You're ignoring all of Google's self-driving data (which includes lots of driving with general-population drivers). And my second -- and more compelling -- point.
I want to see science done with a proper control group, which is group of alert drivers driving in perfect conditions
That's not a proper control group, that's an unrealistic fantasy group.
Again, there has been no publicity on how realistic it would be for Google's self driving array to be put into, let's say, a Chevy Spark
Why would it not be realistic? The Google-made test cars are even smaller than a Spark. All the system requires is some sensors around the perimeter, the lidar unit on top and a smallish CPU case inside.
so it is unknown whether the vast population will ever be able to afford such functionality.
I doubt that most people will ever buy a self-driving car. Not because they'll be priced out of it -- there's nothing inherently expensive in any of the self-driving equipment; it's expensive now but by the time it's made in volume it won't be -- but because fleets of self-driving taxis will make it so that owning a car is simply not necessary the way it is in most of the country today. Calling a car with your phone when you need it will be cheaper and in some cases even more convenient. This will be especially true for the next generation (probably not born quite yet) who will grow up with self-driving taxis and becomes accustomed to having that sort of mobility even before they are old enough to have a car of their own.
Even if it does work perfectly.
It doesn't have to be perfect. It just has to be good enough.
If I'm not driving it, it should only need to be insured against damage caused by others. If it causes an accident, that's the manufacturers liability, not mine, as I wasn't driving it. If it's any way made to be my liability, I'm not going to buy it. I'd rather hire a self driving taxi than own the liability of my own self driving car.
Volvo, Google and Mercedes-Benz have all publicly stated that they will take full liability for accidents caused by their self-driving systems. I expect that this will become the norm, because it just makes sense. If the self-driving system is in control and makes an error that causes an accident, the system is faulty and not fit for its stated purpose, and the maker of the system bears the liability.
However, I predict that many people will never buy a self-driving car. Instead, they'll go from owning a manually-driven car to using a self-driving car fleet, on a pay per use basis.
It's cute how you think AI is strong enough to pick out a deer behind a bush *and* anticipate that it is about to run out into the road
Well, I actually experienced something similar while riding in one of Google's self-driving cars. It wasn't a deer, he was a cyclist, and it wasn't a bush, it was a hedge, but the car saw it and delayed entering the intersection to let him go past, even though no human driver could have known the cyclist was coming.
I'll grant that knowing which way a deer is going to jump is not achievable with AI -- it's not achievable with human intelligence, either. But what AI could do is to decide to slow a bit, or perhaps change lanes to put more distance between it and the potentially-running deer. Of course, in the case of a semi-tractor, the right decision would be to just maintain course and speed. If the deer jumps in the road, well, too bad for the deer. Semis have big and very strong bumpers for just such occasions.
The FACT of the matter is this isn't even real "AI" to start with, it's shitty half-assed "machine learning" crap, and at current it has to 'phone home' to have a remote HUMAN operator take over from it when it runs into something it can't handle.
So what?
If it reduces accidents, and the evidence we have so far is that it does, then bring it on. BTW, we have two forms of evidence: the numbers on accidents that we get from companies making self-driving systems and the willingness of at least some of those companies to accept full liability for any accidents caused by their systems. The former could possibly be dismissed. The latter is pretty compelling.
I'd say that if you're going to buy one of these phones, you're already looking at spending $700-$900 on the device, so you should just plan on spending another hundred or so on the "audio problem". Add that to the purchase price when you decide whether to buy.
The additional cost is annoying, but the main issue is still having to deal with adapters where they otherwise wouldn't be necessary. 1/8" plugs and jacks are ubiquitous. Lightning and USB-C, not so much. I use Bluetooth for certain things, but there are applications where it isn't a reasonable solution.
Well, my experience is that it hasn't been a big problem for me. YMMV, of course.
Could one do software development and testing while offline, with one of these puppies? e.g. Can I have linux in a VM or use docker containers etc in chromeos?
No.
Yes. That one is a chroot system running under ChromeOS' Linux OS, but there are other approaches. Crouton is great as a dev environment.
Also note that ChromeOS works just fine offline as long as the Chrome apps you're using support it. Most do.
In addition, there's always the option of flipping the device to dev mode and replacing ChromeOS with whatever you like (and can get to run; many Linux and *BSD distros work fine on Chromebooks. I'm not sure you could get Windows to run and I would be amazed if you could make OS X work).
Google hopes its new Pixelbook, which sells for $999 to $1,649, will give it a viable challenger to Apple's MacBooks and other premium laptops.
Apple's MacBooks and other premium laptops are OS-agnostic, OSX aside. You can run Windows or Linux on them without having to worry about hitting the wrong key at boot time and wiping out your installation. Google's value proposition is based on collecting data about you and advertising to you; are they going to let you escape their clutches, and install another operating system on the device without extreme hazard at every boot time?
Chromebooks have always had a "dev screw", a switch (originally a screw) that allows you to switch the device into "dev mode". In that mode, all verification of the bootloader and OS is disabled and you're free to install anything you want on it.
Google engineering actually has a pretty strong cultural belief that it's important that users be able to fully "own" their devices. Google can't force that view on Android device makers, but actually has managed to force it on Chromebook makers. So you can be certain that Google's own device has such a switch.
The lack of a 1/8" jack means I would either have to 1) always carry a dongle around with my phone or 2) keep a dongle with every device I *might* connect to.
Or option 3 (which can be mixed with options 1 and 2): switch to wireless. All of the stereos in my house have Chromecast Audio devices connected to them. That cost a few bucks, but it's very nice. Not only can I play music on any one of them, I have also set up a few different groups of devices. And I can do it without having to plug or unplug anything, in fact I usually do it just with my voice, without even bothering to pick up my phone. "OK Google, play <artist / album/ genre / song / whatever> on home group" starts playing throughout the entire house.
I normally use bluetooth headphones anwyay, so that's no big deal. I have one set of high-end Sennheiser over-the-ear headphones that uses the 3.5mm jack, and I also use the audio jack in my car (my truck has BT), so I'll probably buy a pair of dongles to attach to those.
Note that I'm not saying this is the right solution for everyone. The Chromecast Audio costs $35 from Google, $27 from Wal-mart, so buying a half dozen of them costs a little money. And lots of BT audio devices suck (BT audio does not inherently suck, BTW, it depends on the codecs used, which are negotiated between phone and headphone/speaker/whatever). But you can get good ones.
Note that I've been using a Pixel 2 XL as my daily driver (I work for Google, on Android) for about three months now, so I've been dealing with the reality of this problem. It hasn't actually been much of a problem. I already had the Chromecasts.
I'd say that if you're going to buy one of these phones, you're already looking at spending $700-$900 on the device, so you should just plan on spending another hundred or so on the "audio problem". Add that to the purchase price when you decide whether to buy.
Sounds like not really ready for prime time, just cherry picked locations.
Sounds like a cautious start in a state that allows it.
Good engineering mandates aggressive testing, but production deployment should be done slowly and in the most favorable environment first. You don't do your first real surgery on a tricky brain aneurysm, you start with an ingrown toenail or similar. Oh, and you also do it in a state which gives you legal permission to practice medicine.
You should really learn how to say what you mean directly, and to support your position, rather than making snide comments by inference -- and with absolutely nothing to substantiate your position. This sort of post is just the big boy pants version of "Nyah nyah you're a dumbhead!".
I used to work with a couple of NASA subcontractors who talked about when they would code by flipping 8 switches and then pressing a button to push that single byte of code into the computer.
Sure, but they wrote their programs on paper in decimal or hexadecimal (likely decimal), then translated to binary only to toggle it in.
You got that far? My bullshit meter overflowed when he started ranting about binary instead of assembler which they've had from the very beginning as simple text substitution...
Anyone looking over a programmer's shoulder as they pored over line after line like "100001010011" and "000010011110"
No, assembler wasn't available from the very beginning... but no one ever actually programmed in binary. They at least used decimal representations of instruction values and arguments. Even when systems (mostly hobby systems) had no input device other than toggle switches, you wrote out your code in decimal or hexadecimal, then translated to binary only for data entry. You would never "pore over" lines of binary except perhaps to check the translation from the more usable form right next to it.
In addition, even if we that the simulation doesn't deliberately work to make us think it's not a simulation, this still wouldn't prove we're not living in a simulation.
The simulations we create don't operate by trying to simulate every particle. There are simulations that do this, called finite particle simulations, but they're exceptional, not the normal case. Instead, we simulate at a higher level, assuming, for example, that the motion of objects can be modeled without paying attention to the huge numbers of particles that make up the objects. That is, we model the emergent properties that arise from low-level particle dynamics, rather than the particle dynamics directly.
If I were building a universe simulation, I would absolutely do that, use higher-level, emergent property models to model large-scale interactions and events. Only where small-scale interactions make a difference to the observing intelligences in my simulation would I bother modeling the details. This might result in occasional weird edge cases where the presence or absence of an observer appears to affect the outcome of a process.... hmm.:-)
Agree. Americans are just more violent people on average than most civilized nations' populations. No laws will change that.
I'd really like to disagree with this, but I can't. One thing I can say, though, is that the culture of violence is not homogenous throughout the United States. Violence is rare in lots of places, on par with Europe, and in some cases even below many European countries. New Hampshire, for example, has a homicide rate of 0.9 per 100K, on the same order as Denmark, Germany and the UK. Maine, Minnesota and Vermont are on par with France.
It's also worth pointing out that though when you said "Americans" you meant "people in the United States", the fact is that the more literal definition of "Americans" as "people in the Americas" makes your statement even more true. The homicide rate in the Americas (16.3 per 100K) is higher even than the homicide rate in Africa (12.5 per 100K) and this is true even in American countries with strong gun restrictions.
(Aside: Note that total homicide rate is the right measure when trying to estimate the effects of gun control. Measurements of gun homicides put a heavy thumb on the scale and are really quite silly since the tool used doesn't matter all that much to the murder victim.)
My point here is this: gun regulations do affect the amount of deaths by guns
Maybe. I think there's also a significant cultural aspect. One telling statistic: There are more knife murders per capita in Chicago than in Toronto. That can't be blamed on guns, and neither city has any knife restrictions.
Toronto is in Canada, which does have knife restrictions. See also some of the better knife forums where this question comes up a lot.
Bah. Okay, I should have said "no knife restrictions that matter". Lots of places regulate switchblades, gravity-assisted opening blades, long folding knives, etc. In fact Chicago's regulations are roughly the same as Toronto's in that respect. But your average kitchen knife is a perfectly-serviceable weapon, and in many ways superior to the banned knives.
My point here is this: gun regulations do affect the amount of deaths by guns
Maybe. I think there's also a significant cultural aspect. One telling statistic: There are more knife murders per capita in Chicago than in Toronto. That can't be blamed on guns, and neither city has any knife restrictions.
Slashdot Mirror
Your reading comprehension is poor.
Nope. It just has to work better than the average human driver. And that's a pretty low bar.
Automation is actually better for these types of tasks, than strong AI (if we had that) would be
Yeah, strong AI might get bored and decide to focus on its poetry composition, rather than drive.
You're ignoring all of Google's self-driving data (which includes lots of driving with general-population drivers). And my second -- and more compelling -- point.
I want to see science done with a proper control group, which is group of alert drivers driving in perfect conditions
That's not a proper control group, that's an unrealistic fantasy group.
Again, there has been no publicity on how realistic it would be for Google's self driving array to be put into, let's say, a Chevy Spark
Why would it not be realistic? The Google-made test cars are even smaller than a Spark. All the system requires is some sensors around the perimeter, the lidar unit on top and a smallish CPU case inside.
so it is unknown whether the vast population will ever be able to afford such functionality.
I doubt that most people will ever buy a self-driving car. Not because they'll be priced out of it -- there's nothing inherently expensive in any of the self-driving equipment; it's expensive now but by the time it's made in volume it won't be -- but because fleets of self-driving taxis will make it so that owning a car is simply not necessary the way it is in most of the country today. Calling a car with your phone when you need it will be cheaper and in some cases even more convenient. This will be especially true for the next generation (probably not born quite yet) who will grow up with self-driving taxis and becomes accustomed to having that sort of mobility even before they are old enough to have a car of their own.
Even if it does work perfectly.
It doesn't have to be perfect. It just has to be good enough.
My guess is he is a suburbanite who is afraid of meeting new people so never has seen a foreigner.
Fuck you monkey, go back to your own shitty hindustan, I don't want you here.
Guess confirmed. In only 8 minutes!
If I'm not driving it, it should only need to be insured against damage caused by others. If it causes an accident, that's the manufacturers liability, not mine, as I wasn't driving it. If it's any way made to be my liability, I'm not going to buy it. I'd rather hire a self driving taxi than own the liability of my own self driving car.
Volvo, Google and Mercedes-Benz have all publicly stated that they will take full liability for accidents caused by their self-driving systems. I expect that this will become the norm, because it just makes sense. If the self-driving system is in control and makes an error that causes an accident, the system is faulty and not fit for its stated purpose, and the maker of the system bears the liability.
However, I predict that many people will never buy a self-driving car. Instead, they'll go from owning a manually-driven car to using a self-driving car fleet, on a pay per use basis.
It's cute how you think AI is strong enough to pick out a deer behind a bush *and* anticipate that it is about to run out into the road
Well, I actually experienced something similar while riding in one of Google's self-driving cars. It wasn't a deer, he was a cyclist, and it wasn't a bush, it was a hedge, but the car saw it and delayed entering the intersection to let him go past, even though no human driver could have known the cyclist was coming.
I'll grant that knowing which way a deer is going to jump is not achievable with AI -- it's not achievable with human intelligence, either. But what AI could do is to decide to slow a bit, or perhaps change lanes to put more distance between it and the potentially-running deer. Of course, in the case of a semi-tractor, the right decision would be to just maintain course and speed. If the deer jumps in the road, well, too bad for the deer. Semis have big and very strong bumpers for just such occasions.
The FACT of the matter is this isn't even real "AI" to start with, it's shitty half-assed "machine learning" crap, and at current it has to 'phone home' to have a remote HUMAN operator take over from it when it runs into something it can't handle.
So what?
If it reduces accidents, and the evidence we have so far is that it does, then bring it on. BTW, we have two forms of evidence: the numbers on accidents that we get from companies making self-driving systems and the willingness of at least some of those companies to accept full liability for any accidents caused by their systems. The former could possibly be dismissed. The latter is pretty compelling.
I'd say that if you're going to buy one of these phones, you're already looking at spending $700-$900 on the device, so you should just plan on spending another hundred or so on the "audio problem". Add that to the purchase price when you decide whether to buy.
The additional cost is annoying, but the main issue is still having to deal with adapters where they otherwise wouldn't be necessary. 1/8" plugs and jacks are ubiquitous. Lightning and USB-C, not so much. I use Bluetooth for certain things, but there are applications where it isn't a reasonable solution.
Well, my experience is that it hasn't been a big problem for me. YMMV, of course.
Just curious.
Could one do software development and testing while offline, with one of these puppies? e.g. Can I have linux in a VM or use docker containers etc in chromeos?
No.
Yes. That one is a chroot system running under ChromeOS' Linux OS, but there are other approaches. Crouton is great as a dev environment.
Also note that ChromeOS works just fine offline as long as the Chrome apps you're using support it. Most do.
In addition, there's always the option of flipping the device to dev mode and replacing ChromeOS with whatever you like (and can get to run; many Linux and *BSD distros work fine on Chromebooks. I'm not sure you could get Windows to run and I would be amazed if you could make OS X work).
Google hopes its new Pixelbook, which sells for $999 to $1,649, will give it a viable challenger to Apple's MacBooks and other premium laptops.
Apple's MacBooks and other premium laptops are OS-agnostic, OSX aside. You can run Windows or Linux on them without having to worry about hitting the wrong key at boot time and wiping out your installation. Google's value proposition is based on collecting data about you and advertising to you; are they going to let you escape their clutches, and install another operating system on the device without extreme hazard at every boot time?
Chromebooks have always had a "dev screw", a switch (originally a screw) that allows you to switch the device into "dev mode". In that mode, all verification of the bootloader and OS is disabled and you're free to install anything you want on it.
Google engineering actually has a pretty strong cultural belief that it's important that users be able to fully "own" their devices. Google can't force that view on Android device makers, but actually has managed to force it on Chromebook makers. So you can be certain that Google's own device has such a switch.
The lack of a 1/8" jack means I would either have to 1) always carry a dongle around with my phone or 2) keep a dongle with every device I *might* connect to.
Or option 3 (which can be mixed with options 1 and 2): switch to wireless. All of the stereos in my house have Chromecast Audio devices connected to them. That cost a few bucks, but it's very nice. Not only can I play music on any one of them, I have also set up a few different groups of devices. And I can do it without having to plug or unplug anything, in fact I usually do it just with my voice, without even bothering to pick up my phone. "OK Google, play <artist / album/ genre / song / whatever> on home group" starts playing throughout the entire house.
I normally use bluetooth headphones anwyay, so that's no big deal. I have one set of high-end Sennheiser over-the-ear headphones that uses the 3.5mm jack, and I also use the audio jack in my car (my truck has BT), so I'll probably buy a pair of dongles to attach to those.
Note that I'm not saying this is the right solution for everyone. The Chromecast Audio costs $35 from Google, $27 from Wal-mart, so buying a half dozen of them costs a little money. And lots of BT audio devices suck (BT audio does not inherently suck, BTW, it depends on the codecs used, which are negotiated between phone and headphone/speaker/whatever). But you can get good ones.
Note that I've been using a Pixel 2 XL as my daily driver (I work for Google, on Android) for about three months now, so I've been dealing with the reality of this problem. It hasn't actually been much of a problem. I already had the Chromecasts.
I'd say that if you're going to buy one of these phones, you're already looking at spending $700-$900 on the device, so you should just plan on spending another hundred or so on the "audio problem". Add that to the purchase price when you decide whether to buy.
Sounds like not really ready for prime time, just cherry picked locations.
Sounds like a cautious start in a state that allows it.
Good engineering mandates aggressive testing, but production deployment should be done slowly and in the most favorable environment first. You don't do your first real surgery on a tricky brain aneurysm, you start with an ingrown toenail or similar. Oh, and you also do it in a state which gives you legal permission to practice medicine.
You should really learn how to say what you mean directly, and to support your position, rather than making snide comments by inference -- and with absolutely nothing to substantiate your position. This sort of post is just the big boy pants version of "Nyah nyah you're a dumbhead!".
I used to work with a couple of NASA subcontractors who talked about when they would code by flipping 8 switches and then pressing a button to push that single byte of code into the computer.
Sure, but they wrote their programs on paper in decimal or hexadecimal (likely decimal), then translated to binary only to toggle it in.
You got that far? My bullshit meter overflowed when he started ranting about binary instead of assembler which they've had from the very beginning as simple text substitution...
Anyone looking over a programmer's shoulder as they pored over line after line like "100001010011" and "000010011110"
No, assembler wasn't available from the very beginning... but no one ever actually programmed in binary. They at least used decimal representations of instruction values and arguments. Even when systems (mostly hobby systems) had no input device other than toggle switches, you wrote out your code in decimal or hexadecimal, then translated to binary only for data entry. You would never "pore over" lines of binary except perhaps to check the translation from the more usable form right next to it.
If the choice is between an incompetent evil and a competent one, I'd rather endure the former.
The choice is between an unpredictable evil and a predictable one. The latter is much easier to manage.
Also hopefully Trump will trip and fall into a fucking cuisinart,
This is a horribly thing to wish for.
If it happened, we'd have Pence as our president, which would be even worse.
No, it wouldn't. Pence is a more manageable sort of crazy.
In addition, even if we that the simulation doesn't deliberately work to make us think it's not a simulation, this still wouldn't prove we're not living in a simulation.
The simulations we create don't operate by trying to simulate every particle. There are simulations that do this, called finite particle simulations, but they're exceptional, not the normal case. Instead, we simulate at a higher level, assuming, for example, that the motion of objects can be modeled without paying attention to the huge numbers of particles that make up the objects. That is, we model the emergent properties that arise from low-level particle dynamics, rather than the particle dynamics directly.
If I were building a universe simulation, I would absolutely do that, use higher-level, emergent property models to model large-scale interactions and events. Only where small-scale interactions make a difference to the observing intelligences in my simulation would I bother modeling the details. This might result in occasional weird edge cases where the presence or absence of an observer appears to affect the outcome of a process.... hmm. :-)
Not quite - there is a law in Chicago limiting the length of a knife blade. https://chicagocode.org/8-24-0...
Also state laws banning switchblades, etc.
Fair point. As history, though, the claim doesn't stand up in any case.
Agree. Americans are just more violent people on average than most civilized nations' populations. No laws will change that.
I'd really like to disagree with this, but I can't. One thing I can say, though, is that the culture of violence is not homogenous throughout the United States. Violence is rare in lots of places, on par with Europe, and in some cases even below many European countries. New Hampshire, for example, has a homicide rate of 0.9 per 100K, on the same order as Denmark, Germany and the UK. Maine, Minnesota and Vermont are on par with France.
It's also worth pointing out that though when you said "Americans" you meant "people in the United States", the fact is that the more literal definition of "Americans" as "people in the Americas" makes your statement even more true. The homicide rate in the Americas (16.3 per 100K) is higher even than the homicide rate in Africa (12.5 per 100K) and this is true even in American countries with strong gun restrictions.
(Aside: Note that total homicide rate is the right measure when trying to estimate the effects of gun control. Measurements of gun homicides put a heavy thumb on the scale and are really quite silly since the tool used doesn't matter all that much to the murder victim.)
My point here is this: gun regulations do affect the amount of deaths by guns
Maybe. I think there's also a significant cultural aspect. One telling statistic: There are more knife murders per capita in Chicago than in Toronto. That can't be blamed on guns, and neither city has any knife restrictions.
Toronto is in Canada, which does have knife restrictions. See also some of the better knife forums where this question comes up a lot.
Bah. Okay, I should have said "no knife restrictions that matter". Lots of places regulate switchblades, gravity-assisted opening blades, long folding knives, etc. In fact Chicago's regulations are roughly the same as Toronto's in that respect. But your average kitchen knife is a perfectly-serviceable weapon, and in many ways superior to the banned knives.
My point here is this: gun regulations do affect the amount of deaths by guns
Maybe. I think there's also a significant cultural aspect. One telling statistic: There are more knife murders per capita in Chicago than in Toronto. That can't be blamed on guns, and neither city has any knife restrictions.