Elon: "Feature complete for full self driving this year"

[S4WRXTTCS]

They aren’t giving us full refunds, maybe the FSD price, but not the 100k. Not happening

I think that's what Knightshade meant.

I'm fully expecting that they'll either offer refunds for FSD or they'll offer some discount on a new Tesla where the FSD essentially transfers over.

I don't see any other course of action. They sell so many vehicles now that the cost of "trying to make things right" is cheaper than the consequences of the bad press.

Sure AP2 owners probably won't be happy with only $3K, but that's likely the magic number for minimizing bad press. They also have the whole battery gate fiasco to deal with.

 

[Knightshade]

They aren’t giving us full refunds, maybe the FSD price, but not the 100k. Not happening

Uh, full refunds for the FSD. Since that's the thing you wouldn't be getting.

You GOT the car obviously.

 

[daniel]

Oh please, a few months of musk time is years..... This has to be the best ponzie FSD scheme ever....

Nope. Not a Ponzi scheme. A Ponzi scheme is when investors are paid with the money from later investors. The first people in get rich. All the rest lose their money.

What Tesla is doing is not that at all. Tesla has been selling promises to deliver a technology they did not yet have, but which they expected to be able to deliver at a later time. It's a lot like a Go-Fund-Me campaign: Give us money to develop this thing and when we succeed you'll get one at a reduced price. The problem was the claim/promise that the car already had the needed hardware and it would just need an OTA update; and then downgrading the promise from "sleep in your car" or "robotaxi" to "a very very tiny chance it can get you from home to work without your intervention."

 

[DigAdrenaline]

I do feel for people who paid good money trusting Tesla when they said the cars had all the necessary hardware, and now they find themselves with cars that may or may not get the hardware.

Im one. HW2.5/MCU1 bought FSD from new in 2017. Best car I’ve ever owned, worth every cent. I want upgrades, but I also don’t want to lose sight of the cars awesomeness amongst the whining/complaining/hoping for HW3/MCU2.

 

[banned-66611]

To date, Tesla vehicles have driven over 3 billion miles in Autopilot mode.

Still pushing the "but data!" nonsense. Basically an admission that they have made very little actual progress.

So labeling with video and all 8 camera simultaneously.

So they can do some image recognition and are working towards making it efficient enough to run on the car's computers... But have not solved basic issues like building a 3D map of the world which requires depth perception, rather difficult to do reliably with a single camera.

They also have not shown any progress on relating data between multiple video frames. You know the way the images of vehicles and road markings jump around and flicker on the display? They need to fix that.

 

[diplomat33]

Still pushing the "but data!" nonsense. Basically an admission that they have made very little actual progress.

So they can do some image recognition and are working towards making it efficient enough to run on the car's computers... But have not solved basic issues like building a 3D map of the world which requires depth perception, rather difficult to do reliably with a single camera.

They also have not shown any progress on relating data between multiple video frames. You know the way the images of vehicles and road markings jump around and flicker on the display? They need to fix that.

FYI, your quotes make it look like I said those things. I was quoting Elon. Just want to set the record straight.

And by the way, I know it is a dirty word around here but lidar would have easily helped Tesla build a reliable 3D map since lidar is an active sensor that is ideal for depth perception.

 

[diplomat33]

I would caution people, especially new owners. Elon's FSD update was basically "we're still working on the basic fundamentals and I am completely guessing we might have something that has an above zero chance of self-driving you from your house to your work in a few months."

So definitely take it with a big grain of salt. I do think we will see some nice improvements this year with AP3 but Elon is admitting they are still working on the basics. There will be misses and delays. Definitely, don't expect a robotaxi in a few months.

I do think it is very possible that Tesla will launch their ride sharing app before the end of 2020 since Elon says they will probably launch it before they have robotaxi FSD. My guess is that Tesla will want to start getting that revenue and AP on city streets will be deemed "good enough" this year that they will launch the ride sharing.

 

[golfcart]

I have EAP and FSD and I'd gladly take a full refund at this point. What a waste of money so far, I'm glad the previous owner paid for it.

Even if the option were available, I am not going to loan my car out for ride sharing which will rack up tons of miles and let people wear out my interior... I got a performance model because I love to drive it. I'd love to drive it even more with the cost of that (currently) useless feature back in my pocket.

I am supposed to get an email to schedule my HW3 upgrade in the next few weeks. Maybe I should bring it up with the manager at the service center... if feature complete means "some non-zero chance" of getting me from my house to work then I want a refund.

 

[willow_hiller]

something that has an above zero chance of self-driving you from your house to your work in a few months

To be fair, I went back and listened to the call and he said:

"Feature complete just means it has some chance of going from your home to work, lets say, with no interventions. It doesn't mean the features are working well, it means above zero chance." Emphasis mine.

So from that I'm inferring that he defines "working well" as "no interventions." I guess you could argue it wouldn't be "full self-driving" with a few interventions thrown in, but I would be pretty happy with a trip from home to work with 1-2 interventions while they work on refining the features.

 

[Bladerskb]

On the plus side it's so watered down now that future edits should be minor.

Lol I forgot to add the tail end of that quote from the Feb 19th 2019 Podcast about FSD feature complete coming before end of 2019:

"I am certain of that. That is not a question mark."

 

[schonelucht]

You don't seem clear on what a ponzie scheme actually is.

Or even how to spell one.

Most FSD revenue is listed as a liability on Teslas books, and that can't change until they actually deliver it.

Fit the shoe yourself. Madoff's client's money was also booked as a liability. The books don't matter in a Ponzi scheme, they are cooked anyway. What matters is where the money is : free for Tesla to spend.

 

[kkillebrew]

Fit the shoe yourself. Madoff's client's money was also booked as a liability. The books don't matter in a Ponzi scheme, they are cooked anyway. What matters is where the money is : free for Tesla to spend.

Dude you are the one with you shoes on the wrong feet here. You are comparing Tesla to Madoff? Were you a Madoff victim? You need to get off this forum.

Bitter!! Table for one...

 

[Knightshade]

Fit the shoe yourself. Madoff's client's money was also booked as a liability. The books don't matter in a Ponzi scheme, they are cooked anyway.

Except Tesla is a public company where the books are relatively open (Q10 for example) and Madoffs business was not.

But you don't seem super interested in the actual facts of why your ponzi analogy is nonsensical

 

[boonedocks]

I have EAP and FSD and I'd gladly take a full refund at this point. What a waste of money so far, I'm glad the previous owner paid for it.

Even if the option were available, I am not going to loan my car out for ride sharing which will rack up tons of miles and let people wear out my interior... I got a performance model because I love to drive it. I'd love to drive it even more with the cost of that (currently) useless feature back in my pocket.

I am supposed to get an email to schedule my HW3 upgrade in the next few weeks. Maybe I should bring it up with the manager at the service center... if feature complete means "some non-zero chance" of getting me from my house to work then I want a refund.

I wonder how that would go down. I sold my HW2.0 MCU1 car and I got the class action punishment checks and not the current owner after the fact. If they refund some significant amount of $$$$ for the FSD purchase price for cars ordered with FSD I wonder if the current owner or the original purchaser would get the $$$$.

 

[Knightshade]

I wonder how that would go down. I sold my HW2.0 MCU1 car and I got the class action punishment checks and not the current owner after the fact. If they refund some significant amount of $$$$ for the FSD purchase price for cars ordered with FSD I wonder if the current owner or the original purchaser would get the $$$$.

I'd expect original owner as that's who paid Tesla for the feature- which is what you already saw with the previous situation.

It'd only be AP2.0 FSD money though (so likely 3k for most people, and not THAT many people given it's only AP2.0 S/X buyers who also bought FSD)

 

[ChrML]

code-complete can even mean that no unit-testing is done, let alone integration

I really doubt any software team in the world writes all the software, and THEN write the unit-tests. That would make no sense, because you need to know the code you're testing, to write good and valuable tests, which means the only reasonable time to write the tests is the same time and the same person who wrote the unit.

Integration testing is usually at a much simpler level and limited coverage compared to the unit-tests. And can be done when reasonable.

System-level testing however mostly cannot be automatic and I can agree that they may have just started this. At this point the code should work, but has never been tried in real application with human factors involved. In standard TD-development your code often works surprisingly well, and it works more than it does not.

However for such a large project as self-driving, the system-level testing is only the beginning. There will be lots of things to address, which loops back to more development required (with unit-tests, integration tests and back up to new system-level tests).

 

[daniel]

I do think it is very possible that Tesla will launch their ride sharing app before the end of 2020 since Elon says they will probably launch it before they have robotaxi FSD. My guess is that Tesla will want to start getting that revenue and AP on city streets will be deemed "good enough" this year that they will launch the ride sharing.

So basically, the Tesla ride-sharing app will be Uber for Teslas only? If TRUE autonomy is not here yet, then a driver must be in the car. And the only difference from Uber will be that the cars are all Teslas and maybe there'll be a different formula for distribution of the revenue. But if you want to be a Uber driver you can do that now, without an app from Tesla.

"Good enough" AP on city streets means infrequent interventions. The car cannot drive anywhere without a driver.

To be fair, I went back and listened to the call and he said:

"Feature complete just means it has some chance of going from your home to work, lets say, with no interventions. It doesn't mean the features are working well, it means above zero chance." Emphasis mine.

So from that I'm inferring that he defines "working well" as "no interventions." I guess you could argue it wouldn't be "full self-driving" with a few interventions thrown in, but I would be pretty happy with a trip from home to work with 1-2 interventions while they work on refining the features.

As I interpret the statement, there would be a possibility that for some individuals, some of their trips would be completed without an intervention. But the driver is still needed and fully responsible. And the problem is that things come up far more quickly in the city than on the highway. So the driver must be able to react far more quickly when monitoring City NoA than highway AP. Also, signs on the highway are usually just speed limits, which the driver can react to if the car doesn't, and signs about available services, which the car does not need to recognize. In the city there are far more types of signs that the car must be able to recognize. Monitoring City NoA is going to require a much higher level of attention. And the only promise is that maybe once in a while you might complete a trip without an intervention. This is what Elon is calling "Feature Complete."

There is a saying: "Under-promise, over-deliver." If instead of promising robo-taxi capability, Musk had said, "This is going to be a difficult task. We're going to work hard on it. Give us X dollars now and IF we succeed you'll get it for half the price we'll charge after, and if we can't upgrade your car, we'll extend your option to the next Tesla you buy," he'd probably have still sold a lot of FSD packages, but nobody would be complaining, because everybody would have known they were betting on an outcome.

If someone promises to finish a job in two hours but they finish in one, you will be happy. If they promise a half an hour but take an hour, you're angry. Same outcome on the job, but in one case they did better than promised, and in the other, worse. Tesla made a very poor decision when they promised to do a job quickly when nobody could have known how long it would really take.

 

[diplomat33]

So basically, the Tesla ride-sharing app will be Uber for Teslas only? If TRUE autonomy is not here yet, then a driver must be in the car. And the only difference from Uber will be that the cars are all Teslas and maybe there'll be a different formula for distribution of the revenue. But if you want to be a Uber driver you can do that now, without an app from Tesla.

"Good enough" AP on city streets means infrequent interventions. The car cannot drive anywhere without a driver.

Yes, that is implied in "launching ride sharing before FSD is done". Tesla would basically be launching a Tesla-only Uber service and a driver would be required. But the big difference with Uber is that Tesla cars would be improving in their FSD abilities with updates over time. And if you believe Elon, at some point, our cars would become robotaxis with no driver. So the idea would be to start as a Tesla only Uber service with a driver to start monetizing Teslas in a new way sooner and then the service would morph into a robotaxi service at some point in the future when FSD is done. Obviously, that is all based on the huge assumption that our cars can become robotaxis at all. If the hardware is not good enough or if Tesla can't finish the software, then the robotaxi part won't happen.

 

[Knightshade]

. And the problem is that things come up far more quickly in the city than on the highway.

Do they? Speeds are typically much lower in a city.

And the only promise is that maybe once in a while you might complete a trip without an intervention. This is what Elon is calling "Feature Complete."

Well, yes, because it is.

The features all exist at that point. Even if they don't work well enough to never require intervention. He made the point previously it's probably at least a year or two from feature complete to "works well enough you can trust it" (1-2 years might well be optimistic, but point is he's been pretty clear feature complete doesn't mean "done")

 

[diplomat33]

To be fair, I went back and listened to the call and he said:

"Feature complete just means it has some chance of going from your home to work, lets say, with no interventions. It doesn't mean the features are working well, it means above zero chance." Emphasis mine.

So from that I'm inferring that he defines "working well" as "no interventions." I guess you could argue it wouldn't be "full self-driving" with a few interventions thrown in, but I would be pretty happy with a trip from home to work with 1-2 interventions while they work on refining the features.

The original transcript I posted said no interventions. I must have forgotten to write that part in my subsequent reply.

That is a key distinction indeed. Feature complete just means that the driving tasks like traffic lights, stop signs, intersections, etc will be coded in so that there is a non zero chance that the car could complete a given trip with no interventions. Of course, there are a lot of factors that will affect the number of interventions from the choice of route, weather, road conditions, traffic conditions etc... So Elon cannot promise no interventions at all. In some instances, on some routes, feature complete might be able to handle it with no interventions. And Tesla improves their software with regular updates so the number of interventions should decrease over time.