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

[MP3Mike]

Maybe folks here should do a realistic/honest check with the competition as well.

Out of Spec has their "Hogback" test that they run a number of vehicles through: Hogback — Out of Spec Studios

You do have to understand their rating criteria, which I don't fully agree with, like I don't think a capacitive steering wheel should add 5 points, but it is good enough to allow some comparison, and the raw data is available such that you can recalculate the scores if you want to use different criteria:

Though I think they really need to specify the software version for all of their Tesla tests, so you have an idea how old they are. (Same for Ford Blue Cruise since it can be OTA'd.)

 

[scaesare]

Yes, but I don't have enough understanding of GPU's and processors to feel comfortable with it. I looked at it a while back and I think you would have needed to be an expert in chips and AI to honestly make this call.

The good news is the profits Nvidia is making have to come from somewhere at some point or they will not continue. LLM's are interesting but it seems like there will be lots of competition. With FSD I do believe Tesla has a lead which will be hard to overcome.

A quick note: LLM's are the software that needs GPU-type compute to train/run. So, competition amongst LLM's means more demand for compute such as from Nvidia.

 

[DarkandStormy]

Correct. It takes 30 seconds to read the methodology on the report page, and that FUD still abounds:

 

[advocate8]

L2 city streets not from Tesla, door to door, anywhere in China, is available to consumers there-- just as FSD V12 offers L2 city streets door to door, anywhere in NA, from Tesla. You've had this pointed out to you multiple times but keep repeating your inaccurate claim. Why?





I think the disconnect is what the data tells us is:

FSD plus a human is much safer than just a human.

But somehow people (NOT you as you specifically called out supervised)) misread that as:

FSD (by itself) is much safer than a human.

No, the data shows only one thing: that the safety of the supervised system is steadily improving.
Any claim that it shows safety vs solely humans is deceptive and frankly scummy when asserted by a company purporting to have the moral high ground fighting for humanity.
The lie is in the roads where the data
is compiled vs only humans.

 

[MartinAustin]

Interesting, Cramer on CNBC swooning over NVDA and Musk getting the true path for AI breaks biggest for autos.

Very impressed that Jensen is so impressed with Musk.
YMMV

I'm sure Jensen Huang is one of Tesla's biggest fans and very up-to-date on all the products. Jensen was one of the first 15 customers to take delivery of a Model X back in Q4 2015 IIRC... he was handed the keys at the delivery event

 

[scaesare]

Smart CEOs (and simply most, even extremely wealthy ones) don't insult their customers or tell them to F off and say they don't want their $$. Better to simply not upset 50% of the population right and run the company not wading into political/woke debates?

You'd rather CEO's kow-tow to profit over principle?

I, for one, find standing up to folks based on something other than the bottom line a bit refreshing. There's enough potential customers that you don't need to pander to those not acting in goof faith.


Short term profit take a hit?: probably

Long term will it galvanize other folks to staunchly support you? Likely.

Is there value in aligning with the mission despite breaking a few eggs? Definitely (for me anyway).


"Management reserves the refuse service to anyone."

 

[advocate8]

View attachment 1049809

Right. Sure. He’s playing 4D chess. Sure.

 

[scaesare]

Correct. It takes 30 seconds to read the methodology on the report page, and that FUD still abounds:

So, @DarkandStormy , we know from your trying to hand-wave the Pepsi truck thing away that's it was hard to admit you were wrong,

Here's your chance to try again!

 

[scaesare]

Right. Sure. He’s playing 4D chess. Sure.

So give us an example where he's "cozying up" to climate change deniers in order to support such endeavors, as opposed to attempting to build a bridge that might allow inroads to change their stance. Or at least they are the lesser of two (17?) evils...

 

[GhostSkater]

These prices are insane, and makes total sense on the push to make 4680s way cheaper. Crazy to think that I, as a individual consumer have to pay $500/kWh
around here

https://twitter.com/x/status/1792537236388348391

 

[indysurfer]

I've test driven quite a lot of other brands with driving aids just for fun and curiosity, just drove an R1S last week!

Nothing I've driven yet compares to the level of Tesla's free Autopilot, let alone FSD v12 which is leaps and bounds beyond anyone else. Some of the systems I drove were downright dangerous IMHO, but some were okay yet limited in capability compared to Autopilot.

Many of us who proclaim how awesome Tesla AP is have indeed done our homework. If you had done yours you probably wouldn't be making posts like this.

Me too. I’m in rental cars for work travel every month and the adaptive cruise control and lane assist in other brands are no where near as good as Tesla’s basic autopilot. Lots of ping ponging between lanes and generally an uncomfortable experience.

 

[advocate8]

So give us an example where he's "cozying up" to climate change deniers in order to support such endeavors, as opposed to attempting to build a bridge that might allow inroads to change their stance. Or at least they are the lesser of two (17?) evils...

He's basiclaly running with them now and does not in any way seem to be pushing back, since they're so aligned on almost every other culture war thing. You can Google examples, life is too short for me to reiterate the obvious.

 

[Knightshade]

No, the data shows only one thing: that the safety of the supervised system is steadily improving.
Any claim that it shows safety vs solely humans is deceptive and frankly scummy when asserted by a company purporting to have the moral high ground fighting for humanity.
The lie is in the roads where the data
is compiled vs only humans.

The problem with your claim is Tesla has published data for BOTH types of roads-- AP+human used primarily on highways and FSD+human used primarily on city streets- and the accident rate is much lower than humans alone in both cases.

AP+human being lower than FSD+Human, but as I mentioned earlier you'd expect that since non-highway accident rates tend to be higher in general (though less severe due to lower speeds)

Out of Spec has their "Hogback" test that they run a number of vehicles through: Hogback — Out of Spec Studios

You do have to understand their rating criteria, which I don't fully agree with, like I don't think a capacitive steering wheel should add 5 points, but it is good enough to allow some comparison, and the raw data is available such that you can recalculate the scores if you want to use different criteria:

View attachment 1049808

Though I think they really need to specify the software version for all of their Tesla tests, so you have an idea how old they are.

Interesting data- thanks.

TBF on the version thing- it seems they're testing highway only, and IIRC Tesla still hasn't moved the highway stack to V12 so I'm not sure exact version info would matter a ton? I think Elon said 12.5 was the target to get back to single stack local roads and highways both on V12 architecture.

Also interesting they have two different Model 3s in there significantly lower score than the S-- but it seems they call out the S is on FSD, and looking at the detailed data is seems the 3s were basic AP (so no driving modes, auto lane change, etc)

The FSD car is missing 5 points they give for hands-free eye tracking, which it sounds like 12.4 will add and would put it above the Mercedes FWIW.

 

[advocate8]

The problem with your claim is Tesla has published data for BOTH types of roads-- AP+human used primarily on highways and FSD+human used primarily on city streets- and the accident rate is much lower than humans alone in both cases.

AP+human being lower than FSD+Human, but as I mentioned earlier you'd expect that since non-highway accident rates tend to be higher in general (though less severe due to lower speeds)






Interesting data- thanks.

TBF on the version thing- it seems they're testing highway only, and IIRC Tesla still hasn't moved the highway stack to V12 so I'm not sure exact version info would matter a ton? I think Elon said 12.5 was the target to get back to single stack local roads and highways both on V12 architecture.

It has to be the same roads. There is no way to properly compared the data sets. Generalizing by differently-defined categories doesn't do it.

 

[SOULPEDL]

Just when I think there's no more buying opportunity....

Edit... After taking another look at this thread... DUH. Comments like this one tell me the sale is still on.

"The fact is that Tesla has been performing poorly..." Seriously hpiv?

 

[Knightshade]

It has to be the same roads. There is no way to properly compared the data sets. Generalizing by differently-defined categories doesn't do it.

I don't understand how you come to that conclusion.

They tell you here's our accident rate using AP+human primarily highways and here's our accident rate using FSD+human primarily on local roads.

You can't just average these of course as you're missing the info to do so, but both are many times lower than human alone.

There is no road type that's higher.



What circumstance or data weirdness do you think would make it possible for the accident rate on any road to be higher than human alone given that?

 

[scaesare]

He's basiclaly running with them now and does not in any way seem to be pushing back, since they're so aligned on almost every other culture war thing. You can Google examples, life is too short for me to reiterate the obvious.

So you make a vague assertion you are unwilling to support.

Gotcha.

 

[FreqFlyer]

The fact it’s getting safer is important, although not surprising. But the constant, clickbait claim here and elsewhere is that it is safer than human drivers. The available data doesn’t show that.

I am not crazy about FSD tech, but I can understand it being safer. An example, unbeknown to most when you take off in an airliner, SOP is around 500-1000ft engage the autopilot and make course and speed corrections through it. Autopilot gets disengaged late into the final around 1500 feet before landing. Why? FAA has realized the sooner everyone engages autopilot and uses it, the safer the system is. It's the crazy Cessna pilot hand-flying that mucks it up for everyone, yes I am generalizing. Numbers don't lie, flying is the safest mode of travel. To believe otherwise is being contrarian getting a rise out of people.

The problem is, on the highways the "crazy Cessna" pilots are the vast majority of drivers, and those with FSD are in the sliver of the minority. The highway crash numbers prove it.

Will the majority of drivers have FSD of some type in the near future ? Who knows, until then I will continue enjoying driving and blasting my favorite tunes.

 

[NicoV]

Flanders will reach its target of having fast chargers every 25km along its highways by the end of next year. This is ahead of the EU’s target of fast chargers every 60km in 2025:

https://www.tijd.be/politiek-economie/belgie/vlaanderen/Vlaamse-snelwegen-liggen-voor-op-Europese-laadpaaleisen/10547899

 

[advocate8]

I am not crazy about FSD tech, but I can understand it being safer. An example, unbeknown to most when you take off in an airliner, SOP is around 500-1000ft engage the autopilot and make course and speed corrections through it. Autopilot gets disengaged late into the final around 1500 feet before landing. Why? FAA has realized the sooner everyone engages autopilot and uses it, the safer the system is. It's the crazy Cessna pilot hand-flying that mucks it up for everyone, yes I am generalizing. Numbers don't lie, flying is the safest mode of travel. To believe otherwise is being contrarian getting a rise out of people.

The problem is, on the highways the "crazy Cessna" pilots are the vast majority of drivers, and those with FSD are in the sliver of the minority. The highway crash numbers prove it.

Will the majority of drivers have FSD of some type in the near future ? Who knows, until then I will continue enjoying driving and blasting my favorite tunes.

the aviation systems also are getting much much better along the lines of FSD.