Two things. We have automatic high beams but but dipped beams, and it's not 100% reliable. I could make a joke about BMW drivers... And the rain sensor on the Model 3 is not very good, especially when it comes to selecting an appropriate interment speed. So it kind of is a problem when Tesla does it, because it's beta quality software.
The other thing to note is that they also have much higher warranty costs than most manufacturers. They are still having quality issues, especially with paint, so spend more on warranty servicing. Hence they need a higher margin to cover those losses.
Part of the reason Tesla makes big profits off its customers is that they don't do discounts. I've never, ever paid a dealer list price for a car, there are always discounts to be had and sweetners like free servicing and accessories.
That's why it's hard to compare Tesla prices. A $50k Model 3 isn't comparable for a $50k car from anyone else because only a complete chump will pay any other manufacturer $50k for that vehicle. With a pre-reg deal it could easily be $30k.
I seem to recall I had to pretend to be from some island nation for them to allow me to have just one name. The old "real name" policy required you to have two names, a problem for quite a lot of people. They noticed this early on and made an exception for some countries.
Online it's much harder to get noticed too. There are people on Twitch to stream to no-one for weeks on end, just hoping that they will eventually get a viewer. Pirate radio had much higher listening figures.
Rather than try to determine by some complex set of criteria if they are employees, why not decide if she think they should be or not and then make any adjustments to the law that are necessary?
The auto lights aren't perfect, on any car. It's not just a simple on/off with light level detection, it's the high/dipped beams.
I don't get this desire for "busywork" when driving. Another busywork example I don't get: having to "start" the car. If the car has already authenticated you, why make the driver go through an added step of pressing a button or whatnot?
Well in Tesla's case it's because the door locks are not very secure so they had to add a PIN to start the car as a second layer of defence. The less snarky answer is so that you can sit in the driver's seat with no possibility of accidentally moving the car due to stuff like mode confusion.
As for "glancing aside", a quick test for you.... Trivial to type without looking, right?
Most keyboards have little ridges on the F and J keys so that you can locate your fingers correctly without looking, and then touch type. Similarly my car radio has a little nub on one of the channel favourite buttons. I never need to look at either.
But that's only half the problem with touch screens. With buttons you can locate them by feel, and they require a moderate amount of pressure to press down so you can run your fingers over them, use them for a bit of support as the car bounces around etc. Try operating your phone when it is held in a mount and you are in the passenger seat of a moving vehicle, it's not easy. You really need to cup the phone in your fingers to steady both your hand and the display and then carefully hit large touch points. Stuff like swiping is doubly hard. And the Model 3 touch screen is 15" diagonal so you all the touch points need to be near the edges to give you something to grip.
The same design constraints are evident in things like Android Auto.
Put all frequent interactions on the steering wheel. Change track, change station, change volume, mute, pause, change cruise control speed, change follow distance, voice commands, etc.
That's a pretty complex set of steering wheel controls. And what about climate control? Tesla doesn't even have an auto-demist function.
But what are those principles that disallow working with the US government but allow working with the Chinese government?
They cancelled their plan to work with China though. They thought about it, did a bit of prototyping to see what it would be like, and decided it was unethical.
It would have been an interesting story if they had found some compromise that the Chinese would accept and which they thought was overall a benefit to the Chinese people (the spread and democratisation of information is generally a good thing), but that did't happen so there isn't really anything to see here. By the time its existence was leaked it was already over.
$10bn for a military grade cloud system, secure and with high availability doesn't sound unreasonable. They will want their own secure data centres, probably hardened against physical attack as well as connected via dedicated lines to other military systems.
There are lots of additional costs to military grade hardware. Certifications, secure disposal, much higher levels of testing than normal server grade parts.
So what is the situation in the US with vehicle testing?
In most countries vehicles need to have periodic checks to make sure they are safe and meet emissions standards. If you don't want to get the test done you can only drive on private roads, not public ones. Self built cars are the same, if you want to drive on public roads you have to pass the test.
RC enthusiasts go out in groups and cooperate to make it more or less safe. People still get hit or interfere with each other, and it's regulated by gentlemen's agreements.
That only works as long as the number of people is relatively low. Same with driving cars, anyone could do it at first with no test or check on their vehicle. Then everybody wanted to drive and more regulation was needed. Same with radio transmitters and many other things operating in shared spaces.
Lasers don't make great anti-aircraft weapons... You need a hell of a lot of power to do any real damage, and keeping locked on a moving target is hard. Range is quite limited too, due to beam divergence and particles in the air scattering it.
Screening certainly helps, but the problem is that by the time there is enough material to take a sample and screen it you are in a position where the only options are carry to term or have an abortion. Having a third option, correction, is highly desirable.
This of course raises questions about what is considered a defect that needs to be cured. People have abortions for stuff like cleft palette.
I don't think it will really help socioeconomically disadvantaged children though. It won't help the parents find time to read to their kid, it won't make their local school any better, won't buy them books and computers, won't get the lead out of the paint, and most parents probably wouldn't want the colour of their skin changed either.
Having many controls that would be buttons in most cars on the touch screen is going to be a distraction for drivers. Even stuff like the headlight controls and windscreen wiper settings are on the screen, meaning you have to glance aside and hit a touch target with no tactile feedback.
What do the accident stats say? Do we have per-model data on at-fault crashes?
In this case the law is unsatisfactory. Consider the points on which the claim was rejected.
The accusers failed to demonstrate harm. I don't know the arguments used, but it is likely difficult to prove the tracking caused any kind of harm. Since there are no punitive damages in UK law, they can only claim for actual losses and thus it's very hard to prevent companies doing this kind of thing.
Secondly the judge said it was impossible to calculate the number of affected users. Again, this is because UK law doesn't seek to punish, only to put things right, so you can't just point to the fact that Google is very popular and that it's certain a large number of people were affected.
I don't think it's a good thing that people can get away with this stuff by creating enough uncertainty to make legal challenge impossible. Clearly something wrong was done here, Google aren't even denying it, yet there is no recourse.
That's bad op-sec. Any and all metadata that can be collected about you is dangerous, even if it seems trivial now. Everything should be encrypted by default, you should need a really really good reason to use plaintext.
Also consider the potential for interference via MITM attack on HTTP. You could be getting served malware. Some ISPs have injected their own ads and tracking headers.
Actually Firefox is the same. Mozilla have been pushing for this change too.
And Google is somewhat ahead of the curve regarding CAs and security. They know the limitations, that's why Chrome now doesn't display information from enhanced certs. Google knows they are worthless and don't identify the owner of a site reliably, do they don't display them in a little green box next to the address bar any more.
It's actually pissing off a lot of CAs. Now that Let's Encrypt offers basic certs for free, and there is no real difference between basic certs and enhanced certs, they don't have anything to sell.
How big of a deal is 5% better single core performance though, especially when you are being hit with Spectre/Meltdown mitigation and really need to disable hyperthreading too?
Very few tasks are going to benefit from that slightly higher single core performance, because anything that would will get parallelized as soon as possible. It's really just some older games that will run great anyway.
These things tend to go unnoticed because each step is a small and seemingly reasonable one.
Take Google as an example. If they had started with "we are going to photograph every public space, put it all online for public viewing, capture all WiFi SSIDs and map them, and then throw AI at it" someone might have balked.
What actually happened was someone suggested capturing images of streets and using image recognition to enhance their maps with information on street signs. Then the guys doing location services noticed that was happening and asked to add some wifi scanning for their systems, and oops they left packet capture enabled by mistake and got fined over it.
Later someone else suggested that all this imagery could be added to the user facing maps. And then the AI started working with it, teaching their machines to read building numbers and anything else they could.
And all the time they were getting mostly positive feedback. It's incredible how Google Maps shows you a pseudo-3D view now, with accurate building shapes because the AI looked at the satellite imagery and just figured out what the architecture was.
This is what a Chief Ethical Officer is supposed to think about, but apparently it doesn't work.
A nuclear tipped anti-ship torpedo isn't a declaration of nuclear war.
But in any case, a cyber attack is very risky. Russia would likely retaliate, and most of these attacks are against civilians. People will die in hospitals, in accidents, for lack of heat.
So rather than being a tit-for-tat response against military targets, it's an escalation to attacking civilians. Possibly a war crime too, since there is a legal requirement to minimize civilian casualties.
Slashdot Mirror
Two things. We have automatic high beams but but dipped beams, and it's not 100% reliable. I could make a joke about BMW drivers... And the rain sensor on the Model 3 is not very good, especially when it comes to selecting an appropriate interment speed. So it kind of is a problem when Tesla does it, because it's beta quality software.
It's the same price as last year.
If you want cheap then get a Pixel 2 on sale as they shift the last of their stock. I got a Pixel XL 1 for less than half price that way.
The other thing to note is that they also have much higher warranty costs than most manufacturers. They are still having quality issues, especially with paint, so spend more on warranty servicing. Hence they need a higher margin to cover those losses.
Part of the reason Tesla makes big profits off its customers is that they don't do discounts. I've never, ever paid a dealer list price for a car, there are always discounts to be had and sweetners like free servicing and accessories.
That's why it's hard to compare Tesla prices. A $50k Model 3 isn't comparable for a $50k car from anyone else because only a complete chump will pay any other manufacturer $50k for that vehicle. With a pre-reg deal it could easily be $30k.
I seem to recall I had to pretend to be from some island nation for them to allow me to have just one name. The old "real name" policy required you to have two names, a problem for quite a lot of people. They noticed this early on and made an exception for some countries.
Online it's much harder to get noticed too. There are people on Twitch to stream to no-one for weeks on end, just hoping that they will eventually get a viewer. Pirate radio had much higher listening figures.
Rather than try to determine by some complex set of criteria if they are employees, why not decide if she think they should be or not and then make any adjustments to the law that are necessary?
Turning on and off the headlights manually?
The auto lights aren't perfect, on any car. It's not just a simple on/off with light level detection, it's the high/dipped beams.
I don't get this desire for "busywork" when driving. Another busywork example I don't get: having to "start" the car. If the car has already authenticated you, why make the driver go through an added step of pressing a button or whatnot?
Well in Tesla's case it's because the door locks are not very secure so they had to add a PIN to start the car as a second layer of defence. The less snarky answer is so that you can sit in the driver's seat with no possibility of accidentally moving the car due to stuff like mode confusion.
As for "glancing aside", a quick test for you. ...
Trivial to type without looking, right?
Most keyboards have little ridges on the F and J keys so that you can locate your fingers correctly without looking, and then touch type. Similarly my car radio has a little nub on one of the channel favourite buttons. I never need to look at either.
But that's only half the problem with touch screens. With buttons you can locate them by feel, and they require a moderate amount of pressure to press down so you can run your fingers over them, use them for a bit of support as the car bounces around etc. Try operating your phone when it is held in a mount and you are in the passenger seat of a moving vehicle, it's not easy. You really need to cup the phone in your fingers to steady both your hand and the display and then carefully hit large touch points. Stuff like swiping is doubly hard. And the Model 3 touch screen is 15" diagonal so you all the touch points need to be near the edges to give you something to grip.
The same design constraints are evident in things like Android Auto.
Put all frequent interactions on the steering wheel. Change track, change station, change volume, mute, pause, change cruise control speed, change follow distance, voice commands, etc.
That's a pretty complex set of steering wheel controls. And what about climate control? Tesla doesn't even have an auto-demist function.
But what are those principles that disallow working with the US government but allow working with the Chinese government?
They cancelled their plan to work with China though. They thought about it, did a bit of prototyping to see what it would be like, and decided it was unethical.
It would have been an interesting story if they had found some compromise that the Chinese would accept and which they thought was overall a benefit to the Chinese people (the spread and democratisation of information is generally a good thing), but that did't happen so there isn't really anything to see here. By the time its existence was leaked it was already over.
$10bn for a military grade cloud system, secure and with high availability doesn't sound unreasonable. They will want their own secure data centres, probably hardened against physical attack as well as connected via dedicated lines to other military systems.
There are lots of additional costs to military grade hardware. Certifications, secure disposal, much higher levels of testing than normal server grade parts.
So what is the situation in the US with vehicle testing?
In most countries vehicles need to have periodic checks to make sure they are safe and meet emissions standards. If you don't want to get the test done you can only drive on private roads, not public ones. Self built cars are the same, if you want to drive on public roads you have to pass the test.
RC enthusiasts go out in groups and cooperate to make it more or less safe. People still get hit or interfere with each other, and it's regulated by gentlemen's agreements.
That only works as long as the number of people is relatively low. Same with driving cars, anyone could do it at first with no test or check on their vehicle. Then everybody wanted to drive and more regulation was needed. Same with radio transmitters and many other things operating in shared spaces.
Sucks for you but it was inevitable really.
Lasers don't make great anti-aircraft weapons... You need a hell of a lot of power to do any real damage, and keeping locked on a moving target is hard. Range is quite limited too, due to beam divergence and particles in the air scattering it.
Screening certainly helps, but the problem is that by the time there is enough material to take a sample and screen it you are in a position where the only options are carry to term or have an abortion. Having a third option, correction, is highly desirable.
This of course raises questions about what is considered a defect that needs to be cured. People have abortions for stuff like cleft palette.
I don't think it will really help socioeconomically disadvantaged children though. It won't help the parents find time to read to their kid, it won't make their local school any better, won't buy them books and computers, won't get the lead out of the paint, and most parents probably wouldn't want the colour of their skin changed either.
Having many controls that would be buttons in most cars on the touch screen is going to be a distraction for drivers. Even stuff like the headlight controls and windscreen wiper settings are on the screen, meaning you have to glance aside and hit a touch target with no tactile feedback.
What do the accident stats say? Do we have per-model data on at-fault crashes?
Aren't they supposed to have an advantage because they have only a few models to support and tightly controlled hardware?
In this case the law is unsatisfactory. Consider the points on which the claim was rejected.
The accusers failed to demonstrate harm. I don't know the arguments used, but it is likely difficult to prove the tracking caused any kind of harm. Since there are no punitive damages in UK law, they can only claim for actual losses and thus it's very hard to prevent companies doing this kind of thing.
Secondly the judge said it was impossible to calculate the number of affected users. Again, this is because UK law doesn't seek to punish, only to put things right, so you can't just point to the fact that Google is very popular and that it's certain a large number of people were affected.
I don't think it's a good thing that people can get away with this stuff by creating enough uncertainty to make legal challenge impossible. Clearly something wrong was done here, Google aren't even denying it, yet there is no recourse.
That's bad op-sec. Any and all metadata that can be collected about you is dangerous, even if it seems trivial now. Everything should be encrypted by default, you should need a really really good reason to use plaintext.
Also consider the potential for interference via MITM attack on HTTP. You could be getting served malware. Some ISPs have injected their own ads and tracking headers.
Actually Firefox is the same. Mozilla have been pushing for this change too.
And Google is somewhat ahead of the curve regarding CAs and security. They know the limitations, that's why Chrome now doesn't display information from enhanced certs. Google knows they are worthless and don't identify the owner of a site reliably, do they don't display them in a little green box next to the address bar any more.
It's actually pissing off a lot of CAs. Now that Let's Encrypt offers basic certs for free, and there is no real difference between basic certs and enhanced certs, they don't have anything to sell.
If it's your internal network you can just create your own cert and add it to your local machine(s). That's how it's supposed to work.
How big of a deal is 5% better single core performance though, especially when you are being hit with Spectre/Meltdown mitigation and really need to disable hyperthreading too?
Very few tasks are going to benefit from that slightly higher single core performance, because anything that would will get parallelized as soon as possible. It's really just some older games that will run great anyway.
I'll recalibrate my crapometer.
It depends. Where there is an agreement the real number comes up. Where there isn't it says "out of area" or "international".
That's my experience.
These things tend to go unnoticed because each step is a small and seemingly reasonable one.
Take Google as an example. If they had started with "we are going to photograph every public space, put it all online for public viewing, capture all WiFi SSIDs and map them, and then throw AI at it" someone might have balked.
What actually happened was someone suggested capturing images of streets and using image recognition to enhance their maps with information on street signs. Then the guys doing location services noticed that was happening and asked to add some wifi scanning for their systems, and oops they left packet capture enabled by mistake and got fined over it.
Later someone else suggested that all this imagery could be added to the user facing maps. And then the AI started working with it, teaching their machines to read building numbers and anything else they could.
And all the time they were getting mostly positive feedback. It's incredible how Google Maps shows you a pseudo-3D view now, with accurate building shapes because the AI looked at the satellite imagery and just figured out what the architecture was.
This is what a Chief Ethical Officer is supposed to think about, but apparently it doesn't work.
A nuclear tipped anti-ship torpedo isn't a declaration of nuclear war.
But in any case, a cyber attack is very risky. Russia would likely retaliate, and most of these attacks are against civilians. People will die in hospitals, in accidents, for lack of heat.
So rather than being a tit-for-tat response against military targets, it's an escalation to attacking civilians. Possibly a war crime too, since there is a legal requirement to minimize civilian casualties.