity]

The "obvious thing" is to put AutoPilot 1.0 on Model 3.

And put AutoPilot 2.0 on Model S and Model X first.

In part as justification for the higher price and secondly to have a smaller and much more forgiving crowd beta test AP 2.0.

Tesla likely needs AP 2.0 on as many cars as possible to reach "billions of miles". The miles driven on AP 1.0 likely don't count for building the dataset needed to design most AP functionality.

 

[Krugerrand]

Tesla likely needs AP 2.0 on as many cars as possible to reach "billions of miles". The miles driven on AP 1.0 likely don't count for building the dataset needed to design most AP functionality.

That's like saying the first seat belts designed and installed in cars didn't count towards the functionality of the final seat belt design that we currently use today. Of course AP 1.0 miles count! Without those there wouldn't be any data with which to improve and move onto AP 2.0.

 

[electracity]

That's like saying the first seat belts designed and installed in cars didn't count towards the functionality of the final seat belt design that we currently use today. Of course AP 1.0 miles count! Without those there wouldn't be any data with which to improve and move onto AP 2.0.

Regression analysis. They can't do regression analysis on data not recorded on the current sensor suite. There is no valid analogy here related to seat belts.

They have gained experience, but my comment was specific to the dataset. I'm sure they can bring forward the mapping database, and that AP 2.0 will further refine that aspect of the system.

 

[RobStark]

Tesla likely needs AP 2.0 on as many cars as possible to reach "billions of miles". The miles driven on AP 1.0 likely don't count for building the data set needed to design most AP functionality.

The current data set is being used today to design most AP functionality.

People who thought or wished for Model 3 to be a shrunken by 20% Model S at 50% off keep getting disappointed as Model 3 details emerge.

 

[Krugerrand]

AP 1.0 is the 'heads up' data set, not the 'it doesn't count data set'.

Tesla: Here is our AP 1.0 data set of 1 billion miles. It clearly shows a 50% reduction in accidents. We're getting ready to roll our AP 2.0 and we're confident that it'll show an even more substantial improvement based on what we've already recorded and analyzed behind the scenes. We'll be back in 12 months with a 5 billion mile data set.

NHTSA: That's very interesting. What do you want us to do about that?

Tesla: We're giving you this data so that you can get busy organizing and preparing to implement a new set of safety standards and rules. There'll be no denying the data and the public is going to demand their vehicles come with autonomous driving. You'll not want to be caught with your pants down and behind the curve.

 

[electracity]

Too much Harry Potter, not enough STEM.

 

[Krugerrand]

Regression analysis. They can't do regression analysis on data not recorded on the current sensor suite. There is no valid analogy here related to seat belts.

They have gained experience, but my comment was specific to the dataset. I'm sure they can bring forward the mapping database, and that AP 2.0 will further refine that aspect of the system.

Nobody said anything about regression analysis. Of course they'll present (and I believe they've already done such) AP 1.0 data, just like original seatbelt data was presented and thusly changed rules and safety standards for vehicles. Of course AP 2.0 data will be an improvement over AP 1.0 and that data will also be presented, just as seatbelt designed changed, saved even more lives and thusly changed the rules and safety standards for vehicles AGAIN. That's how it works and how it will work. AP 1.0 data is very important because it gets the ball rolling in the bureaucracy.

 

[Krugerrand]

Too much Harry Potter, not enough STEM.

Too much anal retentiveness not enough common sense.

 

[dakh]

From the Code inter

The "obvious thing" is to put AutoPilot 1.0 on Model 3.

And put AutoPilot 2.0 on Model S and Model X first.

In part as justification for the higher price and secondly to have a smaller and much more forgiving crowd beta test AP 2.0.

That's not obvious at all, in fact it doesn't make any sense to me. If they wanted to limit initial exposure they could just use pricing to control for that. There's no version 1 or 2 of AP, it's a continuing development process and I highly doubt they'll be willing to fork the codebase and support two versions at the same time. Not gonna happen IMHO.

The only way this makes sense is if the full auto mode requires much more expensive hardware.

 

[electracity]

From the Code inter


That's not obvious at all, in fact it doesn't make any sense to me. If they wanted to limit initial exposure they could just use pricing to control for that. There's no version 1 or 2 of AP, it's a continuing development process and I highly doubt they'll be willing to fork the codebase and support two versions at the same time. Not gonna happen IMHO.

The only way this makes sense is if the full auto mode requires much more expensive hardware.

They have to support the old AP 1.0 cars, but obviously those evolve to maintenance releases. I agree that all new cars will likely share the same base AP hardware, with the S/X having superior displays.

First the new S/X with AP 2.0, followed by the model 3 when released. I expect a new UI with AP 2.0, as well as a new touchscreen on the S/X (assuming there is something better available). The timing at the end of the year maintains sales momentum, and keeps potential S/X buyers from waiting to see the electronics on the model 3.

The model 3 reveal likely showed an early version of the new UI skin, but it was ignored. It looked to be flat with a lot of black and white, with color used as highlights. This choice seems consistent with where their interior design is going. Plus it is better to show exceptions to the driver than the current colorful screens.

Obviously unfinished:
http://cdn2.autoexpress.co.uk/sites...lic/2016/04/20160331_213047.jpg?itok=7Hdboxmv

 

[MitchJi]

The "obvious thing" is to put AutoPilot 1.0 on Model 3.

And put AutoPilot 2.0 on Model S and Model X first.

In part as justification for the higher price and secondly to have a smaller and much more forgiving crowd beta test AP 2.0.

That's not obvious at all, in fact it doesn't make any sense to me.

The only way this makes sense is if the full auto mode requires much more expensive hardware.

The sensors for full autonomy will initially be much more expensive and the primary obstacle to adding the HW later is the difficulty of adding the wiring harness. The sensor prices will fall dramatically as the production of the chips ramps. Adding the sensors later will be relatively simple because they need to be designed for easy replacement.

It's obvious that if they announce full autonomy for the M3 before they include the HW on the S and X that they risk Osborning sales of the S and X.

So the best way to roll it out is to as soon as they know the requirements, to start installing the wiring harness, and announce that fact on the S and X. They can include the eventual cost of the sensors for current buyers. If they do that and announce the fact that all the required HW is included, that would be a more effective way to drive demand IMO than the referral programs. I think that's obviously the best way to do it, but the idea is probably not obvious.

 

[dakh]

Think about it a different way: I will not follow through with my Model 3 reservation until the AP situation is clear. If Model S gets AP much sooner I'll just buy that or wait depending on my finances at the time. It makes no sense to spend $40+K on a car that will be drastically obsolete in a year or two.

Personally I think the cost increase would be in redundancy of all the control hardware not the sensors. They keep talking about cameras being mostly all you need and yes those are somewhat specialized but they're not expensive at all.

Edit: ah heck nothing but just having this as an outright option has major issues so I'm getting more and more convinced that that's the "obvious".

There's also the issue of non-AP cars still needing the hardware to do accident avoidance. On a similar note other manufacturers would at that point be on-boarding at least the accident avoidance part and I'm assuming Tesla and MobilEye sensor suit would be similar if not the same.

Bottom line, there's hope sensors and other hardware won't be that expensive and they will simply ship all cars full AP ready and charge to enable it.

 

[AlMc]

@dakh: Your last statement is exactly what I believe will happen. S/X gets new hardware and software for AP 2.0 BEFORE the model3 is released. The model3 is released with hardware capability of AP 2.0....when and how much to turn it on is the question.

 

[Papafox]

@dakh: Your last statement is exactly what I believe will happen. S/X gets new hardware and software for AP 2.0 BEFORE the model3 is released. The model3 is released with hardware capability of AP 2.0....when and how much to turn it on is the question.

Bingo. Also, one can theorize that the current referral program is to boost orders through mid-July. What comes next to boost sales? I suspect AP 2.0 hardware.

 

[ev-enthusiast]

First Model X arrived in Europe, they can be found in the UK showrooms.

 

[Bet TSLA]

The "obvious thing" is clearly going to be a heads up display. Nothing to do with auto-pilot.

 

[RobStark]

The "obvious thing" is clearly going to be a heads up display. Nothing to do with auto-pilot.

Elon's answer "the obvious thing" was regarding a question about Model 3 Auto Pilot and Autonomy.

 

[ludicrousspeed]

Elon's answer "the obvious thing" was regarding a question about Model 3 Auto Pilot and Autonomy.

Exactly. He also said "It would be really big news if I start (elaborating) here." His body language alone gives the sense that it's going to be huge news. I think it's very likely that they are going to announce their own ride-sharing platform based on autonomy.