The definition of metadata isn't "useful information about other data" or "data about other data we've decided to store" or whatever. It's just "data about other data". Yeah, frequency histograms are usually pretty useless as metadata, but they are still metadata. I just wouldn't have the system track it.
Another definition for metadata that is insufficiently conclusive is "data outside the fiel describing data inside." What you decide to call a "file" is really a question of convention. They are not necessarily contiguous on disks, for example, and are generally the logical representation the system offers to an application. If the system knows which bytes to offer, it's already read the file size somehow. In mainframe environments they don't even talk about "files" most of the time, generally preferring different ways of representing datasets. So, as you were mentioning the different ways of "representing" sets, you were showing that you actually agree with me but you keep stumbling over the PC baggage of the "file" term. Perhaps it was a bad choice to include it in the conceptual part of the article, since it tends to lock people into a way of thinking.
As for the AFPDS thing - I really don't remember what the hell I was trying to show there. Doing too much crack, I guess. Maybe I'll figure it out later. In the meantime, AFPDS is a flavor of mainframe printstream. I spend a good portion of my day slogging through data and metadata at the byte level, troubleshooting printstream transforms. Mainframe print environments have impossibly convoluted metadata structures.
The reason why I'm being such a jerk about this is that John is being so cool and really quite nice to all you self-impressed punks. Since he's not willing to give you the what-for, someone should write in service of the scales of justice. I know I'm not the only person on/. tired of people imagining themselves such experts that they can talk about "fallacies in fundamentals" and lay down the law. ArsTechnica is generally scrupulously accurate and it's the height of pomposity to be like that when you don't actually know what you're talking about. It's damned disrespectful of soemone who's trying to share his knowledge with you FOR FREE!.
Unless a string of data is floating about in a vacuum, with nothing else to be read, there is nothing to distinguish the bytes in that string of data from bytes in OTHER strings of data. If you add an EOF marker so the system knows where to stop, that's additional information. If you add DCB information with the requisite BDWs and RDWs like on an MVS system, then no EOFs are necessary because the systems counts until it has as many bytes in memory as it was told were there to be had and stops.
SO, file size is most definitely metadata. There's no way of knowing what is in a set of data unless you have additional information which is not the data itself.
Even is this was not true(say, in the data-in-vacuum example), the fact is that META data is "data about data". As others have said, just because the information is intrinsic to the set doesn't mean it is the same as the explicit data itself. I can tell by browsing the hex of an AFPDS file that it's an AFPDS file, so I could write that information somewhere else close by, and it would be metadata, even though it was derived. The simple fact that it is an AFPDS file does nothing within the context of an AFPDS file. By the time it's being processed AS an AFPDS file, the fact that it's an AFPDS file is pretty irrelevant and changes processing not a bit. If it was really a saved character from Everquest that happened to also function as a valid AFPDS file, nothing would change about the result.
As long as you think a file preamble/header/footer/whatever is NOT metadata because it's "in the file", then you're never going to understand what metadata really is. Do you see?
I admit to being an asshole. But that doesn't mean one don't deserve to be pilloried when one tries to pretend one knows one's shit better than the guy writing the article - and fails.
And btw, if you don't know my (rather obscure)acronyms, just ignore them. Their particular meaning is not important to the argument.
AC's dead wrong and it's just silly. I mean, thinking about file size for a second... if the system doesn't know how long that "ordered set of bytes" is, how does it know when to stop reading? You could have a file end marker, of course, but there's more metadata for you.
It just goes to show that the loudly incompetent are too incompetent to have inklings as to their own incompetence.
Since I submit that hardware humans will be behaviorally identical, that's half of your grounds for assumption right there. If the computer simulates the behavior of your physiology (neurons and neural bundles), then that's worth at least a half point as well.
...that we don't know if other bio-humans are conscious, then yes, we won't know if hardware humans are conscious. But then, that's just a position of general skepticism, not an idictment of AI. It (the position) has its champions in professional philosophy, but not very many, and is just about dismissed by most of the feild.
Unless you want to talk about philosophical theology. But, as Ronald De Sousa once said, philosophical theology is like "intellectual tennis without a net".
The "no magic to consciousness" part is is supported by Occam's razor, since there are alternate explanations that require fewer leaps of faith than "magic".
The "nothing about the structure of the brain that's not duplicable" springs from the fact that neuron behavior is pretty simple in the broad strokes - the parts we don't understand are more related to dynamic interactions - which are also straightforward to implement in software.
The 10 teraflop number comes from the number of active neurons in the brain, and their frequency of firing. The 10 teraflops number is an estimate based on how many floating point calculations I think it would take to simulate a neuron with sufficient fidelity.
If you have objections to what I said, I'll address them specifically. There's no "specific" evidence for any of these things, because the claims I'm making are drawn from more than one fact. If you're looking for others who make the similar claims in print, I'm sure you'll find them with little trouble. Steven Pinker, Daniel Dennet, Francis Crick, Douglas Hofstadter, and Rodney Brooks are good folks with whom to start.
Neural nets solve the problem of consciousness like microchips solve the problem of personal computing. You can't just throw a bunch of chips on a board, run electricity through them, and viola! A PC! They have to be chosen for their tasks, the data-paths arranged, and so on. The brain is the same way, but about a thousand times as complex.
I don't mean to say that when we do get AI up and running that we'll never be able to get it to work better than the wetware mother nature gave us. I was just trying to explain why, even though even a 286 8MHz proc has do math a million times faster than we can, it can't do what are brains do very fast at all. And as I mention below, doing what our brains do slowly is worthless.
There's no such thing as "powerful AI systems that are really slow". An agent like a human or an AI interacts with the world in real-time, learns from it in real time.
An AI without a world to live in and learn from would of course be catatonic. An AI too slow to build a useful internal representation of its current situation before the situation changes (thus making the representation worthless) is going to be either catatonic or a moron.
Also, as I mentioned, the human brain is a collection of gadgets, implemented in a big web of neural processing with very complex informational topology. We may be a while yet reverse engineering some of the most clever ones. Maybe it will be much more than the 20 years I give. Maybe it won't be that long. It depends on how important the fine details in are and how easy it is to come up with functionally equivalent "gadgets" that work as well as the brain's more difficult-to-copy architectures.
The important thing here is that we do indeed have some idea how to do Strong AI. To tinker with those ideas until we build something that really does seem a little more intelligent in that wacky "emergent" sort of way, we need some faster hardware. Single-application heuristic gadgets like language translation* are forever going to be bad unless they are embedded in a larger system that can give realtime feedback.
The reason game AIs get better is because the game programmers have more room to work in terms of both time(MHz) and memory. Also, as with ant-colonies, a collection of not-very-bright creatures can make for some pretty intelligent communities.
In the end, though, I like your intuition that agents need to have a life to have something to say. Exactly, I say. Exactly.
In conclusion, yes, it IS a hardware problem, among other things. Without proper hardware we can't expect any actors to have much of a life. And even if we did have the proper hardware tomorrow, we'd still have years of tinkering ahead before we put all the pieces together in a way that works. But right now, even our tinkering is somewhat hobbled by lack of hardware.
* Language translation is probably not a good example because it does not seem likely that any such thing exists in our heads. It's something we're trying to build because we WISH it existed in our heads.
It depends on in what way you mean to challenge me. I can easily back up what I said if you mean to suggest that some homunculus-like structure could just as easily be responsible for consciousness. If you're simply expressing skepticism about my apparent functionalism regarding consciousness, then I will simply refer you to Daniel Dennet, since I couldn't really do that topic any justice here.
However, if you bridled because "emergent" implies a certain "accidental" quality, then I think you have a good point. Consciousness is not an "unintended" side-effect of brain activity. I didn't mean to imply that. I definitely agree that the brain is designed to support consciousness and that "consciousness" and "intelligence" are, while not isomorphic in their ambit, neither are they separable.
When you have emergent phenomena like consciousness, replicating it is a good way of making sure you haven't left out anything important. It means that anything you've changed from the original wasn't essential. It's a way to confirm or disconfirm guesses. Of course, the ungly truth of this is that we're almost sure to make a number of catatonic, retarded, or psychotic AIs before we make our first happy, well-adjusted hardware human. On the other hand, we are sure to learn from our failures.
Even when we do have a working person, even if we had to simulate each individual neuron, at least now we'll be able to look at what's going on in the brain with more detail than ever before - we can track the contents of memory (as in, neural firing dispositions) without sticking a scalple in someone's head.
I should be clear - the familiar "I am a man. All men have Y chromosomes. Therefore, I have a Y chromosome" type of logic is of course uninterestingly tractible. The logic represented by the topology of some of our neural gadgets is not so trivial. Their linkages to one-another are less-than-trivial as well. Still, those are just engineering problems.
However, we won't really be able to abstract the high-level from the low-level (between which there is no real dividing line) unless we understand the topology well enough to know which dynamics can be idealized (or formalized, if you will) and which depend more directly on their fuzzy nature.
I agree totally. Implimenting individuals neurons in code is not likely to be necessary. But I was getting so wrapped up in trying to write a pretty paragraph (in a hurry) that I went overboard. I just hoped no one would call me out on it so I didn't have to explain myself.
And yeah, Minsky isn't exactly the leading edge any more (I don't ever cite him). I probably shouldn't really have defended him so vigorously - the neural net comments just brought me out.
...that one doesn't understand 100%. The brain is a design we know works. It's been stress tested for a million years. Maybe we should just go with that for now and work out the less tractible problems when we can. Because having hardware humans who don't have to die (and are thus concerned about the future), who don't have to eat (and thus use fewer natural resources), and yet can be friends (and more, one assumes) with us bio-humans might be really nice.
Neural nets, on their own, are not very smart in many ways. Sure, there are some problems it seems that only they can solve. But complex, multi-stage problems generally baffle them nearly indefinitely.
The brain is not a big neural amalgam that gets to some critical mass and then suddenly starts doing stuff. It's wired. It's got gadgets. It's really a big collection of them. Some of them are damned complex, composed of sheets of neurons talking to each other in intricate, bewildering arrays.
And modern Connectionists understand that. Certainly the symbolic logic guys were wrong as wrong as those who thought neural nets would solve everything. But that's people like Fodor and Chomsky. The Minksy "Agent" model is very much on the "Connectionist" side of the map. That's not to say that I agree with everything he says, but I think you're unfairly blaming him for the mistakes of others.
Symbolic logic, by itself is no panacea, but neither is the neural net. I'm willing to bet that a lot of the interactions of various neural nets in the brain form very formal symbolic logic gadgets. Also, in the end, it is the formal logic of virtual-neuron microcode in a computer is what will generate Strong AI.
...for game programming, but I'm always happy to have people read Minsky because he tends to crack peoples' preconceptions about what is "obvious" about consciousness and AI and etc. Even better might be Daniel Dennet, author of "Consciousness Explained". Less philosophically sound (and ultimately less satisfying) but still very interesting is Steven Pinker.
I think it's very important to understand that there's no magic to consciousness. It's not something shrouded in mystery about which we know nothing. In fact, we know an amazing amount about individual areas. The topology of the information processing membranes are more complex than we can sort out just yet, but there's nothing about the structure of the brain that's not duplicable by silicon hardware. We just have a lot more mapping to do.
Also important to notice is that to implement the human mind in hardware (as opposed to wetware), we'd need something on the order of a 10 teraflop supercomputer. We just don't have the hardware to pull that off yet. The AI-related optimism of yesteryear was fueled by the misconception that computers are faster than humans. What's really true is that the "programming" that underlies the various gadgets in the mind is the product of millions of years of specialization at small tasks. We have fantastic motor-control gadgets and unparalleled pattern-recognition wetware, for example. Figuring out exactly how many animals are in 15,342 groups of 967 animals each was never all that important, so we never evolved any gadgets to carry out high-speed arithmetic. On the other hand, we're good at seeing how things divide out and how games might be played to our advantage. Idiot savants have been known to find extremely large prime numbers as if by magic - probably the same hardware put to an exotic use.
So in 20 years (or so), we'll have the hardware, and maybe we'll have the information processing topology as well. Some intrepid researcher will put all that in a state-of-the-art cybernetic body. Then it'll be a matter of watching the first hardware human child grow up and meet the world.
PS- make some pretty bold claims here, and also cite a number or two that one might be expected to view with suspicion. I can back it up, just ask.
How long has your friend's kid sister been online? Since she was 9? Or since she was 13? If it was the latter, then small wonder she has yet to branch out. Also, does she have her own computer with her own bookmarks in the browser?
If she STILL does nothing but chat and play games on Yahoo, then I'll count that against what I said. I didn't mean, however, that the youngest of young immediately use everything that the web has to offer. They are just bound to expand, even though our parents might not.
As for entertainment - I think the web time will always be more entertainment-centric than research-centric, because people are, by and large, more interested in being entertained than in learning. Some things will never change.
And most web users have been on the web for 2-3 years. For most people to surf the web, it involves going to an aging, ill-maintained computer in the corner, sitting down in the uncomfortable spare chair they've allocated for the computer desk, waiting for the analog modem to connect to AOL and then clicking about haplessly for a while. Is it really any wonder that the vast majority of people out there aren't venturing beyond a few lowest common denominator sites plus whatever homepage to which Microsoft or AOL has set them to default.
These people don't read somethingawful.com every day. They don't even know what you mean when you say "webcomics". They read dead-tree news and complain about the text on their 15" monitor. They print EVERYTHING.
Given time, though, there's a new generation coming up for whom the web is a second home. Give them another 5 years and we'll see the balance of clicks changing. Of course, older people may not ever really change their habits, but those of us who use the web a lot are responsible for many more pageviews per capita. The only reason we might find this article surprising is because we have gotten used to everything moving at internet speed that we've forgotten that people still move at people speed.
that some genius NSA agents have somehow subverted compilers so that they magically realize that they are compiling security-related code and insert a back door into one's code. If you compile the code, I'm sure it's clean. no need for a "secure" compiler. In fact, I don't even know what that would be. All I was saying is that usually you can't see the code that makes the binary so you dont' know what's in it.
Slashdot Mirror
Here's a link from Harvard:l he .htm
http://www.ksg.harvard.edu/newsclips/001219_wil
It's not a "list", but somehow I think Harvard would know if he weren't actually in the '69 graduating class.
The definition of metadata isn't "useful information about other data" or "data about other data we've decided to store" or whatever. It's just "data about other data". Yeah, frequency histograms are usually pretty useless as metadata, but they are still metadata. I just wouldn't have the system track it.
Another definition for metadata that is insufficiently conclusive is "data outside the fiel describing data inside." What you decide to call a "file" is really a question of convention. They are not necessarily contiguous on disks, for example, and are generally the logical representation the system offers to an application. If the system knows which bytes to offer, it's already read the file size somehow. In mainframe environments they don't even talk about "files" most of the time, generally preferring different ways of representing datasets. So, as you were mentioning the different ways of "representing" sets, you were showing that you actually agree with me but you keep stumbling over the PC baggage of the "file" term. Perhaps it was a bad choice to include it in the conceptual part of the article, since it tends to lock people into a way of thinking.
As for the AFPDS thing - I really don't remember what the hell I was trying to show there. Doing too much crack, I guess. Maybe I'll figure it out later. In the meantime, AFPDS is a flavor of mainframe printstream. I spend a good portion of my day slogging through data and metadata at the byte level, troubleshooting printstream transforms. Mainframe print environments have impossibly convoluted metadata structures.
The reason why I'm being such a jerk about this is that John is being so cool and really quite nice to all you self-impressed punks. Since he's not willing to give you the what-for, someone should write in service of the scales of justice. I know I'm not the only person on /. tired of people imagining themselves such experts that they can talk about "fallacies in fundamentals" and lay down the law. ArsTechnica is generally scrupulously accurate and it's the height of pomposity to be like that when you don't actually know what you're talking about. It's damned disrespectful of soemone who's trying to share his knowledge with you FOR FREE!.
So be quiet.
Unless a string of data is floating about in a vacuum, with nothing else to be read, there is nothing to distinguish the bytes in that string of data from bytes in OTHER strings of data. If you add an EOF marker so the system knows where to stop, that's additional information. If you add DCB information with the requisite BDWs and RDWs like on an MVS system, then no EOFs are necessary because the systems counts until it has as many bytes in memory as it was told were there to be had and stops.
SO, file size is most definitely metadata. There's no way of knowing what is in a set of data unless you have additional information which is not the data itself.
Even is this was not true(say, in the data-in-vacuum example), the fact is that META data is "data about data". As others have said, just because the information is intrinsic to the set doesn't mean it is the same as the explicit data itself. I can tell by browsing the hex of an AFPDS file that it's an AFPDS file, so I could write that information somewhere else close by, and it would be metadata, even though it was derived. The simple fact that it is an AFPDS file does nothing within the context of an AFPDS file. By the time it's being processed AS an AFPDS file, the fact that it's an AFPDS file is pretty irrelevant and changes processing not a bit. If it was really a saved character from Everquest that happened to also function as a valid AFPDS file, nothing would change about the result.
As long as you think a file preamble/header/footer/whatever is NOT metadata because it's "in the file", then you're never going to understand what metadata really is. Do you see?
I admit to being an asshole. But that doesn't mean one don't deserve to be pilloried when one tries to pretend one knows one's shit better than the guy writing the article - and fails.
And btw, if you don't know my (rather obscure)acronyms, just ignore them. Their particular meaning is not important to the argument.
AC's dead wrong and it's just silly. I mean, thinking about file size for a second... if the system doesn't know how long that "ordered set of bytes" is, how does it know when to stop reading? You could have a file end marker, of course, but there's more metadata for you.
It just goes to show that the loudly incompetent are too incompetent to have inklings as to their own incompetence.
Since I submit that hardware humans will be behaviorally identical, that's half of your grounds for assumption right there. If the computer simulates the behavior of your physiology (neurons and neural bundles), then that's worth at least a half point as well.
...that we don't know if other bio-humans are conscious, then yes, we won't know if hardware humans are conscious. But then, that's just a position of general skepticism, not an idictment of AI. It (the position) has its champions in professional philosophy, but not very many, and is just about dismissed by most of the feild.
Unless you want to talk about philosophical theology. But, as Ronald De Sousa once said, philosophical theology is like "intellectual tennis without a net".
The "no magic to consciousness" part is is supported by Occam's razor, since there are alternate explanations that require fewer leaps of faith than "magic".
The "nothing about the structure of the brain that's not duplicable" springs from the fact that neuron behavior is pretty simple in the broad strokes - the parts we don't understand are more related to dynamic interactions - which are also straightforward to implement in software.
The 10 teraflop number comes from the number of active neurons in the brain, and their frequency of firing. The 10 teraflops number is an estimate based on how many floating point calculations I think it would take to simulate a neuron with sufficient fidelity.
If you have objections to what I said, I'll address them specifically. There's no "specific" evidence for any of these things, because the claims I'm making are drawn from more than one fact. If you're looking for others who make the similar claims in print, I'm sure you'll find them with little trouble. Steven Pinker, Daniel Dennet, Francis Crick, Douglas Hofstadter, and Rodney Brooks are good folks with whom to start.
Neural nets solve the problem of consciousness like microchips solve the problem of personal computing. You can't just throw a bunch of chips on a board, run electricity through them, and viola! A PC! They have to be chosen for their tasks, the data-paths arranged, and so on. The brain is the same way, but about a thousand times as complex.
make that, "...doing what our brains do, but slowly, is worthless."
I don't mean to say that when we do get AI up and running that we'll never be able to get it to work better than the wetware mother nature gave us. I was just trying to explain why, even though even a 286 8MHz proc has do math a million times faster than we can, it can't do what are brains do very fast at all. And as I mention below, doing what our brains do slowly is worthless.
There's no such thing as "powerful AI systems that are really slow". An agent like a human or an AI interacts with the world in real-time, learns from it in real time.
An AI without a world to live in and learn from would of course be catatonic. An AI too slow to build a useful internal representation of its current situation before the situation changes (thus making the representation worthless) is going to be either catatonic or a moron.
Also, as I mentioned, the human brain is a collection of gadgets, implemented in a big web of neural processing with very complex informational topology. We may be a while yet reverse engineering some of the most clever ones. Maybe it will be much more than the 20 years I give. Maybe it won't be that long. It depends on how important the fine details in are and how easy it is to come up with functionally equivalent "gadgets" that work as well as the brain's more difficult-to-copy architectures.
The important thing here is that we do indeed have some idea how to do Strong AI. To tinker with those ideas until we build something that really does seem a little more intelligent in that wacky "emergent" sort of way, we need some faster hardware. Single-application heuristic gadgets like language translation* are forever going to be bad unless they are embedded in a larger system that can give realtime feedback.
The reason game AIs get better is because the game programmers have more room to work in terms of both time(MHz) and memory. Also, as with ant-colonies, a collection of not-very-bright creatures can make for some pretty intelligent communities.
In the end, though, I like your intuition that agents need to have a life to have something to say. Exactly, I say. Exactly.
In conclusion, yes, it IS a hardware problem, among other things. Without proper hardware we can't expect any actors to have much of a life. And even if we did have the proper hardware tomorrow, we'd still have years of tinkering ahead before we put all the pieces together in a way that works. But right now, even our tinkering is somewhat hobbled by lack of hardware.
* Language translation is probably not a good example because it does not seem likely that any such thing exists in our heads. It's something we're trying to build because we WISH it existed in our heads.
I won't what?
It depends on in what way you mean to challenge me. I can easily back up what I said if you mean to suggest that some homunculus-like structure could just as easily be responsible for consciousness. If you're simply expressing skepticism about my apparent functionalism regarding consciousness, then I will simply refer you to Daniel Dennet, since I couldn't really do that topic any justice here.
However, if you bridled because "emergent" implies a certain "accidental" quality, then I think you have a good point. Consciousness is not an "unintended" side-effect of brain activity. I didn't mean to imply that. I definitely agree that the brain is designed to support consciousness and that "consciousness" and "intelligence" are, while not isomorphic in their ambit, neither are they separable.
How do you know your parents are conscious? Your best friend?
I expect the same reasons will apply to Strong AI.
When you have emergent phenomena like consciousness, replicating it is a good way of making sure you haven't left out anything important. It means that anything you've changed from the original wasn't essential. It's a way to confirm or disconfirm guesses. Of course, the ungly truth of this is that we're almost sure to make a number of catatonic, retarded, or psychotic AIs before we make our first happy, well-adjusted hardware human. On the other hand, we are sure to learn from our failures.
Even when we do have a working person, even if we had to simulate each individual neuron, at least now we'll be able to look at what's going on in the brain with more detail than ever before - we can track the contents of memory (as in, neural firing dispositions) without sticking a scalple in someone's head.
I should be clear - the familiar "I am a man. All men have Y chromosomes. Therefore, I have a Y chromosome" type of logic is of course uninterestingly tractible. The logic represented by the topology of some of our neural gadgets is not so trivial. Their linkages to one-another are less-than-trivial as well. Still, those are just engineering problems.
However, we won't really be able to abstract the high-level from the low-level (between which there is no real dividing line) unless we understand the topology well enough to know which dynamics can be idealized (or formalized, if you will) and which depend more directly on their fuzzy nature.
I agree totally. Implimenting individuals neurons in code is not likely to be necessary. But I was getting so wrapped up in trying to write a pretty paragraph (in a hurry) that I went overboard. I just hoped no one would call me out on it so I didn't have to explain myself.
And yeah, Minsky isn't exactly the leading edge any more (I don't ever cite him). I probably shouldn't really have defended him so vigorously - the neural net comments just brought me out.
...that one doesn't understand 100%. The brain is a design we know works. It's been stress tested for a million years. Maybe we should just go with that for now and work out the less tractible problems when we can. Because having hardware humans who don't have to die (and are thus concerned about the future), who don't have to eat (and thus use fewer natural resources), and yet can be friends (and more, one assumes) with us bio-humans might be really nice.
Neural nets, on their own, are not very smart in many ways. Sure, there are some problems it seems that only they can solve. But complex, multi-stage problems generally baffle them nearly indefinitely.
The brain is not a big neural amalgam that gets to some critical mass and then suddenly starts doing stuff. It's wired. It's got gadgets. It's really a big collection of them. Some of them are damned complex, composed of sheets of neurons talking to each other in intricate, bewildering arrays.
And modern Connectionists understand that. Certainly the symbolic logic guys were wrong as wrong as those who thought neural nets would solve everything. But that's people like Fodor and Chomsky. The Minksy "Agent" model is very much on the "Connectionist" side of the map. That's not to say that I agree with everything he says, but I think you're unfairly blaming him for the mistakes of others.
Symbolic logic, by itself is no panacea, but neither is the neural net. I'm willing to bet that a lot of the interactions of various neural nets in the brain form very formal symbolic logic gadgets. Also, in the end, it is the formal logic of virtual-neuron microcode in a computer is what will generate Strong AI.
...for game programming, but I'm always happy to have people read Minsky because he tends to crack peoples' preconceptions about what is "obvious" about consciousness and AI and etc. Even better might be Daniel Dennet, author of "Consciousness Explained". Less philosophically sound (and ultimately less satisfying) but still very interesting is Steven Pinker.
I think it's very important to understand that there's no magic to consciousness. It's not something shrouded in mystery about which we know nothing. In fact, we know an amazing amount about individual areas. The topology of the information processing membranes are more complex than we can sort out just yet, but there's nothing about the structure of the brain that's not duplicable by silicon hardware. We just have a lot more mapping to do.
Also important to notice is that to implement the human mind in hardware (as opposed to wetware), we'd need something on the order of a 10 teraflop supercomputer. We just don't have the hardware to pull that off yet. The AI-related optimism of yesteryear was fueled by the misconception that computers are faster than humans. What's really true is that the "programming" that underlies the various gadgets in the mind is the product of millions of years of specialization at small tasks. We have fantastic motor-control gadgets and unparalleled pattern-recognition wetware, for example. Figuring out exactly how many animals are in 15,342 groups of 967 animals each was never all that important, so we never evolved any gadgets to carry out high-speed arithmetic. On the other hand, we're good at seeing how things divide out and how games might be played to our advantage. Idiot savants have been known to find extremely large prime numbers as if by magic - probably the same hardware put to an exotic use.
So in 20 years (or so), we'll have the hardware, and maybe we'll have the information processing topology as well. Some intrepid researcher will put all that in a state-of-the-art cybernetic body. Then it'll be a matter of watching the first hardware human child grow up and meet the world.
PS- make some pretty bold claims here, and also cite a number or two that one might be expected to view with suspicion. I can back it up, just ask.
Nato
How long has your friend's kid sister been online? Since she was 9? Or since she was 13? If it was the latter, then small wonder she has yet to branch out. Also, does she have her own computer with her own bookmarks in the browser?
If she STILL does nothing but chat and play games on Yahoo, then I'll count that against what I said. I didn't mean, however, that the youngest of young immediately use everything that the web has to offer. They are just bound to expand, even though our parents might not.
As for entertainment - I think the web time will always be more entertainment-centric than research-centric, because people are, by and large, more interested in being entertained than in learning. Some things will never change.
And most web users have been on the web for 2-3 years. For most people to surf the web, it involves going to an aging, ill-maintained computer in the corner, sitting down in the uncomfortable spare chair they've allocated for the computer desk, waiting for the analog modem to connect to AOL and then clicking about haplessly for a while. Is it really any wonder that the vast majority of people out there aren't venturing beyond a few lowest common denominator sites plus whatever homepage to which Microsoft or AOL has set them to default.
These people don't read somethingawful.com every day. They don't even know what you mean when you say "webcomics". They read dead-tree news and complain about the text on their 15" monitor. They print EVERYTHING.
Given time, though, there's a new generation coming up for whom the web is a second home. Give them another 5 years and we'll see the balance of clicks changing. Of course, older people may not ever really change their habits, but those of us who use the web a lot are responsible for many more pageviews per capita. The only reason we might find this article surprising is because we have gotten used to everything moving at internet speed that we've forgotten that people still move at people speed.
that some genius NSA agents have somehow subverted compilers so that they magically realize that they are compiling security-related code and insert a back door into one's code. If you compile the code, I'm sure it's clean. no need for a "secure" compiler. In fact, I don't even know what that would be. All I was saying is that usually you can't see the code that makes the binary so you dont' know what's in it.
It's open code, so any back doors are exposed. You can hide that sort of thing in binary, but not in source.