OFFICIAL BUTTON WATCH

[Dan D.]

Button timing of May is aspirational - Elon Musk, Apr 14, 2021

Aspirational - adj

• relating to a strong desire for something
• aimed at or appealing to people who want to attain a higher social position
• Medicine/Medical. relating to inhaling a fluid into the lungs - or you know, a drag of something
• Phonetics. relating to articulation involving an audible puff of breath - as in "Pffft, yeah as if it'll come by May"

 

[diplomat33]

View attachment 653845

hahaha! We have a new timeline here folks, and it's even worse than today's earlier timeline of "hopefully", we're now ASPIRATIONAL

Talk about hedging your bets. If V9 is not released next month, I am sure Elon will say something like "Well, V9 is really awesome but it needs more testing so maybe July."

 

[t3sl4drvr]

Not sure how to read the timelines when Elon goes from "very confident" and "extremely confident" to "aspirational"

 

[Knightshade]

Not sure how to read the timelines when Elon goes from "very confident" and "extremely confident" to "aspirational"

Folks need to remember this quote every time Elon gives a target date for anything

People should not ascribe to malice that which can easily be explained by stupidity." (LAUGHTER) So-- so it's, like, just because I'm, like, dumb at-- at predicting dates does not mean I am untruthful. I don't know-- I-- we've-- I never made a mass-produced car. How am I supposed to know with precision when it's gonna get done?

Sub in "FSD" or "Launch our first reusable rocket to Mars" or anything else he (and to be perfectly fair NOBODY) has ever done, and get used to that.

OTOH, when he gives you a date for something they HAVE done before (start production at a new factory for example), he usually ends up coming in AHEAD of schedule, and with a deliverable noticeably better than the last one.

 

[qdeathstar]

Talk about hedging your bets. If V9 is not released next month, I am sure Elon will say something like "Well, V9 is really awesome but it needs more testing so maybe July."

V9 testing is of limited value, step change improvement is massive with v10!

 

[KArnold]

Button timing of May is aspirational - Elon Musk, Apr 14, 2021

Aspirational:

• Medicine/Medical. relating to inhaling a fluid into the lungs - or you know, a drag of something

Multiple choice? That's what I would choose. He has taken a few too many drags of something.

 

[flyespresso]

V9 testing is of limited value, step change improvement is massive with v10!

Yeah where V10 was a step change, V11 is exponential as it matches the awesomeness in the V11 Software package as well. Beta inbound for that via a side button, stay tuned!

 

[jeremymc7]

It’s all delayed, delayed, further delayed, promised delayed, delayed, and still pending delay again; until it finally, ever arrives.

Multiple Tesla purchaser, multiple Tesla owner, and FSD purchaser at bee car purchase.

 

[EnergyMax]

But what Elon said is "Production Autopilot [probability of no injury] is already above [99.999999%] for highway driving"

I think what he is saying is that there have been Y injuries over the last X million miles - for a very small number Y. But that's a statistic not a probability. I don't see how he gets a probability out of a statistic.

If I have thrown heads 9 times of the last 10 throws that's a result of 90% heads. 90% is not the probability of heads, it's still 50%.

Nope. Once you've sampled a population for an N greater than 30, you can extrapolate. That's the idea behind polls. So Elon's point is valid.

 

[Microterf]

The key word is injury. He didn't say the car wouldn't plow into the back of a cruiser parked on the highway. It will just wake you up instead of kill you like many other cars.

 

[jeremymc7]

Been in a totaled Tesla. Numerous first responders stated the Tesla saved my life. Walked away with only major bruising from the airbag impacts.

 

[AlanSubie4Life]

Nope. Once you've sampled a population for an N greater than 30, you can extrapolate. That's the idea behind polls. So Elon's point is valid.

What exactly is Elon saying? That is my first question. What does the number he quoted mean? I personally have no idea. I will take it as true, for the sake of argument, but can you explain exactly what he is saying? If I drove 100k miles on AP, can I use his probability to calculate my chance of injury? Or do I need to use hours? What exactly does it mean? I honestly have absolutely no idea and I am not trying to be obtuse.

That number of 9s also makes me wonder how many Teslas, exactly, have been sold and how many miles have been driven on AP. And how many accidents have occurred while on AP with the driver paying attention? What are the confidence intervals on that probability he is quoting?

His number is 1 in 100 million chance (if I did not miss a decimal...). Most people would interpret that statistic to mean that if you buy a Tesla and use AP, over the lifetime of the vehicle ownership, 1 out of 100 million people will have an injury accident while using it on the highway with AP engaged, while paying attention (this allows us to ignore the three deaths that have occurred with AP engaged).

That seems very unlikely. I hope it is true though!

Help me out with the interpretation, someone.

 

[AlanSubie4Life]

Walked away with only major bruising from the airbag impacts.

I am very glad you were ok. That still does count as an injury of course! However, I hope this accident did not occur with AP engaged, otherwise you might screw up Elon’s statistic (depending on what he means). My guess is it was not on AP, fortunately for Elon.

 

[EnergyMax]

What exactly is Elon saying? That is my first question. What does the number he quoted mean? I personally have no idea. I will take it as true, for the sake of argument, but can you explain exactly what he is saying? If I drove 100k miles on AP, can I use his probability to calculate my chance of injury? Or do I need to use hours? What exactly does it mean? I honestly have absolutely no idea and I am not trying to be obtuse.

That number of 9s also makes me wonder how many Teslas, exactly, have been sold and how many miles have been driven on AP. And how many accidents have occurred while on AP with the driver paying attention? What are the confidence intervals on that probability he is quoting?

His number is 1 in 100 million chance (if I did not miss a decimal...). Most people would interpret that statistic to mean that if you buy a Tesla and use AP, over the lifetime of the vehicle ownership, 1 out of 100 million people will have an injury accident while using it on the highway with AP engaged, while paying attention (this allows us to ignore the three deaths that have occurred with AP engaged).

That seems very unlikely. I hope it is true though!

Help me out with the interpretation, someone.

Great question. I don't know. I was just pointing out that when you have sampled a large number from a population, you can begin to draw statistical conclusions. Maybe Elon is referring to number of accidents per million miles on the highway. Most likely it's the number of miles per accident, as that is what Tesla reports quarterly. In q4 2020, it was 1 accident per 3.45 million miles on autopilot.

Tesla Vehicle Safety Report | Tesla

At Tesla, we believe that technology can help improve safety. That’s why Tesla vehicles are engineered to be the safest cars in the world. We believe the unique combination of passive safety, active safety, and automated driver assistance is crucial for keeping not just Tesla drivers and...

www.tesla.com

 

[Dan D.]

But what Elon said is "Production Autopilot [probability of no injury] is already above [99.999999%] for highway driving"

I think what he is saying is that there have been Y injuries over the last X million miles - for a very small number Y. But that's a statistic not a probability. I don't see how he gets a probability out of a statistic.

If I have thrown heads 9 times of the last 10 throws that's a result of 90% heads. 90% is not the probability of heads, it's still 50%.

Nope. Once you've sampled a population for an N greater than 30, you can extrapolate. That's the idea behind polls. So Elon's point is valid.

Is N the number of injuries or the distance? How does his methodology work, do you think? You might know better than me.

How did he define injury - on a scale of bruise to fatality? Where is his proof of this number 99.999999% as that is a very specific number of nines? Perhaps he's using some industry-standard way of defining and reporting. How would anyone know?

As usual, Elon has not showed his work so the result is impossible to verify or take as a serious study.


edit - AlanSubie4Life got his response in before me

 

[Dan D.]

How does his methodology work, do you think?

Here's Tesla's methodology from the Tesla Vehicle Safety Report, which BTW contains no actual data lists.
It has some big assumptions and unproven statistics, e.g. "more than 35% of all Autopilot crashes occur when the Tesla vehicle is rear-ended by another vehicle". It is skewed towards the data they have collected from the car, and requires an airbag/restraint deployment. Not a very rigorously collected safety report when you discount certain types of data.
Do all the numbers in the report end up with Elon's number? Perhaps someone wants to figure that out...

Methodology:
We collect the exact amount of miles traveled by each vehicle with Autopilot active or in manual driving, and do so without identifying specific vehicles to protect privacy. We also receive a crash alert anytime there is a crash that is correlated to the exact vehicle state at the time. This is not from a sampled data set, but rather this is exact summations. To ensure our statistics are conservative, we count any crash in which Autopilot was deactivated within 5 seconds before a crash, and we count all crashes in which the crash alert indicated an airbag or other active restraint deployed. In practice, this correlates to nearly any crash at about 12 mph (20 kph) or above, depending on the crash forces generated. On the other hand, police-reported crashes from government databases are notoriously under-reported, by some estimates as much as 50%, in large part because most fender benders are not investigated. We also do not differentiate based on the type of crash or fault, and in fact, more than 35% of all Autopilot crashes occur when the Tesla vehicle is rear-ended by another vehicle. In this way, we are confident that the statistics we share unquestionably show the benefits of Autopilot.

 

[jeremymc7]

I am very glad you were ok. That still does count as an injury of course! However, I hope this accident did not occur with AP engaged, otherwise you might screw up Elon’s statistic (depending on what he means). My guess is it was not on AP, fortunately for Elon.

I am very glad you were ok. That still does count as an injury of course! However, I hope this accident did not occur with AP engaged, otherwise you might screw up Elon’s statistic (depending on what he means). My guess is it was not on AP, fortunately for Elon.

It was AutoPilot. The car actually visibly tried to get away from from the accident. Unfortunately the dash cam didn’t save starting about a 30 sec before the accident. Which I hear was common at the time. And possibly why they now save to internal memory first and then you have to press the button to save to the USB drive after; which is kind of lame considering you might not think or be capable of after an impact.

 

[AlanSubie4Life]

In q4 2020, it was 1 accident per 3.45 million miles on autopilot.

Ok let’s roll with that. Correct me if I am wrong on the math, but that calculates out to:

0.29 accidents per million miles.

So what’s the probability per mile of no accident?

99.99997%, i. He quoted 99.999999%.

So I guess that means that only 1 out of every 30 accidents on AP with the driver paying attention an injury accident? Seems a little low doesn’t it?

So if you drive cross country for 3000 miles on AP, we take that 0.99999999^3000. (Probabilities multiply here.) You get 99.997% chance of NO injury (0.003% chance of injury).

I guess if you do a lot of driving on AP you might rack up 100k AP miles. That works out to 99.9% chance of success.

That seems ok, I guess; 1 out of 1000 people doing that will get injured.

It’s late, might have screwed up the math. Someone should check.

I still wonder what he meant, though, and not sure about that injury/no injury ratio. I feel like freeway accidents result in injury more than 3% of the time.

I also wonder about his confidence intervals (depending on the statistic).

 

[Dan D.]

So taking a look at the videos of FSD Beta up to version 8.2 in city driving. What would we say is the accident rate per million miles?

Should we allow disengagements? It is Level 2, and the driver is supposed to be attentive. I would say that the accidents per million miles is zero then. Blame the driver 100%

If we don't allow disengagements like it is Level 3. Probably one accident per 10 miles on average. (Someone would actually have to NOT disengage to prove this though).

Which methodology will Tesla use? Which methodology will Ford/Mercedes/etc use?

 

[AlanSubie4Life]

In practice, this correlates to nearly any crash at about 12 mph (20 kph) or above, depending on the crash forces generated.

So I guess we’re going with Elon quoting a per mile statistic. (So 1 in 100 million chance of an injury accident per mile driven on AP when paying attention.) That seems fine, though I wish he had specified. That would be really quite good though I think.

I’m surprised that airbag deployment crashes result in injuries only 3% of the time! According to the Tesla info, they deploy 0.3 times per million or 30 times per 100 million miles.