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

[StarFoxisDown!]

Ha! Motor Trend said they could have gone 400 miles! Good enough for me!

Yeah makes it seem like Tesla is sandbagging the range a bit

 

[EVNow]

Once he knows something is possible, he'll seek it and achieve immense success, eventually. But focusing on possible, rather than problems along the way, has lead to massive misses in terms of timing.

There is another aspect that makes him give optimistic timelines.

That is what he is pushing the team to achieve internally as well. If he gives a different timeline outside - he thinks they will not work hard enough.

BUT, important thing to remember is - he has already given multiple timelines on FSD before that have passed. So I think this time it is closer ! So, instead of 2020, we may get robotaxi in 2023 - but the exciting thing is we can all see and experience the progress - unlike something like Model 3, which took 3 years to come out.

 

[UnknownSoldier]

Oh damn. This is Project Raven. They must have really fast-tracked development of this after the S/X sales crashed due to the 3 release.

Here we go boys, strap in because this stock price is going for a wild ride!

 

[neroden]

There is another aspect that makes him give optimistic timelines.

That is what he is pushing the team to achieve internally as well. If he gives a different timeline outside - he thinks they will not work hard enough.

So he shouldn't give any external timelines at all! "As soon as possible!"

 

[Singer3000]

Currently, from what I can get from the presentation, they're using a combination of manual path planning (doesn't scale) and "copying the average driver" (who isn't good enough at it). They'll have to do at least one more iteration on their entire path planning scheme (toss out what they've got and try again with a slightly different approach). Oh, they will do that, but it'll take time...

How many lives would be saved if we took away the tired, drunk and distracted drivers and ensured all drivers on the road can pass a driving test? Is that not what an autonomous system that copies the "average driver" would achieve, even if it initially only worked in certain geographies and conditions? How much would such a system be worth to society? Is it possible for the owner of that tech to monestise that value?

 

[Fred42]

Whew....at the hotel and finally caught up....you all are busy little beavers!

With all the L5 and Robo-taxi talk has anyone been able to prove (to the extent possible) that the Spontaneous fire in the Tesla in China was in fact staged?

No.

 

[neroden]

What a week. And now I have to listen to the earnings call tomorrow? I'm exhausted.

 

[dakh]

Good questions, yes. I don't know where you live. But for me, the answer is "no".

Glitches like occasionally being stopped dead mid-trip and having to call a taxi at random locations? Well, OK, if you're cool with that, fine. In some major cities, it's not a problem, you can just step out and get in the subway.

That was kinda covered too, they predicted that high population density areas will be mostly operating as taxi, and less dense as private vehicles that you can send around and get transported in without having to drive. When cars first came to market, they had to operate in an environment that was designed for horse carriages so I'm sure there were somewhat similar challenges there too. Once the requirements of a self-driving car to its environment are better understood there will be a substantial economic incentive to tweak what we have to make FSD job easier. It doesn't have to happen all at once everywhere. Say, handicapped vehicles would be totally cool to support FSD on, yet it's just too much of a corner case and many more lives would be saved if a more mainstream case is solved first. Same with someone's driveway that you have to ford a river to get to.

So I actually rated this "love", because you're right, "Our use of machine learning is limited by how well can we present real world conditions and outcomes to it".

But this is much, much, harder than you think it is.

One of the things you're wildly wrong about is the difficulty of actually getting the right dataset. In medicine, we don't even know what data we need to collect -- the importance of the gut biome was not recognized until recently and people who talked about it were considered cranks, for instance. Half the work of a good doctor is considering collecting different data. [...]
Driving has a bunch of situations like this too. Tesla hasn't even gotten to those situations where people disagree on the desirable outcomes. With human driving, to some extent, the human is given the decision-making power. Musk even discussed the tradeoff between getting into traffic and risking fender-benders in LA traffic, and basically said the driver would decide. I don't see any way to avoid that, which prohibits FULL self-driving.

Luckily roads are a fairly mechanical, billiard-ball type environments. There are some mitigating factors that make me think FSD computer can be a bit more stupid then an average human and still be a lot safer. One obvious reason is its 100% consistency and focus. Another thing, all Teslas are much more capable cars than average, at least in acceleration and cornering. When you combine good hardware with a split-second decision making and very precise trajectory prediction ability of computers, I think the odds of an FSD Tesla finding a safe(ish) hole in the traffic to merge in are much, much higher than for most human in most vehicles. Another example, if a car is trying to make a right into a multi-lane road, it needs to know that even if the rightmost oncoming lane is empty, someone traveling in the next oncoming lane over can decide to change lanes into the rightmost one while you're pulling out. Humans often don't consider such possibilities. But that's also how plenty of accidents happen, and it will get reflected in the sample data that self-driving predictor will be trained on (because actual accidents will drive inclusion into the training set, not someone's opinion on the subject). There are plenty others. It's a fallacy to think that FSD computer has to be better than humans in most aspects. It will have its strengths and weaknesses, the critical part is that as a whole it needs to be safer than humans and still get the job done.

 

[neroden]

How many lives would be saved if we took away the tired, drunk and distracted drivers and ensured all drivers on the road can pass a driving test? Is that not what an autonomous system that copies the "average driver" would achieve, even if it initially only worked in certain geographies and conditions? How much would such a system be worth to society? Is it possible for the owner of that tech to monestise that value?

I will just say that in the US the average driver probably can't pass a driving test. Maybe in Singapore they can. :-(

 

[Artful Dodger]

Tesloop’s Tesla Model S Surpasses 400,000 miles (643,737 kM)

Better to look at all their X90's on original packs with 250-360K miles
Tesloop's High Mileage Teslas

Yeah, nothing about cycles in those write-ups. Just the pack replacements @ because Tesloop overcharges:

"High Voltage Battery: The Model S has had its high voltage battery replaced twice under warranty at 194,000 and 324,000 miles. Battery degradation over the course of the first 194,000 miles was ~6% with multiple supercharges a day to 95-100%, instead of the recommended 90-95%. Between 194,000 and 324,000 miles Tesloop experienced battery degradation of ~22%"
​

Certainly nothing in there to support any claim to a 3000 cycle life for the battery pack, which would be approx. 680K miles for a 90KWh pack S/X to 80% capacity.

 

[dakh]

I thought Tesla will use a new motor to improve range by 6~8%. Somehow they managed to improve range 10% using the same battery pack. This is impressive. All the other upgrades are very nice too.

I suspect 10% is with the new suspension that will squat for better aero in addition to engine efficiency improvement.

 

[UnknownSoldier]

So we can anticipate S/X sales will go back to normal again now?
I can imagine lot of older-gen (pre-2016) S owners will be wanting to upgrade with the free Ludicrous promotion.
This is pretty impressive but I want to see what the new interior looks like. Are they going to use the single center screen design of the 3 now?

 

[JRP3]

Oh damn. This is Project Raven. They must have really fast-tracked development of this after the S/X sales crashed due to the 3 release.

Or they cut back on S/X production and discontinued the shorter range in preparation for the upgrade they knew was coming.

 

[SOULPEDL]

Due to yesterday's autonomy day and today's refresh, I expect the Q1 numbers to absolutely suck. They wouldn't turn over these cards unless they really needed to. However, I'm more optimistic about Tesla's future than ever.

Or... this will be the short squeeze with a Q1 profit. Amazing just how impossible it is to predict. Never bet agsinst Elon.

 

[StarFoxisDown!]

Or they cut back on S/X production and discontinued the shorter range in preparation for the upgrade they knew was coming.

If they give guidance for Q2 S/X sales that is essentially back to normal, then we know the Q1 was a deliberate hold up, not really demand.

The thing that's hard for me to believe about lack of demand for S/X.....I still interact with a lot of people including a big family gathering a month back(in Texas of all places) and people still covet the S/X......a lot. Who knows though

 

[Chunky Jr.]

It's interesting they didn't increase the battery size to get 400 miles of range. They must be saving the 400 mile card for 2170s or other future refresh.

 

[Artful Dodger]

It's not betting the farm at all. If FSD goes nowhere they still make cars that people like, they still make solar panels, batteries...

Indeed, its leveraging the farmland they're already plowing. Rotating crops? Extra-farm income?

Worst case scenario: FSD owners get the most amazing driver assist ever, and a previously undreamed of safety record.

How are people against that? Is it even possible to be greedy in a hospital bed?

Cheers!

 

[Chunky Jr.]

Or... this will be the short squeeze with a Q1 profit. Amazing just how impossible it is to predict. Never bet agsinst Elon.

Funny, I was about to edit my post to add that perhaps it is actually powder for better than expected numbers and really juice the stock, perhaps for a cap raise, but it felt too wishy washy for one post.

 

[Fact Checking]

Good, enough with the S&X. Back to FSD.

Its not full FSD if it doesn't know how to handle this.

ps : Just to be clear, this is good, well behaved traffic.

View attachment 399990

Firstly, FSD is clearly not feature complete until it can replicate simple traffic maneuvers that Russian soldiers performed in WWII already:

As a second milestone I think FSD must be able to handle light traffic in Mumbai:

Finally, any FSD system that wants to earn the prestigious "TMC Seal Of Approval" needs to be able to handle the following standard traffic scenario, where an ordinary pedestrian removes a stray cat from the road:

 

[dakh]

Or they cut back on S/X production and discontinued the shorter range in preparation for the upgrade they knew was coming.

Yeah, and they had to be quiet about that to not crater S and X sales even further. Kind of a tough deal to manage but I don't see any other way to go about it.