Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autopilot and the meaning of the word 'never'

This site may earn commission on affiliate links.
Firmware 7.0

Like I said, apparently we use the word "never" differently. Also you seem to be misunderstanding the premise that started this whole discussion. It was about a theoretical "never" not something coming from Tesla.

'Never' doesn't mean Tesla must create an unsafe situation by instantly turning off autosteer. You never answered the question on why you keep insisting Tesla is forced to instantly disable autosteer instead of simply warning the driver has driven onto an unsupported road and autosteer will be turning off in a few seconds. Either way this is pointless so continue on.
 
Last edited:
Like I said, apparently we use the word "never" differently. Also you seem to be misunderstanding the premise that started this whole discussion. It was about a theoretical "never" not something coming from Tesla.

'Never' doesn't mean Tesla must create an unsafe situation by instantly turning off autosteer. You never answer the question on why you keep insisting Tesla is forced to instantly disable autosteer instead of simply warning the driver has driven onto an unsupported road and autosteer will be turning off in a few seconds. Either way this is pointless so continue on.

I'm not trying to start things up again, but rather hoping to point out where I think the disconnect between you two is. It is a disconnect I am familiar with, as I've seen it come up in discussions I've had.

I really think this all boils down to the fact that brianman, as he tried to point out in the post I quoted above, has based the entire discussion on the theoretical premise that started the discussion, and that theoretical premise included the "never." brianman is not suggesting that in the real world--if software were to be put into functioning cars--Tesla couldn't or wouldn't do things differently. I expect he probably thinks that they would. But in this particular discussion, brianman is keeping his arguments consistent with the original theoretical limit of never."

dsm363, on the other hand, is taking a more practical approach, and looking at things the way Tesla might actually implement them. That's fine. In reality he is certainly correct (and as I wrote, I expect Brian would agree) that if the software were enabled in cars, it would not function in the discussion's theoretical "never" way.

The disconnect is that those are two different discussions. brianman is sticking to the original premise, and the "theoretical never", and dsm363 is discussing a more practical, real-life implementation.

I really think that is all your disagreement came down to.
 
I'm not trying to start things up again, but rather hoping to point out where I think the disconnect between you two is. It is a disconnect I am familiar with, as I've seen it come up in discussions I've had.

I really think this all boils down to the fact that brianman, as he tried to point out in the post I quoted above, has based the entire discussion on the theoretical premise that started the discussion, and that theoretical premise included the "never." brianman is not suggesting that in the real world--if software were to be put into functioning cars--Tesla couldn't or wouldn't do things differently. I expect he probably thinks that they would. But in this particular discussion, brianman is keeping his arguments consistent with the original theoretical limit of never."

dsm363, on the other hand, is taking a more practical approach, and looking at things the way Tesla might actually implement them. That's fine. In reality he is certainly correct (and as I wrote, I expect Brian would agree) that if the software were enabled in cars, it would not function in the discussion's theoretical "never" way.

The disconnect is that those are two different discussions. brianman is sticking to the original premise, and the "theoretical never", and dsm363 is discussing a more practical, real-life implementation.

I really think that is all your disagreement came down to.
Pretty close. Another point from that...
If your characterization of dsm's position is correct, then that would be in agreement with "the original assumption of <<'never' auto-steer on two-lane>> is not viable" -- which is the point I was trying to make in the first place.
 
Firmware 7.0

Pretty close. Another point from that...
If your characterization of dsm's position is correct, then that would be in agreement with "the original assumption of is not viable" -- which is the point I was trying to make in the first place.

That's about as literal as you can take something. 'Never' could mean Tesla doesn't allow you to start or continue autosteer on a two lane road. Show me where this logic dictates they 'instantly' cut off the feature. I would think a few seconds of warning would be appropriate and not against the law. I've repeated this a few times now. Just re-read this a few times or ask someone else to explain it.
 
Firmware 7.0

Sorry, I take the word "never" literally. I don't plan to change that. Feel free to be cognizant of that in future posts from me.

One more try. And Tesla could still enforce this rule. Just not the 'instant' part of it. That's the part you seem to be struggling with. It would be beyond comprehension for Tesla to instantly disable autosteer when the car still knew what was going on just because it crossed into a two lane road and just realized it. Yet again, they could 'almost never' allow it. There we go. Think we have a winner. I added the word almost to get around this black and white interpretation of the word never.
 
Last edited:
One more try. And Tesla could still enforce this rule. Just not the 'instant' part of it. That's the part you seem to be struggling with. It would be beyond comprehension for Tesla to instantly disable autosteer when the car still knew what was going on just because it crossed into a two lane road and just realized it. Yet again, they could 'almost never' allow it. There we go. Think we have a winner. I added the word almost to get around this black and white interpretation of the word never.
"Almost pregnant" comes to mind.

If/when you change it to "almost never" from "never" to explain your posts, then please do the mental step of considering all of your posts that align with that to be on a completely different topic to the one that I was speaking to. This is apparently why you were not understanding my posts -- you were applying a fudge factor to the premise and then not understanding why you came to different conclusions.

Here's an easy example:
Assertion: No customer-delivered Model S P85D can do 0-60 in under 1.0 seconds.
Evaluation: True.

Now let's fudge it:
Assertion: No customer-delivered Model S P85D can do 0-60 in under 1.0 seconds unless you fudge for 1000ft rollout.
Evaluation: Perhaps false, but kind of ridiculous.


The discussion we were having was about "never" and now you're trying to say "oh, really I meant almost never and so my responses totally make sense". No, they don't. Sorry.


"Never" means "never" not "almost never".


If this is so difficult to see eye-to-eye on, then it totally make sense why these discussion go off the rails constantly. Say what you mean without fudge factor.
 
Firmware 7.0

Dude. Seriously. It's your demand that it instantly shut off which makes no sense. I understand the word never. There is no law that says Tesla couldn't give someone a few seconds to take over once it cross onto a two lane road. Tesla is the one in charge so can do whatever they want. If 'never' means never starting or allowing autosteer to continue on a two lane road but once your cross onto one it forces you to take over then yes. That's my point. Might be difficult to understand though. I added the word almost to try and get past the binary thinking but didn't work.

Address this point.
You must not have fully understood this post as the word almost caused you to misunderstand the main issue.
One more try. And Tesla could still enforce this rule. Just not the 'instant' part of it. That's the part you seem to be struggling with. It would be beyond comprehension for Tesla to instantly disable autosteer when the car still knew what was going on just because it crossed into a two lane road and just realized it.
 
Last edited:
Dude. Seriously. It's your demand that it instantly shut off which makes no sense. I understand the word never. There is no law that says Tesla couldn't give someone a few seconds to take over once it cross onto a two lane road. Tesla is the one in charge so can do whatever they want. If 'never' means never starting or allowing autosteer to continue on a two lane road but once your cross onto one it forces you to take over then yes. That's my point. Might be difficult to understand though. I added the word almost to try and get past the binary thinking but didn't work.

Address this point.
You must not have fully understood this post as the word almost caused you to misunderstand the main issue.
Andy, Flasher, Lola - Go for it. I've lost the ability to get my point across to dsm apparently.
 
Firmware 7.0

Andy, Flasher, Lola - Go for it. I've lost the ability to get my point across to dsm apparently.

Ok. Maybe they can explain why Tesla would be forced to instantly disable autosteer on two lane roads. That's the logic that I don't understand. I've asked you this question multiple times now and you jump on the definition of the word 'never'.
 
Must be a higher form of logic I am simply missing.
Counterproductive and rude. You might want to move yourself to snippiness.

- - - Updated - - -

Ok. Maybe they can explain why Tesla would be forced to instantly disable autosteer on two lane roads. That's the logic that I don't understand. I've asked you this question multiple times now and you jump on the definition of the word 'never'.
Firmware 7.0 - Page 163
 
Andy, Flasher, Lola - Go for it. I've lost the ability to get my point across to dsm apparently.

I like a challenge. I'll take a shot...again. (I actually thought I had done a pretty good job this morning, though.)

In a brand new conversation, starting now, that has nothing to do with the conversation that has taken place thus far, we will discuss the best way for Tesla to implement this software if their goal is to generally, in most cases, to not have the Auto Steer operating on two lane roads. THIS IS A NEW, DIFFERENT CONVERSATION. I believe that Brian will agree that if this is the stated goal--Tesla wants Auto Steer operating as little as possible on two lane roads--that the software could issue a warning when it realized it was on a two lane road, and continue steering the car, allowing the driver time to take over, provided the sensors still had the ability to see lane markings, etc. This is consistent with what dsm363 was suggesting should take place IN THE OTHER CONVERSATION.

So on this point, in this new conversation, you two can agree.

Now here is where it gets tricky.

Going back to the other conversation, Brian was discussing a theoretical situation in which Tesla never, (and never really does mean not ever) allowed for Auto Steer to operate on 2 lane highways. He was not advocating for this. He was not suggesting firmware be written to do this. The conversation was theoretical. But in this theoretical discussion, never means never, so Auto Steer would not operate on a 2 lane highway.

The two conversations--the two premises--are different.

Does this work?
 
I like a challenge. I'll take a shot...again. (I actually thought I had done a pretty good job this morning, though.)

In a brand new conversation, starting now, that has nothing to do with the conversation that has taken place thus far, we will discuss the best way for Tesla to implement this software if their goal is to generally, in most cases, to not have the Auto Steer operating on two lane roads. THIS IS A NEW, DIFFERENT CONVERSATION. I believe that Brian will agree that if this is the stated goal--Tesla wants Auto Steer operating as little as possible on two lane roads--that the software could issue a warning when it realized it was on a two lane road, and continue steering the car, allowing the driver time to take over, provided the sensors still had the ability to see lane markings, etc. This is consistent with what dsm363 was suggesting should take place IN THE OTHER CONVERSATION.

So on this point, in this new conversation, you two can agree.

Now here is where it gets tricky.

Going back to the other conversation, Brian was discussing a theoretical situation in which Tesla never, (and never really does mean not ever) allowed for Auto Steer to operate on 2 lane highways. He was not advocating for this. He was not suggesting firmware be written to do this. The conversation was theoretical. But in this theoretical discussion, never means never, so Auto Steer would not operate on a 2 lane highway.

The two conversations--the two premises--are different.

Does this work?

This makes perfect sense. The problem with the theoretical situation with NEVER on 2 lane highways is that it's basically impossible to implement. So it's an assumption that's not well suited to base a conversation on. Here's why:

The AI system will take a discrete amount of time to determine "I am on a 2 lane highway now". So even if (once it makes that determination) immediately turns of auto steer (and oh btw - there is no IMMEDIATELY in software because instructions need to be executed and that takes TIME), that still means that auto steer has operated on that 2 lane highway. Depending on the quality of visual / radar data maybe just for a very short time - but not NEVER.

That's the problem with such absolute statements. The only real way to implement this is to do it the way I do it. My P85D at this point NEVER uses auto steer on 2 lane highways. That's because I'm still on fw v6.2.
 
This makes perfect sense.

Thanks. I tried.



The problem with the theoretical situation with NEVER on 2 lane highways is that it's basically impossible to implement.


Ahhh, but as Brian pointed out in response to my first attempt to clear things up, that was really the point he was trying to make in the first place:



Another point from that...
If your characterization of dsm's position is correct, then that would be in agreement with "the original assumption of <<'never' auto-steer on two-lane>> is not viable" -- which is the point I was trying to make in the first place.