I Can Think Faster than Auto Pilot

[tandemnw]

Even compared to just Teslas it's roughly 1.5-2 times lower depending if the non-AP folks had active safety features on or not.

I’m sure active safety helps, it gets rid of the entire class of inattentive driving issues. However, I think the stats for AP may include selection bias.

I drive extensively on AP, and every time I’ve had a dicey situation, AP has disengaged on me. If I had gotten in an accident in any of those, I’m betting it’s not counting as “while AP is active” even though it was active milliseconds earlier.

Since AP cherry picks ideal situations, I’m certain it’ll have great stats.

 

[AlanSubie4Life]

and driving on AP is far safer than driving without

The statistics do *seem* to say this, but it’s very hard to say whether this is correlation or causation.

If Tesla even provided the number of miles driven on Autopilot (and no Autopilot, etc.) seasonally (they don’t) we might be able to draw some conclusions about just one factor which might make Autopilot safer. Autopilot got less safe last quarter - was this also coincident with a lower % of AP miles driven by Tesla owners? This would imply that Autopilot is used less in adverse conditions...which would make it appear (potentially) a lot safer.

Obviously this is just one factor. I’m looking forward to Tesla at some point in the future publishing useful data, rather than useless data.

I agree that Tesla’s data shows that lower accident rates when driving a Tesla are correlated with the use of AP. (There’s not enough information to know whether the NHTSA data can be compared to Tesla’s data.) Anyway, that is an extremely weak statement.

 

[Uncle Paul]

Data like this is usually released for marketing/public relations purposes. The cars seem to keep getting better an safer. Tesla is trying to enourage people to buy their cars instead of competitors cars. Safety sells.

 

[IceIceBaby]

Then those folks are provably wrong.

Tesla has published data on this for a while.

Tesla Vehicle Safety Report



In other words the accident rate on Autopilot, compared to the entire fleet of all US cars of all brands, is about 6 times lower

Even compared to just Teslas it's roughly 1.5-2 times lower depending if the non-AP folks had active safety features on or not.

There's also significantly less Tesla's on the road than ice vehicles.... It's really quite hard to compare.

 

[ItsNotAboutTheMoney]

There's also significantly less Tesla's on the road than ice vehicles.... It's really quite hard to compare.

The aim of releasing the numbers is to counteract selection bias of news media.

Tesla points to the statistics and shows up the hand-waving.

 

[Knightshade]

There's also significantly less Tesla's on the road than ice vehicles.... It's really quite hard to compare.

Why?

All the stats are given normalized by miles driven...which doesn't care about # of cars of each type.

 

[IceIceBaby]

Why?

All the stats are given normalized by miles driven...which doesn't care about # of cars of each type.

Lol. Ur trying so hard hahahahaba

 

[Knightshade]

Lol. Ur trying so hard hahahahaba

To explain facts and math? Yes I am.

Why are you trying so hard to not understand them?

Alan raised a valid reason the stats might not be as definitive as Tesla suggests (though we'd need more info to determine if it's correct or not). You did not raise a valid objection and I pointed out why.

 

[IceIceBaby]

To explain facts and math? Yes I am.

Why are you trying so hard to not understand them?

Alan raised a valid reason the stats might not be as definitive as Tesla suggests (though we'd need more info to determine if it's correct or not). You did not raise a valid objection and I pointed out why.

But you've provided no math or numbers at all...

 

[Knightshade]

But you've provided no math or numbers at all...

Except, I did.

You even replied to the post where I provided the link to the full report of those numbers.

Then you raised an objection to those numbers (that you now claim you never saw), an objection that made no actual sense, and I explained the mathematical reason why it made no sense....and you replied to THAT post too and now claim you never saw it.

So increasingly you appear to just be trolling.

 

[IceIceBaby]

Except, I did.

You even replied to the post where I provided the link to the full report of those numbers.

Then you raised an objection to those numbers (that you now claim you never saw), an objection that made no actual sense, and I explained the mathematical reason why it made no sense....and you replied to THAT post too and now claim you never saw it.

So increasingly you appear to just be trolling.

Objective numbers please.

 

[Knightshade]

Objective numbers please.

They're right there in the link.

The only objection you've actually presented to them doesn't make any actual sense since the #s are normalized by miles driven.