Show of hands: who thinks Elon Musk is above having a staffer make up this email or making it up himself?
I think they're too smart to do something like that. The probability of it being found out is low, but the PR damage caused by such fraud would be extreme. On balance, the expected risk of such a move is way too high. Plus, there's every reason to expect they have received some emails like this.
Tesla does not have a good record of repeat customers
I understand the concern, but there's really no evidence for it. Your examples of what Samsung and Microsoft have done aren't evidence... and Google has little more control over Samsung than over Microsoft. Could Google decide that it no longer cares about openness? Sure. But we're actually working quite hard to push it the other direction, and I see no reason to expect that to change.
What is the thing you're saying Google has done "in firmware" for Android for Work, but hasn't "flipped the switch"? Android for Work does nothing in firmware, it's all in Android; the only thing remotely close to that is the use of TrustZone for authentication and crypto key management -- and I'm the engineer responsible for those TrustZone components, and I can't figure out what "switch" you're talking about.
A warning is what we've had for several years now, and it's proven to be inadequate.
I understand this. What I was saying is that there should be a way to disable the new behavior (perhaps a setting in the Developer Options, where ordinary users would never see it) for those who don't need such a muscular approach.
The problem with that approach is that someone selling/giving you a pre-compromised phone would just flip that switch before they give it to you. If you're not going to be bothered by a big warning during bootup, you're definitely not the sort who will dig through the settings and find that problem... or factory reset the device to reset all of the switches.
If the new method really doesn't get in the way, all this is moot.
I thought he was just a pretty average govt. tech employee that decided to leak a bunch of documents. Now he seems to be treated like a leading expert on security? Is there something I missed here? Is his research something beyond a Google search?
How does one become an expert on security? Spend lots of time reading, thinking and studying. What else do you think Snowden has been doing for the last three years? He may not have been a security expert before collecting and leaking the documents, but he's clearly a pretty smart guy, and very motivated to care about security and privacy issues. He's been trying to use the pulpit his fame has given him to highlight those issues, and he's also clearly been doing his homework.
Aside from all of that, though, what's the point in questioning his expertise? If what he's saying doesn't make sense, say so. Your post isn't "insightful", it's just a variation of the argument from authority fallacy... in this case trying to discredit his ideas by citing his lack of authority, rather than addressing the ideas themselves.
Google reports NSL statistics, though with lower precision, based on the deal they worked out with the government. Beyond that, sure, it's possible that the NSA is using covert means without the company's cooperation or knowledge. We know from Snowden's files that at one point they were tapping network connections between Google data centers. Those are encrypted now, but there might be other ways. However, this story provides no evidence one way or the other, and based on this information, the "panopticon" isn't seeing very much and, as I said at the top of this thread, discussions of all-pervasive surveillance aren't related to this story.
I guess I mistakenly implied that freedom is a boolean condition, but it's not; you can be more free, or less free
Ah, so the possession of free will is a fuzzy value. I don't think that change to the definition makes anything clearer... and it also doesn't refute the notion that free will is just an emergent property of quantum randomness, since from the macro level quantum randomness looks fuzzy-valued, defining probability "clouds" of potential outcomes.
If the functioning of that soul deterministic or not and does it matter and why or why not? You've just pushed the question back to the next turtle down.
Indeed, I have. Plus I've introduced a whole new set of questions about the meaning of free will, particularly if we introduce an omniscient God into the discussion. I didn't claim that adding a soul answered anything, though it does allow free will to exist in a deterministic observable universe by providing another degree of freedom where it can live, non-deterministically.
To be sure I'm not painting an overly rosy picture... keep in mind that what I said applies only to devices with unlockable bootloaders. OEMs can choose not to allow unlocking, and most don't. That's their choice. At least Google's design explicitly tells them how to go about allowing unlocking without compromising security, and it pushes SoC makers (who actually write the bootloaders, by and large) to implement support for it so that if OEMs decide to allow unlocking they can do it by flipping a switch.
Automatically preventing the device from booting is incredibly intrusive. I find that objectionable out of the gate -- a warning would be much preferable.
A warning is what we've had for several years now, and it's proven to be inadequate. People purchasing used devices just ignore it because they don't understand it. Supporting the tiny minority who use custom ROMs is good, but supporting the large majority who do not is essential.
However, if Google is really allowing us to use unlocked devices and modify it without getting in our way, then my objection is removed.
Google encourages OEMs to make bootloaders unlockable. Most don't, though, so be careful what you buy. Nexus devices have unlockable bootloaders.
The alcoholic doesn't have any less "free will" than the non-alcoholic. If you put a gun to an alcoholic's head and a glass in his hand and say "If you drink that, I'll blow your brains out", the alcoholic will not drink. If you put the non-alcoholic in an appropriate situation (say, bachelor party), he may drink himself into a stupor. In both, there are a wide range of actions that may be taken by each, and miswiring of the alcoholic's brain simply amounts to a bias (a strong one) in those choices... but depending on the circumstances he will choose differently, and in many of them he'll choose the same as the non-alcoholic.
In the previous paragraph I used the word "choice", not because I think this illustration in any way proves that choices exist, but because it's a convenient label. In all cases the combination of circumstances + wiring determines an outcome. Is there any choice? Do you have a choice in how you're wired? It appears that we do... by consistently exercising "free will" to make one sort of choice we can form habits and create pathways making that sort of choice "easier". Or by choosing what information we consume we affect our brain's wiring.
But there I go, using "choice" and "free will" as though they exist. But who's to say that the initial "choices" to adjust our wiring aren't themselves just a deterministic result of circumstances and wiring? And who's to say the rationale we think underlies those choices isn't just post-hoc explanation of non-rational, pre-determined action? if you assume absolute determinism, then absolutely everything, at every stage, comes about because of the conditions that prevail, including all of the machinery that you believe you use to make decisions.
(I should perhaps mention that I don't actually hold the position I'm arguing here. I believe free will exists, that it doesn't derive from sub-atomic indeterminacy but instead comes ultimately from a soul which science will almost certainly never be able to detect. But I find fascinating the notion that it doesn't actually exist, that it's illusory post-hoc justification of fundamental randomness and emergent properties. It gets even more interesting when you throw in the multiple-world model for QC... the argument then is that you actually make all possible decisions at each decision point, then in each of the branching universes you apply post-hoc justification to explain why you made your "choice". Oh, and then there's the whole "we're actually living in the Matrix" theory:P)
Android is *not* removing openness. I'm a member of the Android security team, and worked around the edges of this feature. We (I'll use that pronoun for simplicity, but please note that I'm not claiming credit) put a great deal of additional effort into making sure that it supported modders who unlock their bootloaders and install custom software. We even made sure that they can use the verified boot feature to ensure that their self-signed images are not modified without their knowledge.
The goal is not to prevent modding, the goal is to improve security by ensuring that malicious images can't be installed.
This is an incredibly intrusive move on Google's part. They should provide a means to disable it.
In what way is it intrusive? All it does is verify that your boot and system image are unmodified... and there's no reason they should ever be modified in a normal device. Now, if you want to get a device with an unlockable bootloader and install different software, that's perfectly fine, and Google supports you in doing it. In fact, in that case you can even sign your own boot and system images and the verified boot system will ensure that *those* aren't modified, that they're exactly what you signed.
Now, this is likely to create problems for people who buy devices that aren't intended to be unlocked, and who have in the past been able to break the OEMs crappy lockdown. That era is ending anyway, though, and I don't think Google's move is significantly accelerating it. And for the vast majority of users it's more important to be able to have certainty that the device they bought on ebay isn't modified in some malicious way.
is there a reason you didn't type own like an adult?
Because "own" and "pwn" are different words, with different meanings. An attacker doesn't not gain ownership of your device in any legal sense or even most practical senses of the word. But the attacker does obtain control over the software running on it, and can use that to influence you in various ways and potentially extend that control to other devices. "Pwn" encapsulates those meanings, "own" does not. Yes, "pwn" started as a silly 733t-speak in-joke, but it has become a term of art in much of the security community (FYI, it's pronounced "pone").
On the other hand the jitteryness of the location reported by GPS that you observed may or may not have been due to GPS jamming by North Korea
I must have written that really unclearly:-)
I did *not* see jitteriness of my location while in South Korea. It was rock solid and very precise, in both Google Maps and Ingress. I mentioned that network location does not have those characteristics to support my assertion that my phone was using GPS, not network location.
If by "free will" you mean, as you should, a certain kind of functionality, an ability to cause your behavior to conform to the patterns you judge that it should conform to, then indeterminism is at most a hindrance and mostly completely inconsequential.
This is the really difficult thing about defining free will. Your definition implicitly assumes that it exists... and that the notion you're trying to define even makes sense. In a perfectly-deterministic universe, it doesn't.
Having free will is the opposite of that: the ability for your wants about [what to want and the efficacy of those wants on your behavior] to be effective.
You're assuming that you're actually making the decisions you think you're making. There's growing evidence that a huge amount of the internal narrative we build is constructed after the fact, to explain what we "chose", rather than as the actual mechanism. That's not to say that all of it works that way, or that we don't ever actually make true choices... but we don't really have any evidence that we do, either.
That doesn't require indeterminism, it just requires a decision-making mechanism built to function that way.
If the mechanism is entirely deterministic, is there any actual choice? To take your example, an alcoholic's bad choices arise from a pretty low-level miswiring of the reward circuits. The difference between an alcoholic's decisions and an occasional drinker's decisions is not one of willpower -- this is pretty well-understood neuroscience -- the alcoholic brains have been "wired" through a combination of genetics and experience to function differently. The difference is actually structural and mechanistic. It's entirely believable that in the future we may learn how to rewire the brain to correct those pathological defects. Will that mean that the former alcoholic acquired free will? Or will it simply mean that the new mechanism operations differently, just as water flows down a new channel when you cut one.
The free will theorem is much more interesting than that. I won't attempt to explain it, though, because I'd undoubtedly screw it up. If you're interested in such things, I highly recommend you read the original paper.
Android Nougat Won't Boot If Your Phone's Software Is Corrupt Or Has Malware unapproved by the device OEM
FTFY.
It's the device OEM's signature that's verified so it doesn't matter what Google thinks, unless it's a Nexus device. If it is a Nexus device you can unlock it and install whatever you like, of course. And you can even sign your own custom images. The bootloader will verify the signature and display the key fingerprint on the warning screen, so you can make sure that (a) the image is what was signed and (b) you are the one that signed it. If the verification of your self-signed image fails, the device won't boot.
This sounds like an excellent complementary feature for malware to trigger for a DoS attack.
If malware can mount the system partition as writable (which is far from trivial) so that it can write changes to the image, it can do much worse than a DoS attack. In particular, it can permanently pwn your device, which would be far more interesting to a malware author than maliciously bricking it.
It's not true that a single byte error will cause verification to fail. Nougat also adds forward error correction (Reed-Solomon coding) to the image structure, so very, very few random errors can cause enough corruption to be unrecoverable and cause verification to fail. It's not impossible that this will happen, indeed given that there are billions of Android devices it probably *will* happen, once or twice. But it will be well below the threshold of other sorts of low-probability catastrophic hardware failures.
Here is a hint: a neural net isn't anything like a brain.
There are some vague structural similarities between a neural net and the basic structure of the brain, but they're certainly very different, yes.
And now they just add "Deep" to the front of "neural net" or "learning" and think that is must be, like, really good now, because it is "deep" learning.
You don't have any idea what the "deep" means there, do you? It's not a content-free buzzword, it has a very specific meaning... and you might consider what the fact that you don't know what it is says about your knowledge in this space.
We aren't any closer to AI than we were 40 years ago.
Right. We've made, for example, no progress at all on voice recognition.
As someone who is a bit of a car enthusiast (always join the forums or car clubs for whatever vehicle I own, etc.) -- the fastest quarter mile results I ever see posted for vehicles taken to the drag strip is 9.x seconds. In most cases, you have people modding various sports or sporty cars to get down into the 12-13 second quarter mile range from wherever they start out at from the factory. Anyone running 11 seconds or under is considered "up there" in performance/speed.
So I'm starting to wonder.... is there pretty much a "hard limit" on how fast a quarter mile you can turn out based on the limitations of physics (tires can only provide so much grip, etc.)? Can you say at some point, "By getting my car to run a 9 second quarter mile, I've optimized it as much as is physically possible for a vehicle that's moving with rolling wheels on the ground?"
Top Fuel dragsters do the quarter mile in 4.x seconds.
The fastest quarter mile ever (according to a Google search) was 3.22 seconds, but that was a rocket car. It had wheels rolling on the ground, but didn't drive them to accelerate.
Since everything we do is driven by our brain, free will, if it exists, must have something to do with neuroscience.
The Conway/Kochen Free Will Theorem says that if free will exists, it derives from the free will possessed by elementary particles. It needn't arise from neuroscience if it's a more fundamental characteristic of the universe.
Note that I'm not claiming that either humans or quarks do or do not have free will, just pointing out that if we do have it, neuroscience isn't the only possible origin. Perhaps what we perceive as our free will is actually the collective free will of the subatomic particles that make up important parts of our brains... though that raises obvious and deep questions about what "free will" even is, since we tend to think of it as being goal-oriented and causal in nature, and it's not clear what kinds of "goals" subatomic particles could even have or how combinations of them could produce what we perceive as free will.
On the other hand, our pattern-matching brains tend to interpret everything in a causal/goal-focused way, to the degree that classical Aristotelian philosophy posited that *all* physical processes were a result of goals (final causes, "teloi", to use Aristotle's word). That is clearly wrong in lots of other cases, maybe this is just another example of our biases misleading us and that the truth is that free will is just how we perceive the macro-level emergent properties that result from quantum randomness. That is the most logical conclusion of the Free Will Theorem, anyway, that free will is nothing more and nothing less than quantum noise, scaled up.
So by weak AI we've had AI since the first programmable device?
No. There's a very large difference between machine learning-based systems and human-programmed systems. One very obvious one is that human-programmed systems can be explained by pointing out the specific rules that are being applied, in what order and for what reasons. With machine learning, especially the deep neural networks that are currently proving so effective, we can explain the structure of the system and the mechanisms used to "train" it, and how inputs flow through it to produce outputs, but we can't explain what logical "rules" are being applied, or when or why.
In theory, it should be possible to derive the rules by examining the weights in the neural network. In theory, it should be possible to implement those rules with traditional, human-defined logical rules. In practice, the details of how the trained system works are usually beyond us, and typically better than we can do with explicit logic.
Ah. I see. Your mistake is simple. You confused what your phone does and what Pokémon GO does.
Dude, your condescension is both obnoxious and completely unwarranted.
I understand perfectly well the difference between cell tower triangulation, Wifi AP triangulation, GPS (2D & 3D, with and without WAAS)... and I'm telling you that my phone had good GPS reception in South Korea. I only once checked to see how many satellites it was seeing... from my hotel room window it could get only four. Outdoors, though, precision and accuracy were both excellent, to the point I didn't bother opening the GPS tools app to check satellite visibility, and this was the case both in cities (where Wifi triangulation can paper over GPS unavailability) and outside of cities.
Slashdot Mirror
Show of hands: who thinks Elon Musk is above having a staffer make up this email or making it up himself?
I think they're too smart to do something like that. The probability of it being found out is low, but the PR damage caused by such fraud would be extreme. On balance, the expected risk of such a move is way too high. Plus, there's every reason to expect they have received some emails like this.
Tesla does not have a good record of repeat customers
Cite?
I understand the concern, but there's really no evidence for it. Your examples of what Samsung and Microsoft have done aren't evidence... and Google has little more control over Samsung than over Microsoft. Could Google decide that it no longer cares about openness? Sure. But we're actually working quite hard to push it the other direction, and I see no reason to expect that to change.
What is the thing you're saying Google has done "in firmware" for Android for Work, but hasn't "flipped the switch"? Android for Work does nothing in firmware, it's all in Android; the only thing remotely close to that is the use of TrustZone for authentication and crypto key management -- and I'm the engineer responsible for those TrustZone components, and I can't figure out what "switch" you're talking about.
A warning is what we've had for several years now, and it's proven to be inadequate.
I understand this. What I was saying is that there should be a way to disable the new behavior (perhaps a setting in the Developer Options, where ordinary users would never see it) for those who don't need such a muscular approach.
The problem with that approach is that someone selling/giving you a pre-compromised phone would just flip that switch before they give it to you. If you're not going to be bothered by a big warning during bootup, you're definitely not the sort who will dig through the settings and find that problem... or factory reset the device to reset all of the switches.
If the new method really doesn't get in the way, all this is moot.
I think that's the case.
I thought he was just a pretty average govt. tech employee that decided to leak a bunch of documents. Now he seems to be treated like a leading expert on security? Is there something I missed here? Is his research something beyond a Google search?
How does one become an expert on security? Spend lots of time reading, thinking and studying. What else do you think Snowden has been doing for the last three years? He may not have been a security expert before collecting and leaking the documents, but he's clearly a pretty smart guy, and very motivated to care about security and privacy issues. He's been trying to use the pulpit his fame has given him to highlight those issues, and he's also clearly been doing his homework.
Aside from all of that, though, what's the point in questioning his expertise? If what he's saying doesn't make sense, say so. Your post isn't "insightful", it's just a variation of the argument from authority fallacy... in this case trying to discredit his ideas by citing his lack of authority, rather than addressing the ideas themselves.
Google reports NSL statistics, though with lower precision, based on the deal they worked out with the government. Beyond that, sure, it's possible that the NSA is using covert means without the company's cooperation or knowledge. We know from Snowden's files that at one point they were tapping network connections between Google data centers. Those are encrypted now, but there might be other ways. However, this story provides no evidence one way or the other, and based on this information, the "panopticon" isn't seeing very much and, as I said at the top of this thread, discussions of all-pervasive surveillance aren't related to this story.
I guess I mistakenly implied that freedom is a boolean condition, but it's not; you can be more free, or less free
Ah, so the possession of free will is a fuzzy value. I don't think that change to the definition makes anything clearer... and it also doesn't refute the notion that free will is just an emergent property of quantum randomness, since from the macro level quantum randomness looks fuzzy-valued, defining probability "clouds" of potential outcomes.
If the functioning of that soul deterministic or not and does it matter and why or why not? You've just pushed the question back to the next turtle down.
Indeed, I have. Plus I've introduced a whole new set of questions about the meaning of free will, particularly if we introduce an omniscient God into the discussion. I didn't claim that adding a soul answered anything, though it does allow free will to exist in a deterministic observable universe by providing another degree of freedom where it can live, non-deterministically.
To be sure I'm not painting an overly rosy picture... keep in mind that what I said applies only to devices with unlockable bootloaders. OEMs can choose not to allow unlocking, and most don't. That's their choice. At least Google's design explicitly tells them how to go about allowing unlocking without compromising security, and it pushes SoC makers (who actually write the bootloaders, by and large) to implement support for it so that if OEMs decide to allow unlocking they can do it by flipping a switch.
Automatically preventing the device from booting is incredibly intrusive. I find that objectionable out of the gate -- a warning would be much preferable.
A warning is what we've had for several years now, and it's proven to be inadequate. People purchasing used devices just ignore it because they don't understand it. Supporting the tiny minority who use custom ROMs is good, but supporting the large majority who do not is essential.
However, if Google is really allowing us to use unlocked devices and modify it without getting in our way, then my objection is removed.
Google encourages OEMs to make bootloaders unlockable. Most don't, though, so be careful what you buy. Nexus devices have unlockable bootloaders.
Ah, you fell into my trap :-)
The alcoholic doesn't have any less "free will" than the non-alcoholic. If you put a gun to an alcoholic's head and a glass in his hand and say "If you drink that, I'll blow your brains out", the alcoholic will not drink. If you put the non-alcoholic in an appropriate situation (say, bachelor party), he may drink himself into a stupor. In both, there are a wide range of actions that may be taken by each, and miswiring of the alcoholic's brain simply amounts to a bias (a strong one) in those choices... but depending on the circumstances he will choose differently, and in many of them he'll choose the same as the non-alcoholic.
In the previous paragraph I used the word "choice", not because I think this illustration in any way proves that choices exist, but because it's a convenient label. In all cases the combination of circumstances + wiring determines an outcome. Is there any choice? Do you have a choice in how you're wired? It appears that we do... by consistently exercising "free will" to make one sort of choice we can form habits and create pathways making that sort of choice "easier". Or by choosing what information we consume we affect our brain's wiring.
But there I go, using "choice" and "free will" as though they exist. But who's to say that the initial "choices" to adjust our wiring aren't themselves just a deterministic result of circumstances and wiring? And who's to say the rationale we think underlies those choices isn't just post-hoc explanation of non-rational, pre-determined action? if you assume absolute determinism, then absolutely everything, at every stage, comes about because of the conditions that prevail, including all of the machinery that you believe you use to make decisions.
(I should perhaps mention that I don't actually hold the position I'm arguing here. I believe free will exists, that it doesn't derive from sub-atomic indeterminacy but instead comes ultimately from a soul which science will almost certainly never be able to detect. But I find fascinating the notion that it doesn't actually exist, that it's illusory post-hoc justification of fundamental randomness and emergent properties. It gets even more interesting when you throw in the multiple-world model for QC... the argument then is that you actually make all possible decisions at each decision point, then in each of the branching universes you apply post-hoc justification to explain why you made your "choice". Oh, and then there's the whole "we're actually living in the Matrix" theory :P)
8.) Openness is formally removed.
Android is *not* removing openness. I'm a member of the Android security team, and worked around the edges of this feature. We (I'll use that pronoun for simplicity, but please note that I'm not claiming credit) put a great deal of additional effort into making sure that it supported modders who unlock their bootloaders and install custom software. We even made sure that they can use the verified boot feature to ensure that their self-signed images are not modified without their knowledge.
The goal is not to prevent modding, the goal is to improve security by ensuring that malicious images can't be installed.
This is an incredibly intrusive move on Google's part. They should provide a means to disable it.
In what way is it intrusive? All it does is verify that your boot and system image are unmodified... and there's no reason they should ever be modified in a normal device. Now, if you want to get a device with an unlockable bootloader and install different software, that's perfectly fine, and Google supports you in doing it. In fact, in that case you can even sign your own boot and system images and the verified boot system will ensure that *those* aren't modified, that they're exactly what you signed.
Now, this is likely to create problems for people who buy devices that aren't intended to be unlocked, and who have in the past been able to break the OEMs crappy lockdown. That era is ending anyway, though, and I don't think Google's move is significantly accelerating it. And for the vast majority of users it's more important to be able to have certainty that the device they bought on ebay isn't modified in some malicious way.
is there a reason you didn't type own like an adult?
Because "own" and "pwn" are different words, with different meanings. An attacker doesn't not gain ownership of your device in any legal sense or even most practical senses of the word. But the attacker does obtain control over the software running on it, and can use that to influence you in various ways and potentially extend that control to other devices. "Pwn" encapsulates those meanings, "own" does not. Yes, "pwn" started as a silly 733t-speak in-joke, but it has become a term of art in much of the security community (FYI, it's pronounced "pone").
On the other hand the jitteryness of the location reported by GPS that you observed may or may not have been due to GPS jamming by North Korea
I must have written that really unclearly :-)
I did *not* see jitteriness of my location while in South Korea. It was rock solid and very precise, in both Google Maps and Ingress. I mentioned that network location does not have those characteristics to support my assertion that my phone was using GPS, not network location.
If by "free will" you mean, as you should, a certain kind of functionality, an ability to cause your behavior to conform to the patterns you judge that it should conform to, then indeterminism is at most a hindrance and mostly completely inconsequential.
This is the really difficult thing about defining free will. Your definition implicitly assumes that it exists... and that the notion you're trying to define even makes sense. In a perfectly-deterministic universe, it doesn't.
Having free will is the opposite of that: the ability for your wants about [what to want and the efficacy of those wants on your behavior] to be effective.
You're assuming that you're actually making the decisions you think you're making. There's growing evidence that a huge amount of the internal narrative we build is constructed after the fact, to explain what we "chose", rather than as the actual mechanism. That's not to say that all of it works that way, or that we don't ever actually make true choices... but we don't really have any evidence that we do, either.
That doesn't require indeterminism, it just requires a decision-making mechanism built to function that way.
If the mechanism is entirely deterministic, is there any actual choice? To take your example, an alcoholic's bad choices arise from a pretty low-level miswiring of the reward circuits. The difference between an alcoholic's decisions and an occasional drinker's decisions is not one of willpower -- this is pretty well-understood neuroscience -- the alcoholic brains have been "wired" through a combination of genetics and experience to function differently. The difference is actually structural and mechanistic. It's entirely believable that in the future we may learn how to rewire the brain to correct those pathological defects. Will that mean that the former alcoholic acquired free will? Or will it simply mean that the new mechanism operations differently, just as water flows down a new channel when you cut one.
I disagree that indeterminism isn't necessary for free will. In a purely deterministic universe free will must be an illusion.
The free will theorem is much more interesting than that. I won't attempt to explain it, though, because I'd undoubtedly screw it up. If you're interested in such things, I highly recommend you read the original paper.
Android Nougat Won't Boot If Your Phone's Software Is Corrupt Or Has Malware unapproved by the device OEM
FTFY.
It's the device OEM's signature that's verified so it doesn't matter what Google thinks, unless it's a Nexus device. If it is a Nexus device you can unlock it and install whatever you like, of course. And you can even sign your own custom images. The bootloader will verify the signature and display the key fingerprint on the warning screen, so you can make sure that (a) the image is what was signed and (b) you are the one that signed it. If the verification of your self-signed image fails, the device won't boot.
This sounds like an excellent complementary feature for malware to trigger for a DoS attack.
If malware can mount the system partition as writable (which is far from trivial) so that it can write changes to the image, it can do much worse than a DoS attack. In particular, it can permanently pwn your device, which would be far more interesting to a malware author than maliciously bricking it.
It's not true that a single byte error will cause verification to fail. Nougat also adds forward error correction (Reed-Solomon coding) to the image structure, so very, very few random errors can cause enough corruption to be unrecoverable and cause verification to fail. It's not impossible that this will happen, indeed given that there are billions of Android devices it probably *will* happen, once or twice. But it will be well below the threshold of other sorts of low-probability catastrophic hardware failures.
Here is a hint: a neural net isn't anything like a brain.
There are some vague structural similarities between a neural net and the basic structure of the brain, but they're certainly very different, yes.
And now they just add "Deep" to the front of "neural net" or "learning" and think that is must be, like, really good now, because it is "deep" learning.
You don't have any idea what the "deep" means there, do you? It's not a content-free buzzword, it has a very specific meaning... and you might consider what the fact that you don't know what it is says about your knowledge in this space.
We aren't any closer to AI than we were 40 years ago.
Right. We've made, for example, no progress at all on voice recognition.
As someone who is a bit of a car enthusiast (always join the forums or car clubs for whatever vehicle I own, etc.) -- the fastest quarter mile results I ever see posted for vehicles taken to the drag strip is 9.x seconds. In most cases, you have people modding various sports or sporty cars to get down into the 12-13 second quarter mile range from wherever they start out at from the factory. Anyone running 11 seconds or under is considered "up there" in performance/speed.
So I'm starting to wonder .... is there pretty much a "hard limit" on how fast a quarter mile you can turn out based on the limitations of physics (tires can only provide so much grip, etc.)? Can you say at some point, "By getting my car to run a 9 second quarter mile, I've optimized it as much as is physically possible for a vehicle that's moving with rolling wheels on the ground?"
Top Fuel dragsters do the quarter mile in 4.x seconds.
http://www.draglist.com/draglist/category.php?VIEW=Extended&CATEGORY%5B%5D=TOPFUEL&x=dragsters&SORTBY=ET%2CYEAR%2CMPH+DESC
The fastest quarter mile ever (according to a Google search) was 3.22 seconds, but that was a rocket car. It had wheels rolling on the ground, but didn't drive them to accelerate.
Since everything we do is driven by our brain, free will, if it exists, must have something to do with neuroscience.
The Conway/Kochen Free Will Theorem says that if free will exists, it derives from the free will possessed by elementary particles. It needn't arise from neuroscience if it's a more fundamental characteristic of the universe.
Note that I'm not claiming that either humans or quarks do or do not have free will, just pointing out that if we do have it, neuroscience isn't the only possible origin. Perhaps what we perceive as our free will is actually the collective free will of the subatomic particles that make up important parts of our brains... though that raises obvious and deep questions about what "free will" even is, since we tend to think of it as being goal-oriented and causal in nature, and it's not clear what kinds of "goals" subatomic particles could even have or how combinations of them could produce what we perceive as free will.
On the other hand, our pattern-matching brains tend to interpret everything in a causal/goal-focused way, to the degree that classical Aristotelian philosophy posited that *all* physical processes were a result of goals (final causes, "teloi", to use Aristotle's word). That is clearly wrong in lots of other cases, maybe this is just another example of our biases misleading us and that the truth is that free will is just how we perceive the macro-level emergent properties that result from quantum randomness. That is the most logical conclusion of the Free Will Theorem, anyway, that free will is nothing more and nothing less than quantum noise, scaled up.
Or not :P
So by weak AI we've had AI since the first programmable device?
No. There's a very large difference between machine learning-based systems and human-programmed systems. One very obvious one is that human-programmed systems can be explained by pointing out the specific rules that are being applied, in what order and for what reasons. With machine learning, especially the deep neural networks that are currently proving so effective, we can explain the structure of the system and the mechanisms used to "train" it, and how inputs flow through it to produce outputs, but we can't explain what logical "rules" are being applied, or when or why.
In theory, it should be possible to derive the rules by examining the weights in the neural network. In theory, it should be possible to implement those rules with traditional, human-defined logical rules. In practice, the details of how the trained system works are usually beyond us, and typically better than we can do with explicit logic.
Ah. I see. Your mistake is simple. You confused what your phone does and what Pokémon GO does.
Dude, your condescension is both obnoxious and completely unwarranted.
I understand perfectly well the difference between cell tower triangulation, Wifi AP triangulation, GPS (2D & 3D, with and without WAAS)... and I'm telling you that my phone had good GPS reception in South Korea. I only once checked to see how many satellites it was seeing... from my hotel room window it could get only four. Outdoors, though, precision and accuracy were both excellent, to the point I didn't bother opening the GPS tools app to check satellite visibility, and this was the case both in cities (where Wifi triangulation can paper over GPS unavailability) and outside of cities.
The 80K and 27K figures were called out by Google as number of user accounts. The request numbers were smaller. It's in the summary.