Todd Burch
14-Year Member
Agree. I cannot fathom the planet in which v12 is only marginally better than v11.Very interesting. Not my experience at all. I have 12.3.3 and it is far and away better than version 11.
You can install our site as a web app on your iOS device by utilizing the Add to Home Screen feature in Safari. Please see this thread for more details on this.
Note: This feature may not be available in some browsers.
Agree. I cannot fathom the planet in which v12 is only marginally better than v11.Very interesting. Not my experience at all. I have 12.3.3 and it is far and away better than version 11.
Replaced a 10 year old Renault Grand Scenic diesel MPV with a M3P off the first boat to the UK in 2019. Transformed my life ! Still grinning from ear to ear. (already had Tesla shares since the first landing of Falcon 9 blew my mind)For years, Tesla has been getting deeper into people's wallets than legacy car companies can. When I bought my Model S in 2013, I had been driving a 2002 Civic for 11yrs and a Hyundai Elantra was one of my options for replacing it. Then my TSLA stock took off, and instead of leaving it in my wallet, I bought a Tesla instead of another gas-guzzler
Yup. Still seeing 2024 massive luxury gas guzzlers driving around, and I am just speechless.While I agree with your points, the very same people have less problems buying other (expensive) cars. This is not only the affordability vs EV issue but also awareness issue.
If some of those folks knew what we know, they'd buy more Tesla cars. 9/10 still have no clue, IMO.
Those should really be taxed out of existence.Yup. Still seeing 2024 massive luxury gas guzzlers driving around, and I am just speechless.
I will reiterate that we are talking about the US here. China and Europe don't seem to be having much trouble.These are very reasonable points. What do you recommend we do to “cross the chasm” and reach the next phase of EV adoption?
Not convinced by this. SAAS companies deserve a higher multiple because they may scale very quickly without any incremental costs. If we are talking FSD for Teslas alone this is not the case, since the max number of FSD subscription is still the same as the number of Teslas with the right hardware. So the long term growth potential should be the same as sold cars (until and unless they license FSD to others). Also, I am not sure about the take rate (even at 99 USD). It seems a lot of people don't even want to buy a car if the monthly cost is slightly higher due to interests rates, which makes me wonder how many are prepared to pay over a thousand USD per year just for added convenience. It seems to me a "nice to have" but not a "need to have". Lastly, I cant see how his excel is set up and I didn't double check it but just from how he describes it, it seems he mixes revenues with profits.Good video by Yashu.
Meanwhile, The Voice of Wall Street, Bary Glack forgets the very basic, fundamental principles of stock valuation
After been reading and watching enviously for a couple of weeks about the hype of V12 I finally installed V12.3.4 yesterday coming from V11. Believing in the hype I was ready to invest more in TSLA, but having experienced the new version I will hold off for the moment.
Compared to V11 I find V12 marginally better. It accelerates and steers a bit smoother and has not stopped randomly in the middle of the road like I had version 11. Also, the wipers seem to work much better.
That being said; it still does the most basic stuff wrong:
- use of indicators. Why is it so hard to turn on indicators when you are entering and exiting roundabouts. Well for FSD it is hard.
- it turns on the indicator because it needs to move to a right lane - that is correct, however, it does it also right before a street on the right side. Drivers in that street think you might be turning right while you are not.
- choise of lanes is odd. For instance when there are two lanes before a traffic light and the left lane is for straight ahead only and the right lane for turning right and straight it is courtesy to use left lane if you want to go straight and leave the right lane for cars that want to go right on red. Not so for FSD
- totally ignores school zones or playgrounds.
- it still has some (slight) phantom braking
One can say that these things might be easily solvable, especially the indicators at roundabouts. If that is so, why haven’t they done that yet? I have used version 10, 11 and 12 and still these things are not solved.
Don’t get me wrong. I like FSD and I use it all the time. But not being able to do the most basic and mondaine things, as mentioned above right, after these many versions — I don’t see it turning into an autonomous system any time soon if ever. I will keep my cash for now.
Interesting. I wonder if these issues would be less obvious or even existing if (i) all cars were on FSD so you would not need to drive "aggressively" because everybody else does (thinking about driving in for instance Italy where you will get stuck if you are too courteous) or (ii) FSD vehicles had better map and fleet awareness, so they were able to better plan their drives. Because as far as I understand, correct me if wrong, FSD is precisely as you describe just reacting to the things instantly around it.I have similar issues. While it is cool tech I still need to disengage multiple times on my 7 mile commute. Especially on the way home out of the city, I don’t even try using it anymore in this situation as it is useless and causes to many problems.
I am not sure how they would even fix the issues I have on my commute home. To describe it, it’s almost like someone that is a decent driver but has never driven in my city, therefore, it doesn’t anticipate which lane it needs to be in to not get stuck in the wrong lane during rush hour traffic. I could totally see a human new to my city driving during rush hour doing the exact same thing.
However, I did drive through a school zone yesterday that had a flashing light indicating a lower speed and it slowed down as I would have. I will drive through there again not during school hours to see if it slows down or maintains speed.
I guess it depends on how much you value safety. One prevented accident at 70+ mph is worth at least the full price of FSD. This is especially true as you get older and your reflexes slow. I consider FSD a "need to have" on road trips and a "really good to have" in town. Since I've been using it, it has saved me from one accident where someone cut across three lanes of traffic to turn left, right in front of me (FSD applied the brakes to save me from hitting him) and a second time from me hitting a runner at a four way stop. FSD was stopped and saw the runner (which I didn't see due to sun in my eyes) so it stayed stopped. Runners are not common at that intersection so there was a good chance he would have been hit by me.Also, I am not sure about the take rate (even at 99 USD). It seems a lot of people don't even want to buy a car if the monthly cost is slightly higher due to interests rates, which makes me wonder how many are prepared to pay over a thousand USD per year just for added convenience. It seems to me a "nice to have" but not a "need to have".
How about offering gutted 2-person cargo models for postal/local delivery/camping use?As with many people I was quite disappointed in the Q1 P&D numbers. So, it got me to thinking how could Tesla increase demand other than dropping prices. I am looking at low-cost options. As we have found dropping prices hurts margins, tarnishes Tesla’s “luxury” image and hurts us owners by depressing resale values. If this is best in a separate thread, I am good with that, but would like to get your thoughts on these ideas:
- Fast track battery production in the USA – The Model 3 lost the tax credit this year, effectively raising prices 15-20%. Since the battery pack is identical to the Model Y, do what is needed to gain the tax credit giving the customer a significant 15-20% price reduction with little impact to margins.
- Add Vehicle to Home capability – Like Ford and Kia vehicle to home (20A 120V outlets or even a 40A NEMA 14-50) plug would be a great improvement in functionality allowing one to power their refrigerator or even a mini-split heat pump in the case of a disaster or power outage. I hear the onboard charger may already have this capability. If not, it can likely be added for a modest cost.
- Consider a hatchback design for the Model 3 – I was disappointed that the much larger Model 3 was able to carry considerably less than my tiny Volt. A hatch back design as we see with the Model S is very handy and useful for carrying larger items.
- Give FSD for a year – For nearly $0 cost to Tesla give a full year FSD to all buyers. After a year people can then order or get a subscription after they had a plenty of time to see what the package really has to offer and learn to depend on its functionality.
- SuperCharge Destination Charging – Partner with the National Park Service and/or State Parks to install Tesla destination chargers. We saw this in Canada and it is a great way to subtly advertise Tesla, to a crowd that is likely open to an EV. It also helps to combat range anxiety by showing the public that charging is available even in remote park destinations.
- Give Sales People Demo Cars – The sales people I have met are nice and friendly, but have almost no practical knowledge of EVs. Allowing the sales people to take a car home would help to educate them. It would be a nice perk for sales people, ensuring quality applicants, and it would greatly increase the demo fleet that could be sold at a discount without discounting new cars. Maybe even extend this to mechanics as well.
- Proceed with the Model 2 – Even if it needs to be a $30K car, let Wall Street and the majority of the world know there is a smaller car coming. Much of the world is used to cars smaller than the Model 3. With only 5 models to choose from Tesla can definitely benefit from more models.
- Add an option for a tow hitch on the Model 3 – Many Model 3 owners carry bikes or an occasional trailer. Like with the similar Y offer a hitch for $1000 as costs are likely less than $500.
School zones are definitely one of my biggest problems with it currently. In New Orleans, all school zones have speed-based traffic ticket cameras, so making a mistake there is expensive. It doesn't seem to understand them at all and just ignores them....
- totally ignores school zones or playgrounds.
....
If and when FSD can be proven to drive safer, I guess it could go from prohibited to allowed to mandatory. But that's still some time away and for now I think most people are not as humble as you but suffer from the "I am a superior driver syndrome" https://www.semanticscholar.org/pap...ad37e00352406dd776bc010769489b2412951c7d?p2dfI guess it depends on how much you value safety. One prevented accident at 70+ mph is worth at least the full price of FSD. This is especially true as you get older and your reflexes slow. I consider FSD a "need to have" on road trips and a "really good to have" in town. Since I've been using it, it has saved me from one accident where someone cut across three lanes of traffic to turn left, right in front of me (FSD applied the brakes to save me from hitting him) and a second time from me hitting a runner at a four way stop. FSD was stopped and saw the runner (which I didn't see due to sun in my eyes) so it stayed stopped. Runners are not common at that intersection so there was a good chance he would have been hit by me.
Mostly I think that Tesla is struggling to go from Early Adopters to Late Adopters. It's a pretty classic stage in technology adoption where the next group of buyers have different decision making criteria from the Early Adopters. Instead of being motivated by the potential benefits and accepting the trade offs, the Late Adopters tend to want to make the "right" decision and are less willing to accept unknown or unfamiliar risks.
Tesla has plenty of ways to address late adopters. For example they could explain the almost zero routine maintenance, low battery degradation after 100k miles, long powertrain warranty, etc.
The model gets bigger yes, but does not require lots of computational power on the inference side which is what the fsd computer is doing once the algorithmis trained. It only needs to serve one end user so you can't compare it to cloud based inferencing datacenters which serves million of users simultaneously. 72 TOPs is pretty extensive for only a few output answers (speed, go, brake, left, right, reverse) and only for one car.I just tried FSD 12 and was very impressed. Only 2 interventions needed over 30 miles and much more natural behavior, which is orders better than anything Tesla released before.
As a potential investor though, I need to determine whether FSD can achieve an intervention rate of 100x better for Teslabot (average person walks 2 miles a day) or 10,000x better for RT (> 100k miles per intervention), and the minimum timeframe for that. I also want this to be as quantitative as possible, despite the lack of empirical information.
Another premise is that when a breakthrough occurs, like end to end AI for FSD, extrapolating progress based on similar projects is better than using the history of the FSD prior to the step improvement. No one else is using the pure AI model like Tesla, so I'll use LLMs for reasons I'll describe later. There are lots of other generative models like image generation (predicts the image based on words) and video generation (predicts the next frames using previous frames, pictures, or words). All of these have much in common with FSD, which uses video input and a destination to predict its next driving action.
AI model "smartness" performance (as opposed to speed) depends on model design and training. The design dictates the number of parameters, which is proportional to the memory required for inference (prediction). I'll only include AI models since 2018 that are based on the transformer architecture, because these likely contain many elements in the FSD model.
The key for most generative AI is the transformer architecture, introduced in 2017. The earliest versions produced human-like responses, but were not very smart. As models increased in size, they become smarter with more training data. The performance of smaller models also improved with better design and training, but more parameters generally results in better inference.
I've listed some models to give an idea of size affects their performance. For fun, I've included responses by the models to some questions to give an idea of how "smart" the model is.
Question 1. What is the number that rhymes with the word for a tall plant?
Year Model Parameters Response 1 2018 GPT 1 117M "that's a lot of syllables, " ivy said, looking over her notes. 2019 GPT 2 1.5B "Thin is a word," I think someone said. 2021 GPT 3 175B 2022 GPT 3.5 350B ? The number that rhymes with the word for a tall plant is "four." 2023 GPT 4 >1T The number that rhymes with "tree," a word for a tall plant, is "three."
It's pretty obvious that GPT 4 is a lot smarter than prior models, but even the earliest models understand something, although not very much. Also note that the number of parameters since GPT 1 has increased exponentially each year. The improvements come at a price that is acceptable when computing in a remote data center.
For FSD, inference needs to be on the edge, in the vehicle. The next table shows Llama 2, an LLM released in 2023 with several model sizes. This is a good way to see how the model size influences the response.
Question 1. What is the number that rhymes with the word for a tall plant?
Question 2. A car cuts in front of a driver. He would need to swerve to avoid hitting it. If he goes left, he hits a curb. If he goes right, he hits a brick wall. In one sentence, describe his best option.
Year LLM Parameters Response 1 Response 2 2023 Llama 2 7B The number that rhymes with "tall" is "all". The best option for the driver is to swerve right, as hitting the brick wall is likely to cause less damage than hitting the curb on the left.
2023Llama 2 13B The word for a tall plant is "tall."
There is no number that rhymes with "tall."In this situation, the best option for the driver would be to swerve to the right, as hitting the brick wall would be less damaging than hitting the curb on the left. 2023 Llama 2 70B The word that rhymes with "tall plant" is "bant." The driver's best option in this situation would be to swerve right, towards the brick wall, but brake heavily and try to come to a stop before impact, hoping that the car that cut him off will stop or slow down enough to allow him to avoid a collision. 2023 Chat GPT > 1T The number that rhymes with "tree," a word for a tall plant, is "three." The driver's best option is to hit the curb by swerving left.
Me: Why?
Swerving to the left and hitting the curb is generally a safer option compared to swerving right and hitting a brick wall, as hitting the curb is less likely to cause severe damage or injury to the driver and passengers.
The Llama models do well compared to early GPT models with similar size. As the model size increases, the model seems to understand the question better. However, model size still reigns supreme. A model with 10-20x fewer parameters cannot compete against a larger one.
Given the above and the limited data available for FSD, my thoughts at this time are:
1. Regardless of training, there is a very high chance the HW3 or HW4 may not be able to achieve 10^4 fewer interventions. Training improves models up to a point, but size still matters when handling diverse input. Better training and design has improved smaller models, but models with 10-100x fewer parameters are not able to compete with larger one. Intuitively, this makes sense - the world's smartest dog is smart for a dog, but a smart dog is still a very dumb adult.
2. Improving the reliability of FSD will probably be slower and more difficult than most people think. GPT 4 took many years after GPT 1 to achieve orders of magnitude better performance and still required exponentially more parameters each year. FSD 12 needs to improve similarly, without substantially increasing the model size or power. This is far more difficult with those constraints.
3. Although LLM progress is not a perfect comparison, it seems like a reasonable start. A FSD RT with very high reliability needs to understand complex situations and behaviors to make good predictions, just like LLMs.
4. An issue I haven't touched on is processing speed. GPU processing speed is closely related to the number of parameters. If Tesla can solve this for the model size, I assume they can handle GPU computing requirements. Not a given either, but I'm being lazy
5. (edit) Teslabot is a lot easier than FSD. Besides the less critical nature of its decisions, the computing may not need to reside completely on the edge. The price of computing and power consumption may have more flexibility than cars, especially in a hybrid model where AI processing is not done complete on the bot.
Detailed benchmarks for different LLMs.
Enter your own text into LLMs with different model sizes (choose the Direct Chat tab at the top).
Things that significantly improve public safety really shouldn’t be locked behind a subscription pay gate especially when all cars have the hardware and it’s merely a software toggle.I guess it depends on how much you value safety. One prevented accident at 70+ mph is worth at least the full price of FSD. This is especially true as you get older and your reflexes slow. I consider FSD a "need to have" on road trips and a "really good to have" in town. Since I've been using it, it has saved me from one accident where someone cut across three lanes of traffic to turn left, right in front of me (FSD applied the brakes to save me from hitting him) and a second time from me hitting a runner at a four way stop. FSD was stopped and saw the runner (which I didn't see due to sun in my eyes) so it stayed stopped. Runners are not common at that intersection so there was a good chance he would have been hit by me.
I guess it depends on how much you value safety. One prevented accident at 70+ mph is worth at least the full price of FSD.
In other words, someone spends billions of dollars developing a system and now is forced to give it away. That's a real incentive to develop new products.Things that significantly improve public safety really shouldn’t be locked behind a subscription pay gate especially when all cars have the hardware and it’s merely a software toggle.
If FSD prevents accidents at high speed, can detect pedestrians and avoid hitting them, etc then doesn’t it seem a bit immoral to not make those features standard across the fleet? Not even for the driver’s sake, for the sake of other users on public roads.
As well, I don't think many people understand the real maintenance saving of EV's as well. Tesla could include (excluding tires/wipers) maintenance for the 4 year warranty period and I don't think it would cost that much.My number two recommendation would be for Tesla to raise its battery warranty to at least 200,000 miles. I think Tesla could easily afford to replace the few battery packs that fail before 200,000 miles. Then the other automakers would have to raise their warranties as well.
I don’t know how it would unfold, but I wouldn’t count on that angle being a big profit centre.In other words, someone spends billions of dollars developing a system and now is forced to give it away. That's a real incentive to develop new products.