Safety Score

[Danilo]

This grading system is a joke. It’s quite dangerous to game it. I found out why I had poor follow distance. It was because people would cut me off at 75mph and AP takes awhile to back off to follow distance 7. What I did was drove 50-55mph on the way home. Every car changed lanes in front of me at 75-80.

Also, hard to time traffic lights so I just cruised through all yellows, late yellows, and just turned Reds. Hit a poor rabbit on the way home at 30mph. Was scared to get a hard braking ding.

Got my score back to 99.

I agree, I was able to see it using my iPad but I didn’t follow too close at all, if I’m too far ppl cut me off and honk at me… is way more dangerous to leave big gaps and ppl cut you off… and I was using navigate on autopilot all the time… I do stop close at the light, maybe that’s the problem?!?

 

[DCEV]

I got a 93 for a 4 mile drive. Complaining about hard braking. I had 2 what I thought were very gentle manual stops and most of it was on TACC.

Similar experience here.

What I find odd is there is no tracking to see if you use indicators before turning or making lane changes. Most reckless drivers I see never or rarely use indicators.

I feel like people who bought a Performance car with FSD are really screwed with this. I mean what's the point of a Performance car with Pilot Sport tires if you can't enjoy a turn or on/off ramp every now and then when it is safe to do so and there are no other vehicles around.

I also really feel for people with grueling commutes where people cut in often, speed of traffic varies very quickly, and you need to drive with the traffic flow in a certain way which this algorithm might not like.

The real kicker here is that your PERSONAL driving preferences have ABSOLUTELY NOTHING to do with how safe you actually are and how well you can monitor FSD beta driving and intervening when needed to keep you and your car safe.

Wish they would just ask permission to look up your law enforcement driving record and insurance claim history because those two factors are a better indicator of your driving. Heck look up my credit score while they are at it.

At this point, I think I'm actually really resentful of the fact that I paid for FSD years ago and waited so long and now just to have a chance of using what I paid a bunch of money for, I need to drive my Performance Model S like I'm leading a funeral procession.

This sucks. The grading scheme is a sick joke. It has completely taken away the joy of driving my car and how I drive my car manually has no bearing on my ability to be responsible when I monitor FSD Beta.

 

[DCEV]

I think probably because I would constantly have that number in the back of my head, constantly measuring every move against it. It would sap the moments of joy these cars currently bring me. Having every aspect of life quantified like this is not something that makes me happy. Optimizing to a score has a lot of malignant failure modes such as we see in ridesharing and online reviews in general. It makes me feel bad to participate in that.

This is absolutely how I feel and I appreciate you sharing this.

We bought a car and then paid a bunch of money for FSD in good faith years ago and then patiently waited for years to be able to use FSD beta, knowing very well what we bought is likely going to be in "beta" for years and years.

I don't think any of us signed up for some BS algorithm to judge and spit some BS score at us each time we drive the car that has no actual basis on safety. I'd say your score is going to vary greatly based on the driving culture of the area you live in and the volume of traffic that actual safety.

What I find really odd is that someone can run red lights, run stop signs, not signal before lane changes, cut people off -- all truly actually dangerous actions -- and none of these count against your score! But oh god forbid you take a turn or onramp to enjoy the fact that you bought a Tesla and not a Prius, and that's a crime that deserves some BS demerit. Basically just about anything you do to enjoy the Tesla you bought could result in a demerit. I mean WTF is this?! Who signed up for this when they bought FSD?

I bought a Tesla so I can safely enjoy what I bought every now and then. Safely enjoying the capabilities of my car has ABSOLUTELY NOTHING to do with my ability to monitor FSD Beta and intervene when needed.

This is just too much patronizing BS. I don't want my driving "gamified." Driving is NOT a game to me and I don't want some BS score spit at me each time I drive my car. When I drive, if I have to stop suddenly or make a quick lane change or do whatever I need to do to stay safe, I do that and I don't want some BS algorithm that barely knows how to drive to grade me on my driving having no understanding of the context.

 

[sperkin]

I agree, I was able to see it using my iPad but I didn’t follow too close at all, if I’m too far ppl cut me off and honk at me… is way more dangerous to leave big gaps and ppl cut you off… and I was using navigate on autopilot all the time… I do stop close at the light, maybe that’s the problem?!?

Distance follow 7 is about 1 to 1.2 seconds. The gap needs to be be bigger. I was dinged for driving on Auto Pilot all day. There should be another option to leave 2 to 3 sec gaps. Also, braking to stay at 7 distance was dangerous as I was almost rear ended three times because the car will stop short to keep the gap.

I had a lot of aggro drivers from LA to SD. However staying in the middle lane at 55mph was fine on the way back. People cut me off at a faster speed that they're not even close to me.

I'm a little disappointed at the grading system, but Tesla is the still the best car out there. Some people may think I'm a troll, but I'm just posting my personal experience and opinion here. Just really disappointed at FS, it's beta roll out, and this grading system. My yearly bonus is coming up, and I can't find a car that I would rather buy than another Tesla. I'm watching YT videos on 911 GT3, 458, and NSX, but still think Tesla is the best car on the market.

 

[EVNow]

Wish they would just ask permission to look up your law enforcement driving record and insurance claim history because those two factors are a better indicator of your driving. Heck look up my credit score while they are at it.

So, basically everything except how you actually drive ?

 

[DCEV]

So, basically everything except how you actually drive ?

Enough of this nonsense false equivalency between this BS grading system and safely driving your car.

I mean WTF is the point in buying a Tesla with FSD if you are expected to take turns like a granny driving to church in a Prius? I just drove to the grocery store and back and I got some BS demerit for a quick turn.

And what does my turns have ANYTHING to do with being able to monitor FSD Beta and intervene as needed? I've monitored Autopilot for tens and thousands of miles safely and I am hyper careful.

If this grading system wanted to score for actual safety, maybe it can give a rats ass about if you change lanes without signaling, run stop signs, run red lights, or cut people off. Those factors would be a lot more relevant than someone who bought a Tesla enjoying a turn or on/off ramp every now and then.

 

[Daniel in SD]

So, the safety score (beta) documentation says:
“The Safety Score formula ignores any events (not just Forward Collision Warnings) that occur while Autopilot is engaged and appropriately utilized.”
And:
“Only use the various Autopilot features as described in your Owner’s Manual.”
And the owner‘s manual says:
"Autosteer is intended for use on highways and limited-access roads with a fully attentive driver. When using Autosteer, hold the steering wheel and be mindful of road conditions and surrounding traffic. Do not use Autosteer on city streets, in construction zones, or in areas where bicyclists or pedestrians may be present."

Is it possible that @Knightshade has finally been proven right? Is Tesla penalizing people for using Autopilot on city streets? I'm seeing a lot of people getting low scores saying "I used Autopilot the whole drive!"

The other factor that people seem to be ignoring is that the hard braking, cornering, and unsafe distance metrics are all proportions. You need to increase the denominator by driving like a normal safe driver, if you drive like a grandma all the time the few times you're forced to brake hard or someone cutting in front of you is going to count a lot more.

I'm sure Tesla has the statistics to back up their formula. Obviously people trying to game the formula will reduce its predictive ability. Though maybe the ability to win the game is predictive of FSD Beta monitoring skills?

 

[AlanSubie4Life]

Today had its ups and downs. In the end, I still feel like I did ok. However, particularly on the freeway have to be very careful to select the right drive, and take those interchanges quite slowly. Take another car if the conditions aren’t right!

Regarding wasting the Performance…It’s tough doing this in a Performance with RE-71Rs on; they just want to be cornered on (gotta have some fun in exchange for the incredible harshness and noise). However, this is just a temporary opportunity to try to qualify for the beta, which is an incomplete L2 system and requires extremely attentive drivers to monitor it. This type of monitoring will likely not last forever (and if it does, I won’t really care, because that would mean the novelty and utility of FSD will be very limited). The main thing is to qualify. Long-term qualification is likely going to be less important for me.

I do think it’s likely there is a positive correlation between those who can work out how to game the system and those who can monitor FSD Beta correctly, even if the skill set is not identical (there is some overlap though). It requires some self-awareness, and a burning desire to acquire FSD Beta, to tolerate the restrictions and optimize the Safety Score. That’s probably fine.

As long as Tesla does not change the terms of the game, I’m ok with this.

I’ve found I’m now hyper aware when I drive. This is probably good, and I’m quite sure I was a much safer driver today than is my habit (just allowed much larger safety margins to avoid having to brake hard). Sadly I had to slow down very gradually for a yellow light and that was a minor strike against me today. I do probably have to optimize the rate of slow down to get the denominator up…slow down slowly, but not too slowly!

Timing was good today - happened to have a lot of driving - very rare that I drive 100 miles in a day, but no special trips for the Beta Queue…

 

[AlanSubie4Life]

Here are the fleet averages. I wonder what set of drivers this includes? It’s not clear whether it’s an active tracking of how the average person in The Queue is doing. I guess they could be continuously collecting anonymized metrics from everyone in the entire Tesla fleet, but not sure what is allowed to be gathered from people who are not signed up for The Queue.

It would be cool to see how you’re doing relative to other Queue participants, in any case. Need to keep it competitive.

 

[mhan00]

Lol, I have Tesla insurance and got a 97 today, with my only ding being “hard braking” when I never once hit the brake pedal. The only time it could have been “hard braking” was when I was coasting down the large hill I live on top of and have gotten good at timing properly to coast to a stop at the light that’s at the bottom without once hitting the brake. This seems to be confirmed because I was at a 95 when i checked after getting to the store after driving down the hill, and got back up to a 97 when I checked after I finished my errands and I got back home. I wonder if I should be calling to complain about how they’ve been calculating my premium?

 

[Daniel in SD]

Lol, I have Tesla insurance and got a 97 today, with my only ding being “hard braking” when I never once hit the brake pedal. The only time it could have been “hard braking” was when I was coasting down the large hill I live on top of and have gotten good at timing properly to coast to a stop at the light that’s at the bottom without once hitting the brake. This seems to be confirmed because I was at a 95 when i checked after getting to the store after driving down the hill, and got back up to a 97 when I checked after I finished my errands and I got back home. I wonder if I should be calling to complain about how they’ve been calculating my premium?

It seems plausible that acceleration is measured by the accelerometer not the derivative of wheel speed. This method makes sense because braking distance will be increased when going down a hill.
A 10% grade (steep hill) is 5.7 degrees.
sin(5.7) * 1g = 0.1g
If regen is 0.2g then the extra 0.1g could push you over the 0.3g limit!

 

[TresLA]

Is it possible that @Knightshade has finally been proven right? Is Tesla penalizing people for using Autopilot on city streets?

I don’t think so, since on all my drives I’ve been using autopilot on surface streets too and am getting 100’s except for one drive where I hit the brakes hard coming out of the parking space (because of a Tesla Model S flying around the parking lot aisle corner heading towards the Supercharger no less!)

From all that I read before pushing “the button”, I gathered that driving with autopilot engaged is the best way to avoid Safety Score dings, so I’ve been doing so including surface streets.

 

[TresLA]

On Android, I installed 4.1.0-... using APKM or whatever mirror. If you have 4.1.0-... on iOS then a compete restart of the phone has reportedly helped some.

I have iOS devices, everything is updated and I am in the queue for FSD beta. No luck.

Me and another just had to force quit Tesla app and open it again. Then it showed right away.

 

[hacer]

Pro tip, it's best to turn on AP on city streets or else you'll be penalized from heavy braking due to green to red stop light events. If AP does it for you, then you will not be penalized. I was penalized today due to two unfortunate light changes.

I've always found AP to be much more rush up to the light and brake hard compared to my own driving style, at least for lights that are and have been red for some time. It should never be necessary to brake hard for a yellow light unless someone in front of you brakes hard or some cross traffic is running their red (for a long time) light. If only hard braking will allow you to stop for the light then you should either being going through the yellow, or you are speeding excessively.

 

[mongo]

I've always found AP to be much more rush up to the light and brake hard compared to my own driving style, at least for lights that are and have been red for some time. It should never be necessary to brake hard for a yellow light unless someone in front of you brakes hard or some cross traffic is running their red (for a long time) light. If only hard braking will allow you to stop for the light then you should either being going through the yellow, or you are speeding excessively.

Or the yellow signal interval is too short.
Yellow time >=( stopping distance for 'normal' braking in 'normal' road conditions)/ (speed limit) + reaction time.
Could replace 'normal' with 'adverse', but drivers should drive slower when roads are bad.

Edit:

Hard braking is defined as backward acceleration, measured by your Tesla vehicle, in excess of 0.3g. This is the same as a decrease in the vehicle’s speed larger than 6.7 mph, in one second. Hard braking is introduced into the Safety Score formula as the proportion of time (expressed as a percentage) where the vehicle experiences backward acceleration greater than 0.3g relative to the proportion of time where the vehicle experiences backward acceleration greater than 0.1g (2.2 mph in one second).

From this definition, a 55 MPH braking event needs to be 8.2 seconds to avoid a demerit. This is 225 feet.
The federal standard for yellow lights is 3-6 seconds:
Yellow Light Time Standards | Stop Short Yellow Lights
6 seconds at 55MPH is 484 feet, 3 seconds is 242 feet, so one hopes the lights are speed adjusted.

Note that the value in the safety formula is proportional to total time braking (more than 0.1G) so one can improve the value if a yellow light got them.
Though, at 1.127294^(hard braking ratio), 100% hard braking should only be a 12.7% increase in the PCF factor, not enough on its own to trigger unsafe ratings (2 points if everything else is ideal).

 

[hacer]

You are correct. I would expect that regulators would want to analyze interventions. Tesla, too, can evaluate how good its system is by the frequency of interventions. Sometimes intervene when AP is taking a curve too wide or too fast for my comfort. This is not necessarily an indication that AP was unsafe, just not smooth enough for my comfort. Even so, as the FSD becomes so sophisticated that occupants are never uncomfortable with the driving, never feel like they need to take over, this is a really good thing.

Still from a safety testing perspective, interventions should probably be analyzed to determine if any immanent safety hazard was present. If so, FSD needs to be fine tuned to anticipate that more quickly than the human driver does. Human drivers are still quite capable of anticipating risks before FSD kicks in. So it is an opportunity for the AI to learn more thoroughly from human drivers.

It's funny that at the same time the safety score heavily penalizes disengagement -as if Tesla would rather have you let the AP or FSD crash the car because that's "safer" than forcibly disengaging. Gali should have just crashed into those railway pillars if he wanted to be a safe driver I guess.

Edit: I see that I misunderstood what is meant by "forced disengagement". I'm sure there would be a better way to describe disabling AP from lack of feedback.

 

[mongo]

It's funny that at the same time the safety score heavily penalizes disengagement -as if Tesla would rather have you let the AP or FSD crash the car because that's "safer" than forcibly disengaging. Gali should have just crashed into those railway pillars if he wanted to be a safe driver I guess.

Forced disengagements is not the driver taking over, it means AP kicked you out, due to ignoring nags (90MPH+ puts you in AP jail normally so that may also apply)
Safety Score Beta

Forced Autopilot Disengagement
The Autopilot system disengages for the remainder of a trip after you, the driver, have received three audio and visual warnings. These warnings occur when your Tesla vehicle has determined that you have removed your hands from the steering wheel and have become inattentive. Forced Autopilot Disengagement is introduced into the Safety Score formula as a 1 or 0 indicator. The value is 1 if the Autopilot system is forcibly disengaged during a trip, and 0 otherwise.

 

[ruralectric]

Have you driven it since opting in? My score showed up 15 minutes after I got back from a drive.

After a day of "sporty" driving out in the "rur" of the Palouse, the score finally showed up. 93, dinged for "agressive turning". I'm gonna have to up my game. I know I can be more agressive than that : /

 

[GWord]

Ha ha ha. Morons. I got the update on an iPad which had the old app installed. When I updated I saw my score was 7 for a deliberateply mellow trip to the hardware store a mile away! Ha ha ha. Morons. Hard braking? Sorry, none, unless you mean when I came to a stop sign at the bottom of a steep hill and regen wasn’t going to be enough so I braked softly enough that Grammy wouldn’t have woken up from her afternoon slumber. Moreover I got a forward collision warning because I passed by a car parked in the side of a wide road. Huh? Or maybe it was because I got a lane departure warning because I was in my goddamned driveway. Larfs! Glad I don’t have Tesla insurance, what a joke.

Every time a NJ resident realizes they’re a terrible driver, an angel gets its wings. Ironically, it’s the places that need autonomous driving the most that make for poor FSD beta testing from a risk perspective.

 

[hacer]

Or the yellow signal interval is too short.
Yellow time >=( stopping distance for 'normal' braking in 'normal' road conditions)/ (speed limit) + reaction time.
Could replace 'normal' with 'adverse', but drivers should drive slower when roads are bad.

Edit:

From this definition, a 55 MPH braking event needs to be 8.2 seconds to avoid a demerit. This is 225 feet.
The federal standard for yellow lights is 3-6 seconds:
Yellow Light Time Standards | Stop Short Yellow Lights
6 seconds at 55MPH is 484 feet, 3 seconds is 242 feet, so one hopes the lights are speed adjusted.

Note that the value in the safety formula is proportional to total time braking (more than 0.1G) so one can improve the value if a yellow light got them.
Though, at 1.127294^(hard braking ratio), 100% hard braking should only be a 12.7% increase in the PCF factor, not enough on its own to trigger unsafe ratings (2 points if everything else is ideal).

Here in Maryland I don't recall ever having encountered a yellow light that didn't provide enough time for gentle braking to a stop from 5 over the speed limit.