Elon Confirms Those Who Purchased FSD get Computer Upgrade Free

[pscarroll]

Sure, it is basically true that none of the FSD features have been delivered, but I'm not sure why you are upset by it so much. I think it was pretty clear last year that it was going to be at least a year before anything meaningful was delivered (I think Elon was saying 6 months or so, with plenty of caveats). So are you surprised? I feel like there wasn't really any lying in this case. Just optimistic timelines from Tesla, which most reasonable people agreed at the time would be delayed. The reason people put up the money was to lock in the price, as far as I can tell. I chose not to buy it because I figured it would be years before it would be useful AND something I would be comfortable using. This was a very common assessment. I'm sorry if you were out of the "Tesla knowledge loop" (or whatever you want to call it) at the time of purchase.
The reclassification of features for AP/FSD was weird, but it doesn't really change anything for you, does it?



Also confused by this (what never happened?). The Twitter posts (there are two of them, you have to take them together) mentioned above has Jung Musk clearly stating that FSD purchasers will get HW3 in the future. So that is a "thing Elon recently said previous FSD purchasers would get".

You don't really have anything to worry about - I think it's pretty likely that in about a year you'll have HW3 with some basic level 2 features above and beyond the existing EAP features that you can play with (and monitor with 100% driver attention!), that non-FSD purchasers will not have. You may have longer to wait and need more hardware upgrades to get to true Level 3, but can't see why Tesla wouldn't follow through on that, if it is necessary. There is no timeline, though; it could be years.

I'm only really upset because I paid $1000 more than everyone else, I bought for $3000 when Tesla lied and said it would cost much more to but later, they won't refund the money and Elon said in Twitter that we'd get EAP (Early Access Program, not enhanced autopilot, Tesla uses EAP acronym for the former, see their blog if you think I'm lying) but he lied about that too. So why should we believe him when he says we'll get free HW3?

 

[AlanSubie4Life]

I wonder how many Oct 2016 leased cars were configured with FSD, and how that will play out at the end of this year if retrofits aren't rolling out soon.





He said "3 months, maybe, 6 months definitely" No caveats unless you define "definitely" differently than I do. I bought FSD in December 2017 because I thought even with Elon-time they HAVE to be getting close to the first feature right? And here we are over two years from that infamous statement with nothing yet.



View attachment 392613

Fair enough. I did say "I think"! Anyway, I think that a significant number of people even at that time in 2017 thought that the reality was quite a bit different, in spite of the Elon pronouncements, which was my point. Even at that point Elon was known to routinely overpromise. You buy early to lock in the price if you want, but in general in life I try not to believe what people say; I have to base my decisions on what I think is technically reasonable/believable.

It looks like the correct scaling time for Elon pronouncements may be 4x. In the engineering world that's fairly standard. So I would expect about 4-8 years for true FSD, which seems fine. There will be lots of cool stuff to play with in the meantime, if you're into that.

 

[AlanSubie4Life]

I'm only really upset because I paid $1000 more than everyone else, I bought for $3000 when Tesla lied and said it would cost much more to but later, they won't refund the money and Elon said in Twitter that we'd get EAP (Early Access Program, not enhanced autopilot, Tesla uses EAP acronym for the former, see their blog if you think I'm lying) but he lied about that too. So why should we believe him when he says we'll get free HW3?

I see. Well, there wasn't a timeline for the EAP thing either, and I haven't been tracking how that promise evolved (after the first announcement). I'm confident that you'll eventually get into the EAP program if they haven't retracted that promise. Have you tried directly contacting them about it? Like I said, really have not been tracking this so no idea where things stand on this front.

I figured that FSD features would get cheaper over time (that's the way technology usually works), but agree that whatever statements they made about that were a bit sketchy and probably just being used to drive demand/up-front payments.

 

[Az_Rael]

There will be lots of cool stuff to play with in the meantime, if you're into that.

Oh, I am not expecting actual FSD and I never was. I just want ONE feature to play with that required buying the package to get. Something that is not released to EAP-only buyers. I figured for sure we would have at least one feature to play with 2.5 years in. That doesn't seem too crazy to think that.

Maybe that first feature will end up being HW3, who knows.

As far as Early Access goes, canned responses from Tesla when asked are that the program is currently at capacity and not accepting additional applications. We will be on the list to get an invite when a space opens up. At some mysterious time in the future.

My poll on the subject currently shows 72 with no invites and 4 with invites. Not really looking good for actually getting that invite. Have any early adopter FSD owners seen an EAP (early access) invite yet?

 

[AlanSubie4Life]

I just want ONE feature to play with that required buying the package to get.

Good things come to those who wait.

I figured for sure we would have at least one feature to play with 2.5 years in.

This anticipation will make the reality even sweeter.

We will be on the list to get an invite when a space opens up. At some mysterious time in the future.

Good things come to those who wait? Eventually a complaint will get through to Jung/Elon on Twitter, he will say "I will look into making this right", then the invitations will go out. That is my prediction. Everything moves at a glacial pace.

 

[rajivpsf]

so much whining... but then we are rewarded .. so please go ahead and do it for the rest of us. I do support you.

 

[SergeyUser]

It looks like the correct scaling time for Elon pronouncements may be 4x. In the engineering world that's fairly standard. So I would expect about 4-8 years for true FSD, which seems fine. There will be lots of cool stuff to play with in the meantime, if you're into that.

The question is - how difficult will be for EM himself to do the 4x scaling? This is what smart and responsible person would do, if he cares to send a message other people will understand.

 

[SammichLover]

I bought for $3000 when Tesla lied and said it would cost much more to but later,

You think they didn't intend to charge more? "Lied" implies this was planned. I seriously think you're giving them waaaay too much credit here on masterminding this. They needed the cash and the sales numbers for Q1, they looked at this as a way to drum up some cash and drive sales in the short term.

P.S. And they still are charging the "much more" now for those that missed the sale.

 

[AlanSubie4Life]

The question is - how difficult will be for EM himself to do the 4x scaling? This is what smart and responsible person would do, if he cares to send a message other people will understand.

Really hard. An optimistic inherently is terribly limited in this regard.

 

[azred]

I'm only really upset because I paid $1000 more than everyone else, I bought for $3000 when Tesla lied and said it would cost much more to but later, they won't refund the money and Elon said in Twitter that we'd get EAP (Early Access Program, not enhanced autopilot, Tesla uses EAP acronym for the former, see their blog if you think I'm lying) but he lied about that too. So why should we believe him when he says we'll get free HW3?

Can’t you stop with the liar crap? I tried to gently prod you in that direction with humor and you told me to stuff it. You are being a jerk.

 

[Daniel in SD]

I'm one of the biggest skeptics on this forum but even I believe that if HW3 provides a "meaningful advantage" then Tesla will upgrade all those that purchased FSD.
I don't believe HW3 will provide a meaningful advantage any time in the near future. Other companies have limitless hardware resources in their autonomous vehicles and no one thinks they're very close to a marketable product.

 

[SammichLover]

Other companies have limitless hardware resources in their autonomous vehicles and no one thinks they're very close to a marketable product.

In which way? The Google, Alphabet company Waymo has a lot of people convinced they are close to proper commercial rollout. They look "reasonably close" to me, too, last time I looked.

Of course they, along with pretty much everyone else's "Level 4/5" system is aimed at a very different market than what Tesla is doing. They are roughly $1/4 million/vehicle systems, so commercial driver replacement systems rather than personal use add-ons to private vehicles.

I don't believe HW3 will provide a meaningful advantage any time in the near future.

Similar question here, what do you mean by "meaningful advantage"? To what in what way?

 

[Daniel in SD]

In which way? The Google, Alphabet company Waymo has a lot of people convinced they are close to proper commercial rollout. They look "reasonably close" to me, too, last time I looked.

Of course they, along with pretty much everyone else's "Level 4/5" system is aimed at a very different market than what Tesla is doing. They are roughly $1/4 million/vehicle systems, so commercial driver replacement systems rather than personal use add-ons to private vehicles.

Waymo is definitely the closest. It seems like the only reason it's a different market is that their system is so expensive. I think a level 4/5 system could easily get buyers as a $30k option on private vehicles so I would expect the technology to trickle down if they can reduce the costs. Most technology is initially very expensive and then becomes cheaper over time. Tesla themselves didn't develop the Model 3 before the Model S and Roadster. That's what makes Tesla's strategy for autonomy so implausible to me.
Is Waymo providing timelines as to when they think they'll deploy without test drivers? For some reason Elon Musk is, even though there's no evidence that they're even remotely close. Arguably "feature complete" autonomous vehicles have been around for a decade and that's where he thinks Tesla will be by the end of this year.

Similar question here, what do you mean by "meaningful advantage"? To what in what way?

You tell me. I don't think stoplight recognition is a meaningful advantage except as a safety feature (providing emergency warning and/or braking). I think that level 2 driving in the city is dangerous and will be banned if Tesla decides to release it. To me a meaningful advantage is a level 3-5 system. A level 3 system that works on the highway at limited speeds would be a great first step.

 

[SammichLover]

Response in two parts, 1 of 2;

I don't think stoplight recognition is a meaningful advantage except as a safety feature (providing emergency warning and/or braking).

Well we definitely disagree here. Although maybe because I use EAP most of the time and on most any road that has marked lanes, and one very regular road has a number of stoplights that I pass and relatively light traffic? Right now my #1 "pain point" with EAP is having to manually get out of it when I'm first to the stoplight, and on this road being first to the stoplight happens quite often as my schedule is such that I don't drive during "rush hour".

I'll take $2K as a "steal of the century" if they get reliable stoplight identification in an actionable way. Just on that one road, even, where it is about as clean a case as there is. I will be entirely unsurprised if they ultimately need a meaningful processing power boost to pull it off, AKA HW3, given it is a pretty challenging problem domain to get something to a public roll out level of reliability.

This assessment is related to the other part.....

 

[SammichLover]

Waymo is definitely the closest. It seems like the only reason it's a different market is that their system is so expensive.

It is actually a different market today, in substance, exactly because of price AND what they are trying to accomplish today. Literally different customers in so many ways. Even at $30K it would be a very different set of customers. They can't going head to head because nobody yet can do both Tesla price and universality AND Waymo capability in the specific target locales, but that the markets are different because they are choosing different priority goals and different paths to get to the ultimate goal doesn't change that the markets are very, very different.

Now where they aim for the more distant future seems far more similar. Maybe I missed it but haven't really seen Waymo talk about going downscale, where Tesla is now, as part of the business plan but Tesla is explicitly is taking aim at the commercial service provider market, even if they are envisioning it as a much larger, "anyone with a car can get into business, some just as a side-business". If Tesla gets there where Musk ultimately envisions, and it is an 'if', it is going to blow Waymo et al up unless that approach's hardware costs have fallen to maybe $1000 cost to manufacture AND they are able to whitelist with their spacial database effectively the whole target country where they're selling.

That's the race in a nutshell. Tesla's software to enable optical based adhoc driving vs cost and data collection for pre-scanned database based driving using LIDAR heavy sensors.

 

[Daniel in SD]

Response in two parts, 1 of 2;


Well we definitely disagree here. Although maybe because I use EAP most of the time and on most any road that has marked lanes, and one very regular road has a number of stoplights that I pass and relatively light traffic? Right now my #1 "pain point" with EAP is having to manually get out of it when I'm first to the stoplight, and on this road being first to the stoplight happens quite often as my schedule is such that I don't drive during "rush hour".

I'll take $2K as a "steal of the century" if they get reliable stoplight identification in an actionable way. Just on that one road, even, where it is about as clean a case as there is. I will be entirely unsurprised if they ultimately need a meaningful processing power boost to pull it off, AKA HW3, given it is a pretty challenging problem domain to get something to a public roll out level of reliability.

This assessment is related to the other part.....

Yeah, I believe stoplight recognition the way you're describing it is extremely dangerous. How much time will you have to notice that autopilot has not seen the stoplight? How will it not inevitably result in an accident where a Tesla runs a stoplight and kills someone? How will that be any different from the Uber incident where a pedestrian was killed?

It is actually a different market today, in substance, exactly because of price AND what they are trying to accomplish today. Literally different customers in so many ways. Even at $30K it would be a very different set of customers. They can't going head to head because nobody yet can do both Tesla price and universality AND Waymo capability in the specific target locales, but that the markets are different because they are choosing different priority goals and different paths to get to the ultimate goal doesn't change that the markets are very, very different.

Now where they aim for the more distant future seems far more similar. Maybe I missed it but haven't really seen Waymo talk about going downscale, where Tesla is now, as part of the business plan but Tesla is explicitly is taking aim at the commercial service provider market, even if they are envisioning it as a much larger, "anyone with a car can get into business, some just as a side-business". If Tesla gets there where Musk ultimately envisions, and it is an 'if', it is going to blow Waymo et al up unless that approach's hardware costs have fallen to maybe $1000 cost to manufacture AND they are able to whitelist with their spacial database effectively the whole target country where they're selling.

That's the race in a nutshell. Tesla's software to enable optical based adhoc driving vs cost and data collection for pre-scanned database based driving using LIDAR heavy sensors.

I don't understand this argument. You don't think that other companies are trying to achieve FSD today? What does targeting different markets have to do with anything? The system has to work just as well for the personal market as it does for the commercial market. It just doesn't make any sense to me to start by trying to develop a cheaper system. What technology has ever worked that way?

 

[SammichLover]

Yeah, I believe stoplight recognition the way you're describing it is extremely dangerous. How much time will you have to notice that autopilot has not seen the stoplight? How will it not inevitably result in an accident where a Tesla runs a stoplight and kills someone? How will that be any different from the Uber incident where a pedestrian was killed?

Well that's the rub of whether they can deploy or not, if they can get it reliable enough. Which is exactly why I expect the HW3 [at the least] will ultimately be required to plausibly get to that point.

You don't think that other companies are trying to achieve FSD today?

Not even within an order of magnitude of the price point at this point, not for private vehicles. That is integral to what FSD is. If you aren't talking the same price, you aren't talking the same customers and the same use, then you aren't talking about a comparable product.

Well Supercruise might be headed that way but they aren't talking about it like that. GM doesn't seem to intend to push into that, which makes since since they've got their own Level 4/5 development project and internal politics probably wouldn't allow cannibalizing like that.

 

[Leafdriver333]

I tried AP for the first time today.
While it was exciting, there were some nervous moments - ie where the 3 lanes become 2 and where the exit lane paint had a recent revision, etc.
It will be interesting to see what FSD will be actually capable of and how soon.

 

[Daniel in SD]

Not even within an order of magnitude of the price point at this point, not for private vehicles. That is integral to what FSD is. If you aren't talking the same price, you aren't talking the same customers and the same use, then you aren't talking about a comparable product.

Yeah that's my point. Tesla is trying to develop the same product at 1/100th the price and by all indications they are WAY behind. The system requirements are exactly the same whether you're selling it for $2k or $200k.

 

[SammichLover]

Yeah that's my point. Tesla is trying to develop the same product at 1/100th the price

Uhhhh. That is where they are heading some ways out. That isn't now.

....and by all indications they are WAY behind.

Ahhhhh. Well you'd conclude that if you were inclined bet be so wrong as too....

The system requirements are exactly the same whether you're selling it for $2k or $200k.

Ignore price as inherently part of the product.

Let me repeat that; Wrong.

Wrong.

Wrong.

They're coming from a different direction, and the gap between these two approaches is quite large right now, which makes it very difficult to judge which is "more ahead" on the way to the intersection point Musk is describing.

Sure it is easier if you just lop off one key aspect, cost. Plus another one on top of that, too, how come coverage the support road domains are which I've only mentioned in passing here and I haven't seen you mention at all. But that just makes the comparison less meaningful.