Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

[KarenRei]

Bold statement. Use of absolutes like "never" are a dangerous path to take. I try to avoid them but to each his/her own.

Dan

Agreed. I'm not a FSD optimist, but again... remember that in 2014, the concept of determining whether there's a bird in a photo without five years and a personal research team dedicated to that specific task was considered "virtually impossible". Now commodity software can identify any common object, often better than humans can.

Self-driving is incredibly-rich in edge cases. But I think they'll nail it eventually. Just not "send the kids to school" / "sleep while you're driving" by the end of the year.

 

[neroden]

A good sign for q2, though expect to hear nonsense from the shorts on increased repair costs.

Tesla Puts an End to Goodwill Repairs

Tesla misclassified many of my warranty repairs as goodwill. Had they not repaired, I would have sued, won, and they would have paid for repairs plus legal costs. If they have been misclassifying routinely, a lot of this will end up in warranty rather than goodwill. That said I have heard of others who got real goodwill repairs.

 

[neroden]

All good, unless there is no demand problem.
I sell this car everyday. Lady yesterday from Montana, never heard of it, now wants one. Who they gonna believe, FUD or Friends, Family, and other owners? My circle likes facts and rides. This next wave is coming from last Thanksgiving diner talk. Easy when its the most amazing ride ever. Sells itself, no change.

Any demand shortfall would end by opening more service centers. I know dozens of people who would buy the car today if the nearest service center was only 100 miles away instead of 250. That is literally the only thing stopping purchases. Management seems blind to this dynamic though.

 

[j6Lpi429@3j]

Full self driving everywhere is a fantasy. It will never, ever, ever happen. Not with humans on the road, and humans will always be on the road, except possibly on freeways.

I understand this view but... I never thought that i would be able to talk to my phone, or my car and they would understand what I'm saying, even if I have an accent, or if i'm a bit drunk, or distracted, and in a slightly noisy environment. And they understand almost anything I say. And they do this for free, at any time, anywhere.

I thought mobile phones were science fiction when I was a kid (star trek communicators), but now I'm on maybe my 5th phone, and its 100x better than kirks. We may not have FSD everywhere before 2020, or before 2030 but NEVER?

 

[Singer3000]

There is some differing between bulls on this issue. Some want Musk in absolute control. Some like him at the helm, but want heavy checks and balances on him. I'm somewhere in between. I want people who will usually cede to his judgement, but push back when there's strong agreement that he's taking bad path.

Elon meets my definition of a genius. But he’s not always right and has gaps in his knowledge, as he would readily admit. “Be less dumb over time”.

Clearing out Elon Friendlies is not the same as appointing Elon Unfriendlies. It means taking away the sycophancy, having more people who will tell him when he should think again about something, or who may have more experience in something than him. From the perspective of retail shareholders, this has the added perk of making the stock more attractive to institutional investors, which is a precursor to any major step up in the valuation.

 

[SO16]

Any demand shortfall would end by opening more service centers. I know dozens of people who would buy the car today if the nearest service center was only 100 miles away instead of 250. That is literally the only thing stopping purchases. Management seems blind to this dynamic though.

I live in Michigan with no access to service centers in this state. Fortunately mobile service has been great. I have had my car over 2 years and 30k miles. Haven’t had to bring it to a service center yet. People need to get over the mindset of “having to bring in their vehicle to a service center” all the time.

I’m sure there will be a case where I need to. But that should be few and far between.

 

[neroden]

Regarding the FSD computer, I have been thinking that at 200 frames per second there is a lot of analysis happening very quickly. If the NN can take 5 of those sequential frames and look for the emergence of a safety risk based on those 5 frames (my guess) for example, then in one 40th of a second there begins a sort of recognition/reaction.

If traveling at 60ish mph, about 90 ft is covered one second. One fortieth of a second means a distance traveled of 2.25 feet. This is an amazing reaction time if I were to be correct in my assumptions.

Human reaction time is about 0.7 to 1.5 seconds so in that time the vehicle at 60 MPH has traveled 63 to 130 feet. The FSD computer equipped vehicle may be capable of something like 25xish faster reaction than a human.

If traffic regulations were to begin to be based on reaction times then it "might" be that certain vehicles might have much higher speed limits. It is a big "might" but it is interesting to think of an "Augmented Safety" lane with a speed limit of 120 mph or whatever.

Even at 120mph the NN recognition/reaction time might be (speculative on my part) in the range of under 5 feet of travel which is still 12 times better than a human at 60 mph.

Imagine if similar to an HOV lane, there was a lane established for only one hour where Augmented Safety vehicles could travel at much higher speeds (maybe platooning). There would be a considerable increase in throughput and possibly overall reduced congestion. Interesting to wildly speculate about what this FSD computer might change.

Reaction time i provement is nice, but.... Weather. Even with instant reaction time, in poor weather the safe following distance is... A lot shorter than what most humans do. On the Mads Max Chicago expressways, any self driving car will have to keep a longer spacing than the highly irresponsible human drivers. Result, lower throughput, more congestion, though certainly lives saved.

 

[neroden]

Communication seems to be worst in class at Tesla, I'm 6 months of trying to solve a 5 min issue. I have never seen a more messed up company on communication and CS and marketing blunders. Where to they get their managers, Paypal? This concerns me more than other factors out there, it's like a perpetual glass ceiling.

Yes, comms at Tesla is about as bad as it gets... I have only seen worse in a few monopolies and businesses which were actually trying to chase away customers. That said I think Tesla is getting slightly better. Still have a ways to go.

 

[neroden]

I live in Michigan with no access to service centers in this state. Fortunately mobile service has been great. I have had my car over 2 years and 30k miles. Haven’t had to bring it to a service center yet. People need to get over the mindset of “having to bring in their vehicle to a service center” all the time.

I’m sure there will be a case where I need to. But that should be few and far between.

I have had 5 issues in 5 years which required a lift, IIRC. Just open the damn center, even if it only has one bay. Even a chance of this deters purchasers... It is about peace of mind. Like Superchargers, which are rarely used but sell cars.

 

[SO16]

I have had 5 issues in 5 years which required a lift, IIRC. Just open the damn center, even if it only has one bay. Even a chance of this deters purchasers... It is about peace of mind. Like Superchargers, which are rarely used but sell cars.

In my state, it’s not by Teslas choice.

 

[AndyH]

Swedish paper Börsveckan recently did an analysis of Tesla. Translation of first paragraph:

"What is Tesla's enormous growth worth?
Tesla's stock has made a breathtaking journey since it's IPO in 2010 with an upswing of almost 1700 percent. For the most part the upswing is warranted, but there are still challenges and risks."

The rest of the article mentions actual facts without too much prejudice. They're ignoring autonomus tech and overestimate other car makers electric efforts but apart from that it's fair. Their projection:

Revenue is too low beyond 2019. Börsveckan estimates 20-30% continued growth.

 

[neroden]

I believe the video sampling is at 100 Hz. The 200 frames per second sometimes mentioned I believe comes from including the previous frame (from 10 msecs ago) as an input to the neural network. (This is very useful as it makes the NNs basically stateless: the output doesn't depend on previous outputs.) So yes, the NNs process 200 frames per second, but the real physical sampling rate of video information is 100 frames per second.

So if the vehicle control software requires 5 frames to be certain that emergency evasive action is unavoidable, then that's 50-60 msecs of delay for 5 frames to arrive (depending on the offset of the first frame which can be anywhere between ~0.01 and ~9.99 msecs), so the action to take - such as to emergency brake - would be after these 50-60 msecs:



So under the "5 frames assumption" this is 4.5-5.4 feet traveled, but there's also some actuation delay: brakes can be applied very quickly, but emergency steering takes more time.

Still amazingly fast, and it's much, much faster and much more consistent than human drivers. Once FSD takes off most humans won't be permitted to drive on public roads without FSD assistance. We humans really suck at driving safely, and there's about ~1 million avoidable traffic fatalities every year globally, ~37,000 every year in the U.S. alone.

Put differently: every single day there's 2,700+ preventable traffic deaths on the roads, basically a 9/11 amount of people murdered by lack of an alternative, every single day.

Human drivers are committing probabilistic mass murder in plain sight, which is only tolerated due to the incredible utility of road transportation. But it's a very bloody trade-off and once there's a better driver the crowding out of humans will be swift and brutal.

Therevhave been better options all along. As a society, we chose the mass murder, especially in the US. Driversinsist on tailgating on expressways, speeding... Because to them, saving 10 seconds is worth more than people's lives. Judges give licenses back to convicted reckless drivers and people with a history of hundreds of moving violations, bwecause it is apparently more impprtant to give the killer "mobility" than to prevent him from killing. A woman deliberately drove on the sidewalk, killed someone, in Mamnhattan, NO prosecution, kept her licence...

In short, as a society, the US just does not care about the mass murder. Same as we let gun massacres happen. I hope society will change for the better as it did in the Netherlands with the Stop Der Kindermoord campaign. If it does not, then the future of self driving cars is Uber running over pedestrians and shrugging. If that nightmare scenario happens I will sabotage the self driving cars myself.

 

[KarenRei]

Reaction time i provement is nice, but.... Weather. Even with instant reaction time, in poor weather the safe following distance is... A lot shorter than what most humans do. On the Mads Max Chicago expressways, any self driving car will have to keep a longer spacing than the highly irresponsible human drivers. Result, lower throughput, more congestion, though certainly lives saved.

This is backwards. Vehicles maintaining more space decreases congestion, not increasing it.

Stop Tailgating, It Only Makes Traffic Jams Worse

“Our work shows that, if drivers all keep an equal distance between the cars on either side of them, such ‘perturbations’ would disappear as they travel down a line of traffic, rather than amplify to create a traffic jam,” says Horn.

Of course, asking drivers to pay attention to their bilateral control is easier said than done. Instead, the researchers suggest that auto manufacturers add rear-facing sensors to cars and update their adaptive cruise control software. According to the study, if only a small percentage of cars used this feature it could dramatically reduce traffic jams.

Also, the more "smart" vehicles there are out there, the easier it becomes to do smart routing and maintain consistent high speeds, including planned spaces for people to merge in and out (mergers without space left for the drivers to move into are a major cause of spontaneous traffic jams - researchers have shown (with real world testing) how a single car merging poorly or braking spontaneously can cause a long traffic jam with no other contributing factors).

Autonomous driving would significantly increase road throughput. And the higher the market penetration, the more it will - you can start for example having smart traffic lights that know when there's going to be a bunch of cars that need to go through in one direction, and prepare in advance. Or in the ultimate case, even smart lane reversals. If you can significantly reduce the rate of freeway accidents, making them safer, then you can also increase the maximum freeway speeds.

 

[jerry33]

How are so many i3’s (relatively) selling?
Slashed prices/huge discounts?
It’s basically an EOL car.

I thought it was because of the "girls can't ride in an i3 video" which I can no longer find.

 

[mongo]

This is backwards. Vehicles maintaining more space decreases congestion, not increasing it.

Stop Tailgating, It Only Makes Traffic Jams Worse

Also, the more "smart" vehicles there are out there, the easier it becomes to do smart routing and maintain consistent high speeds, including planned spaces for people to merge in and out (mergers without space left for the drivers to move into are a major cause of spontaneous traffic jams - researchers have shown (with real world testing) how a single car merging poorly or braking spontaneously can cause a long traffic jam with no other contributing factors).

Autonomous driving would significantly increase road throughput. And the higher the market penetration, the more it will - you can start for example having smart traffic lights that know when there's going to be a bunch of cars that need to go through in one direction, and prepare in advance. Or in the ultimate case, even smart lane reversals.

That paper address tailgating (asymmetrical gaps) @neroden is refering to symmetrical, but tighter, gaps. (That sim looks like the red car slows which results in a Tesla backup)

“Our work shows that, if drivers all keep an equal distance between the cars on either side of them, such ‘perturbations’ would disappear as they travel down a line of traffic, rather than amplify to create a traffic jam,” says Horn.

Assuming a steady speed, less gap results in higher vehicle density and vehicle throughput with the extememe being a train. Anecdotally, people set TACC to a shorter distance on freeways to prevent cars from cutting in and slowing the effective rate (and in the short term) decreasing the gap...

 

[jbcarioca]

There is some differing between bulls on this issue. Some want Musk in absolute control. Some like him at the helm, but want heavy checks and balances on him. I'm somewhere in between. I want people who will usually cede to his judgement, but push back when there's strong agreement that he's taking bad path.

This is well-said. Everyone should be pleased that Tesla is maturing, so improved governance is an essential part of healthy maturation. That in no way diminishes Elon Musk; rather, successfully executed it will permit him to concentrate on the areas in which he excels so spectacularly.

On the face of this it seems that the role of Robyn Denholm at Tesla is acting as a close analogue to that of Gwynne Shotwell at SpaceX. Both are highly regarded as adept corporate leaders, with the atypical strength of forcefulness coupled with an apparent calm assurance that ends out being a positive influence on a less calm person supplying innovation and :’impossible’ goals.

In my opinion SpaceX would probably not have succeeded without Shotwell, and Tesla will be positively transformed with Denholm. The oddity is that one is COO and the other Chairman but both supply similar benefits. If we are really to be successful with TSLA and continue it’s transformative success we will be in desperate need of Straubel, Hocholdinger and a long cast of other outstanding people. It will only be likely to be so if the ‘emotional state’ of Tesla stabilizes.

It seems to me we might consider the less-successful transition of Ford Motor Company after it peaked 100 years ago. There never was anybody to act as a facilitator and stabilizer for the excesses of Henry, who gradually self-destructed.
We might also consider how Steve Jobs, in the end, needed Tim Cook to make his visions work.

Long winded, as usual. Still, this reorganization foresees a fundamental transformation. It remains to be seen if it will be successful. I plan a substantial increase in my TSLA position on the next large dip.

Finally, we all must realize that shorts, the SEC and market makers are joined with auto dealers, most oil companies, distributors of gasoline, auto makers and the political beneficiaries of their collective largesse. Their joining is to impede, at any cost, transition to reduce fossil fuel use. We repeat all this endlessly but...

If TSLA can actually become a major home of institutional investment there might be a fighting chance to change the broader outlook. We need to make TSLA spread as infectious diseases do, from the inside, usually not perceived until strong enough to resist all defenses. Hopefully with more favorable outcomes.

 

[jerry33]

Any demand shortfall would end by opening more service centers. I know dozens of people who would buy the car today if the nearest service center was only 100 miles away instead of 250. That is literally the only thing stopping purchases. Management seems blind to this dynamic though.

I wouldn't say blind, but it takes a significant amount of money to open and staff a service centre. Doing so before there are enough cars in the area to support one is gambling, and eats into cash flow. While everyone wants a nearby SC, until there are enough cars to warrant one, it just doesn't make good business sense--especially when there can be a local Ranger.

 

[Silent Ludicrosy]

Yesterday, at a stop light, a bus turned right in front of me into my lane and expected me to back up. Luckily there was nobody behind me. Then there is the crazy mess I went through in a parking lot with illegally parked cars, cars stopped with blinkers on, a lane too narrow for two way traffic (with two way traffic)... And the other parking lot which combined this with a blind corner and cars making K turns to reach gas pumps. Plus the pedestrians of course. All in one day...

I agree, solving vision is not enough for all situations. Cars will need to perform other human behaviors like planning, having situational awareness, communicating with humans, contingency planning, and predicting behavior of cars/people/animals to achieve level 5 FSD. I don’t think it’s impossible, but we’ll probably be stuck at level 4 for a long time.

 

[KarenRei]

That paper address tailgating (asymmetrical gaps) @neroden is refering to symmetrical, but tighter, gaps. (That sim looks like the red car slows which results in a Tesla backup)


Assuming a steady speed, less gap results in higher vehicle density and vehicle throughput with the extememe being a train. Anecdotally, people set TACC to a shorter distance on freeways to prevent cars from cutting in and slowing the effective rate (and in the short term) decreasing the gap...

Need another study?

One Driver Can Prevent a Traffic Jam

The techniques are simple, though some of them—such as leaving a large gap between your car and the one in front and freely letting other drivers cut in—feel counterintuitive to most drivers.

This is well known among traffic engineers. More space equals fewer traffic jams. It may feel counterintuitive, but it's reality.

 

[Reciprocity]

I assume you mean 18650, not 21650... And I agree the architecture is more important than the cells themselves.

They were not going to go 90% of they way there and not finish the job. All the savings is from the new format and the fact it's made at scale at GF1 they won't go 90% of the way to that savings and stop.

Elon said something that parked my interested he said cells for China would initially come from Japan, gf1 and China. This could mean that model S/X gets upgraded and Osaka gets upgraded and those cells go to China in Nov-dec. Osaka to Lingang much cheaper then to Nevada.

You can't upgrade Osaka if you're reliant on those cells in the US. The issues at GF1 are yield related, meaning they have to equipment and they know how to do it right, they just have to do it.

One possibility is that Tesla comes out an LR+ that is 120KWh based on 2170 while still offering the 18650 SR at 100KWh or even smaller. For a time period until Osaka can be converted. If the packs are compatible then it makes some sense, but it might be way too complex to keep them compatible for production complexity.