Poll: 81% of Prospective Model 3 Owners Say They Won’t Pay Upfront For Full Self-Driving

[vc_row][vc_column][vc_column_text]It seems most prospective Model 3 owners aren’t willing to shell out cash upfront for a $3,000 “full self-driving capability” option that is likely years away from becoming available to engage.

In a poll posted by jsraw 81.3% (347) of respondents said they will not pay for the feature at purchase. Adding the option later will cost an additional $1,000. Of respondents, 18.7% said they will pay for FSD upfront.

According to Tesla’s website, FSD “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.”

Elon Musk has said that level 5 autonomous driving is possible with second generation Autopilot and the FSD option, meaning the car is fully autonomous in any and all conditions. During his TED talk in April, Musk said the company plans to conduct by the end of 2017 a coast-to-coast demo drive from California to New York without the driver touching the wheel.

Obviously, there will be regulatory hurdles ahead and Musk has said it will likely be two years before owners will be able to engage FSD capability.

See a few comments on the poll below, or go to the thread here.

Swift

EinSV

jason1466

Waiting4M3

Enginerd[/vc_column_text][/vc_column][/vc_row]

 

[JeffK]

I´m a software developer myself and you seriously underestimate the complexity of those systems......
Nothing about this whole thing is trivial seriously nothing!
Systems under heavy development for years and years can get thrown off by minor details.
............this is not a thing where errors are acceptable and I really don`t think you understand what it means to have a system that has to combine a wide range of sensor input data to a coherent and 100% correct picture.

What you`re imagining is "ok" for something like a photo-edit-program, it`s absolutely inacceptable for a system that`s steering a 2 ton vehicle. We don`t have room for errors in the industry.....
If i make a mistake some 10 bar pipe might kill someone, or if an autopilot makes a mistake it might cause a huge crash.

Autopilot can and will crash statistically speaking. The goal is to simply be safer than a human (2x and then 10x are Tesla's goals) .

Find a traffic sign dataset, add a whole bunch of images of graffiti painted stop signs to the stop sign set. Train a CNN on it or retrain a model already trained on signs with your messed up stop signs. I'd bet a large percentage of those stop signs get labeled as stop signs after that.

FSD will not be absolutely safe and accident free.... ever. It doesn't have to be.

 

[Dynastar]

Impressive amount of negativity for people that are buying model 3s. Kind of feels like SpaceX fans sitting around saying "sure, love the Falcon 9, but landing a rocket? That's rediculous." Tesla has already accomplished lots of "impossible" things to make the 3, in 2017, a reality. Not always on time, sure. The 3 was later then hoped (vs original plan years ago), AP1 took longer to activate then promised, AP2 equality with AP1 took longer, but they all happened. Tesla isn't BSing you when they say FSD is on the way.

If they can do an LA-NY demo drive in the next 9 months then I'm sure we'll start seeing FSD features trickling out in the same time period. Doesn't mean we will be sleeping in our cars, but that in urban settings you'll be able to let the car do more of the work, while still being ultimately responsible for its actions. Maybe that won't be helpful for some, but I look forward to it.

 

[landis]

So even if it gets automated in some way this will require you to always verify the perceived reality of the car...basically making it useless.

the whole ap things is nonsense unless it gets to level 3 and higher because that`s the point where the driver actually gains freedom/time. Before that some people might "abuse" the systems and do stuff other than watching their car

Not at all, it's the same level of complication as stop and go traffic. It's even more simple if, as the rumor stated, the driver would be required to hit the accelerator to go again. I don't think AP is nonsense at all... like I said it's ADAS (advanced driver-assistance systems).

Yes, I also expect FSD features will add (at least some) additional 'convenience' to EAP even while waiting for regulations to allow >L2.
Yes, it could be a long wait for the regulations to catch up but as a technologist, I will have more fun watching from the 'inside'.

What use does such a system have? the driver doesn`t gain anything from it. Worst case he even has to aggressivly correct fail-decisions of the system.

Easy, it will increase the safety and convenience of 'AP'.

I drive two hours on non-telecommuting days mostly on 4-6 lane highways and have already been eagerly anticipating EAP.
Since half of the journey is not limited access and has many traffic lights it is reasonable to expect that even very limited FSD features can manage to increase the convenience for this portion. I expect to remain mentally engaged and alert, mostly by increasing my awareness of the surrounding traffic.

 

[Phrixotrichus]

Autopilot can and will crash statistically speaking. The goal is to simply be safer than a human (2x and then 10x are Tesla's goals) .
FSD will not be absolutely safe and accident free.... ever. It doesn't have to be.

You`re talking from the viewpoint of an optimistic customer, but you have absolutely no idea about the legal and technical side of things.
To have the authorities approve of an automated system we have to go through really harsh tests every time, proving that our system is safe in basically every possible and impossible situation a human can think of ...and our systems are stationary and don`t drive around at xy mph with pedestrians and other moving objects around them.

I seriously don`t think that most people here have ANY idea what it means for a system to get approval to be truly autonomous. There`s billions of $ in insurance behind it.....

I absolutely do believe that FSD will come at some point, I just seriously doubt that it will happen during the lifetime of the model 3, or that it`s even possible with its hardware....

Find a traffic sign dataset, add a whole bunch of images of graffiti painted stop signs to the stop sign set. Train a CNN on it or retrain a model already trained on signs with your messed up stop signs. I'd bet a large percentage of those stop signs get labeled as stop signs after that.

That stuff looks nice in documentations or study-papers....practical use is limited since reality is a teensy weensy bit more complex than most datasets, still.......believe me when I say that immediate pattern recognition and interpretation is still the one discipline where humans are way superior to any computer.

 

[EVNow]

Before FSD can become a reality first EAP has to be working nearly flawlessly. From all the posts we can read here, that is far from reality. Step n can't come quickly if step 1 hasn't happened yet.

Humans regularly overestimate what can be achieved in the short term and underestimate what can be done in the long term. So, while being optimistic that FSD can come in 10 years time, I see no evidenced to suggest it will come within the next 5 years. I'll most probably have a different car (newer 3 or Y or whatever) by the time FSD is available.

 

[tander]

But, you won't get to use FSD until regulators authorize its use...

IIRC regulators have already put laws on the the books in many areas years ago, Google has had cars on the road for many years now. As I understand it the main reason for the wait on FSD is they are collecting an arsenal of data for liability reasons, they have to be able to prove beyond a reason of doubt that it's safer than people before releasing it, otherwise they will have an army of lawyers suiting up every time something goes wrong.

 

[ELRev]

I'm not a scholar on this subject but have done a fair amount of my own research, and I'd like others' opinions on the following:

Many in this thread have referenced the uphill legal battle facing FSD capability, but what is the nature of the laws currently in place to prohibit or outlaw such technology?

My research has led me to understand that there are no such laws that make FSD "illegal" and the legal language around the subject is vague enough that there's more than a healthy argument to be made for the technology's legality as legislation currently stands, rather than requiring a long, uphill battle.

Tesla has been on the forefront of this conversation already with EAP. It seems likely that, once Tesla's FSD tech has had ample time to learn and refine itself, Elon would give the green light and put the ball in the court of legislators, rather than trying to skirt around a potential legal battle that may currently have no justification for stopping that train as legislation currently stands.

 

[Flyshacker]

I read several places that a new communication standard is being developed to be used between self-driving cars and the road and other surfaces. Once that is developed and approved by governments, a car will need to have the equipment to communicate according to this standard in order to be legally self-driving. IOW, it won't be enough to have all the cameras and software to use a self-driving car legally on the road. AFAIK, Tesla has never mentioned communication equipment in any of their car models. That may mean that all existing Teslas (and other cars) will NEVER be street-legal for self-driving until the necessary communication equipment is purchased and installed. So, don't get your hopes up!

 

[claudelaval]

My friend, you are an optimist, but I'm afraid you're deluding yourself. Do you really think Tesla has FSD working in their test builds but somehow they are just keeping it secret to themselves? They are utterly unable to deliver FSD at the moment and probably not for many years with new hardware.

There is no "fleet learning" going on. Tesla has no way to transmit and receive camera feeds from the cars in any usable capacity.

IMO EAP & FSD won't be done by their own. I mean the owner of AP1.0 really believe that their car would be able in the near future of doing it but we know now that it won't. By making customers pay for FSD right now Tesla may get sue if it's not done in a near future.

 

[EVNow]

IIRC regulators have already put laws on the the books in many areas years ago, Google has had cars on the road for many years now. As I understand it the main reason for the wait on FSD is they are collecting an arsenal of data for liability reasons, they have to be able to prove beyond a reason of doubt that it's safer than people before releasing it, otherwise they will have an army of lawyers suiting up every time something goes wrong.

This is not quite correct. Google and others have got limited approval to test self driving cars. NHTSA has published guidelines for the industry & states, but their approach is - bring the vehicle to us, we'll test it before approving (paraphrasing a WSJ podcast interview). Various states have enacted some laws as well, establishing working groups and such.

http://www.ncsl.org/research/transp...elf-driving-vehicles-enacted-legislation.aspx

 

[claudelaval]

So, in your mind, would the intermittent-FSD you describe still be entitled to use the unmitigated FSD name? (Note: You have identified absolutely no organizationally-accepted standard for FSD...)

Resounding no. I don't preorder software, full stop. Besides, if the car gets totaled before the first FSD feature gets released, then you just pissed away $3000 for nothing. As for government regulation, you can be assured its going to take a while, and I'll bet that the topic of autonomous vehicles with no human required is going to be politicized every which way.

Absolutely right AceSkywalker. You can count on the association for the blind, cyclist,deaf people will jump on the bandwagon and politicians will put roadblocks forever...hope we're wrong!

 

[JeffK]

Various states have enacted some laws as well, establishing working groups and such.

Various other states have not and therefore have no legislation making it illegal.

 

[gregincal]

And I´ll stick to my opinion that anything that`s below level 3 is useless. It´s just a gimmick. The driver doesn`t gain anything from it since he still has to focus on the road.

Wow, I guess you should tell all the clueless car manufacturers to get rid of cruise control because you have deemed it worthless.

 

[JeffK]

To have the authorities approve of an automated system we have to go through really harsh tests every time, proving that our system is safe in basically every possible and impossible situation a human can think of ...and our systems are stationary and don`t drive around at xy mph with pedestrians and other moving objects around them.

This is simply not true and there's no way to do this. Instead your proof is in statistics. Same is true with FDA approved drugs, etc. Stats are behind a number of safety related laws and regulation.

I absolutely do believe that FSD will come at some point, I just seriously doubt that it will happen during the lifetime of the model 3, or that it`s even possible with its hardware....

Cool beans... but that doesn't stop smaller features from being activated only for people who purchased the FSD upgrade, like I've mentioned.

That stuff looks nice in documentations or study-papers....practical use is limited since reality is a teensy weensy bit more complex than most datasets, still.......believe me when I say that immediate pattern recognition and interpretation is still the one discipline where humans are way superior to any computer.

We are talking about image recognition and the imageNet challenge has shown that computers can be better than humans at image recognition. We also use machine learning at my place of employment every single day for use in a highly regulated industry. It's way more complex than image recognition and yet we still get the job done with machine learning, simulation, and real world testing.

 

[siggyfreud]

I'll be purchasing it. I believe in the company, and believe they'd make us (close to) whole should it fall through. Yes, it's essentially a corporate kickstarter, and I'm okay with that. I believe that as features get added they'll be rolled out to those who purchased FSD up front. I honestly don't expect FSD's implementation to be a light switch, but more of a gradual rollout of features that gets us there. Maybe it's a $3k gamble, but I've spend far more on far dumber and riskier things in my lifetime. Perhaps not the best excuse, but it's one I'm okay with .

 

[Foxtrotter]

I'll get it. Elon has said that it would be released in stages so that suggests that parts will be available sooner.
I want to support Tesla in their move to autonomy and in general.
We know its hard, otherwise everyone would already have it. We know there are legal hurdles. So?

 

[Sudre]

I'll start with the laws thing. Many states like MO have a law that states the driver must have at least one hand on the wheel at all times. This law in MO will remain in place even after FSD is federally legal. I know how slow my state is at getting rid of old laws. Typically they usually do the opposite and try to resist all change.

Then there is the whole recognizing stop signs thing. You can tell the person who is not a real/good programmer. Facebook can pick my face (wife, friends, anyone's) out of a picture at almost any angle and distance even with sunglasses on. I don't think a supercomputer will have any issues picking out a stop sign with time to stop. Heck even I have run a few stop signs hidden by trees. I bet there is a much higher chance FSD would have stopped at that sign because it would have learned from the other 574546748 drivers that a stop sign was behind that tree.

If you want EAP to exit a highway you will have to purchase FSD. That is only my opinion but I think that's where that is going. Long ago Elon had said EAP was from on ramp to off ramp. That in my mind includes the off/on ramp to another highway.

 

[FalconFour]

I love how this thread basically consists of:
- it's so easy, just (do thing that's obvious like reading signs)
- it's not so easy, because science, brains, AI, humans, software.

-or-
- it's easy because Tesla
- it's not easy because regulations

Two polar opposite ends of the spectrum, occupying two different camps. And somehow, nowhere in the middle, is the reality of this.

I'm in the middle. I understand things are hard, but I also understand the vast amount of resources being dedicated to getting it right. Just think that every moment you're working at your job, there are thousands of people at work thinking about every shallow little thought you might come up with in 10 minutes of browsing this topic. Thinking about, and developing, ways around the hurdles in government and technology, ways to get things done.

If there's one thing I know for absolutely certain, it's that buying FSD with the car will not go to waste. Even if it's not immediately available, or perfect right away.

 

[Phrixotrichus]

This is simply not true and there's no way to do this. Instead your proof is in statistics. Same is true with FDA approved drugs, etc. Stats are behind a number of safety related laws and regulation.

Right ....now what I´m doing on a daily basis isn`t "true"...okaaaay.

Cool beans... but that doesn't stop smaller features from being activated only for people who purchased the FSD upgrade, like I've mentioned.

Last time I checked full self driving meant "full self driving" and not "maybe you get some extra features sometimes in the future...maybe"-driving.
Sorry but the amount of goodwill to accept that and pay upfront for an almost unkeepable promise is borderline dumb.
As someone else here called it, it`s a "corporate kickstarter".

We are talking about image recognition and the imageNet challenge has shown that computers can be better than humans at image recognition. We also use machine learning at my place of employment every single day for use in a highly regulated industry. It's way more complex than image recognition and yet we still get the job done with machine learning, simulation, and real world testing.

sigh...you`re still comparing apples and oranges. The cars aren`t supercomputers (yet). Their dataset is fixed and their capability to deduct and guess severly limited. You´re acting like every car was a massively powerful portable neural network. Well, they`re not and they`re dumb .....and even neural networks are still far away from being capable to actually perceive reality as we see it. You bringing up some image analysis challenges here is quite telling about your understanding of the complexity of the problems at hand......

For now, just go and test the newest AP versions of any manufacturer and see what happens when the road markings go missing or change color, or signs got some graffiti on it......then you´ll see how "smart" these cars really are as of now.....

We in factory control/implementation wouldn`t trust an ai to run a single pump at the moment without a human double checking it, and the complexity in a fixed system is laughable compared to the ever changing dynamic an automated car has to deal with.

 

[landis]

Two polar opposite ends of the spectrum, occupying two different camps.

Umm,
A bit of a gross generalization. ;-)
I read it is as an argument between FSD or bust v.s. various accommodations for various sorts of a phased reality.

And somehow, nowhere in the middle, is the reality of this.

Those panels are hilarious and not unique to this thread. That is like a day in the life of TMC.

If there's one thing I know for absolutely certain, it's that buying FSD with the car will not go to waste. Even if it's not immediately available, or perfect right away.

Yeah, pretty much what I said a page or two ago...

Last time I checked full self driving meant "full self driving" and not "maybe you get some extra features sometimes in the future

Well sure, but not really.
Now that you mention it, I contend that this is the central flaw in your noisy argument.
There is absolutely no way that 'FSD' feature means nothing until it does everything.
It is a pretty fair bet that Tesla won't be asking you about what intermediate milestones are appropriate for actual delivery.

One little example discussed previously on this forum is using all of the cameras while EAP will be specifically limited to four.
So EAP+'FSD feature' could require current nags, but result in a more capable and confident experience while regulation jurisdictions fight (or not) over the new realities.

The cars aren`t supercomputers (yet).

By what definition?
Certainly not by the current standard, but you don't need to go back too many decades...
Forty times prior power, or whatever it is, qualifies in my book.