Unfortunately it seems the best (not safest) approach with yellow lights is to run it instead of trying to stopped. It seems you likely be hit more for the harsh breaking on a yellow versus a little tap to run it. Be it on AP or not.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Safety Score
- Thread starter ScottNY
- Start date
Unfortunately it seems the best (not safest) approach with yellow lights is to run it instead of trying to stopped. It seems you likely be hit more for the harsh breaking on a yellow versus a little tap to run it. Be it on AP or not.
I mean according to the subscription autosteer on city streets is by the end of the year right?!!! So what about two months of beta. Everybody is getting it fleet wide who bought it according to the Tesla website, app and Elon.
Anyway, still have a 100 even after my DT to home crazy commute somehow. See how the next several days go.
Anyway, still have a 100 even after my DT to home crazy commute somehow. See how the next several days go.
KArnold
Active Member
By chance do you park in a tight garage? See Getting Dinged on Safety Score for Parking in my Garage?
This is exactly what you want to hear from someone using a system which is supposed to improve safety
Be glad you don't live in Oregon with their restrictive yellow lights, where hard braking is legally required. You MUST come to a stop the instant it turns yellow if it is possible and safe to do so, and ending up in the intersection while it's red is illegal. I really doubt a police officer will consider 0.31G of braking "unsafe".
I am still entertained over here with all of you trying to game your scores against an arbitrary and capricious standard to get something you paid for. I'm still in awe of what Tesla has done.
How many of you hoping to get into the beta have had an at fault accident in the last 5 years that triggered Tesla's accident metrics (airbag or other active safety deployment)? Is this really making you a safer driver, or are you just playing a game against a completely uncaring opponent?
Seatbelts reduce injury during an accident. They don't prevent them.
Keep up. The Safety Score is a highly scientific formula honed over 6 Billion miles of data to accurately estimate your Predicted Collison Frequency.
AlanSubie4Life
Efficiency Obsessed Member
As I said, driving at or below the limit makes this a non-issue for the most part in California. No I don’t live in Oregon which is good. The only accident I have been in was in Oregon where I stopped short for a yellow light and got rear-ended, as a first-year driver.This is exactly what you want to hear from someone using a system which is supposed to improve safety
I will just say I am driving much more slowly now, and my RE-71R tires will last a long time now (which is not a great thing - since they are so terrible it is best to destroy them ASAP). I should probably put on my CrossClimates to limit my cornering forces for the Beta Queue though. Less harsh too; might prevent g limit excursions at the margins.
There’s no question in my mind that the Beta Queue has improved the safety of my driving. For that I can thank Tesla.
Sadly the improvement will only be temporary, though the extreme practice of defensive driving techniques is probably never a bad thing, and may help fine-tune those skills.This is a good point. Am I parking too close to the garage wall. The garage is barely moving but I do park quite close. Is this an issue I wonder ???
Interesting observation about how the "safety score" is calculated.
The "PCF" is predicted collision frequency per 1M miles.
The safety score is calculated as Safety Score = 115.382324 - (22.526504 x PCF)
Ok, so literally having no collisions ever only gets you a safety score of 115.
Tesla's own collison report says a Tesla not on AP has an accident every 2.05M miles (0.488 / 1M miles).
Thus, the average Tesla car has a "safety score" of 104!
So for everyone with a score of less than 100, you are on average, lower than the average Tesla driver. Ouch.
A score of 80? That means they are predicting a collision every 640k miles (1.56 / 1M miles). Literally 3.2 times the rate of the average Tesla.
The difference between a score of 90 and 100 is 892K miles vs 1,471K miles PCF. It's completely non-linear.
This whole formula is meant to pack most drivers in the 80-100 range, capping the number at 100, making everyone feel good, even though the actual fleetwide average is above 100.
Genius Marketing.
The "PCF" is predicted collision frequency per 1M miles.
The safety score is calculated as Safety Score = 115.382324 - (22.526504 x PCF)
Ok, so literally having no collisions ever only gets you a safety score of 115.
Tesla's own collison report says a Tesla not on AP has an accident every 2.05M miles (0.488 / 1M miles).
Thus, the average Tesla car has a "safety score" of 104!
So for everyone with a score of less than 100, you are on average, lower than the average Tesla driver. Ouch.
A score of 80? That means they are predicting a collision every 640k miles (1.56 / 1M miles). Literally 3.2 times the rate of the average Tesla.
The difference between a score of 90 and 100 is 892K miles vs 1,471K miles PCF. It's completely non-linear.
This whole formula is meant to pack most drivers in the 80-100 range, capping the number at 100, making everyone feel good, even though the actual fleetwide average is above 100.
Genius Marketing.
Last edited:
If they were to add this to the app for everyone and tie it to Tesla insurance to use as advertising it could be good free advertising. If I were to see a daily update that days (or trips) would get Tesla insurance for 50% or less vs my current insurance I might jump ship. But I would want to see daily or week related quotes like we’re getting stored right now for AP/FSD.
Any street tire can do way over 0.4G. There's no additional penalty for 1.2G vs .9G.
Technically a higher performance tire increases your score. You can complete the turn faster, and all Tesla cares about is time, not G force. 1 second at 1.2G has less impact than 0.6G for 2 seconds. So stupid.
But you're worried about how "safe" a driver you are and you're driving RE-71's around on the street and believe going slower is safer and 0.3G braking or 0.4G turns are dangerous? It takes all kinds, I guess. I consider myself an aggressive driver and would never run RE-71's on the street- I take them off right away after every event. Of course, it actually rains in Seattle.
Last edited:
I just did the math. To have 1/2 the risk of an average Tesla driver would require a safety score of...
109%.
Well, I’m assume that there potential liability for damages would be less if it’s just the destroyed vs a fatality
AlanSubie4Life
Efficiency Obsessed Member
Obviously not. Slower IS safer though, in general. It’s hard to get around that, no matter how good the tires are. It’s just physics.
Yes, I know. I was kidding.
Yes that is part of the optimization strategy if you do have to stop for a yellow light. Slam brakes and get to maximum g briefly then release and regen. Requires excellent car control, it is quite an optimization problem! You are right, I will leave the RE-71Rs on, for minimum pulse width.
Last edited:
This is not a potential liability score. It's a potential collision frequency.
Also, what liability? All failures of Tesla's autopilot or FSD are unquestionably the driver's fault, right?
You did the math wrong. Top possible score is 100 as calculated by some very intelligent posters. You made a bad calculation somewhere and then jumped to an erroneous conclusion.
And yet, Tesla doesn't include speed in any of their calculations, despite having many measures of it.
Kinda conflicts with the first things you see when you click on the Model S:
Link?
The Score is literally 115 - 22 x PCF, and PCF is rate per 1M miles. It's right there.
If your PCF is 0 (no accidents ever), your score is 115. It's that easy. Your actual PCF is 0.5 (1:2M miles)? Your score is 104. Fact.
I am doing the math by actual rate of collisions per million miles. Not going through their insane guestimates based on their random factors. This just proves their inputs don't ACTUALLY calculate a useful PCF, because a PCF of 0.48 (the actual fleet average) gives you a score of above 100. Fact.
‘Braking score is a ratio ( time over 0.3g / time between 0.1-0.3g). So if you drive around the neighborhood doing a bunch of 3 mph / s decelerations from 30 mph, you’re adding 10s a pop to the denominator. Count 1001, 1002… as you watch the speedometer to get a feel. So if your hard braking was only 3 seconds, then 30 gentle neighborhood decelerations would reduce it to 1%, even if you had no other gentle decelerations that day.Observations after a few more drives - if you live anywhere around steep hills, you're screwed with respect to the hard braking score because you truly can't use the friction brakes without getting dinged for it. Got a 3.4% hard braking score this morning after stopping gently for pedestrians crossing the street at the bottom of a hill using the brake pedal. Either have to crawl down every hill too steep for regen or run stop signs to allow regen to stop you once the hill tapers off. Or find routes with different geography
I did this the first night and it worked great but I must have taken one turn a little too fast, even though I thought I was taking it very slowly, so I got dinged for fast cornering.
Note this only works if done on the same day.
Similar threads
