sundaymorning
Active Member
I was told by an employee that Tesla’s website crashed twice on the day Elon announced the $35k, so if your order hasn’t gone through, it’s likely bc the server was down.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
I was told by an employee that Tesla’s website crashed twice on the day Elon announced the $35k, so if your order hasn’t gone through, it’s likely bc the server was down.
sundaymorning
Active Member
Does he know many visitors walk into the Tesla stores vs the rest of the industry? Asking for a friend..
Buckminster
Well-Known Member
Regarding 2 wk delivery
In support of Karen:
"All publicity is good publicity"
In support of Karen:
- Vins are high - Tesla could be squirrelling them away for a major delivery event
- Tesla must realise that many headlines will be negative if they don't deliver on time. Fred is going to have a field day never mind the rest of the shorts/mainstream media.
- Suits Tesla to give customers an optimistic delivery - stops them buying the imminent Tesla killers available almost now
- Forces media to produce headlines - "Tesla sold out due to incredible demand"
- We haven’t even seen an SR or SR+ in the wild yet have we?
- 2 weeks seemed wrong to start off with - it takes 2 weeks to deliver....
- M3Rider estimated 60k would order now. Even if we halve that it is probably 12 weeks plus worth.
"All publicity is good publicity"
In 2016, Elon went on stage to claim that all their cars are now equipped with the hardware to enable FSD. And not just 150ft summoning in a parking lot or following a bit of navigation input (aka what's now called FSD). But the real deal, coast to coast any route you pick. Didn't happen. We are already 2 iterations further and that hardware from 2016 will never do what he promised. (In typical fashion, they are now coming out with other features that aren't available for that hardware and no upgrade path, letting is slowly languish like AP1 hardware)
Let's turn the tables here : is there any prediction by Elon so far on FSD that has come to fruition? What's the multiplier on 3 maybe 6 certainly months these days?
I don't know if anyone already mention this but Elon Musk on recode decode podcast (5th November) mention that the orders for the supply chain are placed 6 month in advanced.
Also i remember that in this article tesla, after for the first time manage to reach 5.000/week peak volume production of model 3, asked from suppliers to reduce the cost.
It has been around 7 month until tesla announced the 35.000 model 3 and other price cuts.
Maybe there is a connection here?
Also i remember that in this article tesla, after for the first time manage to reach 5.000/week peak volume production of model 3, asked from suppliers to reduce the cost.
It has been around 7 month until tesla announced the 35.000 model 3 and other price cuts.
Maybe there is a connection here?
I know, I know, quoting myself. But I have one more thing to add:
Making it right doesn’t necessarily mean paying us all back some amount in cash. It could mean anything from a $1000 voucher for Tesla store and/or off a car to intangibles like early access to updates and/or invitations to Tesla unveilings. Something that doesn’t really cost the company much or anything but provides value to the likely (relatively) small number that bought FSD at order time.
It seems like the only thing "Tesla killers" end up actually killing is themselves
Antares Nebula
Active Member
You may be right. I know Musk's policy re advertising. But when you have revolutionary new products, you can't rely on the media or word of mouth only. The former may confuse or give negative spin, and the latter may not reach scale quickly enough. Revolutionary and/or controversial products need proper communication and education by the company. Apple advertised the first Mac (1984 commercial) and also the latter generation of revolutionary products with the second coming of Steve Jobs (ipod, imac, iphone, etc.). Apple and their products are well know, but they knew they have to advertise and even create their own stores. They can't rely on the media, the public, or other store clerks (Best Buy) to sell their revolutionary products.Sat in Cyprus McDonald's, having a coffee, and hewr two blokes discussing " Musks company's are in trouble" talking about store closures and how Volvo will be releasing a new EV this year, very poorly informed, Tesla needs to begin advertising!! And push out a message to the more common people who don't follow tech news like we do..
The lack of communication from the PR team is poor, and people's understanding of Tesla is poor, I mentioned to the wife yesterday I want to buy a model 3, she flat out refused as in her words " cars that drive them selves are dangerous I keep seeing stories on the news about people dieing because of the autopilot"...
TIME to start advertising!! That should out us over 420
I know two people personally who are very smart but don't follow Tesla closely and when I was talking to them about it, one thought that Tesla was going bankrupt and another thought that Musk was a psychopath. But once I explained things to them rationally and with evidence, they were convinced, with one of them even buying stock in Tesla.
The fact of the matter is that the shorts and media smear campaign has unfortunately been very effective to at least create doubt and hesitation on the part of your average consumer (and investor). Maybe Tesla advertising (and beyond just twitter) is the only way they can change this narrative. They need to hire some really good ad agency, like Apple did, and communicate a cool, slick message that shows the full advantage of a Tesla EV over other EVs and over other cars period.
Last edited:
adiggs
Well-Known Member
I'm also very long, and feel about the same about FSD. If you're looking for unattended cars to be roaming about public highways, delivering themselves, showing up for test drives, or doing anything really this year or next as an important component of your investment thesis, then:
a) I really hope you're right and I'm wrong - that'll be a stunning amount of incremental value over and above what I'm expecting out of the company
b) FSD, without vehicles running around unattended, is going to be hugely valuable in an incremental way to the existing driver assist technologies (this is my thesis).
Lane changes on restricted access highways without intervention - that'll translate to nearly SC to SC self driving by the car (with a driver that keeps a hand on the wheel and remains attentive). That's huge for road tripping.
Recognizing stop lights / stripes / signs, and presumably stopping for them appropriately on autopilot. THat'll be immediately valuable to me on a 60 mile commute I make reasonably often.
I can imagine these technologies getting proven enough over the next year or 2 so that general FSD isn't enabled, but targeted FSD becomes enabled. Here I'm thinking the car watches over my shoulder and notices my morning and evening commute to work each day, figures out about where I like to park, and after enough repetition, offers to drive me to work and park in the morning, and then drive me home and park in the evening. I have, in effect, trained it on the edge cases that make up a common drive for me, and now the technology is ready to step in and help.
These are steps along the road that are all a long ways away from unattended cars wandering around on public highways. They can provide consumers, and therefore TSLA, value in the next year or 2. And if I'm right, they will continue to reset the public conversation about what FSD really means, and how it gets deployed into the wild and starts reshaping society.
If I define FSD as a car that roams around unattended on a public road, then I'm with @anthonyj - that's vaporware for at least the next 2 years, and I suspect it's vaporware for the next 5.
Not really a good metric. When I google BMW here, I get the local BMW site with my country specific TLD, not BMW.com. With Tesla, I get Tesla,com (since many of the country-specific tesla domains are not owned by Tesla)
adiggs
Well-Known Member
I agree with the sentiment. The threshold you're using is way too low. 99% of cases means that 1 mile in 100 is undriveable by FSD.
This is one of those places where the order of magnitude is way, way beyond that. I don't know what the right threshold is, but one way to think inform your intuition is accident statistics when they first started collecting and reporting them, were reported in terms of "million miles driven". Over the decades that metric changed and has become "hundred million miles driven".
And the numbers are getting so small again, I sort of expect the metric will change again to "billion miles driven".
For FSD, my first guess is that a vehicle needs to handle the stuff that happens over 10,000 miles of driving pretty reliably. For many drivers, that would translate to a year's worth of driving. For something on the road continuously, that'll translate to several breaks / need for help a year. That's 2 orders of magnitude above 99% (translating the 1% failure as 1 mile in 100).
Fact Checking
Well-Known Member
That's a rather cheap shot. As I said:
Elon is a hopeless optimist when it comes to features that don't exist yet, and that 3-6 months statement was a stellar example of that - but we might also list other missed projections, like:
- Falcon Heavy launch, "6 months away" for like 5 years, but delivered February last year.
- Rocket landings - wildly optimistic timings, eventually delivered.
- Model 3 ramp-up, timing was 2x off there too - but eventually they delivered.
- Autopilot 2 behavioral parity with AP1 - promised to arrive in 2017, delivered early 2018, after a messy divorce from MobilEye.
- Navigate on Autopilot: this one was actually delivered almost on time.
You did not cite a single example of Elon being wrong about the capabilities of existing software and hardware, which is pretty amazing given how much Elon is talking about all the technical details, compared to other CEOs.
bhtooefr
Active Member
It'd be pretty trivial, given that it's an induction motor - just don't send power to that motor. (It could still be used for additional regenerative braking to recover some of the efficiency lost from carrying it and the bearing drag that it causes.)
However, I agree that it simply doesn't make sense, given how simple it'd be to leave the motor out (and the fact that the SR, SR+, and MR don't have it, meaning that particular production line variation still needs to exist), and the cost, weight, and drag penalty of the second motor.
In the computing world, the term commonly used is "nines of reliability", and Musk has used that term around it.
If the car is running continuously for a year, 99%, or 2 nines, means that it'll be incapable of handling 3.65 days worth of a year's driving.
"Five nines", or 99.999% percent, a common threshold for reliability in computing, means that it'll be incapable of handling 5.26 minutes worth of a year's driving. Better, but what's the failure mode?
Honestly, I suspect that somewhere between 7 and 8 nines is where things need to be to make regulators happy. Six nines will probably beat many, but not most, human drivers.
Last edited:
Fact Checking
Well-Known Member
Note that the metric for "economic success of FSD" is even more favorable than the '99%' metric: FSD has to work much better than humans, on a significant percentage of routes that matter economically to a specific customer.
For example FSD working well on highways is already a big step forward, because it already covers around 50% of Vehicle Miles Traveled (VMT):
Note how big proportion of vehicle miles traveled highways are, compared to 'local' roads, yet they are just a small part of the road network.
Also note that highways are also the most dangerous ones in terms of traffic accident fatalities (due to the much higher speeds and quadratic kinetic energy), so Tesla starting with NoA also improved safety disproportionately.
Likewise, if for example a trucking company's fixed routes of travel are handled well by FSD then it's only those ~1,000 miles of routes that matter to them, out of trillions of miles of possible routes. If FSD works on those routes then then FSD will be 100% functional and 100% economically useful to them.
Tesla will also be able to concentrate their efforts on the AP disengagement events that a larger percentage of the Tesla fleet is reporting, so instead of going for a 99.999999% solution they can concentrate the effort for where it matters most.
So FSD is a statistical problem on a per customer basis, with an automated feedback channel of disengagement events to Tesla, which feedback loop I expect to drive FSD market penetration a lot faster than many would expect.
Last edited:
Antares Nebula
Active Member
Face recognition is an easier problem than locomotion. Human babies can recognize faces before they can move around. The agricultural robot example is an example of autonomous tech, and so supports my argument. But that is an easier autonomous problem than self-driving.
Planes, drones, even the SpaceX dragon capsule are all autonomous to a certain extent. But driving on the road, with all the signs, rules, other nearby cars and people, is a much harder autonomous problem than all the above. But nonetheless it is still in the larger class of autonomous (or locomotion, maneuvering) problems, which are still easier than harder AI problems that require symbolic and abstract reasoning, just like the latter are harder for humans than locomotion or driving is.
It is really very simple. We will create an autonomous car before we create an AI doctor or AI engineer. And all of these will eventually happen.
My Italian not yet ex-wife (who makes me the closet Ascetic seem like a big spender) became very excited on the news of the Model Y reveal. That's one base version sold.
Fact Checking
Well-Known Member
That's not universally true. Here's an iguana hatchling's first minutes in life, fresh out of the egg, behavior and locomotion purely inherited via genes, with zero real-world experience to survive:
No, that problem has little to do with the complexity of locomotion: human babies are mainly facing a "bootstrapping problem": their brains and heads are way too large for the birth channel (a hipbone limitation) so all of them are born via preterm birth, many, many months before being fully developed.
So the limbs and even brains of human babies has to grow a lot after birth (while most of the ~86 billion neurons are present at birth, most of the axons [weights] only grow in the first few years, increasing the mass of the brain massively. Less than 10% of the 'neural network' is present at birth.), and locomotion skills have to adapt constantly to changing limb and body mass proportions and muscular abilities. Muscle mass is way too low at birth as well - all basically compression tricks done to support a huge brain.
This is basically an evolutionary inefficiency that hasn't been optimized yet, because human brain size grew way too fast compared to typical evolutionary timescales:
So it doesn't make sense for human babies to have locomotion skills encoded in genes - human babies are not independently viable for about two years after their preterm birth, and they are growing rapidly and in a non-proportional matter that makes early locomotion skills mostly irrelevant. Facial recognition on the other hand is body proportion independent and can be encoded at birth.
Most animals have it much easier with locomotion, they are born basically proportionately and are able to move around effectively within hours - sometimes minutes after birth. As they grow they scale up proportionately, keeping birth-encoded locomotion abilities in balance.
Last edited:
I fundamentally disapprove of the 'app' concept because running someone else's code on my device means that one more unknown entity can take control of my device (as such I run only open-source apps on my non-Android/non-IOS phone).
As our digital selves become an even greater part of our lives, I find the app concept increasingly disconcerting.
With that said, the buy part of the Tesla app (which naturally should be open source) could simply be a redirect to the Tesla website.
bhtooefr
Active Member
I feel like the "app" versus "web" thing is a false dichotomy.
People have gotten an idea into their heads that the web isn't code running "on the device", but of course it is, and there's tons of exploits to break out of browser sandboxes. (And re: JavaScript, while View Source is an option (although often not on mobile browsers), does minified JS really count as usable source? I mean, a disassembly or decompilation of a native binary is about as useful...)
Similarly, people have gotten an idea into their heads that native code can't be sandboxed... but there are ways to do that. Granted, many OSes don't expose proper permissions controls for keeping something confined to the sandbox, or don't make the sandbox as easily available as they should, but...
If you truly want to prevent someone else's code from running on your device, disable JavaScript, but I'm pretty sure Tesla's site will break to the point of uselessness if you try. (Granted, I think JavaScript was a mistake, but hey.)
People have gotten an idea into their heads that the web isn't code running "on the device", but of course it is, and there's tons of exploits to break out of browser sandboxes. (And re: JavaScript, while View Source is an option (although often not on mobile browsers), does minified JS really count as usable source? I mean, a disassembly or decompilation of a native binary is about as useful...)
Similarly, people have gotten an idea into their heads that native code can't be sandboxed... but there are ways to do that. Granted, many OSes don't expose proper permissions controls for keeping something confined to the sandbox, or don't make the sandbox as easily available as they should, but...
If you truly want to prevent someone else's code from running on your device, disable JavaScript, but I'm pretty sure Tesla's site will break to the point of uselessness if you try. (Granted, I think JavaScript was a mistake, but hey.)
Antares Nebula
Active Member
I agree @Fact Checking. Encoding is different than learning. The question so boils down to how complex is the neural process whether encoded or learned.* And face recognition is hard for some animals. What is true is that object recognition is a prerequisite to locomotion. But, of course, differentiating finer and finer nuances in objects can become an ever harder AI problem, and in some cases can be harder than certain forms of locomotion. And all creatures that have locomotion don't need to differentiate every object in their environment. But all things equal, locomotion, I would say, is in general a harder AI problem than object recognition, since it requires the latter plus added functionality.
*I could take a pill one day (or like Nio in the Matrix) get a download of certain skill sets. And so being born with it or even acquiring it can be easy. That doesn't mean the actual neural process is simpler (computationally in the brain) relative to other learned processes.
*I could take a pill one day (or like Nio in the Matrix) get a download of certain skill sets. And so being born with it or even acquiring it can be easy. That doesn't mean the actual neural process is simpler (computationally in the brain) relative to other learned processes.
Last edited:
Similar threads
- Locked
- Locked
- Locked
