Tesla and Uber Unlikely to Survive...
Discussion
RobDickinson said:
Elon, still more credible than DonkeyApple
So you think renting out cars for minicabbing is a new idea? 
And that the zero reaction in share price on a reveal that could save the company by creating product demand is a sign that investors believed what he was saying? RobDickinson said:
If you can make 20-30k PA renting out your tesla so long as you can cover the loan/repayment/lease you can effectively buy one very cheaply
problem will be if the market gets flooded
It's a least a decade away and it won't be as popular as one might think for a whole host of reasonsproblem will be if the market gets flooded
RobDickinson said:
If you can make 20-30k PA renting out your tesla so long as you can cover the loan/repayment/lease you can effectively buy one very cheaply
problem will be if the market gets flooded
A 50% yield? Chap, the model already exists and has done for decades. It’s not just boats, planes and super cars that get rented out to generate a yield. People have been putting their Mercs and the likes out to rent overnight for donkeys years. problem will be if the market gets flooded
Do you really think the yields are 50%? That’s just stupid thinking. No one is that naive or deluded.
Western yields are single digits across the board for basic asset rental. And you actually think that Musk can mysteriously create 50%? Please entertain us with just the most basic explanation as to how? A keep the explanation based on planet Earth.
DonkeyApple said:
Western yields are single digits across the board for basic asset rental. And you actually think that Musk can mysteriously create 50%? Please entertain us with just the most basic explanation as to how? A keep the explanation based on planet Earth.
I thought its pretty obvious, take away the most the expansive part of system, which for virtually every business is the human.What is a complete guess is IF Tesla can actually develop a system to remove the human cost of providing personal transportation.
Musk clearly thinks so, and has pretty much bet his entire lifes work behind it.
Crazy or not, theres not many people I know with so much self confidence they are willing to go 'all in' on tech pretty much everyone else is saying is impossible to at the commercial level.
sambucket said:
Surprise suprise, miserable old gits who decided years ago to value Tesla's tech at zero, continue to value Tesla Tech at zero.
And the sun comes in the morning.
Who’s valuing it at zero? It’s value is linked to the number of cars they sell. And that’s if it actually exists. And the sun comes in the morning.
The point is that it clearly has a value but that many people appreciate that the value is not infinite and that arguably it is less than being proported in some quarters.
Most people realise that the timing of last nights mystic meg show was a function of wanting association to the Uber ipo.
Fully autonomous driving is going to be permitted in certain global locations and if you want to compete against the incumbent minicab market without needing humans on your books then the Tesla car looks a good bet. Buy 50-100 and put them on the game 24/7. And your neighbour will also buy a fleet and put them to work. A minicab is much better than BTL as the unit entry cost is so much lower. I could take out the whole local minicab industry where I live and I didn’t do it then someone else will. Got the parking space and the ability to borrow and for a few K you’ve got a £50k asset generating yield 24/7.
Will the yield be 50%? No, of course not. That’s just loony thinking. Markets simply don’t work like that. Yields are governed by risk amount over the risk free rate and asset values and rents instantly adjust to fit.
Just ask yourself this simple question: why would Tesla sell a car for $50k if all buyers are willing to pay $250k? And that’s just assuming that it’s not consumer rates that plummet as the streets are flooded with minicabs that can completely undercut all competition including private ownership costs.
Just basic common sense makes it clear that institutional money would flood the market leaving next to zero demand so next to zero yield for the chap who just has one car, uses it when everyone else is using cars and then wants to rent it out when the least number of people are in transport.
RobDickinson said:
Honestly I must be missing where DonkeyApples multi billion dollar companies are...
Rob, try thinking your thought process over. If you are pinning intelligence to wealth then by your own system you must be borderline cabbage.
Maybe Musk is 100x smarter than me but that means that you’re 100x dumber than me.
Tuna said:
The most interesting thing in the announcement is that it's still just over the horizon. That's an investors pitch, not a manufacturers promise and it's designed to keep the cash flowing.
That was my view having watched the presentation too.There is also a key issue difference between Musk's approach and the other leading companies trying to implement FSD which came to light.
The other companies are all producing detailed maps of the roads. So the car is told via the map the road layout, the position of the signs, the tunnels, the traffic lights etc etc. The car can then focus its attention solely on figuring out the obstacles. This is why the cars currently need to be geo-fenced for FSD, to ensure the car has the correct map detail available.
Whereas with Musk, he is trying to get the car to figure out everything on the fly, including the likely path of the road, just by looking with its various cameras.
Whilst maybe that approach gives the potential for greater autonomy across longer distances, the method used by the other companies seems far more pragmatic and more likely to result in a solution sooner.
It also makes the computational tasks far simpler - if the car knows everything already about the route, the lane positions, the speed limits, which traffic lights correspond to which lanes etc, focusing on the obstacles becomes 100% of its task. It also means changes in the weather are far less likely to confuse the car. Whereas with Tesla, its cars are needing to solve the same problem again and again just to assess its physical surroundings and to guess which way the road is going to curve. I think that will prove to be a big mistake by Musk.
I was also a little disappointed in the presentation to learn more about how Tesla were training their neural networks. There was a lot of talk previously about how Tesla had the most real-world data from billions of miles from their "fleet" driven on auto-pilot. But in reality, they seem to be collecting a very small amount of data, and mainly from situations when they tell the software in the cars to report certain occurrences to give them a broader range of examples for training their neural network.
But because Tesla are not aiming to produce detailed mapping, the amount of training their system needs to do to figure out the basics of its physical surroundings, makes their task massively more difficult.
Why should they collect a whole bunch of 'its OK' data, its pointless. You only grab the relevant data when the driver has to do something or something unexpected happens.
As for detailed maps why? The car still needs to figure it all out, teslas do have maps just not down to the inch because its not needed.
As for detailed maps why? The car still needs to figure it all out, teslas do have maps just not down to the inch because its not needed.
Job kennison understands this stuff
Eg
https://twitter.com/jebkinnison/status/11204122770...
https://twitter.com/jebkinnison/status/11204134615...
The vast majority of bulls think this “millions of miles of data” is stored somewhere for training. It’s not. $tsla staff working on one issue set up criteria for examples, get examples, train network, run in shadow mode to test. One feature at a time, not holistic...
Hint that they send filter instructions to fleet and it returns example videos that fit filter. Admission that returned data is selective - there’s no mult-petabyte library of all data of all sensors and actuators. Which we knew because $tsla doesn’t pay for that much LTE.
Eg
https://twitter.com/jebkinnison/status/11204122770...
https://twitter.com/jebkinnison/status/11204134615...
The vast majority of bulls think this “millions of miles of data” is stored somewhere for training. It’s not. $tsla staff working on one issue set up criteria for examples, get examples, train network, run in shadow mode to test. One feature at a time, not holistic...
Hint that they send filter instructions to fleet and it returns example videos that fit filter. Admission that returned data is selective - there’s no mult-petabyte library of all data of all sensors and actuators. Which we knew because $tsla doesn’t pay for that much LTE.
Edited by anonymous-user on Tuesday 23 April 11:23
gangzoom said:
I thought its pretty obvious, take away the most the expansive part of system, which for virtually every business is the human.
What is a complete guess is IF Tesla can actually develop a system to remove the human cost of providing personal transportation.
Musk clearly thinks so, and has pretty much bet his entire lifes work behind it.
Crazy or not, theres not many people I know with so much self confidence they are willing to go 'all in' on tech pretty much everyone else is saying is impossible to at the commercial level.
What happens to any asset in a free market that has an above average yield for the risk?What is a complete guess is IF Tesla can actually develop a system to remove the human cost of providing personal transportation.
Musk clearly thinks so, and has pretty much bet his entire lifes work behind it.
Crazy or not, theres not many people I know with so much self confidence they are willing to go 'all in' on tech pretty much everyone else is saying is impossible to at the commercial level.
Musk was attempting to make out that private individuals would be able to put their car on the game when they are not using it and make a nice turn. They won’t and can’t. It’s simple economics. The market will be set by institutional money that will run fleets of cars 24/7 and that will drive yields back into line with the market. What that means is that the private owner will earn far less per hour of labour and on an asset that is depreciating quicker due to use and costing more to hold due to cleaning and maintenance.
Just like any such model on the planet it works for those who can run multiple assets which are financed at the lowest rates and who can sweat that asset all day long. The chap who has one car that he’s paying 4% or more on, with higher running costs and lower tax efficiency and only working a few hours a day doesn’t stand a chance.
What Musk is trying to get retail punters to believe is the equivalent of someone being able to run a small village shop as competitively as Mike Ashley or Philip Green can run a shop. The world doesn’t work like that. He knows that, everyone with money and in business knows it. It’s just the basic economic laws that we all live by.
The reality is that a single, private owner will almost certainly not earn enough net income to cover the cost of putting the asset to work because they cannot be competitive.
RobDickinson said:
Why should they collect a whole bunch of 'its OK' data, its pointless. You only grab the relevant data when the driver has to do something or something unexpected happens.
As for detailed maps why? The car still needs to figure it all out, teslas do have maps just not down to the inch because its not needed.
On your first point, that isn't the basis on which they are collecting the data. It is a far more limited dataset, mainly based on which bit of neural net training they happen to be working on at that point in time.As for detailed maps why? The car still needs to figure it all out, teslas do have maps just not down to the inch because its not needed.
As for the rationale for detailed maps - it's obvious. If the car is told the exact lane positions, the details of the road furniture, which traffic lights apply to which lane, the future path of the road etc etc the car then doesn't need to deduce that information in real time on the road. Hence it can focus its sensors on the difficult bit which are the obstacles in its path.
My general impression is that autopilot has a very slight real world lead, but Musk is making up all the timelines and taxi stuff on the fly. But that is fine with me, if it publicises HW3 and raises awareness of Tesla's slight edge, etc he is doing his job.
It's hardly fraud compared to what other automakers get up to. And the safety concerns are a joke. How many people have VW needlessly killed?
It's hardly fraud compared to what other automakers get up to. And the safety concerns are a joke. How many people have VW needlessly killed?
Gassing Station | EV and Alternative Fuels | Top of Page | What's New | My Stuff