I think battery swap can be reserved for super high end EVs..perhaps the Rolls Royce of the world where you get that exclusive feeling having such concierge service. It's absolutely not viable as an actual mode of energy delivery to the masses.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
StealthP3D
Well-Known Member
Tesla is not waiting for anything. In fact, in Q1 2022 new Superchargers were turned on at a rate double that of Q1 2021 and their stated plan is to continue adding Superchargers at an accelerated rate:
Tesla Almost Doubled The Rate Of Supercharger Deployment In Q1 2022
Tesla continues fast global expansion of the Supercharging network, achieving strong growth of new installations compared to the previous year.
insideevs.com
I prefer to travel the rural backroads of the American West so, personally, I would welcome more expansion of Superchargers in rural areas. This would serve my own personal selfish needs in an ideal manner.
However, I am also a fan of Tesla using first principle's thinking to remain fiscally responsible. It costs a lot of money to expand into the wide open areas with low population and low traffic density, not to mention the on-going higher costs of servicing, maintaining and providing security for electrical equipment in more remote areas.
First principles thinking will illuminate the fact that Superchargers exist to drive EV adoption (not to cater to my personal desires) and, as long as Tesla can sell every car they can make to people who do not demand to travel in areas poorly serviced by Superchargers, they should only expand in a strategic manner. It would be a colossal waste of money to build and maintain remote Supercharger sites that would be lightly utilized while driving adoption by people living/travelling outside populated areas and increasing order wait times unnecessarily. It makes complete sense to encourage early adoption by those whose charging needs can most easily be met by servicing the most highly travelled routes.
It's easy to forget that people buying BEV's are still early adopters. As the market matures, Tesla will buildout Superchargers on increasingly rural routes, as much as needed to maintain more than enough demand to sell every car they can produce. This is how first principles thinking works and is a major driver of Tesla's incredible success. I am stunned at their ability to build such an impressive charge network to service their sales in such an efficient manner while their competitors struggle to just build EV's at a profit, without any significant expenditures on providing fast charging for customers. And I love the fact that Supercharging is a fraction of the cost of private charge networks (even though I have free fast charging for life).
I would love a more robust rural fast DC charging network and it will come before it is needed to keep the company performing like a well oiled machine in terms of driving EV adoption and fulfilling it's stated mission. We live in amazing times!
AMD based Frontier will be the new #1 (announcement pending till Monday)Weekend discussion:
Brain performance in FLOPS
The computing power needed to replicate the human brain's relevant activities has been estimated by various authors, with answers ranging from 1012 to 1028 FLOPS. Details Notes We have not investigated the brain's performance in FLOPS in detail, nor substantially reviewed the literature since...aiimpacts.org
View attachment 810160
Tesla's Dojo supercomputer - sorting out fact from hype
Tesla's Dojo supercomputer was announced with fanfare - and a specific use case in mind. But how does it stack up against other high-performance computers? Will Dojo constitute a step forward for the HPC industry, and for Tesla's own automotive AI pursuits?diginomica.com
"the fastest supercomputer today is clocked at 450 petaFLOPS, ten thousand times faster than the Playstation and Tesla claims Dojo will reach exascale: an exaFLOP is one quintillion (10^18) double-precision floating-point operations per second (FLOPs)."
- Frontier, powered by AMD Epyc CPU and Instinct GPU is the new #1
- Rmax: 1.1 exaFlops, Rpeak: 1.686exaFlops, power: 21.1MW
dhanson865
Well-Known Member
There have been posts here that have gotten 40+ disagrees. Do you really want a disagree reply storm for every stupid post?
A good post is then seen once and liked/informatived/helpfulled.
A bad post is then seen dozens of times as it is quoted and replied to to explain how bad a post it is.
The signal to noise ratio would plummet and I am astonished that any mod here would think that is a good idea.
I really would not attribute the zero accidents (touch wood) to the quality of FSD. Rather oddly zero accidents because it is just not very good, rather it sucks. So much that drivers are paying a lot of attention and I am guessing only a few are actually using it for city driving.
The situation is exactly the same as the L2 offerings from all other manufacturers. They all perform poorly, or work only under specific conditions that very few people use it at all. Hence no accidents using L2 assistance from Mercedes, Ford, GM etc..
On the other hand, Tesla's L2 (Autopilot) is so bloody damn good that it is used extensively in all situations and people drop their guard and ends up in some accidents occasionally.
If you don't use it, or forced to pay enormous focused attention when using it, there is no possibility of accidents. If it is used extensively, lets your relax and it works really good 99% of the time, the other 1% ends up in an accident.
It's attributed to the quality of the FSD program. Who knew volunteers are more cautious about their own car and all the consequences that come with a crash vs "professional engineers" spending their time texting and slacking off while their car kill pedestrians. Also I understand how going around in pre-determined routes hundreds of times a week can cause boredom.
But you know, regulators are screaming at Tesla for using volunteers and not "professional drivers" which statistically has proven they are an order of magnitude more dangerous than these volunteers.
Last edited:
Most owners are AP only. FSD and in particular FSD beta is a modest fraction of the owner base. Even among FSD drivers I suspect most use simple AP or even traffic assist cruise most of the time.
For a lot of us, these tools are amazing. And definitely prevent some accidents.
Last edited:
Who knew that drivers know they'll be sued from here to the moon if they get in a wreck and kill someone while playing with their car? Apparently they do! Thankfully everyone is being very safe and FSD program is progressing very well.
We will know when it's close to being ready when it can identify school zones and take the appropriate action. It would also be nice if Chill Mode wasn't Chill Mode for F1 drivers.
StealthP3D
Well-Known Member
I am a huge fan of using data properly to make informed analysis.
That said, the world is more complex than many appreciate and in many investment-type situations, data is difficult to interpret correctly in a manner that is productive and useful. Most often, I think, it is used in a misleading manner, intentional or not. More often the misuse of data is unintentional due to complicating factors that are not understood or appreciated enough.
I find intuitive knowledge from someone who is well-informed and has demonstrated superior understanding of the dynamics involved more useful than data-driven facts and conclusions from people who have a poor grasp of the under-lying realities. In my lifelong investment career, I have found data-driven investment decisions to be less useful and less accurate than good vision and understanding of the subjects, topics and principles that come into play. Data can help form this kind of knowledge and understanding but only on a very low-level and most problems are too complex, too interconnected, and too filtered through human complexities for a data-driven conclusion to be more useful than the "fuzzy logic" conclusion of someone who has a good understanding of most of the dynamics involved. Accurate data is often misleading to a majority of people due to under-appreciated realities.
I was making a light joke re the wife swapping comment. But I agree with you re actual batter swapping.
2daMoon
Mostly Harmless
What you are saying then is,
"You've never really been ignored until you have been ignored by a 
."
."
Gigapress
Trying to be less wrong
I agree, but to clarify, I wasn’t making an attribution of FSD Beta by itself leading to the result, as posters above have mentioned. By design the FSD Beta development program includes careful integration of the machine and human factors. The program as an integrated whole is what’s been extremely safe. Humans alone or FSD Beta alone would lead to collisions, but evidently when combined properly they can go hundreds of millions of miles without even getting a scratch on the paint.
FSD Beta cancellation advocates are claiming that the program is a dangerous experiment on public roads and should therefore be cancelled or slowed down until it’s better. This claim of danger is just plain wrong, so wrong it’s ludicrous. In the beginning they had more of an argument, but by now we have the data to prove it and the critics still keep making noise, just as they’ve been complaining about Autopilot for years despite irrefutable safety data.
Even for the top 1% of safest drivers in general, I would be willing to bet that they are not three orders of magnitude safer than the average driver in the USA, especially if you heavily bias the sample pool of good drivers to ones who live in accident hotspots like LA and SF where the behavior of other drivers increases risk. I don’t know of any other safety technology or human behavioral influence campaign (like “Click it or ticket” ads, DUI laws, speeding enforcement, etc.) that can drop collision risk by 99.9%+ relative to average, irrespective of how the result was achieved.
Last edited:
MartinAustin
Active Member
I'm sure they've already thought of it... I'm sure they've already looked at the processes and looked at what it takes to accomplish it. They have probably isolated a few key enablers that need to be implemented before it can go into action.
What we see today in Austin is a lot of cars with open rear hatches, engineers doing weird testing and messing with unfinished cars. Perhaps once the parts shortages go away there is a better chance for it to work.
Perhaps they will wait until the south end of the Austin factory has gotten its extension. I'm not even sure that "new vehicle staging" will remain where it is today... might move to one of the larger parking lots as production quantities increase.
FSD as it is today can probably accomplish the driving around at low speeds. Eight-way cameras on every new car - even the parked ones - can be analysed periodically to look for anomalies - for example, if strong winds blow a piece of debris into the parking lot. Or if a "human employee" or member of the public somehow drives a car into the lot and parks it, and walks off without telling anyone. These anomalies can then be investigated and straightened out. Tesla cars could notify Facility Security if they see people walking around that aren't expected to be there. Who needs Knightscope?
ZachF
Active Member
I have to imagine the eventual record holder will be the Optimus training system.
2daMoon
Mostly Harmless
Agreed
Posts like you describe are exactly what I think the "Eye roll" emoticon, or something like that, would be better suited for, were it an available option. It doesn't say "I disagree" which could imply there is a potential for correction, rather, it would say, "I can't believe you even posted that chit"
Last edited:
Obviously a productive Sunday morning, when I glean something avatarable in earlier posts.
I don't know...it seems to me that drivers that qualify for the FSD beta are self selecting as safe drivers. I bet no one is the program has been in a driver induced accident in quite a while. They don't speed too much, they have safe following distances, they don't txt and drive, etc. I further bet that the daily route is safe and that there are no teenage drivers and no new drivers and no drivers for whom english is difficult.
The "average" US driver category is pulled down dramatically by just a couple of variables (age, experience driving, etc). If you remove the bottom 10% of the driving fleet (I think you'd find that you don't have a single FSD Beta tester in there) I bet you already have a much safer pool of drivers- by an order of magnitude. . Further once removing that 10% I bet you'd find the Tesla FSD beta drivers are only marginally safer than the general population (before the FSD beta). Then I bet you can chalk alot of that up to just general AP driving safe following distance on interstate commutes, etc. So pull out all the people with any AP system and compare Tesla FSD to the population of cars with some sort of advance AP system and I suspect you'll find that there is a difference but dramatically less when comparing to the general population. Does that make sense? Compare FSD Beta users to Tesla users with safety score of 90%+ with AP. That's the number that would show how much improvement there potentially could be. If I am a regulator thats the start of the numbers I would like to see. Then compare to car owner specs to similar car specs ( MY performance to MY performance owners with Beta with the same safety driving scores).
Gigapress
Trying to be less wrong
These points are important factors, but consider that human-supervised FSD Beta is approaching commercial aviation levels of safety. We are talking about a 1000x difference from average.
The program has 100k drivers, which as @Singuy estimated is around 7% of Tesla’s entire American customer base.
Age and driving experience does not have nearly enough effect to explain this result. 16-17 years olds are a small minority of the overall driving population and their risk is around 4x higher than 30-40 year olds. That leaves at least a ~250x difference out of the 1000x total to explain with demographic factors.
https://aaafoundation.org/rates-mot...-relation-driver-age-united-states-2014-2015/
Other demographic factors make us expect more baseline risk for FSD Beta testers.
I would guess that around a third of them are in the LA and SF areas, where collision rates are much higher than the US average. In these locations maintaining a big following distance is almost impossible because of aggressive, opportunistic drivers who will take that as an invitation to cut in.
Tesla’s driver population also skews heavily male, which is not good for collision risk because men crash cars at almost triple the rate women do. (Source).
Is 7% of Tesla’s owner pool elite drivers who can match the safety of commercial aviation all by themselves simply by picking the right routes and exercising the normal level of caution they previously had before becoming a Beta tester? I highly doubt it.
Last edited:
Similar threads
- Locked
- Locked
- Locked
