Safety Score

[ConSaab]

It’s not only mileage. If you drive 100 miles on AP, it won’t dilute your unsafe following distance. You’ll need to drive 100 miles manually keeping safe distance with a vehicle in front of you in order to dilute that. Pretty sure the only thing mileage on AP dilutes is the FCW.

I wish this was mentioned to me before I did my 272 drive which was needed for me to reach 100… I did it mostly on AP…

So if I understood this correctly, if I were to get a cumulative 100 score on main page, I have to work on reducing all the percentage digs I had… example follow to closely and hard braking…

 

[Ramphex]

I wish this was mentioned to me before I did my 272 drive which was needed for me to reach 100… I did it mostly on AP…

So if I understood this correctly, if I were to get a cumulative 100 score on main page, I have to work on reducing all the percentage digs I had… example follow to closely and hard braking…

Refer to what @valem said above. Those stats don’t really matter in order to get overall 100. But if you wanted to drop those numbers then you would have to manually drive following someone (not too closely ((which who knows what that means))) in order to dilute the following too closely strikes. Just driving around on AP doesn’t decrease that number.

It would really help if the car provided guidance. Like if you were following someone too closely, it would give you some sort of a bar that would give you a strike if it filled up, but if you resolved the issue then it wouldn’t strike.

Braking too hard is outright ridiculous. AP borderline locks up brakes for yellow lights, even worse the yellow flashing lights, where are the strikes against it and positive points for my driving?!?! Honestly though the AP/TACC are wearing out my brake pads and rotors with their nonsense, it’s almost like they don’t know how to take an advantage of regen.

 

[insaneoctane]

I don't think Elon was counting on the 2 day delay when he tweeted that only scores of 100 would initially get the beta. Since telling us that, there has been motivation to find ways to get to 100, when before people were happy with their 99s. Now with 2+ more days, the number of 100s is growing and maybe approaching a big enough number to move the goal posts again. TeslaFi reported 9% of their fleet had 100% when Elon tipped the 100 requirement, now it's 12%, which is a 33% increase. If it was 1,200 before, it might be 1,600 now and that's 1.6x Tesla's initial goal....guess we'll see....

 

[Ramphex]

Whenever you have a unsafe following percentage for the day try to find a spot to accelerate and then slowly brake. Do this over, and over. After a bit of this it will drastically reduce the unsafe following distance.

AP usage tends to exaggerate the unsafe following distance not because anything on AP is counted, but that the time that is counted typically involves being close (like merging in).

Are you referring to… hard braking? Not unsafe following? Lol I don’t see accelerating and braking diluting unsafe following… unless it does?

 

[a13x_k]

The TeslaFi 100 scores have jumped again to 12% as of now and that's a jump from 11% about 29 hours ago.
Just for reference here are all of the screen captures I've posted since Thursday morning (2 days ago) with the most recent at the top. Note that their FSD fleet vehicle count that they report on has gone from 1,569 2 days ago (far below) to 1,880 just now.

View attachment 719662

Not sure if that was discussed before. So given that we know that there are ~1200 drivers with 100 and if teslafi is a somewhat representative sample, then there are ~1600 people with score of 99. ~1500 with 98. That doesn’t seem right to me. Perfect 100 is so hard to get, so I would have thought that there are significantly more people in the high 99s.

 

[S4WRXTTCS]

That 96 hurts me so bad! I had no idea at the time that I could so easily fix it before the end of the day. I've had only perfect scores since...

Yeah, the really messed up part is everything can be fixed as long as you do the fix that day or that drive.

Unsafe following distance -> Drive on the freeway without AP on for some amount of time well behind someone without following too closely while going above 50. I don't think it will take a lot of time because you only need to counter-act the tiny amount of time you weren't on AP while above 50. Things like merging which may have dinged your score.

Hard braking -> Lots of acceleration and slow braking. Don't worry about miles, but just keep doing this over and over.

Hard cornering -> Find some parking lot or roundabout in the middle of the night, and keep going around and around

FCW -> Do a reboot

AP Lockout -> Do a reboot (this one is totally preventable though).

I believe you can even Opt-Out before you take a drive you know will screw you, and then you can opt back in after the drive. There might be unintended consequences of this though.

 

[Ramphex]

These will be the guys testing out FSD beta

 

[S4WRXTTCS]

Are you referring to… hard braking? Not unsafe following? Lol I don’t see accelerating and braking diluting unsafe following… unless it does?

I don't know what happened to my brain on that post

I deleted the post because there was no recovery from that impressive of a brain fart.

PS - I wrote a more complete post about all situations.

 

[S4WRXTTCS]

View attachment 719728

These will be the guys testing out FSD beta

tapout sure tapped out of this one.

 

[ConSaab]

Refer to what @valem said above. Those stats don’t really matter in order to get overall 100. But if you wanted to drop those numbers then you would have to manually drive following someone (not too closely ((which who knows what that means))) in order to dilute the following too closely strikes. Just driving around on AP doesn’t decrease that number.

It would really help if the car provided guidance. Like if you were following someone too closely, it would give you some sort of a bar that would give you a strike if it filled up, but if you resolved the issue then it wouldn’t strike.

Braking too hard is outright ridiculous. AP borderline locks up brakes for yellow lights, even worse the yellow flashing lights, where are the strikes against it and positive points for my driving?!?! Honestly though the AP/TACC are wearing out my brake pads and rotors with their nonsense, it’s almost like they don’t know how to take an advantage of regen.

Gotcha…so need a clarification on the spreadsheet and score app that’s been shared here.

The miles required to reach 100… does that need to be driven manually or can it be driven on AP?
Obvious it’s easier to get those miles on AP.

 

[Ramphex]

You can do miles on AP if you just need miles and 100 daily scores.

Safety Score Calculator <— simple app for tracking miles/score

 

[Sporty]

I had a 96 over 149 miles which alone would have required over 1000 miles to fix back to 100

Ya.. getting to 100 is brutal. I didn’t start worrying about it until a day or two after they said the mats all they were taking and then it was too late. Stuck at 99

 

[Ramphex]

1435 miles to go because of that one day with false FCW lololol, this will never happen.

That was also like day 2 of the games, wish I knew about the reset trick. This sugar being beta seems impossible to get 100 without some sort of cheating: resets, park the car, parking lot turns, etc

 

[valem]

View attachment 719728

These will be the guys testing out FSD beta

That's what I'm talking about. They should have had a minumum total mileage requirement. Very sad

 

[thewishmaster]

Should get up to 98 tomorrow. 99 is within reach but would take me another 200 miles which would be another couple weeks for me; it might take less time to wait until they expand the rollout to 98s

 

[AlanSubie4Life]

OK, so how to you get 100% score without cheating?

I don’t use the cheat. I have a 99.73. If you exclude the first two days I’d have a 99.92. They definitely should fix the cheat. It would probably be most fair to delay things another week, send out a software update that is required in order to log scores in the app, and give people seven days, resetting all scores to date. Then everyone is knowledgeable about the game, and it will come down to ability to do the miles cleanly. They should have a mileage minimum too, and a certain minimum number of days of driving (not necessarily 7). They could even rescale the scores or add the first decimal place and make the bar a little higher if they really wanted to. They're not going to do this of course, and that's fine.

I've only done "score repair" on a couple of early days. It's been over a week since I've had to worry about that. I'm ok with score repair though. Requires commitment, haha!

-- By driving very little (lowers chance of mistake, and formula doesnt seem to factor in mileage so far as we know)

-- By living in a very quiet area with few/no traffic signals, hills, other drivers etc.

I’ve driven 533 miles over 73 trips. I live in a hilly area. I am a bit careful about which routes I take now - there are certain hills with traffic lights that are outside my control for hard braking (video earlier), so I do not take them anymore. Driving with traffic signals is stressful - it’s the only place I’ve had hard braking events.

If you log a bunch of good miles it actually protects you from future errors. The problem in the early going is that your scores are quite sensitive to one-time events when you're talking about getting to a 100 score. They should have just stuck to a 30-day evaluation period prior to making the decision on admissions. My main mistake was driving so much (105 miles) the first day - but it wasn’t for the Safety Score - just had a lot of driving to do!

Does the car filter out gravity from it's accelerometer reading for braking and turning? Because it sure feels like it doesn't. Turning slowly around a high grade corner seems to ding aggressive turning.

No.

The scoring factors have been going down over time as I accumulate more "normal" braking, turning and following time ... but only mileage can improve the score... so what is the point?

Just for your information on overall performance. They don’t count in your score.

I wish this was mentioned to me before I did my 272 drive which was needed for me to reach 100… I did it mostly on AP…

So if I understood this correctly, if I were to get a cumulative 100 score on main page, I have to work on reducing all the percentage digs I had… example follow to closely and hard braking…

You’re fine. No worries. Just get miles at 100 each day.

it’s almost like they don’t know how to take an advantage of regen.

Maybe they are using FSD beta 100 score drivers to get clips for training AP. They had been training on Tesla drivers before, obviously. Now they’ll be training on Tesla drivers driving properly.

These will be the guys testing out FSD beta

I think Tesla has some wiggle room to make sure people have scores on seven days based on earlier Elon tweet. Though this requirement has definitely not been made explicit. And may not happen, and probably is not intended to be a requirement, nor was it meant to be implied.

It does seem like they should have had a 25 mile per week minimum or something. On the other hand these people may be low risk if this is their typical mileage.

 

[hamoneaster]

Honestly, I think all bets are off right now. I feel like elon won’t hold his word. ESP because 3+ extra days will add several hundred more 100s and that will be way more than they are willing to admit into beta first wave. So the rules will change.

 

[ConSaab]

Honestly, I think all bets are off right now. I feel like elon won’t hold his word. ESP because 3+ extra days will add several hundred more 100s and that will be way more than they are willing to admit into beta first wave. So the rules will change.

Most likely will be the case but we can all just try and cross our fingers…

 

[AlanSubie4Life]

I feel like elon won’t hold his word. ESP because 3+ extra days will add several hundred more 100s and that will be way more than they are willing to admit into beta first wave. So the rules will change.

I don't understand this. There's allegedly 2000 Tesla employees as testers. I doubt they're really any better than the people who have been selected from at least 10 thousand owners (it's probably more like 30-40k minimum, though no one knows how many people are really participating).

So why does it matter whether there are 3300 beta testers, or 4000? It all seems about the same. Obviously there is slight increased risk but it's not really that much.

It would definitely be better if the pool were composed of people who were only very rarely cheating, though. It's no longer clear how much that is the case.

I'd actually be ok with the ability to opt-in/opt-out on a drive-by-drive basis (and it may be possible already). This obviously has major issues as well (you might only be getting data from people on certain types of driving), but it's not as egregious as clearing out driver-inattention mistakes whenever they happen. Opt-in/opt-out would be best paired with some sort of minimum number of non-negligible trips (say more than 2 miles) or minimum mileage requirements because that would create an incentive to mostly stay opted in.

 

[ChimpledPot]

View attachment 719731
1435 miles to go because of that one day with false FCW lololol, this will never happen.

That was also like day 2 of the games, wish I knew about the reset trick. This sugar being beta seems impossible to get 100 without some sort of cheating: resets, park the car, parking lot turns, etc

This is exactly why I’m not concerned that the safety score can be gamed to get a 100 score. The type of people who are willing and able to do this are exactly the type of person who will be able to quickly figure out the pitfalls of FSD and adapt very quickly, and be the most conscientious users of FSD.