Wiki MASTER THREAD: Actual FSD Beta downloads and experiences

[coreylista]

Well so much for my 99% score!!

5 perfect days of driving... suddenly this... Not a single warning/alarm/alert/disengagement... NOTHING... FCW?!?! and hard braking?!?! I've learned to only let regen slow down the car, I never use the brakes... 99% of this drive was in AP!! I live about a half a mile away from the highway, and I work about a half a mile from the exit...

Such a stupid short 15 mile drive, everyday, but its been perfect... now suddenly I get hit with what most certainly is an error by the system. All to screw up my chances of getting the beta tomorrow...

 

[nvx1977]

So for those wondering about the process of being kicked off of FSD Beta. Looks like you get one written warning, after which point it's removed:

https://www.reddit.com/r/teslamotors/comments/qcka24

I'm fully supportive of this. I wonder if there are other triggers that can get you kicked out.

 

[impastu]

Well so much for my 99% score!!

5 perfect days of driving... suddenly this... Not a single warning/alarm/alert/disengagement... NOTHING... FCW?!?! and hard braking?!?! I've learned to only let regen slow down the car, I never use the brakes... 99% of this drive was in AP!! I live about a half a mile away from the highway, and I work about a half a mile from the exit...

Such a stupid short 15 mile drive, everyday, but its been perfect... now suddenly I get hit with what most certainly is an error by the system. All to screw up my chances of getting the beta tomorrow...
View attachment 723922

Put on extra miles today...you can get those FCW and HB scores averaged way up...maybe back to 99% for the day.

 

[wtwieder]

Agreed. That describes my last FSD experience theee days ago. Don’t plan on letting it drive again until 10.3; and that may be short lived as well. By software development methodology definition, “beta” software would never be this unrefined. This is Alpha at best, running on the development server, not production server.

I totally agrer. In fact, I had the best day yesterday, in almost 2 weeks of driving with 10.2. Why? I turned FSD beta off and drove the entire day without it. What a relief: No heart pounding accelerations during turns, no hard braking, no steering wheel jerking back and forth in my hands, no interventions needed to keep FSD from killing me, etc.

I am glad to have had a chance to have driven with 10.2 but see no reason not to return to the safe driving habits that easily gave me the perfect safety score needed to get 10.2. I missed not driving safely. Furthermore, I love my M3 and worry about all the wear and tear on it, not to mention the possibility of it getting damaged in an accident,.

Moreover, I think I have given enough feedback to Tesla in 10 days that show my issues with 10.2. To continue reporting the same issues is just repetitive. I will turn FSD beta 10.3 on again when it is downloaded, but don't intend to use it for more than a few days if none of my concerns have been fixed in any consistant way.

To put all this another way, I realized when accepting the job of being a test pilot, I was accepting risks. But to keep flying with the same plane, repeating the same dangerous flights over and over again, doesn't make sense to me. However, I am happily willing to test fly a new plane, 10.3, when it gets delivered. Hopefully it will be better.

 

[nvx1977]

What does the second thing mean? Sometime at a stop light I look at my phone and when he green light comes I get an alert to keep my eyes on the road even though the car has barely rolled any… is that what they are talking about?

my interpretation is that this purely AP warnings to pay attention. We now know that certain behaviors detected by the cabin camera can cause an AP warnings, so I would assume all of those warnings are potentially "Strikes". Seems like you can maintain 1 strike per 3 miles without earning a "strikeout?"

AP jail is an automatic strikeout. And I suppose if you pick up 3 of those (in one drive? in one day? ever?) then you're out. The message is still not super clear on how it kicks you out. Better err on the side of caution and minimize any sort of AP nag.

Anecdotally I was doing a long highway slog yesterday and did get the apply torque to the wheel message a lot. Not sure if I got more than one in 3 miles, but I'll be a bit more vigilant about that from now on.

 

[nvx1977]

I totally agrer. In fact, I had the best day yesterday, in almost 2 weeks of driving with 10.2. Why? I turned FSD beta off and drove the entire day without it. What a relief: No heart pounding accelerations during turns, no hard braking, no steering wheel jerking back and forth in my hands, no interventions needed to keep FSD from killing me, etc.

I am glad to have had a chance to have driven with 10.2 but see no reason not to return to the safe driving habits that easily gave me the perfect safety score needed to get 10.2. I missed not driving safely. Furthermore, I love my M3 and worry about all the wear and tear on it, not to mention the possibility of it getting damaged in an accident,.

Moreover, I think I have given enough feedback to Tesla in 10 days that show my issues with 10.2. To continue reporting the same issues is just repetitive. I will turn FSD beta 10.3 on again when it is downloaded, but don't intend to use it for more than a few days if none of my concerns have been fixed in any consistant way.

To put all this another way, I realized when accepting the job of being a test pilot, I was accepting risks. But to keep flying with the same plane, repeating the same dangerous flights over and over again, doesn't make sense to me. However, I am happily willing to test fly a new plane, 10.3, when it gets delivered. Hopefully it will be better.

Yeah it would be good to know how/when our telemetry is used. It could be our current driving is being used for training for 10.4 or beyond. Hard to know. I don't think I'll start sandbagging just because a new release is coming out. I may stop actively submitting AP snapshots and just rely on certain disengagements to be what Tesla is looking for.

I tend to use FSD less when I have passengers (unless they want me to). When I don't use it, I drive like I was going for 100 safety score. Not because I feel like I need to, but to show the passengers how far FSD needs to come before it can drive buttery smooth and let everyone take a nap without being jolted awake every few seconds.

 

[Phlier]

The car was tailgating the vehicle ahead on the freeway,

Did you adjust your follow distance to 7?

 

[queue]

I'm fully supportive of this. I wonder if there are other triggers that can get you kicked out.

I am too, but with all the false positives in the Safety Score Beta, I'm a little concerned about how sensitive this is or isn't. And not publishing full criteria will end up just creating a lot of noise in the testing community.

 

[jebinc]

I totally agrer. In fact, I had the best day yesterday, in almost 2 weeks of driving with 10.2. Why? I turned FSD beta off and drove the entire day without it. What a relief: No heart pounding accelerations during turns, no hard braking, no steering wheel jerking back and forth in my hands, no interventions needed to keep FSD from killing me, etc.

I am glad to have had a chance to have driven with 10.2 but see no reason not to return to the safe driving habits that easily gave me the perfect safety score needed to get 10.2. I missed not driving safely. Furthermore, I love my M3 and worry about all the wear and tear on it, not to mention the possibility of it getting damaged in an accident,.

Moreover, I think I have given enough feedback to Tesla in 10 days that show my issues with 10.2. To continue reporting the same issues is just repetitive. I will turn FSD beta 10.3 on again when it is downloaded, but don't intend to use it for more than a few days if none of my concerns have been fixed in any consistant way.

To put all this another way, I realized when accepting the job of being a test pilot, I was accepting risks. But to keep flying with the same plane, repeating the same dangerous flights over and over again, doesn't make sense to me. However, I am happily willing to test fly a new plane, 10.3, when it gets delivered. Hopefully it will be better.

@wtwieder Exactly. Same here. Tesla has plenty of feedback from me, re the 10.2 release.

 

[FSDtester#1]

@wtwieder Exactly. Same here. Tesla has plenty of feedback from me, re the 10.2 release.

I just put my Plaid in the shop for about a week for full PPF, so there won't be anymore butt clenched driving for a bit. I think I already grew back 4 new hair follicles I lost in the last 2 weeks

 

[Bill Price]

I am too, but with all the false positives in the Safety Score Beta, I'm a little concerned about how sensitive this is or isn't. And not publishing full criteria will end up just creating a lot of noise in the testing community.

Yes, the false positives are problematic to be sure. But of course, the SS IS beta they say, but I'm thinking more alpha, or pre-alpha? The one MAIN problem I have with SS dings is the hard braking. You're driving along and some typhlotic driver pulls right out in front of you. How does it make you a bad driver if you jump on your brakes in order to avoid a collision that would have been caused entirely by the other driver? It's a catch-22 situation and I choose to use my brakes to avoid these clueless drivers, which mitigates a collision and the subsequent fallout of the collision, such as time spent at the collision site asking the other driver for pertinent information (insurance info) and just why they failed to see you and pulled out in front of you, time spend with the police officer, time spent with the insurance adjusters, and time spent at a collision center having your car fixed. Of course, I could have just slammed into the other cars/trucks and proven in court that the other party was at fault because of the video saved by TeslaCam. I have avoided no less than three collisions in the past month due to driver inattention, precipitancy, old age, or a combination thereof. I fail to see the logic that avoiding these collisions precipitated completely by an unsafe driver, makes ME an unsafe driver.

 

[jebinc]

I just put my Plaid in the shop for about a week for full PPF, so there won't be anymore butt clenched driving for a bit. I think I already grew back 4 new hair follicles I lost in the last 2 weeks

@Ryan27 Yup, I took the Plaid out for a non FSD and spirited drive yesterday; nary a worry or care for safety score or anything else. Best drive in the past six weeks; like getting a new car!

 

[EVNow]

So for those wondering about the process of being kicked off of FSD Beta. Looks like you get one written warning, after which point it's removed:

https://www.reddit.com/r/teslamotors/comments/qcka24

Who are these knuckleheads ?

I'm hanging on dearly for my life clutching the steering wheel when driving on Beta.

BTW, my last night drive was slightly "better". Had to disengage only because of a sudden slowdown and a round-about. No interventions otherwise. Even that slow down happenned almost at the foot of the hill - we almost made it through the twisted road following a car without FCW. Interestingly, I never get FCW/sudden slowdowns on the same road going uphill.

Google Maps

Find local businesses, view maps and get driving directions in Google Maps.

www.google.com

 

[mxnym]

Zach says 150k cars are using safety scores now. Thats a good proxy for # of US customers who have purchased/subscribed to FSD !

I'm not sure who Zach is, but as another user already mentioned, Texas insurance users may count. Also, since Zach doesn't appear to have said "150k more cars," California insurance users, too.

Congratulations! I’ve seen reports that the app calculates and indicates score over 30 days, but the algorithm to grant beta access is a 7 day aggegrate. Yet several people have reported getting the beta simultaneous with attaining a 100 score. Not sure what the algorithm really is.

Not sure if I will ever get it; every couple of days my score gets dinged due to hard braking, despite us never touching the brake pedal. Regen is enough to trip it.

Meanwhile, driving like grandpa, sitting at a 98 in the app, 99 if I hand calculate a rolling 7 day average. We’ll see what happens at the end of the week, providing wife doesn’t mess up the score….;-)

I'm not sure why people assume Elon is so literal in his tweets (for instance the lack of an "at least" before "99/100" doesn't automatically imply "rounds to 99 [and not 100] in at least 100 miles" even if "100/100" meant "rounds to 100 in at least 100 miles, which isn't necessarily what that meant, either). Even if Elon would generally be super literal, Twitter's character limit may restrict his ability/desire to do that. I think it's another KISS thing and the average over (up to) the last 30 rolling days is what is used. I would have originally assumed that you needed at least 7 days before being granted access, but a post in the last page or two mentioned getting it within 2-3 days of enrolling, so the 7 days may not be relevant at all now even if it was originally. Also, FWIW, my safety score over the last 12 days would be 100 (no dings at all) over 681 miles, but I haven't been offered the beta yet because it's 99 from day 1 (or at least I'm assuming that's why, 2017 X100D upgraded from MCU1/AP2.5 to MCU2/AP3).

 

[EVNow]

I'm not sure who Zach is, but as another user already mentioned, Texas insurance users may count. Also, since Zach doesn't appear to have said "150k more cars," California insurance users, too.

Zach is the king of coin, Tesla (a.k.a CFO).

Yes, I understand it might include TX/CA insurance customers. There will be a significant overlap of those with FSD owners - but - more importantly I don't think the number of insurance customers is anything significant. Zach has always dodged the question about uptake in insurance and made it sound like an experimental small venture.

 

[MP3Mike]

I'm not sure who Zach is, but as another user already mentioned, Texas insurance users may count. Also, since Zach doesn't appear to have said "150k more cars," California insurance users, too.

No, California insurance users don't count as Tesla isn't allowed to use the Safety Score for the insurance products offered in California.

 

[EVNow]

No, California insurance users don't count as Tesla isn't allowed to use the Safety Score for the insurance products offered in California.

But - are their security scores still collected / calculated ?

 

[ahoen117]

My beta FSD experience today was the worst day I’ve had so far - and on well marked lightly traveled rural roads. Phantom braking, missed stop signs, full on with stopped school busses, wild yoke shaking back and forth, crossing over the center line, driving on the left side of a private road as if it were in the UK, etc. Very poor showing; not sure what changed. Will likely wait for 10.3 before I do any more beta FSD driving. Plaid time, until Saturday!

Yesterday on my way to work I had *horrible* phantom braking for seemingly no reason. Ended up manually taking over for the rest of the drive. While at work I got to thinking that it had been a few weeks since I've been to the car wash. Sure enough my right B pillar camera was filthy. I went ahead and cleaned all of them including the windshield cluster with a microfiber/water which seemed to fix the issues I was having.

 

[aronth5]

What does it matter if FSD is in beta or not?
It could go to widespread usage like the other beta products. It would still be out there.

Because Tesla is demonstrating to all potential regulators they are implementing a controlled rollout with the "Safety Factor" a key metric in determining who is eligible to participate as a beta tester. It's actually an very smart approach whether owners agree or not. Risk Mgmt 101. At least this approach provides Tesla some cover when accidents happen which of course everyone knows are coming. A wide open beta rollout would be a free-for-all that Regulators to limit or stop FSD.

Now people can argue the validity of the safety factor scoring methodology but Tesla can mitigate that by improving how the scoring is done which I suspect they are already working on. Remember they want to rollout insurance nation wide and having an improved safety factor tied to insurance rate will help profit margins for the program.

 

[aronth5]

Yesterday on my way to work I had *horrible* phantom braking for seemingly no reason. Ended up manually taking over for the rest of the drive. While at work I got to thinking that it had been a few weeks since I've been to the car wash. Sure enough my right B pillar camera was filthy. I went ahead and cleaned all of them including the windshield cluster with a microfiber/water which seemed to fix the issues I was having.

Now just imagine what will happen with a heavy wet snow storm? For all the promise FSD has, camera maintenance with the exception of the front camera is a serious flaw in design. Perhaps intentional though to mitigate cost. Last winter there were times when my "Y" cameras were covered with snow/ice while driving even with the camera "defroster/heat" supposedly working. Hopefully newer VIN's are better.