0:02

All zone media. Wake

0:06

up, everybody, I've got a new podcast. This is

0:08

Better Offline and I'm your host ed ze Tron.

0:22

Last week, the National Highway Traffic Safety

0:24

Administration released a damning report

0:26

about Tesla's autopilot and its full self

0:28

driving systems, which the nhtdsay

0:31

referred to as not adequately ensuring

0:33

that drivers maintain their attention on the driving

0:36

task. One second, though, to delineate

0:38

between the systems, autopilot is more like

0:40

a sixty version of cruise control keipping

0:42

your car and lanes, changing lanes when you hit a thing,

0:45

hitting the indicator, and following the car in

0:47

front of you on the highway very basic. Full

0:50

self driving is when your car drives

0:52

itself. It makes turns, you tell it where to

0:54

go on the GPS, and it goes through

0:56

intersections, follows lights, and

0:58

a bunch of other things it appears to not

1:00

really be capable of doing, with the NHTSA

1:03

saying that both autopilot and full self driving

1:05

created a trend of avoidable crashes involving

1:08

hazards that would have been visible to an attentive

1:10

driver. The report, which

1:12

covers the period between January twenty eighteen

1:14

and August twenty twenty three, described a critical

1:16

safety gap to quote CNBC's Laura

1:19

Colodney in the autopilot's system,

1:21

which contributed to at least four hundred

1:23

and sixty seven collisions resulting in at

1:25

least thirteen fatalities and forty

1:28

nine injuries. Musk

1:30

has recently tried to convince investors that Tesla

1:33

is now all in on AI and his flimsy

1:35

dreams of having a robotaxi company. This

1:38

somehow also resulted in Musk firing

1:40

the majority of the team behind the one Tesla

1:42

product Everybody Likes It's supercharger

1:45

network, leaving the status of the North American

1:47

charging standard created with Tesla's

1:50

help in jeopardy. To

1:52

explain what the hell is going on with Tesla,

1:54

I brought in Ed Niedermeyer, who has been writing

1:57

and commenting on the auto industry and mobility

1:59

text since two an a. He's the author

2:01

of Ludicrous, The Unvarnished Story of Tesla Motors,

2:04

and the co host of the Autono cost He

2:06

lives in Portland, Oregon. I'm really happy

2:08

to talk to him,

2:11

okay, Ed, So please tell

2:14

me what has been happening in the world of elon

2:16

Musk. In the last two weeks, I

2:18

want to.

2:18

Say, there's

2:22

a lot, a lot of a lot has been

2:24

happening, and I

2:26

think, you know, a lot

2:28

of it has really deep roots. Like

2:30

a lot of the things that are happening are are

2:33

so dramatic right now, but they've been sort

2:35

of been building towards us for a really long time. I

2:37

would say, like at a high level, what

2:40

appears to be happening is essentially,

2:44

you know, Tesla as a

2:46

as a stock, as a perception,

2:48

as a set of stories and dreams

2:50

that that Elon Musk has been weaving

2:53

for for many years now has essentially overtaken

2:56

Tesla as as a as a

2:58

company, as a real thing. And I think, you

3:00

know, from the very beginning of sort of stumbling

3:02

onto this company, the defining characteristic

3:05

of Tesla is this gap between perception

3:07

and reality. And you know,

3:09

I've written a lot about the problems

3:12

that Tesla has as a company, and

3:14

yet you know, for five years, much

3:16

longer really five years since my book came out, you

3:18

know, Elon Musk has been able to sort of use perception

3:21

to kind of make a lot of those issues

3:23

not matter through raising money or you

3:25

know, sort of creating diversions and all

3:28

kinds of other things. And I think what's happening

3:30

now is that, you know, the the problems

3:32

with the car business are so fundamental

3:37

the and and there's nothing

3:39

in place to sort of solve them. And

3:42

in the car business, you know, things take time.

3:44

Problems take time to solve, especially problems

3:46

like we don't have good new product that can

3:48

compete in the market. And so you

3:51

know, we've sort of reached this point now where

3:53

it seems like Elon is not even

3:56

really trying to save the reality

3:58

of this business and is sort

4:00

of all in on perception. And

4:03

that's sort of taking a bunch of different

4:05

forms.

4:06

So in practice, though, what does

4:08

it mean that it's escaped reality?

4:11

So you know, Tesla's a very

4:14

nearly a two million unit a year car

4:16

business, and having built that up

4:18

from nothing is like incredible,

4:20

Like it's a historic achievement in the

4:22

world of cars. The problem is

4:24

is that that seems to have peaked, right,

4:27

So they did about one point eight five million last year,

4:29

and essentially sales are going down and

4:31

they are slashing prices and slashing margins

4:34

as fast as they can to keep that decline

4:36

from getting any worse. But

4:39

really what happened is that

4:41

is that during COVID they

4:43

were able to print through a number

4:45

of sort of unique set of circumstances,

4:48

they were able to sort of print really impressive

4:50

profits that made this business

4:52

seem very very real, and it was very

4:55

very real. The problem is has created this complacency.

4:58

In order to record those profits, they've basically

5:00

been starving research and development and new product

5:02

investment in new product. And so now

5:05

now that sales are declining, you know, they can cut

5:07

the prices to kind of slow that decline,

5:09

but the only thing that's going to actually turn it around

5:11

is new is actual new product, and

5:14

those investments haven't been made. And and he

5:16

had the opportunity on the call to say, you know, we

5:19

understand the problem. We're taking it seriously. And

5:22

he sort of vaguely mentioned like new product

5:24

is coming, but he did not

5:26

sort of describe it in in the

5:28

kind of credible way that that is, you know,

5:30

sort of makes it clear that that there is

5:32

actually a plant to solve that problem.

5:35

And that was the latest invest the cult just to be specifics

5:37

yeah.

5:37

And and so instead of that, right, he's

5:40

he's gone all in on this idea that Tesla's an

5:42

AI company, that it's a you know

5:44

that's self driving this. They're going to show

5:46

a robotaxi, you know, and

5:48

and and that's that's pumping the stock,

5:50

right, that's doing the thing that that traditionally happens.

5:52

The problem is is that those things

5:54

don't make any money, Like, they don't even make revenue,

5:57

let alone profit, right.

5:58

And uh, the AI site lose

6:01

the money.

6:01

Yeah, yeah, and and again you know what's

6:04

what's really puzzling and

6:07

troubling about what's going on right now is

6:09

is, you know, Tesla has on paper

6:11

something like thirty billion dollars in cash

6:13

a little bit less, right, which

6:16

is enough to really solve like a lot

6:18

of these problems. And if and if he gone

6:20

on on that call and said, you

6:22

know, we're gonna we're gonna take ten billion dollars and we're

6:24

gonna use it to to really like invest

6:27

in a whole new generation of products and

6:29

give them some detail about what that was, you know,

6:31

I think things would be we would

6:33

be having a very different conversation right now. And

6:35

instead they're spending what are they spending money on? They've

6:37

been spending on since the pandemic, you know, the

6:39

cyber truck, you know,

6:41

and and now sort of lots of GPUs.

6:44

They they're in like this race to buy more

6:46

of an Nvidia's production than you know,

6:48

these other big and by the way, very very profitable

6:51

tech companies.

6:52

Can I just ask a quick question though? You say these

6:54

GPUs so like the ones used

6:56

to train models and run models like Cope

6:58

and Ai. All these the same GPUs

7:00

that Mosk bought for Xai his AI

7:03

company attached to Twitter, or these

7:05

a completely different set.

7:07

Presumably they're completely different. You know, Tesla

7:09

is a publicly traded a publicly traded

7:11

company and X's is private.

7:14

So you know that said he

7:16

does also very much, you know, blur

7:18

these these boundaries within his sort of empire,

7:20

and frankly, you know, in ways that are not

7:22

always strictly legal. So there

7:24

may be it may be that Tesla is using some

7:27

of Xai's hardware and

7:29

vice versa. It's

7:31

hard to know for sure.

7:33

So walk me through this NHTSA report

7:37

what happened there? Because it looks bad?

7:39

Yeah, so some history is kind

7:41

of important here, right, So, so first

7:44

of all, you know, the Tesla has

7:46

admitted that crashes have involved

7:48

autopilot including fatal one since twenty sixteen.

7:51

The very first time, right, and people don't always know it was

7:53

a guy in China whose family had

7:55

the brilliant idea of, you know what if we like

7:57

maintain chain of custody on the vehicle

8:00

data. And that was the first time

8:02

that Tesla admitted, oh yeah, out of pilot was actually

8:04

involved. And so this may have been going on

8:06

for even longer than anyone realizes.

8:09

NTSB, which is, you know,

8:12

this investigative body. They don't have any regulatory

8:15

power, but they're really good investigators. They mostly look

8:17

at air crashes. They were really

8:19

early and looking at three fatal crashes

8:21

that happened between twenty sixteen and twenty eighteen.

8:24

And they concluded essentially that

8:26

Tesla's system looks sort

8:28

of self driving enough, but

8:31

it operates in areas you're allowed to use it in places

8:33

where it doesn't it's not designed for, and people

8:35

become complacent and they stop paying attention.

8:37

It's not good enough to trust your life to. It's

8:40

just good enough to kind of make you complacent

8:42

and not paying attention when when it runs

8:44

into something it can't handle.

8:45

And these are the findings of the old report.

8:47

This is the NTSB. Yes, there's a different body, and

8:50

they recommend it to NITSA, who has maybe

8:52

less good at investigating stuff like this, especially

8:55

with human factors, but has all the regulatory power.

8:58

And they said, look, this is a problem, you know, And

9:01

knits A had an

9:03

enforcement guidance at the time that said, you know,

9:05

if the a system like autopilot,

9:09

you know, is prone to foreseeable misuse,

9:12

you know, that can be a defect and we can recall it. And

9:14

yet somehow that was never used from

9:16

twenty sixteen, and it wasn't until twenty

9:18

twenty one that

9:22

NITSA finally said. You

9:24

know, they did two things essentially the summer of twenty twenty

9:26

one, they opened an engineering analysis

9:28

of autopilots, sort of the first step

9:30

towards identifying a defect and

9:32

ordering a recall. They also, at the same

9:34

time, and the connection here was

9:36

not always obvious, they started collecting

9:39

data from across the industry and so

9:41

now when you have crashes

9:43

that involve any of the other

9:46

sort of level two driver assistant systems out

9:48

there, you have to report that to the government.

9:50

So they've been simultaneously since

9:53

the summer of twenty one investigating

9:55

Tesla specifically, but then also

9:57

collecting data from the rest of the industry to kind of get

9:59

a sense of of, you know, is this a unique

10:01

problem to Tesla, And pretty

10:04

clearly the answer to that was yes, because

10:07

what we've learned is that in December

10:09

of twenty twenty three, you know, they

10:12

they basically took you know, the these

10:14

findings that showed, you

10:17

know, a good deal of of

10:19

crashes happening, including fatal ones,

10:22

and basically forced Tesla to do a recall.

10:25

They did that in December over two million

10:28

vehicles, and they did it with an over the air

10:30

software update. And for

10:32

me, you know, having watched nits a sort of

10:34

drag its heels frankly or at least move

10:37

very very slowly to address what I think is a it's

10:39

been a pretty understandable problem

10:42

and and a recall worthy problem

10:44

for for a long time now. I

10:47

kind of thought, you know, okay, they're going to take let Tesla

10:49

do a software update, pretend like something has happened,

10:51

and sort of move on. The thing is is that

10:53

we were still seeing crashes happen. In

10:55

fact, there was a fatal crash literally the day

10:57

before. Uh, you know this this most

11:00

recent earnings call. And so

11:02

now what what Nitsa's

11:04

is doing is they're actually looking into the

11:06

remedy to that recall. They're saying, you

11:08

know, well, we're we because these things

11:10

keep happening, we may we think that maybe

11:13

just updating the software wasn't enough to

11:15

actually fix this problem. That to

11:18

me is it's a very rare for NITSA

11:20

to do one of these it's called like a recall

11:22

query. Uh, it's very

11:24

rare for them to do that. That strongly suggests

11:27

that that that they're really actually going to hold Tessel.

11:29

To account on recall query something that

11:31

happened before this new report, or is that

11:33

what this current report is.

11:34

So this so what's really interesting is is that

11:37

this report is essentially the

11:39

results of nitsa's investigation

11:41

over since since twenty twenty one essentially,

11:44

And and they go through and they describe sort of, you

11:46

know, how they found out about all these different kinds

11:48

of crashes and how they sort of analyze them and

11:50

basically about half of them they were able to kind of throw

11:52

them out right away, and then from the

11:55

other half they're able to drill down and identify,

11:58

you know, a number of kinds of crashes that to

12:00

keep happening that are all indicative

12:02

of this problem that you know NTSB

12:05

identified, you know way back in twenty

12:08

eighteen, twenty nineteen, twenty twenty and

12:11

and so so what was interesting is is that data

12:13

is they had taken that to Tesla and

12:15

basically, reading between the lines,

12:18

strong arm them into the recall. And that's

12:20

often how recalls happen is that either

12:22

either the automaker does it voluntarily

12:25

or the or the investigator or the regulator

12:27

brings a body of evidence to them and says, listen, like

12:30

you either do this or we're gonna or we're gonna do it

12:32

for you.

12:33

So, so this nh TSA

12:35

report, is it

12:37

does it do anything? Or is it just is this

12:40

something Tesla received before it came out?

12:42

Like how was this delivered and what happens

12:44

as a result of it.

12:45

So, so usually this stuff usually

12:48

what happens is that right they build up

12:50

this this body of evidence, they take it to the automaker,

12:52

they essentially use it to force them into a

12:54

recall, and then once the recall happens,

12:56

usually you never see it. It doesn't become public.

12:59

So the fact that this is public is

13:01

huge.

13:02

Yeah, that was what confused me. This feels like

13:04

a strange document for everyone to see.

13:06

Yeah, and it is absolutely

13:09

I mean, look, everything about this is kind

13:11

of novel territory. NITSA has not really

13:13

gotten into this sort of automated driving driving

13:15

assistant stuff before, so there's no playbook here.

13:18

But within the context of automotive

13:20

regulation, it is rare and and

13:22

and what it implies is that they

13:25

forced this recall. Tesla

13:27

did the easiest thing possible,

13:29

which is, we'll just update the software over the air.

13:32

And NITSA, what did that update

13:35

do if you if you've used it?

13:37

So so it appears have been a couple

13:39

of them. And and I think and that's one of the things

13:41

Nissa's looking into is exactly what what did

13:43

you do? But but fundamentally, the only

13:46

thing that they really could do was

13:49

to essentially create a lot more nags

13:51

in the system. And nature when you know, you have

13:53

the hands off the wheel or whatever, and

13:55

and the system is like, you know, take control, take

13:58

control, take control. What people love, what

14:00

consumers love about autopilot is that

14:03

it doesn't do that very much, right. It kind of

14:05

lets you sort of sit back,

14:07

which is which is the problem. Right, people like

14:09

the unsafety.

14:10

The feature and the problem it seems exactly

14:13

yeah.

14:13

And so and so you know, and and by

14:15

the way, Tesla has has gotten around that

14:17

sort of by making a very very

14:20

misleading safety claims, statistical

14:22

safety claims about autopilot. So people they're

14:25

getting their cake and they're eating it too, right, They're they're

14:28

the system is designed really to look

14:30

self driving, so that kind of helps the

14:32

stock price. It's it's designed

14:34

to enable you to kind

14:36

of look away and do other things that you shouldn't

14:39

be doing while you're driving, but it comes

14:41

with this statistic that is comforting

14:43

where it's like, no, this is actually safer

14:45

than a human too. So, so Tesla is

14:47

kind of that. This is why it's so popular. It's

14:49

the lack of safety with the

14:52

veneer of a of a fake you know,

14:54

safety statistic is what's made it so popular.

14:56

And frankly, this is why I've been skeptical

14:59

that and it's a wood really do anything.

15:01

But the fact that they forced

15:04

the recall. Tesla took

15:06

the easy route. NITSA could have just said,

15:08

yeah, okay, we we've gone through the motions here,

15:11

let's let's move. We've done yeah,

15:13

let's move on right, But instead, because

15:15

these crashes are still happening, you

15:17

know, NITSA feels

15:19

the need to to not just

15:22

say, you know, we need to look at at

15:24

what you did to address

15:26

this recall and make sure that it's actually solving

15:29

the problem. Implicitly, we

15:31

we don't think it is. But then it also

15:33

released this this data that

15:36

that to the public now, so now now all

15:38

of us can go and look and say, okay,

15:40

yeah, like there's a reason that this

15:42

recall happened. This isn't just you know, you

15:45

know, dark Brandon, you know, cracking

15:47

down on Elon because because you know, you

15:49

can't handle his realness or whatever. Yeah,

15:52

yeah, like this is not just some politically motivated

15:54

thing like like and this is the flip

15:57

side of NITSA taking their time on this. As frustrating

15:59

as it's been, They've built up a lot

16:01

of data not just about Tesla but about

16:03

the rest of the industry that shows Tesla

16:06

does have a unique problem here. And

16:09

I think you know what they're going to show is that is

16:11

that the update they've done so far isn't

16:13

going to be enough, and that really leaves Tesla

16:15

in a pickle because there's not a

16:17

lot of other great options for fixing these problems.

16:30

And it seems also that this report basically

16:33

gives plaintiffs the ability

16:35

to sue Tesla on some level,

16:37

it seems like this will create a bunch

16:39

of litigation oh against the company.

16:41

Yeah, and there already has been you

16:43

know Tesla just right also right

16:46

before this most recent call, they

16:48

settled a lawsuit dating back

16:50

to a crash, back to a twenty eighteen that

16:53

you know that that lossuit have been going on for a really long time.

16:56

And I'm sure what nis

16:59

Is done has has played a role in that.

17:01

Absolutely all this data,

17:03

everything that KNITSA has put out in the public is

17:06

just it's just like handing sort of loaded ammunition

17:09

clips to to all the lawyers out there. And

17:11

frankly, that's kind of how regulation in this

17:13

country works. You know, Uh,

17:15

our regulators will do what they

17:17

have to when they have to, when when it becomes unfeasible

17:20

for them to sort of ignore stuff. But generally speaking,

17:23

a lot of regulation effectively happens

17:25

through uh, you know, lawsuits,

17:27

through civil civil legal litigation.

17:31

So do you think that this leads to them

17:33

actually having to do something with the autopilot or

17:35

are they just gonna hope that that don't get

17:37

sued too much. It feels like they may.

17:40

This feels like a normal company would

17:42

just pull Auto Pilot entirely.

17:44

Yeah, So, so I think that's it's

17:47

gonna it's gonna come to something like that. So

17:49

because this over there update

17:52

didn't so, so so they did over

17:54

their update that that kind of keeps people being

17:56

nagged more, right, and and A

17:59

it's anest stopping the crashes from happening, but B it's

18:02

really eroding what people like about the product.

18:05

And and essentially again like it gets back to

18:07

Tesla, you know, is willing to

18:09

create products that lawyers at other companies

18:12

would just put the kabash on. They wouldn't

18:14

let it happen. So so Tesla kind of assuming

18:17

that nits a you know, is going

18:19

to find that that the current fix is

18:22

insufficient. There are sort of two basic

18:24

routes that that Tesla can take. One

18:27

is they can dramatically

18:29

reduce sort of the it's called

18:31

the control authority and the and the capabilities

18:34

of the system. Essentially, the

18:36

the you know, a system, a driver

18:39

assistant system should be designed to assist

18:41

the driver and the fundamental

18:43

flaw of of autopilot is

18:45

that it actually is more designed to look

18:47

like it the car is almost self driving, and those are

18:49

two different things. So instead of the automations

18:52

assisting you.

18:53

Yeah, how would they different actually really get into

18:55

that?

18:55

Yeah, so okay. So the

18:58

feature that has the best proven

19:01

record of improving safety outcomes is called

19:03

automated emergency braking, And

19:06

essentially people don't even know it's on the car.

19:08

But if you get yourself into a really sticky

19:10

situation, someone cuts you off something

19:12

like that, the car will there's forward collision

19:14

warning. The car will warn you, and then the car

19:17

will actually break itself that combination

19:19

of those two things, and again people don't even really know

19:21

they're there for the most part. You know,

19:23

something like a forty percent reduction in frontal

19:25

crashes. Okay, so like proven

19:27

statistical safety advantage,

19:30

and it's because

19:33

the critical piece of it is that you

19:36

can you don't rely on it. No one sits there and says,

19:38

oh, it's cool if I just accelerate into you

19:40

know, this semi truck.

19:41

Or whatever, I'm just going to crash into various

19:43

objects and see what happens.

19:44

Yeah, you don't get this over reliance.

19:47

But when with these level two systems, and

19:49

it's a combination of how the system is designed,

19:52

right, it's designed to look as if it's

19:54

self driving, convince people itself driving,

19:57

and then puts the driver in what's called

20:00

a you know, a vigilance task, which

20:03

which means, you know, and people are constantly talking about

20:05

what bad drivers humans are. The reality

20:07

is we're actually considering we're not evolved to move

20:09

at these speeds and everything. We're actually pretty darn good

20:11

at it. We just drive a lot of miles

20:14

and over those miles, bad things happen. The

20:16

the what we're worse at than

20:18

driving are these vigilance tasks. And we know

20:21

this from research going back one hundred years.

20:23

And like you know, radar operators in World War

20:25

Two. You know, you sit there and you

20:27

force someone to watch a screen and then

20:29

when you know, one little, you know thing

20:31

happens, you know, you have to respond within

20:34

a very very short amount of time. This is called a vigilance

20:36

task. We're terrible at that. And if you think

20:39

about, you know what that can

20:41

mean. You know, things happen fast on the freeway if

20:43

you're even slightly not paying attention, and

20:45

all of a sudden there's a situation. You have to be able

20:48

to read what's going on, decide

20:50

on the right course of action, and then implement

20:52

it properly. I mean, this is insanely hard

20:54

for people to do and and

20:56

and it creates, you know, these

20:59

kinds of thing fafety problems. So

21:03

so you know, I think fundamentally this is

21:05

this is the challenge that tessels up against, right,

21:07

is that is that they're selling this

21:10

as a safety thing, but there's no safety

21:12

record. You know, IAHS has all of the insurance

21:14

data and they say there's no record, there's

21:16

no evidence that that any level two

21:19

system, which is what what these are called, has

21:21

any measurable impact on safety.

21:23

And is level four completely autonomous?

21:25

Yeah, Level four is completely autonomous within

21:27

a restricted area.

21:30

Two.

21:30

Level two is essentially uh

21:33

automated assistance of

21:36

of of two axis of control.

21:38

It's it's kind of confused, but essentially it's

21:40

it's adaptive cruise control. A lot of cars have adaptive

21:42

cruise control, and that's where you you have cruise

21:44

control. It holds the speed and then if there's

21:46

an obstacle in front of it and matches speed with that.

21:49

So that's the longitudinal control. And

21:51

then and then you have lane keeping, which

21:53

essentially keeps you within a lane. And and then you

21:55

know, they build on that a

21:57

little bit by by having you know, a nowtigation.

22:00

You know, so if you're going to take an exit, it'll start

22:02

getting you over into the next lane instead of just

22:04

holding you in one lane. But essentially,

22:07

you know, these level two systems

22:09

exist in other brands. Tesla's

22:12

is the most popular because

22:15

they've implemented it in a way that makes it seem

22:17

more self driving than others. And again, right

22:20

that that element

22:22

is what makes it unsafe. And so it

22:25

sounds hyperbolic, but there's really a

22:27

very direct through line between you

22:29

know, design decisions

22:31

that endanger people and

22:34

you know, the stock, because that's what it's

22:36

for, right, It's not to keep people safe, it's to convince

22:39

people that Tesla's a leader in self driving car technology,

22:41

which they actually aren't.

22:42

Seems not kind of a sham, like he's just

22:44

trying to make it seem autonomous when

22:46

it's not even good at autonomy. Because

22:49

the people who buy stocks don't seem to pay attention.

22:51

Yeah, So I mean it's it's a fascinating situation

22:54

where yeah, like the word right, So then there's

22:56

the Really it's a way

22:58

of building up this this frankly

23:00

scam of self driving, right, because that's

23:02

they're getting people to pay ten fifteen

23:04

thousand dollars a car for this full

23:06

self driving you know add

23:09

on. And people wouldn't do that

23:11

unless they had some reason to think

23:13

that that, you know, this

23:15

is something that is somewhat near and

23:19

essentially, again, like they have to endanger

23:21

people to create that perception. So a

23:24

scam is just sort of ripping

23:26

people off. This does more than that. This

23:28

endangers people in order to rip them

23:30

off. It's almost like we need a new word for for

23:33

how bad this is. You know.

23:36

Yeah, it's interesting

23:38

because any other company, someone

23:40

would be in prison, someone would be arrested,

23:43

maybe someone would be sued by the government for like billions

23:45

of dollars. This feels like it will

23:48

continue to kill people unless

23:51

something changes. But it also doesn't sound like Elon

23:53

Musk will actually change anything.

23:55

Well yeah, I mean he's going all in on self

23:57

driving, right, I mean.

23:58

He doesn't seem to be. When he's going all in on it, it

24:00

doesn't seem to be doing anything to might get better.

24:03

Well yeah, and that's because fundamentally,

24:06

the you know, Tesla's approach to the technology.

24:09

They had to find a way to make self

24:11

driving technology work with their existing business

24:13

model, and they did that by

24:15

using very cheap hardware. Essentially, so

24:18

if you look at way Wayme's really the only company

24:20

that is really actually doing self driving.

24:22

They have robotaxis in San Francisco, and like

24:25

you may love them or hate them, but like it is incredibly

24:28

impressive to be in a completely drivers vehicle

24:30

in somewhere like downtown San Francisco. They're

24:32

doing it, and they do it through two things,

24:34

like fundamentally, one is that they

24:37

limit the domated operates in so it only

24:39

operates in San Francisco. And then they're you know, they're

24:41

expanding to new markets. But it's not a general

24:43

solution. You create a model that works

24:46

in a specific area, and

24:48

then the and the other thing you have to do. It's sort of like

24:50

turning it into a board game, right, Like

24:52

like you beat a human, you have to bound

24:54

the complexity. Like in the bounded

24:56

complexity of go or chess, an AI

24:59

can beat us. The other thing that AI needs to

25:01

beat us at a game is perfect intelligence,

25:04

I'm sorry, perfect information rather which

25:06

means right in chess or go or whatever. You know,

25:08

everyone knows exactly where each piece

25:11

is. There's no confusion about it. And WEIMO

25:13

does that with these really really robust

25:15

sensor systems incorporate ldar and

25:17

radar and.

25:18

Also and

25:21

where it goes.

25:22

Yeah, so it's the combination of those two things.

25:24

And and the problem is that those two things are incompatible

25:27

with cars. No one is going to spend

25:30

you know, maybe someone would spend three hundred thousand dollars on

25:32

a on a car that they didn't have to drive,

25:35

but but not if it only works in

25:37

San Francisco. Right that live cars

25:39

have to be able to go wherever we want them to go, and

25:41

they have to have a market. You know, they can't be too

25:44

expensive. And so the

25:46

things that you need to make real self driving

25:48

work just aren't compatible with self driving.

25:51

And and so essentially

25:54

what they've done is is just use

25:57

you know, cheap hardware that that isn't

25:59

too expensive and that

26:03

does just enough to make people think

26:05

that it's self driving. And I think you know, one of the things

26:07

we've learned is is, you know, people

26:11

we bring forward our ideas about driving from

26:13

humans. If you see a human, like

26:15

if you see a kid driving

26:18

and like doing a driver's test, right, Like

26:20

if you if you can do a driver's test, that

26:23

means for a human that you have the basic skills

26:25

that you can sort of generally apply them everywhere. But

26:27

the problem is that machine learning doesn't work that way.

26:30

Right with machine learning, you

26:32

there's nothing the generalizability of

26:35

AI is you know, it's

26:37

a huge topic, right, everyone wants to

26:39

believe in it, but certainly

26:41

when it comes to something that is safety critical,

26:44

where you know, it's it's one thing for a large

26:46

language model to screw up and hallucinate and everyone

26:48

laughs and you know, haha, that's

26:50

funny when it comes to driving, right, what you're

26:52

doing is you're reconciling a probabilistic

26:55

system with the need for ninety nine point

26:57

nine nine nine nine percent reliability.

27:00

And that point one zero zero

27:02

zero one percent is where people die, yes.

27:05

Because because we drive millions and millions and millions

27:07

of miles, right.

27:08

And Americans drive too much as well, so

27:11

that's dangerous to our roads are

27:13

worse and so

27:16

yeah.

27:16

So what's amazing is

27:18

is that is that you know humans are bad at babysitting.

27:22

You know, that's essentially what it would have forced you to do right, you're

27:24

babysitting like a teenager. Essentially, who's

27:26

driving. You put a teenager behind the wheel, you

27:28

babysit them. We're not necessarily good

27:30

at that, But from Tesla's perspective, it doesn't

27:32

matter, because what's important

27:34

is that we get the liability. Right

27:37

they so and like a vehicle.

27:39

Except this report is going to potentially change

27:41

how that is viewed by judges.

27:43

Well, yes, so, so you know Tesla

27:46

is not so so a vehicle

27:48

becomes self driving when the when the manufacturer

27:50

or the owner operator takes legal

27:53

liability for it, right and and

27:55

and so essentially, none of what Tesla's doing is

27:57

is is self driving,

27:59

because it's stakes humans with

28:00

the consequences. And Madeline

28:03

and Claire at least had a great paper a number of

28:05

years back called moral crumple zones.

28:07

And that's essentially what Tesla's doing is easy humans

28:09

as moral crumple zones. And

28:11

and and that's you know, the

28:14

way they've architected the system. It's it doesn't

28:16

give us a good chance of catching

28:18

the system's mistakes. But again

28:21

it doesn't matter because we're just there as the crumple

28:23

zone. We're just there to take

28:25

the blame for the system's mistake

28:27

from from Tesla's perspective.

28:29

Taking a step back, what does Tesla

28:32

actually do now? What will will

28:34

they change? Will they just keep

28:36

sitting there? And Ela mus say, it's actually epic that

28:38

the cause kills people. It's good we need we

28:40

need more babies, but we need less adults. Yes,

28:43

Like what is it?

28:44

So? So, what's happening right now? Right? So sales

28:46

have peaked and they're cutting into

28:48

their prices in their margins in order to keep it,

28:51

you know, sort of from from falling even

28:53

further. Uh. The problem

28:55

is is that what's left of those margins

28:57

depend very heavily and and just the

29:00

so the volume, demand and the and the profit margins

29:02

depend very heavily on autopilot and full self driving

29:05

because these are like unique to Tesla,

29:07

and they're unique reasons to buy a Tesla.

29:10

And and frankly, you know, with with full self

29:12

driving, that's ten fifteen thousand

29:14

dollars per car in an industry

29:17

where like you know, people have massive

29:19

fights at the at the development level over pennies

29:22

per car. To add ten thousand dollars

29:24

in pure profit on a car, this paper's

29:26

over a multitude of sins financially,

29:29

you know. And and so if the

29:32

prospects here is that not only

29:34

are Tesla sales falling, will continue

29:36

to fall, you

29:39

know, without because there's nothing, there's no new product

29:41

to turn that around. But then also you take

29:43

away this incredible

29:46

uh, you know, advantage in terms

29:48

of profit margin, right this ten to fifteen thousand,

29:50

even at the take rate is ten percent

29:52

ten thousand and fifty thousand dollars per car. Is you spread

29:54

that over your whole fleet, and and it's

29:57

it's by auto industry, and it

29:59

is a huge profit.

30:00

But will they have to pull autopilot out.

30:02

If they do? And I again my sense

30:05

is that this is where N's is going. If they pull autopilot,

30:08

you know, all of those same problems also apply to

30:10

full self driving. All of a sudden, Tesla

30:12

sees another reason for

30:15

their volume to go down. Right, there's people who

30:17

are now not going to buy the car because they

30:19

know it's it's autopilot, isn't safe, and

30:22

it doesn't have those things. But then it turns

30:24

into a negative margins. And the thing is their margins

30:26

have been compressing, compressing, compressing, and

30:28

they're at the point now.

30:29

It turns into negative margins. Though the auto

30:32

the car business might be the car business. So is

30:34

there a delineation between full self driving and

30:36

autopilot or are they the same thing?

30:38

They're fundamentally the same thing. It's just the autopilot

30:40

you're only supposed to use it sort of on freeways,

30:43

and full self driving you use

30:45

it sort of on city streets and everywhere

30:47

else. So it's the differences in what's called operational

30:50

design domain. But also the difference

30:52

is autopilot is like

30:54

the price is built into the cost of a Tesla, whereas

30:56

full self driving is an optional extra on

30:59

top of that.

31:00

So what would get pulled out then?

31:02

I mean, in theory both of them, because they both have the

31:04

same sort of fundamental problem, right,

31:06

and.

31:07

And Elon actually do this? Though would he

31:09

actually do it is my question because

31:11

he's he's a dickhead, as we well

31:13

know. But also this is the

31:15

one thing he spent the last month or

31:17

so, just like autopilot is the best part of

31:19

the business. We don't even need the car,

31:22

like he seems to be selling

31:24

this hard Yeah.

31:25

Yeah, no, So so the Nitsa

31:28

does you know they move so slowly,

31:30

they're so tentative. They're so uh, you

31:32

know, hesitant to confront an

31:34

Elon Musk type character, which is why this is

31:36

all taken so long. But they

31:39

have an immense amount of power essentially

31:41

they can do. They can order a mandatory recall, and

31:43

then they can even order a mandatory

31:46

stop sale. They can say it's illegal

31:48

to sell Tesla's in this country until

31:50

you essentially deactivate

31:52

the system or or or implement some kind

31:54

of fix that we deem is appropriate.

31:57

So that's probably impossible to

31:59

fail.

32:00

Though, so so fixing

32:02

right, So, so there's two ways to fix. One

32:05

is that you dramatically reduce the capability

32:07

of the system and probably

32:10

increase the nags even further and so basically

32:12

destroy the value that people want. That's

32:14

one way, and that's probably the most likely

32:16

way. The other way is to actually implement like

32:19

better hardware for driver monitoring, because

32:21

Tesla.

32:21

Uses which would cost tests or a great deal of money.

32:24

And it's really hard, if

32:26

not impossible, to essentially pull your entire

32:28

fleet in and like actually install hardware

32:30

that you know, you don't have the wiring harnesses for it,

32:32

you don't it In theory, it could be done,

32:34

but I think it's basically economically impossible,

32:37

and so then you have this prospect of all of a

32:39

sudden, Tesla can no longer do

32:42

autopilot and full self driving, and given

32:44

what's happening with their margins now, Tesla then becomes

32:46

a negative margin business, which means

32:49

you don't make it up on volume. Right, every

32:51

car you sell you lose money on and

32:53

and one of the ways that I'm

32:56

you know, and I'm I'm sort of along with everyone else,

32:58

trying to puzzle through sort

33:00

of some of the decisions that are being made here.

33:03

But one one scenario that

33:05

kind of potentially makes this all make sense

33:07

is that Elon kind of gets it

33:10

that this is going to happen, that that

33:12

NITSA is not screwing around. They don't want

33:14

more deaths on their hands. They're going to they're

33:17

going to force the issue on this, and

33:19

that that that the margins

33:21

will go negative and there's no new product to turn it

33:23

around. Maybe,

33:25

you know, Elon really like it

33:27

could explain some of his behavior if he's

33:29

just sort of come to terms with with, you

33:32

know, the core business is going

33:34

to die and sort of like with Twitter, right,

33:37

you know, he thinks in the way his mind works

33:39

he thinks he can sort of go and tell

33:41

Earth, like, oh, the regulators or the adverage.

33:43

Right in the case of Twitter, he's like, he's like, I'm just gonna

33:46

tell you know, Earth that the advertisers

33:48

killed Twitter. I don't remember that

33:50

that interview.

33:51

Yeah, so oh oh, I've got it engraved in my brain.

33:53

He he may pull that like. That's

33:55

that's one of the you know, we

33:57

have to like sort of put ourselves in the mind

33:59

of of someone who's obviously not

34:02

very normal.

34:04

Yeah, I'll say it's my podcast, I

34:06

don't care.

34:07

Yeah, so so I mean that that may

34:09

be one way to explain all of what's going on, because.

34:12

Otherwise I would just go out and say that the wokes

34:14

of stop Torto pilot, but

34:16

he would actually pull it. You think, well,

34:18

I mean I.

34:19

Think you would have no choice but to put like like

34:21

I think he's I think right,

34:23

because because if

34:25

it gets pulled, if he if you can't have autopilot

34:27

and full self driving, it hits the

34:30

right, it hits the volume of sales, it has, the

34:32

hits the profit margin on each sale, and it

34:34

hits the stock.

34:35

It hits and it kills the Robotaxi idea which

34:37

was already completely stupid. It just kills

34:39

that you can't have that anymore.

34:40

Yeah, yeah, no, and and and even that

34:43

is a weird thing to pivot to as

34:45

well, because.

34:46

It's just a stupid It's just to be

34:48

clear, every single journalist who wrote about the robotaxi

34:50

thing without rolling their eyes and writing that they

34:52

were doing so committed some level of

34:55

malpractice. In my opinion, it's just not

34:57

going to happen. It's complete bullshit.

34:59

Yeah, yeah, yeah, I completely agree.

35:02

I mean, look the idea that

35:04

you know you can do sort of

35:06

level five self driving, which means fully

35:09

you know, automated, no human monitoring

35:12

or anything, and not in a limited

35:14

domain, but everywhere, Like, no one else

35:16

is even selling this. Tesla's been selling

35:18

it since twenty sixteen. No one else

35:20

is even selling it. If it were possible, wouldn't

35:22

one company want to compete on it? Right?

35:25

Yes, economics

35:27

logic, surely this seems very like

35:29

if you're saying that cars are generally

35:32

thin margin businesses and you suddenly have a

35:34

way to sell software on it, surely someone else

35:36

would try. Someone like Forward, for example, who

35:38

has tons of cash, government subsidies,

35:41

tons of brand power. They would also do

35:43

this, except if it was too dangerous.

35:45

Yeah, exactly. And so what you do see

35:47

from Ford in general motors and others is that they do they

35:49

have level two systems like autopilot that are

35:51

for the freeway, but they have much more robust driver

35:54

monitoring and things like that. And they're not going around

35:56

saying, you know, you'll be able to you

35:58

can buy a car that will some day drive itself

36:01

completely by its own, uh, anywhere

36:03

you want to go. Everybody

36:05

in the business knows

36:07

that that is not serious and has known for

36:09

a really long time. And unfortunately

36:12

we're in a weird situation where it's

36:15

like no one has called it out, and so you

36:17

know, he's again. It'll be eight years

36:19

this fall that they've started since they've

36:21

started taking money for for

36:24

this just blatant scam.

36:26

And and I think the reason that people have

36:28

got that they've gotten away with it so far. Obviously

36:31

it's nothing to do with with technical plausibility,

36:34

although it does sort of tie into, you know, the AI hype

36:36

that we see elsewhere, so people think that AGI

36:38

is near. You know, it kind of makes sense why

36:40

they might think that it might.

36:41

Be because the shadows on the walls of the cave that

36:44

suggests it's possible.

36:45

Yes, But but the real reason,

36:47

I think is that you know, when

36:50

you say the word self driving car, what

36:53

people would Americans in particular, think is

36:55

a car that drives itself and

36:58

what is it?

36:58

I mean, that's probably because those the gold damn

37:00

words. But that's the

37:02

thing. No, No, it's misleading in its face.

37:05

Yeah, yeah, exactly. But but what people don't

37:07

understand is that where

37:09

autonomous driving actually does work

37:12

is in these robotaxis that

37:14

are fundamentally different than cars.

37:17

Right, a car you have

37:19

to be able to buy it and own it, and

37:22

a car also goes anywhere you want to go, right

37:26

and so and so where this technology

37:28

works, the way in which it works is

37:30

fundamentally different than a car. So

37:32

that's why no one else is trying to

37:35

make a self driving car, right,

37:37

It's because it's impossible. But

37:40

but Elon's willing to do it. And he gets away with it because

37:42

it's like he's selling a piece, a

37:44

puzzle piece that fits into the empty spot

37:46

in people's brain. He's the only person

37:48

who's selling the mental model that people

37:50

have for self driving car. And

37:53

this is why with like Wimo is actually

37:55

doing it. They have actual driverless vehicles,

37:57

you can get actual rides in and

37:59

and no one stops and is like, wait

38:02

a second, how is how is it that they're able

38:04

to do this? You know? But

38:06

every year Elon says this is ready and then

38:09

and then it's not, like what's the.

38:10

Disconnecting IM saying it would be ready since like twenty

38:13

nineteen as well, he has been talking at his US

38:15

whole about this a minute.

38:16

Yeah he has, he's I mean, even before

38:19

the official announcement of full self driving, he had

38:21

a number of quotes saying he thought it was going to be you

38:23

know, within a year or two. And

38:26

there's just these long, long lists of his

38:28

quotes. No one has been as

38:30

wrong about this technology as he has.

38:33

And yet he continues to get more

38:36

confidence and faith, you

38:39

know, certainly from financial markets then

38:41

even the companies that are like proving it and

38:43

doing it the right way. And

38:46

I think that's a really like troubling

38:48

commentary on sort of the relationship between

38:50

technology and capital and in our society these days.

39:05

So changing subjects

39:08

slightly, Elon also very

39:10

recently laid off most of the Supercharger

39:12

team, which to me is one of the funnier things

39:14

he's done because the supercharger network

39:17

for Tesla appeared to be a completely

39:19

unregulated monopoly where everyone gave

39:21

him free money, and the entire industry

39:24

had started to like, most of the industry had started buying

39:26

into his charging standard. And then they'd said,

39:28

you know what, Elon, we'd actually like you to have more power.

39:31

Please just run this whole thing, and

39:33

then he fired most of them. What the hell happened?

39:35

Yeah, So, at a high

39:38

level, the weirdest thing about all of this is

39:40

that you know, Tesla

39:42

has money, Like Tesla on

39:45

paper at least there's almost thirty billion dollars in

39:47

cash, and so it should

39:49

be spending its way out of its problems and not cut

39:51

it. But instead Elon And this

39:54

is why I think this is as much

39:56

just sort of about him and sort of

39:58

who he is now rather than anything to

40:00

do with the business. Is that he

40:02

he likes He just cuts stuff that's his. It's

40:05

like, oh, you know, we're having trouble, Like, let's

40:07

get rid of the dead way, let's get rid

40:09

of cost. The problem is is

40:11

that is that this doesn't solve any of his problems.

40:14

I think with supercharging,

40:17

it's a little different. So so supercharging,

40:20

like Tesla itself is, was

40:22

a really critical piece of this sort

40:24

of going zero to one with the EV business.

40:27

Like I think you know, when Tesla at first got started,

40:30

they had to do something like the supercharger network,

40:32

right, it was there was it wasn't really optional,

40:34

like if it was key part of growing their market.

40:37

Yeah, because none could judge the cause otherwise.

40:39

Yes, And as long as it was Tesla exclusive,

40:42

it was an increasingly over time

40:44

as competition got got better, it

40:46

became sort of the reason to buy

40:48

Tesla's When he opened

40:51

it to others, that changed,

40:53

right then all of a sudden, it's no longer

40:56

you know, well, if you want access to Tesla superchargers,

40:58

you have to buy a Tesla.

40:59

No.

40:59

No, you can buy a Rivian, you can buy a Ford and

41:02

you get access to those things. It's no longer

41:04

a unique selling point for Tesla,

41:06

and I think that's one of the reasons you've seen demand

41:09

fall off. Right. There's a lot more competition now,

41:11

and a lot of that competition can

41:14

use the same chargers. I

41:16

think, you know, because it's

41:18

been so good and so reliable, people

41:21

assume that it's also a good business I'm

41:23

not sure that that's actually the case. Real

41:25

estate is very expensive. It ties up a lot

41:27

of capital. Yes, they're making

41:30

probably some gross margin on the electricity.

41:32

They probably sell the electricity for more than they buy it

41:34

for. So there is some kind of a business there.

41:36

But whether or not that's paying off the cost

41:38

of capital in a way that would actually be attractive

41:41

as a standalone business is not clear, And

41:43

I think there's reasons to suspect it may not because

41:45

it is essentially a it's

41:48

like a feature for Tesla cars. It's

41:50

like sort of putting cash on the hood or something

41:53

like that. Potentially, so the

41:55

business of supertruging has never really

41:57

been fully disclosed, and

42:00

you know, essentially it seems

42:03

like maybe this may actually

42:05

be one of the more rational decisions. And in

42:07

that, you know, the business itself

42:09

may just not be that good, you

42:12

know, Tesla. By opening it up, they'll

42:14

they'll get revenue from other, you know, owners

42:16

of non Tesla vehicles,

42:19

and in order

42:21

to make that business look good, they simply can't invest

42:24

more. There's no incentive in them to grow.

42:27

There have been estimates that suggest it's worth like

42:29

ten to twenty billion dollars, like it's actually

42:31

generating real revenue, but we just don't know,

42:33

do we.

42:34

Yeah, Like a lot of things with Tesla,

42:36

the accounting is very opaque, right,

42:38

and again, like you know, they supposedly have thirty billion

42:41

dollars in cash or almost thirty billion dollars of cash

42:43

on their books, and yet they somehow don't seem

42:45

able to like meaningfully spend their way out of

42:47

some of these problems. So I you know, I'm

42:50

not a friendsic account and I'm not going to make any allegations

42:52

about them like cooking their books, but I do know there's been a

42:54

lot of suspicion

42:56

about that over the years, and certainly

42:59

when it comes to Superchargers, it's never

43:01

been broken out in a way that would allow you to say,

43:03

oh, yeah, this is definitely something that can

43:05

stand alone. Frankly, if it were

43:08

like like if he, if

43:10

it were a standalone business, it would be something

43:12

you could spin out right now or

43:14

sell to a competitor or something.

43:16

Surely it would be something you would volunteer the information

43:18

for as well to show how good Tesla was.

43:20

But you also wouldn't fire the entire team

43:22

before doing it.

43:23

You also would not do that. No, but talking

43:25

of broken how about that cyber truck?

43:28

What the hell is going on there?

43:30

Yeah?

43:30

So so again, you know, it's

43:32

a question of priorities. You know, the

43:36

car business is a capital in tons of business.

43:38

And you know Tesla's been They've

43:41

had a lot of things go well for them, but

43:43

they've been resting on their laurels and

43:46

and you have to invest your money

43:48

in something in order to keep growth going in

43:51

the car business either, right, you keep explaining the superchargers

43:53

or the you invest in.

43:54

New cars or must cars.

43:57

Yeah, and so the soup, so

43:59

the the cyber truck like

44:03

in some ways there was it was

44:05

it was a brilliant idea to make a big truck

44:07

because if you can only compete in one

44:09

car market in the world, the combination

44:12

of volume and profit margin in full sized trucks

44:14

in America is the number one best business

44:16

period right right that

44:18

that business keeps the Detroit automakers

44:21

going. They barely make any money on anything other than trucks.

44:23

The trucks essentially subsidize most

44:25

of the rest of their businesses, and

44:27

so and so, you know, targeting that segment was

44:30

smart. Uh, Targeting with

44:33

essentially a meme was not smart. And

44:35

I think that to me, the cyber truck

44:38

is a symbol of sort

44:40

of uh, you know, Elon's ego,

44:43

sort of hitting escape velocity, uh,

44:45

and essentially reaching a point where he

44:47

can no longer take

44:50

advice, like whether it's just in terms of the

44:52

styling, you know, like like you

44:54

could have you could have done a wedgie

44:56

truck like that and and not made it look

44:58

like such garbage. If if he'd listened to

45:00

his his dialist Printsman whole thousand

45:03

is certainly talented enough to have made

45:05

a much better design or version

45:07

of that concept. It was clear Elon was

45:09

like, no, I wanted to be flat

45:11

and straight lines, and I wanted to look

45:14

like low polygons, like like.

45:15

A video game, like the game of Flashback.

45:19

Yeah. And I think it's like in his mind

45:21

the difference between a rendering

45:23

and reality is like blurry to

45:26

him. I think for him, like if it looks good

45:28

in a rendering, well of course it's gonna it's

45:30

gonna look good in reality.

45:32

And and you know, the rest of all of Tesla's

45:35

designs have been you know,

45:37

you can definitely tell the quality

45:39

problems if you know what to look for. But they've

45:41

got these curves and these different panels and these things

45:43

going.

45:43

On, and the guy other than designing

45:46

them.

45:47

Yeah, well, I mean obviously, as you know, they

45:49

have a designer. It's just that Elon

45:52

won't listen to anybody. I think that's I

45:54

don't know of any other way to explain, right,

45:57

Like, like the product

45:59

play, any team would tell him. Listen, Elon,

46:02

Like, we know this full sized truck segment. It's

46:04

so huge. We can build our next wave of growth

46:06

if we get the right product to this segment.

46:09

Right, Yeah, they made the model s

46:11

for trucks like an eight eight or one hundred

46:13

grand fucking brilliant truck,

46:15

which they are fully capable of doing, surely.

46:17

Yeah, well except that, except that Elon doesn't

46:20

believe in market research, rightft

46:22

true, and and and he goes with

46:24

his gut. And it's like if he thinks it's cool,

46:27

then it's going to work. And I think,

46:29

right, like, this has been true enough

46:33

for a long time.

46:34

But has it? Okay, I actually want

46:36

to push back on that. Has that actually been true

46:39

since like twenty sixteen? Because

46:41

what cool idea has Elon Musk had

46:43

that's worked in that time?

46:47

Well since twenty sixteen, So

46:49

I mean I've always argued that that, you

46:51

know, full self driving was also one

46:54

of the It was really the first time where he sort

46:56

of hit this escape velocity, like he's always had

46:58

these these hypeie kind of things.

47:00

Yeah, it's a good idea. I'm not saying it's not. It's

47:02

just the way he manifest today is the problem.

47:05

Yeah. And and I mean, you

47:07

know, I think throughout the history of the company, right,

47:09

like so like the Tesla Roadster, you

47:12

know, like originally it was just gonna be uh, the original

47:14

one was just gonna be like a lotus of lease with batteries

47:16

and an electric motor shoved into it. Elon

47:19

kind of both turned it into

47:21

a better product and a product

47:24

that really established Tesla's brand, but

47:26

also killed the financial viability of

47:28

it at the same time. Uh, you

47:30

know, and and and and at that point,

47:33

right it was all about building up hype. It

47:35

didn't have to perform as a business. He could he

47:37

could emphasize sort of brand

47:39

building and looks and feel over

47:41

the profitability because it was an early

47:44

stage. It was you know there they're pure play.

47:46

You know, it's it's experimental. Yeah,

47:48

yeah, so we just use it to raise more money. And then and

47:51

then we'll get serious right the process.

47:52

They've just never released the second Road stuff. It's

47:54

just never coming out. No, No, it's fucking

47:57

insane. Like there's so many people have written about this

47:59

thing.

47:59

Well, and people put money down on it. Remember

48:01

the Founder's edition, people put down fifty at

48:03

the entire full price two hundred fifty

48:05

thousand dollars up.

48:06

It's good. That is bonkers. That

48:08

is come on.

48:10

But but like this is the thing with Tessel too,

48:12

though, is that like, once you get away with something

48:14

for a while, how do you then

48:17

decide, Okay, this is not actually acceptable? And

48:19

I think this is kind of the problem that full self driving

48:22

an autopilot are having, right, and I think hopefully

48:24

knits A, you know, will take action and that will

48:27

be it. It's the government, right, like they

48:29

should be the ones who step in. And

48:32

frankly, I think, you know, if you want to look at

48:34

you know, who's what

48:37

what does this whole story sort of point

48:39

to in terms of being the underlying problem.

48:41

It's not Elon really in the sense that

48:44

Elin's just following incent his incentives,

48:46

Like he's come out of a situation where

48:48

it's okay to kind of lie and and exaggerate

48:51

and in order to make yourself wealthy. And

48:53

he's been good at that, and so he's just kind of

48:56

continuing to follow the incentives there. The

48:58

real failure is is law

49:00

enforcement and regulators, right, Like,

49:02

as a society, you know, you

49:04

have someone who's endangering what's already

49:07

a very dangerous activity, making roads

49:09

even less safe, while lying

49:11

about, you know, claiming that it is safe

49:14

and doing it to become wealthy. Like even

49:18

if you think it should be okay

49:20

for Elon to do this because he's magic and special,

49:23

the example that it sets right

49:25

then it tells everyone else this is an okay

49:28

way to become the richest man in the world is by

49:30

endangering other people and lying. You

49:33

know, that is an example.

49:35

There's a society. I don't think we can afford to

49:37

let sort of sit unchallenged, And so I think

49:39

the real failure there here is is just

49:42

it's the government.

49:51

So I thought that was a really good point to end the interview,

49:54

because while Musk has

49:56

and continues to be and probably will

49:58

for WHOA, He'll continue to a horrifying

50:00

man that constantly tests how far a billionaire

50:03

can go. The failure to hold Tesla accountable

50:05

is one that lands at the feet of the government

50:08

and really the media. For years,

50:10

the press has given Musk fairly unquestioning

50:13

press even to this day, though there

50:15

are some exceptions, people like Ednita Bayer,

50:17

Lourical, Odny Thenett, Lopez al and Nonsmann,

50:20

Preston Grind Ryan Mack. If I left

50:22

you out, I'm really sorry, but this sounds

50:24

like a lot of people. But there are so

50:27

many more members of the media who have just huffed

50:30

Musks farts. Even last

50:32

week, Alex Cantrowitz of Big Technology.

50:35

I really really respect Alex.

50:38

I've loved his work since BuzzFeed. I think he's phenomenal,

50:41

except for this. He uncritically published

50:43

an email from Mask about his plans for Grok,

50:45

the large language model he's bolted onto Twitter,

50:48

and how it will interface with Twitter's news feed,

50:50

generating stories stories

50:52

from tweets. Now, the big

50:54

miss here, other than just copy pasting

50:57

something Elon Musk said, was leaving out

50:59

the fact that Grog has been doing

51:01

stories already. It's been summarizing

51:03

the news, including multiple hallucinated

51:06

stories like one about basketball player

51:08

Klay Thompson allegedly vandalizing

51:10

places with bricks after Gok

51:12

misunderstood that people were referring

51:15

to him bricking shots in a basketball

51:17

game very basic English, and even

51:19

then Grok can't get it. Alex, what the hell

51:21

are you doing?

51:22

Mate?

51:22

I said this on Twitter and I'll say it on here. What are

51:25

you doing? You are smarter than this. I do not know

51:27

why you're doing this. No one should be doing this.

51:29

If Elon must send you something, you give

51:31

it a critique, you look at it through the fair

51:33

lens. Because this man is not trustworthy.

51:36

Elon Musk he half asks everything,

51:39

He rushes, he tricks, he cheats, he

51:41

half explains him moreover, he lies.

51:44

Elon Musk is not someone to take it a word

51:47

or to treat with the benefit of the doubt.

51:49

Every time that we buy into whatever

51:51

weird narrative or made up thing he

51:53

has about how Tesla will work itself out

51:56

of a jam or how X will be big. We're

51:58

helping a man who has acted disingenuously

52:00

and dangerously and will continue to

52:03

do so. A man who platforms actual

52:05

nazis a man that retweets

52:08

anti Semitic things. This is who

52:10

we're dealing with. And as I've said before,

52:12

though, governments must also take

52:14

Musk a lot more seriously, and they

52:16

should cut him out of subsidies and programs.

52:19

I'd say permanently, but at least as

52:21

long as he continues to release this buggy

52:24

dangerous software and platforms,

52:27

racists and insane freaks

52:29

who would kill people like me. I am

52:31

Jewish, and I'm confident that some of the people

52:33

he shared would absolutely murder my ass dead.

52:36

And that's the thing. This is the guy. This

52:39

guy has billions of dollars. Elon Musk.

52:41

He's a liar, he's a scam artist.

52:44

And it doesn't matter that he's got

52:46

billions of dollars. One can still

52:48

be corrupt, selfish, and a complete fucking

52:50

idiot with that many zeros in the bank.

52:53

While the threat of Elon Musk is something to take

52:55

very seriously, though his ideas

52:57

are most certainly not so, I challenge

52:59

you, as a member of the media, as a

53:01

listener, as a consumer, to

53:04

look at everything he does in the same way

53:06

you would a teenager that has been lying

53:08

to you for months. That's who Musk

53:10

is, and he's been lying a lot longer than a few

53:12

months. He's been lying for the best part of a decade,

53:16

and he manages to make money and spiked

53:18

Tesla's stock every single goddamn

53:20

time. People like Jim Kramer and

53:22

his ilk fuel his murderous,

53:25

genuinely dangerous ideas. I

53:28

challenge you, whoever this is listening,

53:31

to think very critically about this man. Thank

53:41

you for listening to Better Offline. The editor

53:43

and composer of the Better Offline theme song is

53:45

Matasowski. You can check out more

53:47

of his music and audio projects at Mattasowski

53:50

dot com, m A. T. T OsO

53:53

w Ski dot com.

53:56

You can email me at easy at Better Offline

53:58

dot com or visit better off Line dot com to

54:00

find more podcast links and of course, my newsletter.

54:03

I also really recommend you go to chat

54:05

dot where's youreaed dot at to visit the discord,

54:08

and go to our slash Better Offline to check

54:10

out I'll Reddit. Thank you so

54:12

much for listening. Better Offline

54:14

is a production of cool Zone Media. For more

54:16

from cool Zone Media, visit our website

54:19

cool Zonemedia dot com, or check

54:21

us out on the iHeartRadio app, Apple Podcasts,

54:23

or wherever you get your podcasts