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

[Curt Renz]

True. The other thing I follow is how just much local auto dealers dictate the automotive coverage in newspapers.
It's nuts to open up any paper today and only see articles on ICE cars and nothing at all on Tesla.
Because of this, many people just had (and maybe still have) no idea what is really going on.

You have to look real carefully to see at the bottom of all of the car sections in local papers, in very very small print:

"advertising supplement"

The local dealers run the car section print section.

Indeed. And it's the dealerships that run auto shows. That's why Tesla doesn't often participate.

 

[powertoold]

Right. Because we all know that people never vandalize cars, throw objects off overpasses, shoot random people etc.

So you’re saying if FSD doesn’t deal with super rare criminal activity situations, regardless of how many normal accidental deaths it prevents, it’s not worth deploying or using FSD? And you don’t think these situations can be avoided simply by including a “Return to pickup location” button on the display?

not sure if you’re agreeing with my point or not lol

 

[Andy O]

I believe he was talking about decreasing space while increasing speed so something like speed/space = 1000-10000 percent increase.

I believe Elon was comparing factory efficiency to chip efficiency. He spoke of increasing chip efficiency as measured in production (computations) per unit volume. I don't see any real value in measuring a factory by that metric. Stacking machines on top of one another doesn't sound like the best way to gain an increase in productivity per dollar invested.

 

[StarFoxisDown!]

So you’re saying if FSD doesn’t deal with super rare criminal activity situations, regardless of how many normal accidental deaths it prevents, it’s not worth deploying or using FSD? And you don’t think these situations can be avoided simply by including a “Return to pickup location” button on the display?

not sure if you’re agreeing with my point or not lol

At the risk of joining into this discussion which SHOULD be moved to the FSD thread......it's beyond silly to judge FSD by such random and rare occurrences, like you're saying. FSD progress and safety should 100% be judged by natural every day occurrences that the FSD will encounter. On the highway road blockage/robberies.....rioters.....come on people. If you're driving a car and you're stuck in a riot, doesn't make a difference if you're driving or FSD is driving, you're stuck in a bad situation.

 

[Lessmog]

I believe Elon was comparing factory efficiency to chip efficiency. He spoke of increasing chip efficiency as measured in production (computations) per unit volume. I don't see any real value in measuring a factory by that metric. Stacking machines on top of one another doesn't sound like the best way to gain an increase in productivity per dollar invested.

AWS?

 

[scaesare]

What is Max Headroom doing in that picture?

Hey!

 

[ByeByeJohnny]

Of course, and by that I think he meant production speed.

I'm pretty sure he means some other measurement. 10.000% would mean a car would be built in minutes. From the first bolt to finished product.

More likely he means human working hours per car built. Or return on working capital or something.

 

[Curt Renz]

Benzinga - this afternoon: Tesla Adding New Safety Features With Upcoming Software Update

 

[Nocturnal]

So you’re saying if FSD doesn’t deal with super rare criminal activity situations, regardless of how many normal accidental deaths it prevents, it’s not worth deploying or using FSD? And you don’t think these situations can be avoided simply by including a “Return to pickup location” button on the display?

not sure if you’re agreeing with my point or not lol

No, the opposite. If it has the net effect of saving lives that's what matters, not crazy "what if" scenarios around hackers and terrorists.

I'm pretty sure he means some other measurement. 10.000% would mean a car would be built in minutes. From the first bolt to finished product.

More likely he means human working hours per car built. Or return on working capital or something.

Gotta be all inputs. Time, human capital, energy, space etc.

 

[Andy O]

AWS?

AWS ? Amazon Web Services ?

 

[Curt Renz]

Electrek - this afternoon: Tesla is considering a home 'energy package' with solar, Powerwall, EV charger bundle - Electrek

 

[Knightshade]

Benzinga - this afternoon: Tesla Adding New Safety Features With Upcoming Software Update

That's not a new feature- it rolled out well over a year ago originally.... this update just rolls it to more geos... (also it's one feature, not features plural)


Tesla pushes Dynamic Brake Lights that warns other drivers of rapid braking

 

[oldTAVguy]

With the way this is being forced down on low volume I think: 1 - They have taken the FOMO out with the side. 2 - Lowered the SP significantly from the ATH. 3 - The price is not going into free-fall. This leads me to believe that the S&P decision will come out today or Wednesday.

 

[LTC_RRR]

Well, Adam Jonas will be retired by 2040, nobody will care whether he was right or wrong.

Hell, nobody much cares NOW!

 

[Lessmog]

I believe Elon was comparing factory efficiency to chip efficiency. He spoke of increasing chip efficiency as measured in production (computations) per unit volume. I don't see any real value in measuring a factory by that metric. Stacking machines on top of one another doesn't sound like the best way to gain an increase in productivity per dollar invested.

AWS ? Amazon Web Services ?

Exactly, as a counter example.

 

[M|S|M MYP]

Chalking up today as a down day along with the Naz...normally Monday's are $TSLA rocket days but with this anemic volume, the MM's can push it any way they want.

 

[Paracelsus]

Electrek - this afternoon: Tesla is considering a home 'energy package' with solar, Powerwall, EV charger bundle - Electrek

Love the concept, Curt...................but Fred painting TSLA in a positive light while he is once again peddling for others to use his code is painful. My crystal ball says that Fred will be writing angry dialog towards TSLA about the Energy Package and the Reference Code program shortly after he gets his Energy Package for free.

 

[dl003]

There is one concern that I can't shake regarding Robotaxi: it's the difference between imitation and understanding. You can get to 99% with imitation. I don't think you can get to 100. Elon mentioned during the Shanghai interview that AI still hasn't progressed very far in terms of understanding IIRC. Take the recent riots in many of our cities. Imagine that your granddaughter and a few of her friends were taking a RT to a bar late at night and stumbled into one of these riots. A driver would know to get them the hell out of there. What is the RT algo going to do? Does it have a databank of similar situations to draw from? No. Can it rely on some rules based decision to leave? No. It won't understand the danger. What if thieves block a road to steal from motorists? Will the computer know that it's OK to run somebody over if they are pointing a gun at you? No. So what is the solution to that? Geo-fencing around bad neighborhoods? Good luck with that solution.

So my concern is that we need AI that can do more than imitate of follow rules ... it's got to understand. Maybe Apple or Google or Amazon os some Chinese company don't have our data or our chip, but maybe that isn't what gets you over the hump. Maybe the answer is in the next generation neural net?

If you think I don't understand all this stuff you are correct. I don't really understand any modern technology. I don't trust myself to evaluate whether Elon can get across the goal line ... and if so, faster than everybody else. Yes, I could just trust him. I trust him a lot. Just not enough to have 40% of my net worth riding on this working. At the current 8%, if it triples in the next few years then great, if it gets cut in half then no big deal. (You'll appreciate the logic of that better when you're retired.)

I could just as easily be talking about all the things I love about TSLA. On other boards, where people don't understand all the wonderful things happening I regularly praise TSLA and Elon. But it's important to force yourself to critically examine your assumptions. Most on this board don't seem to want to do that. Easier to put someone on ignore than to consider their argument.

The problem is not that you don't have confidence in AI but it's that you overestimate humans. Majority of people won't be able to deal easily with the .0000000001% cases either. Humans also just follow rules. What we perceive as judgment is simply the outcome of accumulated experience. Tesla's AI learns in the same manner as humans: through vision. This is First Principle 101: Tesla is ahead not because of its years of experience and money, but because it seeks to solve a human's problem using a human approach. LIDAR is not what human relies on. Mapping is not how we make driving decisions. We learn through our eyes and our experience and that's precisely what Tesla AI is doing. Currently, Tesla is the only one taking this approach. Dangerous areas should be easily identified by the computer via the web. Just like flooding or traffic jam. When was the last time thugs blocked a road to rob people? What are the chances someone in an RT is the first motorist to run into that blockade before the authority is alerted? If a human driver got into that situation, what would they do differently from a computer? I bet you anything 100% of people would just stop right there to see what's going on.

 

[Nocturnal]

I'm pretty sure he means some other measurement. 10.000% would mean a car would be built in minutes. From the first bolt to finished product.

More likely he means human working hours per car built. Or return on working capital or something.

The problem is not that you don't have confidence in AI but it's that you overestimate humans. Majority of people won't be able to deal easily with the .0000000001% cases either. Humans also just follow rules. What we perceive as judgment is simply the outcome of accumulated experience. Tesla's AI learns in the same manner as humans: through vision. This is First Principle 101: Tesla is ahead not because of its years of experience and money, but because it seeks to solve a human's problem using a human approach. LIDAR is not what human relies on. Mapping is not how we decides how to drive. We learn through our eyes and our experience and that's precisely Tesla AI is doing. Currently, Tesla is the only one taking this approach.

Human drivers are morons. There were 1.35 million road traffic deaths globally in 2016

Source: Road traffic deaths

Even if AI was only as good as an average driver, that would be a huge gain. AI doesn't get drunk. AI doesn't see a pretty girl jogging. AI doesn't get distracted by a phone call, or a bee. AI isn't exhausted from being up late. Also, as we reach critical mass of AI drivers, then the cumulative safety gains would build off each other. When 1 human drives defensively, it's not uncommon for other humans around to drive offensively to overtake or merge. 2 AIs wouldn't have that issue.

 

[Unknown_Wizard]

We need Form 10-Q to be published...