17.26.76

[oktane]

As one of those p85d peeps, I believe @oktane is perfecty fine talking through their disappointment, and even though they have a propensity for continuing to state this a tad too often for some, the lessons I can relate to as the reality of sadness when the Tesla tarnish occurs.
Things I generally recommend now to mitigate this are:
a) don't believe the accuracy of anything Tesla says ahead of it being independently verified (e.g. AP1, HP claims, battery capacity, etc). I don't believe they are purposely lying all the time, rather appear to be very creative with the realization of Elon's unencumbered enthusiasm as reality hits.
b) never evangelise, or perpetuate the evangelism from others, to anyone else for anything you haven't seen. This way lies the erosion of friendship and credibiiity
c) concentrate on getting your money back for functionality not available on the due sheet/Monteray sticker of the car - from @oktane 's previous posts they appear have more than enough to spend $100 or so and talk to an attorney on options assuming your have exhausted all options with Tesla. Either way, it's not worth stressing about any more. I sometimes wish I'd gone after Tesla for the missing HP, though it's plenty quick already.
d) Most importantly: remember you have a great car in its own right.

Tesla's mission is important and the early adopters helped move that forward. Sometimes it is hard for those who have been significantly invested to think of Tesla critically. Doubly so when combined with hindsight-biased historical recollection. Elon himself has stated that constructive criticism is key to making better products. This will be needed as the masses buy Model 3s.

Net: We can't presume poor intent when others don't see the world from your own perspective. I just wish I had the ability to think this objectively more often, rather than with emotional response

My 2c

Thanks for your wise post.

 

[MarcusMaximus]

I'm shaving with Occam's razor on this.

TomTom both has different geo-fences and different road-type categorizations than Garmin. This is true and can explain all the weirdness. I know absence of evidence isn't proof of absence, but there is nothing showing that HW2 reads signs nor is there any added explanatory value to thinking it does. I mean, I'd like it to be true that my HW2 car is getting more capable, but hope isn't enough.

I don't know if this is something resident awesome poster @verygreen can lend insight on or not.

What I saw yesterday was a little too accurate to be explained purely by a map change. The speed was perfectly synced with passing the sign. Like, down to the foot. I've never seen GPS be anywhere near as accurate as it'd have to be for what I was seeing. And not just once, it was perfect like that every single time.

I'd note that it's not totally outlandish that they'd use speed limit signs in this way. It'd make perfect sense if they can recognize speed limit signs pretty well, but have some difficulties with them, e.g. reading the actual speed, problems with false alarms(reading an address or something as the speed), or being able to consistently figure out what 25mph when children present or END 55 means. Triggering a map check to see what the speed is expected to be and just using the sign as a post for where to set it to that is always safe.

 

[alcibiades]

What I saw yesterday was a little too accurate to be explained purely by a map change. The speed was perfectly synced with passing the sign. Like, down to the foot. I've never seen GPS be anywhere near as accurate as it'd have to be for what I was seeing. And not just once, it was perfect like that every single time.

I'd note that it's not totally outlandish that they'd use speed limit signs in this way. It'd make perfect sense if they can recognize speed limit signs pretty well, but have some difficulties with them, e.g. reading the actual speed, problems with false alarms(reading an address or something as the speed), or being able to consistently figure out what 25mph when children present or END 55 means. Triggering a map check to see what the speed is expected to be and just using the sign as a post for where to set it to that is always safe.

I have experienced the "instant" change on certain spots ever since the update to 17.24 -- but that's because of precision of the database. There are other spots where it is completely off routinely -- because the database isn't as precise there yet. Call TomTom. I did.

 

[alehbaba]

why do some ap2 cars have 17.17.17? Shouldnt the update come over wifi for the car? That seems way out of date.

 

[SomeJoe7777]

I test drove AP1 cars extensively and was impressed. There was no way for me to reasonably conclude or know that AP2 would be starting again from zero, rather than building on an already functioning (great) product.

This I think is the crux of the matter. You and others fall into a narrow sales window where you were researching and shown AP1 cars and were delivered a (non-working) AP2 car. Then you were told that the functionality you saw during the test drive would be delivered with your new car by the end of 2016.

For this, I am empathetic. Indeed, I was well aware of the shortcomings going in, but you thought you were past that period.

I do feel that you will eventually receive everything you wanted, albeit many months later than you were told. And, I think you will have the most technologically advanced vehicle on the road for years to come as the improvements keep coming.

 

[MikeBur]

After a longer drive yesterday on the 91 and 15, I would call 17.26.76 the Rodeo version. It's the first release to do bucking bronco style turns and acceleration/deacceleration. Everything is instantaneous with jerky motion as the car ahead slows and speeds up or lane lines change. This isn't progress, it's now a discount amusement ride. What was Disneyland quality has dropped to a carnival. At this rate, I expect a future version to feel like a child's rocket ride at the supermarket.

I am feeling more sad than amused. I truly don't know how anyone can say AP2 is close to AP1 across the board, though my immediate compare/contrast was significant and makes me happy I've stuck with AP1, though If not addressed in next 6 months will definitely go in the minus column when considering upgrade.

From couple of days, few hundred miles, X90D AP2 loaner compared to p85d AP1: auto steering on interstates/freeways, yes it's very similar. Though the lane changing, random false-positives on early earning/phantom cars/etc scared the bejesus out of me and I couldn't trust it... maybe the loaner was calibrating still, though it was a couple of days and hundreds of miles.
I'm unafraid of treating AP as a test vehicle (from earliest videos of initial updates and videos). This just felt... "random", aka unpredictable to me. I really, really, want to see this improve significantly soon

 

[jddssc121]

Did roughly 500 miles over the last couple of days with AP2, 17.26.76

My observations are mostly the same as other posters...

• Highway auto steer is very good; set and forget at this point and probably the same as AP1
• There’s still some occasional jitter sometimes in the longitudinal control when going 85/90mph - it’ll sometimes slow ever so slightly, seemingly for nothing at all. It causes the passengers to look at me and go “is she ok?”... it’s only a 1 or 2 mph slow down, or hesitation every so often, but it is noticeable
• Lane changing is too abrupt, a bit startling. It also doesn’t really come out of the change very nicely - it could be better. Also, I’ve noticed that if there’s a follow car that changes out of your lane in front (i.e you’ll be able to pass), the car will actually slow down a bit whilst that car in front changes lanes. I suppose this might be intentional, but it’s a slightly odd feeling.
• I wouldn’t describe the longitudinal control behaviour in traffic jams as silky smooth - it can be a bit heavy on the acceleration and braking; not as smooth as I’d drive it, and could make some passengers a bit nauseous. But, it’s not terrible - perfectly acceptable.
• Lane holding is great now - doesn’t seem to bias as much to the side. It could still be smarter here, but it’s holding the center line really nicely the vast majority of the time
• Steering round curves is great - vastly improved
• Ghost braking under bridges seems gone; no sudden brake events at all on this trip - thank goodness.
• Longitudinal control at highway speeds is pretty smooth overall, bar the occasional jitter here and there.

Overall, very happy. It’s not quite as smooth and solid as AP1 still, but it’s got to be really, really close at this point. I’m excited for the additional improvements from now on - I’m sure there’ll be more tweak updates, but the next major update could mark the diversion point of AP vs EAP feature sets, perhaps true automatic lane changing (rather than turn signal activated)? I also expect quite major auto-steer improvements when they activate the front facing wide angle camera, though I expect we won’t see that until August, given that they only recently started gathering imagery/video from us, and will want to gather lots of data in all conditions (I’m not sure many Tesla’s will have seen lots of snow yet) to train the vision models - a process that can take a couple of weeks of number crunching anyway, depending on the dataset size.

One thing I'd add. The lane change, apart from being abrupt, is also dangerous. I was curious how well the side monitoring was working, so I experimented a few times when i had a car next to me. I held the wheel firmly and then turned on the blinker to lane change. 3 out of 4 times it would have changed lanes into a car directly next to me had a not kept the car straight..... unacceptable.

 

[NerdUno]

@SomeJoe7777 said: "I do feel that you will eventually receive everything you wanted, albeit many months later..."

My money is on years, not months, given the pace of progress on the AP2 rollout. The Model 3 buyers are in for quite a surprise. If you think some of us are inpatient, just wait...

 

[jammertwo]

With this update, and coming from 17.17.17, I've found autopilot is more responsive on local roads (haven't had a chance to test it on the highway). The previous version would put the vehicle over the line before moving back on sharper turns. With this one, the vehicle stays in the lane. It's super jerky at times but does a much better job of keeping the vehicle in the lane than the previous version.

That said, it's still more of novelty on local roads since it's not a pleasant driving experience with all the jerking around. Getting there!

Tried to get perpendicular to work but no luck yet.

 

[mrkisskiss]

One thing I'd add. The lane change, apart from being abrupt, is also dangerous. I was curious how well the side monitoring was working, so I experimented a few times when i had a car next to me. I held the wheel firmly and then turned on the blinker to lane change. 3 out of 4 times it would have changed lanes into a car directly next to me had a not kept the car straight..... unacceptable.

Ah yes - people should know that there is no monitoring - it's up to the driver to decide when it's safe to change lanes. Otherwise, it'll just change lanes whenever, assuming there's nothing detected by the ultrasonics (which aren't good to rely on).

This is partly why I'm wondering if one of the first EAP features we'll see will be true automatic lane change, using the rear and side cameras to verify if it's safe or not to proceed, then automatically activating the turn signal and changing lanes for you when a slow car is detected in front of you, or if there's no car in the other lane (i.e keep the fast lane free whilst keeping you at your set speed). This kind of tech has been demonstrated by Audi already, and looks pretty useful!

 

[SomeJoe7777]

The Model 3 buyers are in for quite a surprise. If you think some of us are inpatient, just wait...

I'll be in that group as well. I have two M3's reserved. I am preparing myself for the inevitable bugginess.

 

[whttiger25]

I mostly agree with the first part of your statement in that bullying is unacceptable. Whether oktane has really been "bullied" in this thread is debatable, however.

But the 2nd portion of your statement above comes across as entitled. You Q4 2016 purchasers seem to believe you're the only ones who didn't get what you were promised from Tesla. That's nonsense. There were thousands of cars sold from Q4 2014 to Q3 2015 with AP1 hardware and no autosteer. October 2014 P85D purchasers waited a year for any autosteer functionality whatsoever, and what arrived in October of 2015 wasn't all that great. Only in the last 6 months has AP1 become the gold standard reference to which everything else is compared (for me, I pinpoint this moment as firmware 2.52.22 in January of 2017).

So, for an October 2014 P85D purchaser, they waited a year for any autosteer functionality, and another 15 months for something fairly solid and trustworthy. That's 27 months of AP1 software development, and half of the problem had already been implemented in hardware by Mobileye. Meanwhile, you Q4 2016 purchasers have waited 9 months and have 90% of that same functionality.

What that adds up to is that there are people who waited three times longer than you have for their paid-for feature, yet you guys are on this forum complaining way louder than they ever did. That comes across as entitled.

I knew what I was getting into when I decided to purchase my 85D, which was delivered April 2015. I waited 6 months for autosteer, and another 15 for 2.52.22. And I've seen the advances that Tesla has made during that time, including the refresh body style, AP2 hardware and software, glass roof, more powerful CPUs, bigger batteries, etc. I knew going in that my car would have bugs, missing functionality, and be surpassed in technology and capability almost immediately. My choice under those conditions was reduce all of that risk by leasing the car, and that choice is going to pay off for me. In April of 2018 I will get a new one, with all the latest stuff (whatever it may be at that point).

What surprises me is that some of you Q4 2016 purchasers who are so vocal on this forum expected something different. When I researched my purchase, it was clear to me that all of those above-mentioned items were risks that come with a revolutionary new car design, and a brand new car company that hasn't found their footing yet. Was this not clear to you? Something is different about you guys that I'm having trouble figuring out, because other Tesla buyers have had it worse, yet they aren't as vocal as you and they still love their car.

I think the problem people have isn't entitlement, it's regression. Tesla doubled the price and delivered a product half as good as what was already out there.

I just installed the update and am looking forward to trying it on my way home from work. I drove an AP1 car for 10 months, and loved it. I've had my AP2 car for a month and haven't felt comfortable using it for more than a minute at a time, until it does something unnerving and I decide to just turn it off and pilot the car on my own.

Perhaps they should have kept the price at $2500 until parity was reached?

 

[chillaban]

FWIW, I'm a tesla fan and satisfied AP1 -> AP2 upgrader despite only recently seeing ANY instances at all where AP2 was better than AP1.

I definitely criticize Tesla for the way EAP/AP2 was represented at announcement. They did warn that the cars will not ship with any safety features or automation features, but claimed that was because the software was pending validation and that EAP was to roll out in December pending validation.

Most reasonable people read that to mean that the software is basically ready to go, but needs bake time / real world proving. But what we saw and what Tesla eventually admitted was that basically at time of announcement, there was zero work done, and they've been racing to rebuild AP from scratch since then. This seems highly dishonest to me and I completely understand why the early adopters of AP2 (or those who got volunteered into an AP2 upgrade at time of order) feel deceived.

I wish Tesla wouldn't do business this way. I speak for myself but I bet there are others that feel this way: If Tesla was up front about this, I would still have viewed Tesla favorably. If they would've said "MobileEye and us decided to head separate ways rather abruptly, we believe this new hardware is superior in capability but the software will take months to reach parity"… I think they would've had more support from their customer base. Most of the folks here are fans of both the products and the company's long-term mission.

 

[whttiger25]

FWIW, I'm a tesla fan and satisfied AP1 -> AP2 upgrader despite only recently seeing ANY instances at all where AP2 was better than AP1.

I definitely criticize Tesla for the way EAP/AP2 was represented at announcement. They did warn that the cars will not ship with any safety features or automation features, but claimed that was because the software was pending validation and that EAP was to roll out in December pending validation.

Most reasonable people read that to mean that the software is basically ready to go, but needs bake time / real world proving. But what we saw and what Tesla eventually admitted was that basically at time of announcement, there was zero work done, and they've been racing to rebuild AP from scratch since then. This seems highly dishonest to me and I completely understand why the early adopters of AP2 (or those who got volunteered into an AP2 upgrade at time of order) feel deceived.

I wish Tesla wouldn't do business this way. I speak for myself but I bet there are others that feel this way: If Tesla was up front about this, I would still have viewed Tesla favorably. If they would've said "MobileEye and us decided to head separate ways rather abruptly, we believe this new hardware is superior in capability but the software will take months to reach parity"… I think they would've had more support from their customer base. Most of the folks here are fans of both the products and the company's long-term mission.

To me, it seems like the best possible way Tesla could have handled this situation would have been to suspend the $1000 post delivery markup on EAP for all vehicles that were ordered while AP2 was not on par with AP1. That way, we could have ordered our cars without it, and waited until the product was actually worth $5000 to upgrade our vehicles without penalty. Only those thin on financing would be affected.

 

[chillaban]

To me, it seems like the best possible way Tesla could have handled this situation would have been to suspend the $1000 post delivery markup on EAP for all vehicles that were ordered while AP2 was not on par with AP1. That way, we could have ordered our cars without it, and waited until the product was actually worth $5000 to upgrade our vehicles without penalty. Only those thin on financing would be affected.

I feel like any level of financial incentive won't make up for the fundamental feeling of deceit in the way that Tesla originally announced/advertised AP2/EAP features. There's already a class action lawsuit of plantiffs who feel they would not have bought Teslas at all if they had known of the state of AP2, so just mild goodwill gestures of package discounts would not make those people happy.

Whether or not the packages are priced appropriately and whether early adopters have been "rewarded" by enough incentives, IMO, is a separate discussion from whether or not it was right of Tesla to paint a rosy picture of EAP in the first place. Those of us that have been following Tesla for a long time seem to know their MO and feel inherently wary of a new feature announcement before early adopters take the bite. But, it's worth remembering that is not at all how the rest of the automotive industry works. Their advertised features for a particular model year's press release truly represent what the car is already capable of, at time of delivery. Of course the concepts they present at auto shows are a completely different story, but customers understand that difference between concept car and new model year ordering catalog.

 

[MarcusMaximus]

If Tesla was up front about this, I would still have viewed Tesla favorably. If they would've said "MobileEye and us decided to head separate ways rather abruptly, we believe this new hardware is superior in capability but the software will take months to reach parity"… I think they would've had more support from their customer base

Some of them, at least. The people now complaining, and claiming fraud, over FSD not being here 6 months ahead of any time suggested by anyone at Tesla likely wouldn't have felt much different regardless of statement of a timeline.

 

[whttiger25]

I feel like any level of financial incentive won't make up for the fundamental feeling of deceit in the way that Tesla originally announced/advertised AP2/EAP features. There's already a class action lawsuit of plantiffs who feel they would not have bought Teslas at all if they had known of the state of AP2, so just mild goodwill gestures of package discounts would not make those people happy.

Whether or not the packages are priced appropriately and whether early adopters have been "rewarded" by enough incentives, IMO, is a separate discussion from whether or not it was right of Tesla to paint a rosy picture of EAP in the first place. Those of us that have been following Tesla for a long time seem to know their MO and feel inherently wary of a new feature announcement before early adopters take the bite. But, it's worth remembering that is not at all how the rest of the automotive industry works. Their advertised features for a particular model year's press release truly represent what the car is already capable of, at time of delivery. Of course the concepts they present at auto shows are a completely different story, but customers understand that difference between concept car and new model year ordering catalog.

I definitely meant they should have done that at the very beginning when advertising the system. I.e., make the same statement you presented above about time to parity from the very beginning, AND offer than $1000 waiver as well. It's definitely too late for Tesla to fix this one and as we approach parity it will become a moot point. However, they have certainly damaged their relationship with many customers, and as the EV landscape becomes more competitive they may suffer from that as we have more and more EV options in the future.

And meanwhile, I'm still waiting on being able to listen to my USB music in track order, not alphabetical, a regression bug introduced in 8.0 and one that I've emailed them about after every single firmware update. It's pretty relevant for many genres of music but especially classical and electronic, two of my favs. Maybe I'm OCD on this but I think I may be more upset about that one than AP2 parity, lol. I mean...it's SUCH an easy bug to fix, and they did manage to program the easter egg module in there while this bug remained unaddressed.

Gapless playback would also be nice.

 

[MarcusMaximus]

I definitely meant they should have done that at the very beginning when advertising the system. I.e., make the same statement you presented above about time to parity from the very beginning, AND offer than $1000 waiver as well. It's definitely too late for Tesla to fix this one and as we approach parity it will become a moot point. However, they have certainly damaged their relationship with many customers, and as the EV landscape becomes more competitive they may suffer from that as we have more and more EV options in the future.

And meanwhile, I'm still waiting on being able to listen to my USB music in track order, not alphabetical, a regression bug introduced in 8.0 and one that I've emailed them about after every single firmware update. It's pretty relevant for many genres of music but especially classical and electronic, two of my favs. Maybe I'm OCD on this but I think I may be more upset about that one than AP2 parity, lol. I mean...it's SUCH an easy bug to fix, and they did manage to program the easter egg module in there while this bug remained unaddressed.

Gapless playback would also be nice.

Also, keeping multi-artist albums, e.g. Soundtracks, together. I have like 10 different "Moana" albums right now(I have a 2 year old)

 

[MikeBur]

FWIW, I'm a tesla fan and satisfied AP1 -> AP2 upgrader despite only recently seeing ANY instances at all where AP2 was better than AP1.

I definitely criticize Tesla for the way EAP/AP2 was represented at announcement. They did warn that the cars will not ship with any safety features or automation features, but claimed that was because the software was pending validation and that EAP was to roll out in December pending validation.

Most reasonable people read that to mean that the software is basically ready to go, but needs bake time / real world proving. But what we saw and what Tesla eventually admitted was that basically at time of announcement, there was zero work done, and they've been racing to rebuild AP from scratch since then. This seems highly dishonest to me and I completely understand why the early adopters of AP2 (or those who got volunteered into an AP2 upgrade at time of order) feel deceived.

I wish Tesla wouldn't do business this way. I speak for myself but I bet there are others that feel this way: If Tesla was up front about this, I would still have viewed Tesla favorably. If they would've said "MobileEye and us decided to head separate ways rather abruptly, we believe this new hardware is superior in capability but the software will take months to reach parity"… I think they would've had more support from their customer base. Most of the folks here are fans of both the products and the company's long-term mission.

Absolutely agree. It may have been a short term hit, though under-promising and over-delivering is something that appears to be more of an optional aspiration now, rather than a principle.
Saying that, I still believe some of this was extremely poor communication to the sales and service center teams. The only way someone would *know* if this was purposeful deception would be through legal means, unfortunately. Tesla was, and should be, better than this.

 

[whttiger25]

Also, keeping multi-artist albums, e.g. Soundtracks, together. I have like 10 different "Moana" albums right now(I have a 2 year old)

I've always encountered many issues in systems entirely dependent on ID3 tags to organize music. It's my conclusion, personally, that none of these systems will ever do exactly what I want. This is why I appreciated folder view in version 7.x. I could just navigate the folders I've organized to find the exact album I wanted. In version 8.0 it was great that they added the full tag based database for situations when I wanted to search for something specific, but then then messed up folder view. Ugh.