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

[boonedocks]

@diplomat33 remember that Elon himself declared FSD feature complete on the ARK podcast. Take that with a giant bolder of salt, but with the current climate at Tesla and the re-wording, yes dumbing down, of FSD I am actually reasonably concerned.

 

[BinaryField]

@diplomat33 remember that Elon himself declared FSD feature complete on the ARK podcast. Take that with a giant bolder of salt, but with the current climate at Tesla and the re-wording, yes dumbing down, of FSD I am actually reasonably concerned.

And how about a money argument? If Tesla truly believed that they had a real L4 solution, shouldn't they be charging more for it, not less, as seen in the recent pricing debacle? I guarantee that people would be clamoring for it and would be willing pay tens of thousands of dollars for the ability alone. Saying that Tesla has an L4 solution now or the reasonable future is saying that the company does not want to make money.

 

[mongo]

@diplomat33 remember that Elon himself declared FSD feature complete on the ARK podcast. Take that with a giant bolder of salt, but with the current climate at Tesla and the re-wording, yes dumbing down, of FSD I am actually reasonably concerned.

NO HE DIDN'T.
He said he expects it to be feature complete this year with driver supervision.
big difference
HUGE
9:50
On the Road to Full Autonomy With Elon Musk — FYI Podcast

 

[mongo]

And how about a money argument? If Tesla truly believed that they had a real L4 solution, shouldn't they be charging more for it, not less, as seen in the recent pricing debacle? I guarantee that people would be clamoring for it and would be willing pay tens of thousands of dollars for the ability alone. Saying that Tesla has an L4 solution now or the reasonable future is saying that the company does not want to make money.

Tesla's mission is not making the most money. It is accelerating the transition to sustainable transportation. Making the feature affordable increases demand for Teslas. Mission is also making the safest cars, if FSD is safer than humans, then the mission is better supported by have it be affordable enough for people to buy it.

 

[BAven8068]

It’s hard for me to wrap my head around people not seeing what is happening with the watering down of FSD. I was a full believer in the early claims that they would achieve real FSD with the current set ups. After the restructuring and pricing of the AP/FSD packages and then Elon saying that FSD will be feature complete this year, I think its clear where this is all going. I mean they just made the feature of manually changing lanes with your turn signal while in Autosteer part of the “FSD” package. How is that part of that package when its been on AP1 cars? Their definitions of of AP and FSD are changing and there is a reason for that... they overstepped on their FSD claims and actually had people fork over large amounts of money based on those claims.

It’s my opinion by seeing these events unfold that the only thing we will see added to FSD is a version of Navigate on Autopilot for city streets that will make an attempt to recognize stoplights. It will never achieve reliability to the point the steering wheel nag and driver attention can be relaxed from its current state. We will then be left with just a slightly more advanced driver assistant package thats marketed as FSD and then the blame will always be reverted back to regulations.

 

[tinm]

Ugh, 23 pages to slog through. I scanned a bunch of them and notice someone did bring up that "feature complete" has a particular meaning in software development. Some thoughts on that:

Elon's being sneaky using the term "feature complete." When I was CEO of a tech company and I'd ask the engineers "when will it be ready" they would say it'll be "feature-complete" on date X, some time way out in the future. I knew what trick they were pulling: it didn't mean you could ship on date X. It meant it could be handed over to QA on date X so they could start rigorous testing and bug-reporting. Then there would be a period where the engineers would have to evaluate the bug list, prioritize it, we'd all agree on what bugs had to be fixed before shipping and which ones were lower priority, they'd go fix those important bugs, QA would re-test, and eventually we'd ship. Some time after date X. Maybe weeks, months, depended on the project.

But here's the thing. EAP/FSD tech is no ordinary software. it's not a dumb website, or some app. It is like AI. It has a voracious deep-learning appetite. Hence all the talk about billions of miles of fleet driving, making the thing so robust. That's deep machine learning. It's gotta have that to learn.

Arguably, a human baby is "feature complete" when it's born. Its brain has all of basic wiring in place. But has it learned yet? No. Takes years for the neurons and synapses and pathways to figure stuff out through day-to-day, real-world experience. So it seems kinda misleading for Elon to trot out the phrase "feature complete" because with FSD that's basically like saying the baby will have its brain ready by end of this year.

Do you want to turn your driving over to a digital baby? How about a digital two-year-old?

Finally, I would argue that regulatory approval is a feature, a critical-path feature, and that FSD isn't really feature complete until it has regulatory approval.

 

[diplomat33]

FSD was once a future thing. I bought into that and invested in that with my order.
Now the definition of FSD has changed to be a present thing: "you already have it! it's called NoA, Summon and AutoPark! More stuff coming soon!"

How can you possibly defend Tesla in this case?

Give me a break.

I am not defending Tesla in everything. But you don't understand what the website is saying if you think it is saying that FSD is just NOA, Summon and Auto Park now and we already have FSD. That is not what it is saying. Sorry.

Perhaps because there is actually some merit to them?

I would like to know what your background is that justifies your unlimited defense of Tesla. Are you involved in safety-of-life engineering? Do you have insider knowledge of some type? The technical users of this forum have the exact opposite assessment of the Tesla AP program from you. How do you explain this discrepancy?

"unlimited defense of Tesla". HA HA HA HA. That is hilarious. I have been critical of Tesla many times. I am merely expressing my cautious optimism when it comes to FSD. And yes, "cautious optimism" because I have said many times that I think L4 autonomy is not going to happen for some time. But apparently, even that position is not good enough. The only acceptable position allowed on this forum is that Tesla will never achieve FSD. Heck, we are not even allowed to be happy when EAP improves. We are supposed to say that EAP still sucks, and is still worse than AP1, right?

No, I don't have any technical expertise that relates to self-driving. I am merely expressing my opinion based on what I have read and based on my own experience with EAP everyday. I have the right to express my opinion. The critics are merely expressing their opinion as well. And none of them work for Tesla. So they can't know for sure what Tesla's progress is or whether Tesla's method will fail. They are merely expressing their opinion based on the public information that we all have.

But I have been fair from the start that Tesla will most likely need more hardware to get to L4 autonomy, so I am not disagreeing with the experts on that point. I am merely expressing optimism that Tesla can achieve this first version of "FSD with driver supervision" on the AP3 hardware. That is not a crazy position to take IMO.

 

[john85775]

I am not defending Tesla in everything. But you don't understand what the website is saying if you think it is saying that FSD is just NOA, Summon and Auto Park now and we already have FSD. That is not what it is saying. Sorry.

Features previously scoped into "Enhanced Auto Pilot" are now part of the "Full Self Driving" package.

If that isn't moving the goal posts, I don't know what is.

 

[mongo]

Elon's being sneaky using the term "feature complete." When I was CEO of a tech company and I'd ask the engineers "when will it be ready" they would say it'll be "feature-complete" on date X, some time way out in the future. I knew what trick they were pulling: it didn't mean you could ship on date X. It meant it could be handed over to QA on date X so they could start rigorous testing and bug-reporting. Then there would be a period where the engineers would have to evaluate the bug list, prioritize it, we'd all agree on what bugs had to be fixed before shipping and which ones were lower priority, they'd go fix those important bugs, QA would re-test, and eventually we'd ship. Some time after date X. Maybe weeks, months, depended on the project.

Not sneaky at all, in the ARK podcast he is very explicit that the system will NOT be hands off, that it will take time/ miles/ and validation to achieve the end goal.

 

[mongo]

Features previously scoped into "Enhanced Auto Pilot" are now part of the "Full Self Driving" package.

If that isn't moving the goal posts, I don't know what is.

What it is is altering the option packages to better differentiate them and make the adaptive cruise type feature more affordable/ cost competitive.
Those who had purchased EAP previously are still going to get those features.

 

[john85775]

What it is is altering the option packages to better differentiate them and make the adaptive cruise type feature more affordable/ cost competitive.
Those who had purchased EAP previously are still going to get those features.

AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING

(If it was, my 2013 Ford Fusion had "full self driving" features on it)

 

[mongo]

AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING
AUTOPARK ISN'T FULL SELF DRIVING

(If it was, my 2013 Ford Fusion had "full self driving" features on it)

Even if it does it without you in the car or monitoring it?
That is part of the option differentiation, otherwise EAP would just be FSD with nags...

 

[tinm]

Not sneaky at all, in the ARK podcast he is very explicit that the system will NOT be hands off, that it will take time/ miles/ and validation to achieve the end goal.

So "Full" self driving won't quite be "full" when reaches the point where they'll start calling it "full". It'll be "FSQWSHIS" (full self driving with some human involvement still).

For me "Full" means car knows how to drive anywhere anyhow, with driver asleep or no driver at all.

 

[BAven8068]

Nothing in the FSD package is FSD right now that’s the problem.

I was fine with the old FSD package being basically nothing with a promise of real FSD features “coming soon”, even though I thought it was crazy to spend money on it when there were no benefits at the time of purchase.

Now they have shifted EAP features over to FSD package and are still calling it FSD. If this isn’t lowering expectations I don’t know what is.

 

[mongo]

So "Full" self driving won't quite be "full" when reaches the point where they'll start calling it "full". It'll be "FSQWSHIS" (full self driving with some human involvement still).

For me "Full" means car knows how to drive anywhere anyhow, with driver asleep or no driver at all.

Nothing I the FSD package is FSD right now that’s the problem.

I was fine with the old FSD package being basically nothing with a promise of real FSD features “coming soon”, even though I thought it was crazy to spend money on it when there were no benefits at the time of purchase.

Now they have shifted EAP features over to FSD package and are still calling it FSD. If this isn’t lower expectations I don’t know what is.

Right, at this point in time, nothing in FSD is Fully Self. However, just like the old description of FSD was of the future capability, so is the current one. And like before, the features will improve via OTA over time (and via HW3 swap). So, down the road, theoretically, the FSD option package will live up to its name.

Take NoA, if you take away the confirmation clicks, then isn't that, on ramp to off ramp, a FSD function (when nags are removed)? EAP is now just adaptive cruise control, which seems like a good delineation from FSD. It, EAP, is a fixed function set that does will not expand into the FSD space.

 

[john85775]

Nothing I the FSD package is FSD right now that’s the problem.

I was fine with the old FSD package being basically nothing with a promise of real FSD features “coming soon”, even though I thought it was crazy to spend money on it when there were no benefits at the time of purchase.

Now they have shifted EAP features over to FSD package and are still calling it FSD. If this isn’t lower expectations I don’t know what is.

+++++++

Exactly this.

 

[Kant.Ing]

In my firm belief, FSD is what the famous 2016 video is. After 100 years, when human look into their history and ask the first ever FSD achievement, it's gonna be Tesla and in 2016! /s

 

[BinaryField]

"unlimited defense of Tesla". HA HA HA HA. That is hilarious.

Indeed it is.

I have been critical of Tesla many times. I am merely expressing my cautious optimism when it comes to FSD. And yes, "cautious optimism" because I have said many times that I think L4 autonomy is not going to happen for some time. But apparently, even that position is not good enough. The only acceptable position allowed on this forum is that Tesla will never achieve FSD. Heck, we are not even allowed to be happy when EAP improves. We are supposed to say that EAP still sucks, and is still worse than AP1, right?

Seeing your posts in other threads, your comments have been very far from what I'd consider "cautious". I'll also add that what you have said is a very clear mis-representation of what has been going on with EAP development, which has been legitimately controversial.

No, I don't have any technical expertise that relates to self-driving. I am merely expressing my opinion based on what I have read and based on my own experience with EAP everyday. I have the right to express my opinion. The critics are merely expressing their opinion as well. And none of them work for Tesla. So they can't know for sure what Tesla's progress is or whether Tesla's method will fail. They are merely expressing their opinion based on the public information that we all have.

To say that people are "merely expressing their opinion" is disingenuous. It should be clear that an expert opinion, whether or not they work for Tesla, is worth substantially more than what you or I can come up with (though I actually do work in safety-of-life systems, just not autonomous cars). And some of them have much more than "public information" on the matter, e.g. the ability to get into the hardware / software. I also feel the need to point out that forming opinions based on reading summaries or pop culture articles without the requisite technical foundation does not increase the value either.

 

[Bladerskb]

But the bottom line is that I take issue with the idea that Tesla won't achieve FSD. Tesla is nearing "feature complete" FSD that works in development mode. And Tesla is improving the hardware with AP3. Tesla is close to "FSD with driver supervision". Of course, that is a long way to L4. I am not naive on this. Driver supervision will be required at first but Tesla will continue to improve it until it does get to L4. Again, we can whine that Tesla is not at L4 yet like Musk promised years ago but I think Tesla is in a good place now in their FSD development.

And the idea that Tesla will release a FSD that will kill people is sensationalism. Heck, the reason that Tesla has delayed releasing some features is precisely because they are not going to release features that could be unsafe. For example, Tesla released NOA with driver confirmation for that very reason. I have no doubt that when Tesla does release FSD that will be safe and the driver confirmation will just be a safety precaution.

I am glad to hear that you want Tesla to succeed. And yes, I do agree with you that Musk should be a more responsible with his promises and tweets. But I have to disagree a bit that Tesla is redefining FSD just so that they can pretend like they are keeping their promises. The FSD on the website will be feature complete and will able to self-drive on city streets with enough confidence to release to the public. So Tesla is not releasing half-FSD and calling it a day. Tesla is just planning to release FSD with driver supervision as a precaution and as an intermediary step on the path to L4. It is not Tesla's final FSD. As you well know, being a long time Tesla owner, Tesla uses OTA updates to gradually improve their products. Tesla will surely improve FSD over many years to come.

 

[mongo]

To say that people are "merely expressing their opinion" is disingenuous. It should be clear that an expert opinion, whether or not they work for Tesla, is worth substantially more than what you or I can come up with (though I actually do work in safety-of-life systems, just not autonomous cars). And some of them have much more than "public information" on the matter, e.g. the ability to get into the hardware / software. I also feel the need to point out that forming opinions based on reading summaries or pop culture articles without the requisite technical foundation does not increase the value either.

You say expert opinion, I say who predicted the single seat semi, or that Roadster 2020 was ready for reveal?
Even an expert can't do anything without accurate reliable data.