Tesla and Uber Unlikely to Survive (Vol. 2)
Discussion
Tuna said:
Whilst I'd never bet against a particular technology, and don't like the taste of hats, I will be genuinely surprised if Tesla deliver either of those for general use by the end of the year. If the cars still occasionally phantom brake, getting them to reliably (and safely) respond to traffic lights seems like a very tall ask. Cars that have trouble manoeuvring at slow speed out of your garage/around a car park are not (yet) suited to safely entering and exiting complex junctions.
In terms to the biggest risk to drivers, moving away from 'fancy lane keeping' to making key, safety critical decisions might even make Musk pause.
If people want to remind me of my cynicism in a year's time, I'll be happy to be proven wrong - there are only a handful of groups throwing the sort of effort at this problem that Tesla are, so if it can be solved, it will be a huge step forward.
Reacting to traffic lights is relativly trivial once visual is there? requires no complex steering, you just need reliable stop line and light recognition, which seems increasingly likley in 2020. If it can all be reliably visualised then the rest is simple? AFIAK In terms to the biggest risk to drivers, moving away from 'fancy lane keeping' to making key, safety critical decisions might even make Musk pause.
If people want to remind me of my cynicism in a year's time, I'll be happy to be proven wrong - there are only a handful of groups throwing the sort of effort at this problem that Tesla are, so if it can be solved, it will be a huge step forward.
Sambucket said:
It seems to me like the more bearish posters are allowed to have complex opinions, but at same time get to accuse the fanbois of only seeing in black and white.
Seems a bit patronising to me to when bears state as gospel that all the fanbois are blind cultists and in aware of the risks and issues Tesla face. Which is obviously not the case.
The realiTy Is that most of fanbois on the thread are equally informed on the negative aspects. But that consideration is no fun for trolls Who just want to trash talk loser fanbois.
All a bit sad really.
Isn't the point that the 'fanbois' absolutely reject any criticism of Tesla as being invalid, and an attack on the company? Seems a bit patronising to me to when bears state as gospel that all the fanbois are blind cultists and in aware of the risks and issues Tesla face. Which is obviously not the case.
The realiTy Is that most of fanbois on the thread are equally informed on the negative aspects. But that consideration is no fun for trolls Who just want to trash talk loser fanbois.
All a bit sad really.
It's hard not to take those (few) posters' views as being black and white when a response to "this is difficult" is "no it isn't, you just want Tesla to fail, the company has proven you wrong in every possible way".
Here's a reminder of how this all started on PH over 2 years ago 
yipper said:
Great article in The Economist this week, showing how Tesla has a 75% to 99.9% chance of failure, due to its gigantic debt load.
History suggests companies with big debt and big losses, over multiple years, like Tesla and Uber, rarely thrive or survive.
Such firms have charismatic leaders just winging it on debt.
Tesla and Uber are statistically doomed...
https://jalopnik.com/it-d-be-a-business-miracle-if...
https://www.economist.com/news/business/21730446-f...
What happened to Yipper anyway, he vanished.History suggests companies with big debt and big losses, over multiple years, like Tesla and Uber, rarely thrive or survive.
Such firms have charismatic leaders just winging it on debt.
Tesla and Uber are statistically doomed...
https://jalopnik.com/it-d-be-a-business-miracle-if...
https://www.economist.com/news/business/21730446-f...
Tuna said:
Isn't the point that the 'fanbois' absolutely reject any criticism of Tesla as being invalid, and an attack on the company?
It's hard not to take those (few) posters' views as being black and white when a response to "this is difficult" is "no it isn't, you just want Tesla to fail, the company has proven you wrong in every possible way".
Ever wonder why the fanboys are fanboys? It's hard not to take those (few) posters' views as being black and white when a response to "this is difficult" is "no it isn't, you just want Tesla to fail, the company has proven you wrong in every possible way".
Could be that tesla are doing the right things and that most of the criticism was bulls
t? Could it really be? RobDickinson said:
Ever wonder why the fanboys are fanboys?
Could be that tesla are doing the right things and that most of the criticism was bullst? Could it really be?
Maybe I'm wrong, i dunno. Rob what do you think the snapshot percentage probability of Tesla restructuring, both start of thread and today?Could be that tesla are doing the right things and that most of the criticism was bulls
t? Could it really be? Maybe tuna has a point in that the downside risk of FSD failing, and Y not scaling, is probably not fully priced into the stock.
But the bears are guilty of absolutism too. Musk is 100% definitely a fraud who does no innovative engineering. Tesla will never ever turn a profit...... same kind of black and white crap.
Check my posts from early 2018, there was real chance at that time of going bust based purely on running out of cash before scaling model 3.
I said repeatedly all they needed to do was hit 5k production a week and they'd be fine.
We are not blind or stupid, but most of the thread has been about tesla killers or autopilot lawsuits, demand cliffs etc etc m
I said repeatedly all they needed to do was hit 5k production a week and they'd be fine.
We are not blind or stupid, but most of the thread has been about tesla killers or autopilot lawsuits, demand cliffs etc etc m
Tesla have a fair few barriers to entry into their market now.
They have a battery factory
They have their own supercharger network that is on another level, the closest thing that exists is Ionity but its still miles behind (although they have done a terrific job imo)
Excuse the puns below -
They are miles and miles ahead on the software front.
They are miles ahead on branding
They are miles ahead on advertising costs (I think Ford spend $800 a car on advertising)
They have their own factory in China (all other car manufacturers are stuck in JVs) which they made in 12 months and they claim they can make the next factory even quicker.
They have a battery factory
They have their own supercharger network that is on another level, the closest thing that exists is Ionity but its still miles behind (although they have done a terrific job imo)
Excuse the puns below -
They are miles and miles ahead on the software front.
They are miles ahead on branding
They are miles ahead on advertising costs (I think Ford spend $800 a car on advertising)
They have their own factory in China (all other car manufacturers are stuck in JVs) which they made in 12 months and they claim they can make the next factory even quicker.
Sambucket said:
Reacting to traffic lights is relativly trivial once visual is there? requires no complex steering, you just need reliable stop line and light recognition, which seems increasingly likley in 2020. If it can all be reliably visualised then the rest is simple? AFIAK
It's one of the specific reasons Google cited for getting out of self-driving research - that traffic lights were an 'unsolved' problem.Consider temporary lights by the side of the road, reflections in street signs, occasions when you can see other lane's lights, high level brake lights, illuminated signs, coloured dots on trucks, advertising hoardings...
How does a computer know that a bright red thing it sees is definitely a stop sign that applies specifically to it? Imagine the phantom braking from that going wrong! And if it doesn't reliably stop at every single red light, you risk serious accidents.
Similarly, if it sees a green light (reflected on the side of a truck, for a filter lane etc. etc.) can it be sure that it's safe to go? Again, getting that wrong could result in a serious accident.
It's true that spotting a red thing or a green thing is 'easy'. What is difficult is being sure that the red thing or the green thing applies to you. That's ignoring adversarial issues, like people shining laser pointers at you.
Tuna said:
It's one of the specific reasons Google cited for getting out of self-driving research - that traffic lights were an 'unsolved' problem.
Consider temporary lights by the side of the road, reflections in street signs, occasions when you can see other lane's lights, high level brake lights, illuminated signs, coloured dots on trucks, advertising hoardings...
Have you seen youtube videos of Tesla traffic light visualisation in action in USA right now? Looks pretty close to solved to me. Maybe 95% accuracy rate in one vid I saw. Consider temporary lights by the side of the road, reflections in street signs, occasions when you can see other lane's lights, high level brake lights, illuminated signs, coloured dots on trucks, advertising hoardings...
I'm not saying 95 is good enough for anything but experimentation. But it's on the way.
As with all TEsla stuff, it's all supposed to be used as an extra set of eyes under supervision, not as a replacement.
You might argue that is not FSD, but Tesla are still selling the functionality and people are buying it.
RobDickinson said:
Could be that tesla are doing the right things and that most of the criticism was bullst? Could it really be?
Let's be clear here - some of the criticism was indeed BS. t? Could it really be? That does not automatically mean that all (or even most) is BS. Nor does it mean Tesla should be immune to criticism, or people saying they are still an extremely risky bet. Using the fact that they're still here as evidence that they are 'doing the right things' - that's like claiming an end to boom and bust.
Feel free to point out any of my critical posts that turned out to be wrong.
jamoor said:
Tesla have a fair few barriers to entry into their market now.
They have a battery factory
They have their own supercharger network that is on another level, the closest thing that exists is Ionity but its still miles behind (although they have done a terrific job imo)
Excuse the puns below -
They are miles and miles ahead on the software front.
They are miles ahead on branding
They are miles ahead on advertising costs (I think Ford spend $800 a car on advertising)
They have their own factory in China (all other car manufacturers are stuck in JVs) which they made in 12 months and they claim they can make the next factory even quicker.
Ford afik spend billions a year on advertising. They have a battery factory
They have their own supercharger network that is on another level, the closest thing that exists is Ionity but its still miles behind (although they have done a terrific job imo)
Excuse the puns below -
They are miles and miles ahead on the software front.
They are miles ahead on branding
They are miles ahead on advertising costs (I think Ford spend $800 a car on advertising)
They have their own factory in China (all other car manufacturers are stuck in JVs) which they made in 12 months and they claim they can make the next factory even quicker.
One expert (Monroe? Autozine?) said its a minimum 7 year project for any other manufacturer to integrate their cars to a level tesla has. That's fundamental to how teslas work, physically and as a business proposition
Having not looked at this for a while , good to see it is still going on and yes Tesla is now4
540 a share .... so still around .
As 911 Gt fan I would never thought EV but 3 years ago got one and have been fully converted , it does not make me a Tesla fan boy , just that I have pulled my head out of the sand and seen the future .
540 a share .... so still around .
As 911 Gt fan I would never thought EV but 3 years ago got one and have been fully converted , it does not make me a Tesla fan boy , just that I have pulled my head out of the sand and seen the future .
RobDickinson said:
jamoor said:
Tesla have a fair few barriers to entry into their market now.
They have a battery factory
They have their own supercharger network that is on another level, the closest thing that exists is Ionity but its still miles behind (although they have done a terrific job imo)
Excuse the puns below -
They are miles and miles ahead on the software front.
They are miles ahead on branding
They are miles ahead on advertising costs (I think Ford spend $800 a car on advertising)
They have their own factory in China (all other car manufacturers are stuck in JVs) which they made in 12 months and they claim they can make the next factory even quicker.
Ford afik spend billions a year on advertising. They have a battery factory
They have their own supercharger network that is on another level, the closest thing that exists is Ionity but its still miles behind (although they have done a terrific job imo)
Excuse the puns below -
They are miles and miles ahead on the software front.
They are miles ahead on branding
They are miles ahead on advertising costs (I think Ford spend $800 a car on advertising)
They have their own factory in China (all other car manufacturers are stuck in JVs) which they made in 12 months and they claim they can make the next factory even quicker.
One expert (Monroe? Autozine?) said its a minimum 7 year project for any other manufacturer to integrate their cars to a level tesla has. That's fundamental to how teslas work, physically and as a business proposition
When do you expect Tesla to show profit?
coetzeeh said:
When do you expect Tesla to show profit?
In what 3 of the lat 5 quarters? This quarter could see them into the Sp500. They now have a market cap of $80bn, 3rd only to VW and Toyota.
you are an idiot to make excessive profits in a capital intensive growth business whilst being the runaway market leader
IQ's are still damned low in this thread.
Sambucket said:
Have you seen youtube videos of Tesla traffic light visualisation in action in USA right now? Looks pretty close to solved to me. Maybe 95% accuracy rate in one vid I saw.
I work in machine learning. 95% accurate does not mean 95% solved. You can download free software that will give you 90-something percent accuracy rate for identifying visual scenes - it's been available for a decade.To give you a sense of scale... my daily commute involves 10 sets of traffic lights each way. At 95% accuracy, a Tesla would try and kill me every single day of my life.
If you have a million drivers with similar commutes, and achieve 99.999995% accuracy, you will have a hundred accidents a day.
And remember, visualisation has still not solved the decision process - it just means you have reliable information about one part of the problem. (the other parts being things like oncoming/crossing traffic, safe stopping distances, other vehicles/obstructions in the junction, pedestrians, unexpected light locations.and so on)
Nearly 40% of all traffic accidents happen at junctions - humans find this a difficult problem. Why should computers be any better?
Tuna said:
Nearly 40% of all traffic accidents happen at junctions - humans find this a difficult problem. Why should computers be any better?
Because they are capable of processing information from 4 different directions at once and have better spacial and relative speed awareness than the average human?Have you seen most pensioners at T-Junctions?
SWoll said:
Because they are capable of processing information from 4 different directions at once and have better spacial and relative speed awareness than the average human?
But struggle to tell the difference between a traffic cone and a pedestrian? (I know, it's better now.. nothing to worry about there, oh no)Gassing Station | EV and Alternative Fuels | Top of Page | What's New | My Stuff