The Chinese room experiment does not demonstrate what you think it does. Serle's argument is (being generous) circular. If the set of rules the man implements is sufficiently nuanced, then the entire *system* is intelligent. There is nothing distinct about a "mind" aside from a set of rules for symbolic information processing.
I particularly liked this reply, from the article:
Churchland's luminous room Consider a dark room containing a man holding a bar magnet or charged object. If the man pumps the magnet up and down, then, according to Maxwell's theory of artificial luminance (AL), it will initiate a spreading circle of electromagnetic waves and will thus be luminous. But as all of us who have toyed with magnets or charged balls well know, their forces (or any other forces for that matter), even when set in motion produce no luminance at all. It is inconceivable that you might constitute real luminance just by moving forces around! The problem is that he would have to wave the magnet up and down something like 450 trillion times per second in order to see anything.
The International Commission on Non-Ionizing Radiation Protection published a document containing all the information one could possibly want.. Read the conclusions section if nothing else.
Bullshit. The brain is a computer. Sure, it's a strange architecture: it's made of billions of impressively energy-efficient gates each operating at the order of tens of hertz. Fan-out is huge --- a gate on a microprocessor might be connected to 50 others, but a neuron can have tens of thousands of connections. A CPU has one fast, global clock, while the brain has overlapping and distributed clock signals for synchronizing neuron firing. The short term memory system uses the equivalent of old-fashioned delay lines, while long-term storage is implemented with redundant, distributed rewiring. It's content-addressable and has a storage capacity in the terabyte range, though it has really lousy indexing. Input and output are essentially memory-mapped, with lots of special purposehardware acceleration.
There are a lot of similarities too: both our computers and our brains run software, with only a few basic features baked into the hardware. Both parse raw environmental input and parse it into abstractions that can be manipulated symbolically according to software-defined rules. Both can evaluate the lambda calculus and run a universal Turing machine. Neither can solve the halting problem in all cases. Both have large data stores. Both have networked inputs. Both crash. Both employ algorithms and data structures to process information. Both eventually fall apart.
Our brains are not magical devices somehow above scientific inquiry. They are ordinary, pedestrian objects in that obey the same laws of physics that govern baseballs and light switches. That we don't completely understand all the brain's mechanisms is no reason to believe it's qualitatively different from any other computer. Have you read every line of code in the web browser you're staring at?
So, you can't say you are 100% sure...Maybe the interaction is too small to have an effect on one's health, but nobody proved that yet.
SCIENCE DOES NOT WORK THAT WAY.
Any demand to be "100% sure" that an effect does not exist is unreasonable nonsense. Nobody can do that. All we can do is make observations, derive mathematical patterns from them (we call these "theories") and make predictions based on these generalizations with the aim of finding conflicts between the theory and the observation. If we find any, the theory is wrong and we begin again. If we don't find any contradictions, we provisionally accept the theory. As the theory continues to accurately predict reality, our confidence in it increases. With respect to a long-established theory like that of gravity or evolution, it's overwhelmingly likely that any discrepancy between the theory and observation is due to an error in the latter, and one would need extraordinary evidence to overturn such a theory, e.g. the famous hypothetical "fossil rabbit in the precambrian".
Now, with respect to electromagnetic fields, we have very specific and accurate theories to describe their behavior, effects, and interactions, and these theories have lasted over a century. The neurons that make up our brains and that form our consciousness are not special, and like all matter, they are also subject to these same theories. Neurons create electrical voltages by changing the concentration of ions inside themselves relative to their environment, and this mechanism is well-understood: the mysteries of the brain are in the emergent phenomena. The chemistry is pedestrian.
Now, when we plug the numbers for consumer electronics and neurons into electromagnetic theory, we see that there is simply no effect. The radiation is too weak to disrupt the bonds that join molecule in DNA and proteins, and at the intensity used in mobile phones, the heating effect is weaker than that of a pillow at night. The theory predicts that nothing should happen.
Now, just to be sure, various researchers have looked for an effect anyway, and have overwhelmingly failed to find a link between normal levels of mobile radiation and cancer. The few spurious positive results can be attributed to bad experimental design (e.g., uncontrolled and self-selected survey responses) or simple publication bias (if you perform a hundred studies at a 95% confidence interval, five of them will show spuriously positive results!).
Combined, the robustness of electromagnetic theory, our understanding of the chemistry of the cell, and the failure to find conclusive causal evidence paint as certain a picture as one can paint using the canvas of science. The theory says that mobile phones shouldn't cause cancer. When we look for cancer, we find nothing. We have no plausible explanation for how they could cause cancer. The only reasonable conclusion to draw is that absent further and significant evidence, cell phones do not cause cancer.
No, it's not 100% certain. It's also not 100% certain that there's no luminfiferous aether, and we can't be 100% sure that rotting fruit doesn't turn into insects by itself. If you want to believe in a cancer risk anyway, you're no better than someone who believes in hexes, astrology, or homeopathy.
All right --- Sony appears to be guilty of perjury after filing a takedown notice for someone else's work.
Who is going to do something about it? Selective enforcement is wonderful, isn't it? If Sony succeeds in this, it'll embolden others to file takedown notices against anything they dislike for any reason whatsoever.
Oh, for fuck's sake, have you seen LWN's "security" page? Every week, there's some remote code execution vulnerability or another. At least distributions regularly push updates --- Apple usually waits for its next minor release. I'm sick and tired of this puerile and reflexive Microsoft-bashing.
The Microsoft of today is nothing like what it was 10, 15 years ago when it became fashionable to hate. If MS is an evil empire today, it's the empire of Brezhnev, not of Stalin. It's generally pretty reasonable and a decent citizen of our software community. It's perfectly legitimate to expect consistency from people, but companies are composed of people, and to a large extent, the people at Microsoft are different these days.
Full disclosure: I work for Microsoft. I hack on Windows.
It's not that internationalization is the work of the devil, but rather that it should happen at a higher level than an individual PDF. Allowing different content to be displayed in different language environments raises serious questions about document integrity: imagine a international contract PDF that displayed one payment for German users and another for French ones.
PDF-the-document-format is a good thing in that it allows perfect reproduction of a printed document anywhere. PDF-the-generic-container, on the other hand, is both frightening and of dubious utility, but I can see why Adobe might have a business case for trying to drive this approach anyway. This is why we can't have nice things.
No, at least Mozilla blocks Flash popups too. The issue is that these "popups" are created in response to user clicks, and the browser can't tell the difference between Live Jasmin spam and a legitimate, requested pop-up because both are run from the click event handler.
The only solution is to disable popups entirely, which will cause compatibility issues. This is why we can't have nice things.
We're not talking about IE6, and this isn't 2003. It's time to update your prejudices. IE9 is a decent standards-conforming browser. It's not all that exciting, but it's not awful, and I can understand why people are perfectly content with it.
When teaching students how to program (which is entirely different from teaching them computer science), you should begin with the most fundamental concepts: talk about raw memory and opcodes. Discuss briefly how these instructions are actually interpreted and implemented (how a half-adder works is fascinating, even if most people never have to build one in real life).
Once your students understand how to make computers do basic things with raw instructions, teach them jumps, conditionals, loops, and even subroutines. After that, introduce higher-level languages and compilers, and demonstrate that the compiler merely automates what your students have already been doing. From there, teach progressively higher-level constructs, including second-order function references, data structures, and so on. Object-orientation falls out naturally once you get to structures and function pointers.
If you follow this approach, your students will have an understanding of the entire abstraction hierarchy, which is not only of immensely practical value, but also underscores the principle that nothing in this field is "magical". You can always pierce an abstraction, and even more importantly, erect new abstractions where appropriate. The most common flaw I find in programmers is the inability or unwillingness to build new abstractions. The only way we make progress in this field is by the old reductionist approach of breaking a hard problem into smaller parts and attacking each individually. When you teach your students how to do that by demonstrating the power of abstraction, you make them better programmers.
Programmers shown UML, Java class graphs, and so on right away become too familiar with that level of abstraction. They think of lower levels as some kind of magic and don't realize they can and should build their own levels on top of what they're given. The result is often incoherent, rambling, brittle, and ugly code. Don't let that happen.
That's a good point, and one I hadn't considered. Does anyone more knowledgeable know whether that's possible? If you use a conventional hash, you can simply increase the number of iterations as high as you want.
Salting addresses some attacks, but as CPU time becomes cheaper, it becomes increasingly feasible to brute-force even salted hashes. To address this issue, you need key strengthening as well.
Or, better yet, just use the system designed to store passwords: bcrypt.
*sigh* Then again, I'm confident that we'll see incompetent web application developers using unsalted MD5 for decades to come. People don't learn from others' mistakes it seems.
Pointer typedefs were a bad idea in the 1980s. They're just terrible today. One pet peeve of mine is this: typedef struct _FOO { int Blah; } FOO, *PFOO;
void SomeFunction(const PFOO);
That const doesn't do what you think it does. There was never a good reason to use pointer typedefs. There is certainly no good reason to do so today. Just say no. If your coding convention disagrees, damn the coding convention.
This is exactly the crap that Microsoft's genuine advantage is designed to stop. Small-scale personal piracy is one thing, but I fully support efforts to squash unctuous commercial enterprises like this one.
I can't believe that this point hasn't been made yet: the reason for leaving torrent out of autocomplete is to prevent the knowledge of torrents from spreading. They don't want normal, non-geeky people seeing the word "torrent" next to their favorite song, clicking it out of curiosity, then learning about the wonderful world of file-sharing.
This isn't about existing torrent users. It's about slowing down the creation of new ones.
True enough, but encryption works as a digital signature
No, it doesn't. Encryption without authentication is always subject to terrible attacks. Always include an authenticator in an encrypted message. An attacker not being able to decrypt a message is no barrier to his being able to manipulate it for profit.
Slashdot Mirror
The Chinese room experiment does not demonstrate what you think it does. Serle's argument is (being generous) circular. If the set of rules the man implements is sufficiently nuanced, then the entire *system* is intelligent. There is nothing distinct about a "mind" aside from a set of rules for symbolic information processing.
I particularly liked this reply, from the article:
The International Commission on Non-Ionizing Radiation Protection published a document containing all the information one could possibly want.. Read the conclusions section if nothing else.
Bullshit. The brain is a computer. Sure, it's a strange architecture: it's made of billions of impressively energy-efficient gates each operating at the order of tens of hertz. Fan-out is huge --- a gate on a microprocessor might be connected to 50 others, but a neuron can have tens of thousands of connections. A CPU has one fast, global clock, while the brain has overlapping and distributed clock signals for synchronizing neuron firing. The short term memory system uses the equivalent of old-fashioned delay lines, while long-term storage is implemented with redundant, distributed rewiring. It's content-addressable and has a storage capacity in the terabyte range, though it has really lousy indexing. Input and output are essentially memory-mapped, with lots of special purpose hardware acceleration.
There are a lot of similarities too: both our computers and our brains run software, with only a few basic features baked into the hardware. Both parse raw environmental input and parse it into abstractions that can be manipulated symbolically according to software-defined rules. Both can evaluate the lambda calculus and run a universal Turing machine. Neither can solve the halting problem in all cases. Both have large data stores. Both have networked inputs. Both crash. Both employ algorithms and data structures to process information. Both eventually fall apart.
Our brains are not magical devices somehow above scientific inquiry. They are ordinary, pedestrian objects in that obey the same laws of physics that govern baseballs and light switches. That we don't completely understand all the brain's mechanisms is no reason to believe it's qualitatively different from any other computer. Have you read every line of code in the web browser you're staring at?
SCIENCE DOES NOT WORK THAT WAY.
Any demand to be "100% sure" that an effect does not exist is unreasonable nonsense. Nobody can do that. All we can do is make observations, derive mathematical patterns from them (we call these "theories") and make predictions based on these generalizations with the aim of finding conflicts between the theory and the observation. If we find any, the theory is wrong and we begin again. If we don't find any contradictions, we provisionally accept the theory. As the theory continues to accurately predict reality, our confidence in it increases. With respect to a long-established theory like that of gravity or evolution, it's overwhelmingly likely that any discrepancy between the theory and observation is due to an error in the latter, and one would need extraordinary evidence to overturn such a theory, e.g. the famous hypothetical "fossil rabbit in the precambrian".
Now, with respect to electromagnetic fields, we have very specific and accurate theories to describe their behavior, effects, and interactions, and these theories have lasted over a century. The neurons that make up our brains and that form our consciousness are not special, and like all matter, they are also subject to these same theories. Neurons create electrical voltages by changing the concentration of ions inside themselves relative to their environment, and this mechanism is well-understood: the mysteries of the brain are in the emergent phenomena. The chemistry is pedestrian.
Now, when we plug the numbers for consumer electronics and neurons into electromagnetic theory, we see that there is simply no effect. The radiation is too weak to disrupt the bonds that join molecule in DNA and proteins, and at the intensity used in mobile phones, the heating effect is weaker than that of a pillow at night. The theory predicts that nothing should happen.
Now, just to be sure, various researchers have looked for an effect anyway, and have overwhelmingly failed to find a link between normal levels of mobile radiation and cancer. The few spurious positive results can be attributed to bad experimental design (e.g., uncontrolled and self-selected survey responses) or simple publication bias (if you perform a hundred studies at a 95% confidence interval, five of them will show spuriously positive results!).
Combined, the robustness of electromagnetic theory, our understanding of the chemistry of the cell, and the failure to find conclusive causal evidence paint as certain a picture as one can paint using the canvas of science. The theory says that mobile phones shouldn't cause cancer. When we look for cancer, we find nothing. We have no plausible explanation for how they could cause cancer. The only reasonable conclusion to draw is that absent further and significant evidence, cell phones do not cause cancer.
No, it's not 100% certain. It's also not 100% certain that there's no luminfiferous aether, and we can't be 100% sure that rotting fruit doesn't turn into insects by itself. If you want to believe in a cancer risk anyway, you're no better than someone who believes in hexes, astrology, or homeopathy.
Dammit, stop modding my comment up and mod this one up instead.
All right --- Sony appears to be guilty of perjury after filing a takedown notice for someone else's work.
Who is going to do something about it? Selective enforcement is wonderful, isn't it? If Sony succeeds in this, it'll embolden others to file takedown notices against anything they dislike for any reason whatsoever.
Oh, for fuck's sake, have you seen LWN's "security" page? Every week, there's some remote code execution vulnerability or another. At least distributions regularly push updates --- Apple usually waits for its next minor release. I'm sick and tired of this puerile and reflexive Microsoft-bashing.
I use one everywhere. The Microsoft mouse is to pointing devices what the Model M is to keyboards. It's amazing.
The Microsoft of today is nothing like what it was 10, 15 years ago when it became fashionable to hate. If MS is an evil empire today, it's the empire of Brezhnev, not of Stalin. It's generally pretty reasonable and a decent citizen of our software community. It's perfectly legitimate to expect consistency from people, but companies are composed of people, and to a large extent, the people at Microsoft are different these days.
Full disclosure: I work for Microsoft. I hack on Windows.
It's not that internationalization is the work of the devil, but rather that it should happen at a higher level than an individual PDF. Allowing different content to be displayed in different language environments raises serious questions about document integrity: imagine a international contract PDF that displayed one payment for German users and another for French ones.
PDF-the-document-format is a good thing in that it allows perfect reproduction of a printed document anywhere. PDF-the-generic-container, on the other hand, is both frightening and of dubious utility, but I can see why Adobe might have a business case for trying to drive this approach anyway. This is why we can't have nice things.
The laws of physics aren't any different for planes made in China. Why would they be cheaper to operate?
No, at least Mozilla blocks Flash popups too. The issue is that these "popups" are created in response to user clicks, and the browser can't tell the difference between Live Jasmin spam and a legitimate, requested pop-up because both are run from the click event handler.
The only solution is to disable popups entirely, which will cause compatibility issues. This is why we can't have nice things.
Modern Internet Explorer:
We're not talking about IE6, and this isn't 2003. It's time to update your prejudices. IE9 is a decent standards-conforming browser. It's not all that exciting, but it's not awful, and I can understand why people are perfectly content with it.
When teaching students how to program (which is entirely different from teaching them computer science), you should begin with the most fundamental concepts: talk about raw memory and opcodes. Discuss briefly how these instructions are actually interpreted and implemented (how a half-adder works is fascinating, even if most people never have to build one in real life).
Once your students understand how to make computers do basic things with raw instructions, teach them jumps, conditionals, loops, and even subroutines. After that, introduce higher-level languages and compilers, and demonstrate that the compiler merely automates what your students have already been doing. From there, teach progressively higher-level constructs, including second-order function references, data structures, and so on. Object-orientation falls out naturally once you get to structures and function pointers.
If you follow this approach, your students will have an understanding of the entire abstraction hierarchy, which is not only of immensely practical value, but also underscores the principle that nothing in this field is "magical". You can always pierce an abstraction, and even more importantly, erect new abstractions where appropriate. The most common flaw I find in programmers is the inability or unwillingness to build new abstractions. The only way we make progress in this field is by the old reductionist approach of breaking a hard problem into smaller parts and attacking each individually. When you teach your students how to do that by demonstrating the power of abstraction, you make them better programmers.
Programmers shown UML, Java class graphs, and so on right away become too familiar with that level of abstraction. They think of lower levels as some kind of magic and don't realize they can and should build their own levels on top of what they're given. The result is often incoherent, rambling, brittle, and ugly code. Don't let that happen.
So much so that he managed to jump the line for a liver transplant.
That's a good point, and one I hadn't considered. Does anyone more knowledgeable know whether that's possible? If you use a conventional hash, you can simply increase the number of iterations as high as you want.
Salting addresses some attacks, but as CPU time becomes cheaper, it becomes increasingly feasible to brute-force even salted hashes. To address this issue, you need key strengthening as well.
Or, better yet, just use the system designed to store passwords: bcrypt.
*sigh* Then again, I'm confident that we'll see incompetent web application developers using unsalted MD5 for decades to come. People don't learn from others' mistakes it seems.
California only had power problems after deregulating its utilities.
And how, pray tell, do you differentiate legitimate users from malicious ones? Karma works because a trusted third party maintains it.
Ah, hysterical raisins. I never wrote 16-bit code; that's a good explanation now that you mention it. Thanks.
Ever see Windows code?
Pointer typedefs were a bad idea in the 1980s. They're just terrible today. One pet peeve of mine is this:
typedef struct _FOO { int Blah; } FOO, *PFOO;
void
SomeFunction(const PFOO);
That const doesn't do what you think it does. There was never a good reason to use pointer typedefs. There is certainly no good reason to do so today. Just say no. If your coding convention disagrees, damn the coding convention.
This is exactly the crap that Microsoft's genuine advantage is designed to stop. Small-scale personal piracy is one thing, but I fully support efforts to squash unctuous commercial enterprises like this one.
I can't believe that this point hasn't been made yet: the reason for leaving torrent out of autocomplete is to prevent the knowledge of torrents from spreading. They don't want normal, non-geeky people seeing the word "torrent" next to their favorite song, clicking it out of curiosity, then learning about the wonderful world of file-sharing.
This isn't about existing torrent users. It's about slowing down the creation of new ones.
No, it doesn't. Encryption without authentication is always subject to terrible attacks. Always include an authenticator in an encrypted message. An attacker not being able to decrypt a message is no barrier to his being able to manipulate it for profit.