If you don't include yourself in your mental model you would not be able to identify potential risks, as having no self-awareness means not being aware of risks to self. That makes for a very quick "natural de-selection."
I'm not saying it's not important or useful to have a model of yourself. Clearly it is. (Although your risk example isn't complex enough to show that. An amoeba could have a simple rule that says "If you detect a [member of 'dangerous thing' list] then move away", no nontrivial self-modeling necessary. Where having a model of the self becomes more useful is when you want to take into account your own effects on the world, so you have to calculate what those effects are likely to be.)
But that's beside the point. The point is that having a model of yourself doesn't have to be related to consciousness in any way, and probably isn't. It's easy to give a robot a model of its environment, and equally easy to include a representation of the robot itself in the model. But the robot isn't conscious or aware in the usual sense. Modeling yourself is not noticeably different, consciousness-wise, than modeling somebody else and their effects on the world.
Then at some point your "map" has grown so complex that it is only meaningful when you include yourself, your needs and preferences into its model equations. And then you have... consciousness, or a functional equivalent thereof.
Well... it sounds nice, but only if you don't think about it too hard.
The thing is, why should including a model of yourself in the map be so special? It's just another model. You could even argue that it ought to be easier, because the mind has more reliable information about you than about most things it has to model (say, other people).
Consciousness is about awareness, period, not just awareness of the self.
Douglas Hofstadter's book "Gödel, Escher, Bach - an Eternal Golden Braid" has some HEAVY examination of (human and machine) consciousness. My favorite metaphor he uses for consciousness is an ant hill. The ant hill has many layers of emergent, recursive properties.
Careful with this, though. While this is an excellent book, its treatment of mind and consciousness is often a little dubious. (Also, the anthill is more a metaphor for how complexity can arise from simple components rather than consciousness.)
The main point I'd bring up is that there's no particular reason to think that consciousness is related to recursion in any interesting way. It's actually odd how many philosophers define consciousness as "knowing about oneself", as the original poster did. Really, if I know about (have a mental representation of?) my desk, or if I know about you, knowing about me doesn't seem very different.
A better idea about consciousness is one that grows out of what gwernol is talking about: Consciousness is what lets you focus on something, the thing that looks at all the stuff going on in your brain and chooses one item to work on specifically.
The thing about Cyc is, it's all done by a private company, and they don't really release much of anything about their work. So there's no way to know just how much progress they're making, what their strengths and weaknesses are. Which is a shame, because I think they have a promising idea that no one else currently has the resources to execute in the same way: enter into the system, more or less by hand, some ten million or so common-sense facts about the world, together with algorithms for making uncertain, defeasible deductions from these.
(Maybe the closest alternatives are systems that try and automatically learn facts from text that's been annotated somehow for the purpose by people. I don't know of any particularly good ones, though.)
This reminded me of Anna Wierzbicka too. I read a paper of hers once. (Oh, look, I actually have it right here because I'm a big dork.) The whole project of reducing language to a few semantic primitives seemed... sort of mildly interesting in an intellectual-challenge kind of way, but overall completely pointless. It's not like brains actually define words in terms of primitives like this; if they did we'd see big processing-time differences for primitives vs. non. So it's just a weird little exercise.
(Also, aren't there words for family relations in there somewhere? Hmm, maybe not.)
(Also also, the paper has three pages of references, 1.5 of which are to herself. I think the rest disagree with her.)
"A computer would never be able to compete with competent chess player, and could certainly never compete with a Grandmaster. Chess requires true human intelligence."
Douglas Hofstadter put forth basically this point of view in his Pulitzer winner "Godel, Escher, Bach" in 1979. He predicted that computers wouldn't be able to beat grandmasters until they'd achieved human-level intelligence in general.
There was a competition for children to design a bit of sculpture for the cathedral. Personally, I think this choice is great: I suspect it's high and out-of-the-way enough, and similar enough to other random gargoyles, that you'll miss it unless you look carefully and think, "Hey, wait a minute..."
I also enjoyed their rationale. Heh. It's even sort of true, if you replace "evil" with "mythical monsters".
"Centuries ago, on our cathedrals, grotesques were intended to symbolize the evil that existed outside the church. Today, Darth Vader is an excellent example of evil in our times."
A very tough problem, involving considerable trial and error and pretty much insolvable in interview-time; I'm guessing the reports that this particular problem was used are probably apocryphal.
The statement of the problem: Given 12 objects, exactly one of which is a different weight than the others, determine a) which object is different and b) whether it's lighter or heavier than the rest. Use only 3 weighings on a balance scale.
It can be solved, but the solution boils down to a bunch of if-then directives.
------------
Weigh 4 balls vs 4 balls.
Equal:
One of the remaining 4 is heavier or ligher.
Weigh 3 "good" balls against 3 of those 4 balls.
Equal:
Weigh the remaining ball against a "good" ball
to see if it's heavier or lighter. []
Different:
Observe whether the three balls were heavier or lighter
than the good balls. If heavier, weigh 1 of the 3
against another of the 3. If these are the same, the
last ball is different and heavier. If they're
different, the heavier ball is the odd one out.
Similarly if the three balls are lighter than the
good ones. []
Unequal:
Take (A) 3 balls from the heavy side and 1 from the light side, and weigh against (B) 3 "good" balls and 1 from the heavy side.
If they're equal, we know one of 3 balls from the light side is lighter than the others. Weigh 2 against each other. []
If side A is heavier, we know one of the heavy-side balls on side A is heavier than the others. Weigh 2 against each other. []
If side B is heavier, we can limit our search to the light-side ball on side A and the heavy-side ball on side B. Weigh one of these against a "good" ball. []
I think there might also be one variation on part of this solution (maybe on the second weighing, if the first turs out unequal).
No... that just means Hotmail receives a lot of spam. So many people use it that a reasonable proportion of possible usernames are taken, and that means spammers can and do use "dictionary" attacks, where they send e-mail to random usernames and then just hang onto the addresses that don't bounce.
I believe that big providers like Hotmail and Yahoo try reasonably hard to prevent people from sending spam from their accounts, as it uses up bandwidth and creates ill will, so they do things like limit number of recipients per message, or recipients per day, that sort of thing. (Can anyone confirm that?)
But a spammer can make their e-mails appear to come from whatever address they want, and if there's a URL in the message they don't need to worry about whether people can reply.
Well, you don't need Chomsky to see that a simple lookup table won't work; sentences can be arbitrarily long, so you can always make a sentence longer than the longest sentence your lookup table knows. Searle only talks about a "set of rules", not a lookup table, so his room really could be a Turing machine.
Then he says that rule application can never produce understanding, and tries to make this claim sound obviously correct. (I think it's not, if only because everything above the quantum level, presumably including the brain, seems to be deterministic and in principle describable by rules.)
An intelligent AI would do whatever it was programmed to do. No more, no less. Any goals - or, equivalently, conditions that make the AI "happy" - would have to be installed by its creators. Intelligence in no way implies direction.
I am more amused that she seems to have been using this as an argument for her hiring practices.
"No, you tired old executives, we need to hire more kids! We'd be following the business practices of the most successful chocolate factory in the world! Or do you want to end up like Slugworth?"
(And it apparently worked, too.)
Hell, the simplest would be an easy reading comprehension or logic test with a short-answer blank - the computer would never get it, and all humans would.
The only problem with that is that it's nontrivial to make up such tests. And that means that no site will ever have more than a relatively small number of them, in the hundreds or thousands at most. So all an attacker needs is a list of question types and answers.
Part of the reason "read these letters" tests are so widely used is that the number of variations is combinatorially large -- you can automatically vary background, rotation, slant, color or pattern, font, size, baseline, and so forth, individually for each letter.
Amazingly enough, none of these seriously impede people's ability to read the text, amazingly enough. That's some impressive OCR.
Rather, the argument that Spitzer used seemed to be that because the contract was included in the box of a mass market software product and had terms that were anathema to the public interest, those terms should be thrown out (i.e. as in with a contract of adherence).
Um, I don't think it's automatically illegal to do something against the public interest. According to the article:
Last spring, Mr. Spitzer sued Network Associates, asserting that the company's software included an unenforceable clause that effectively violated consumers' free speech. The clause, which appeared on software products and the company's Web site, read: "The customer will not publish reviews of this product without prior consent from Network Associates Inc."
Justice Marilyn Shafer of State Supreme Court in Manhattan ruled that the clause was deceptive...
The argument seems to be that it's a free speech issue, unless the NYT is oversimplifying.
Are we supposed to be impressed with a computer that can serve 8 [actually 11] hits and 4 pages per second?
If that's what a slashdotting consists of, then probably yes.
Re:Performance Critical
on
Hacker's Delight
·
· Score: 2, Insightful
aimed more at getting more features then having a fast, stable program
As I understand it, "stable" is kind of in opposition to both "fast" and "features". The reviewer's point is that super-optimized code tends to have strange tricks in it that are difficult to read and understand. That makes the code hard to maintain and increases the likelihood of bugs, so performance tweaking isn't a great idea unless speed is really important to an application.
Excellent points. I think Brin was pretty much making stuff up as he went along, like the rings representing tools...
>>>This yearning makes sense if you remember that arbitrary lords and chiefs did rule us for 99.44 percent of human existence. It's only been 200 years or so -- an eye blink -- that "scientific enlightenment" began waging its rebellion against the nearly universal pattern called feudalism
>Not to break it to you Einstein, but democracy was invented in ancient Greece. That's not a couple of hundred years, it's a couple of thousand years... just about as old as christianity itself.
Not to mention that feudalism is a lot more recent than Brin seems to think - it requires agriculture (a fact which he mentions and then ignores) that only developed in the last several millenia. Before that - which is the real 99.x% of human history - you pretty much just had small bands of hunter/gatherers, maybe some herders, and you can't be an arbitrary chief when you see everybody you're ruling every day. Maybe we should be yearning for that, huh?
What I meant was, if pi had repeating digits, then you'd be able to show that it was rational. But it isn't rational, therefore it can't have repeating digits without creating a contradiction.
Although nobody's looking for exact repeating patterns, I think some people are on the lookout for digits and patterns that occur with greater probability than chance would predict, like if 12% of the digits of pi seemed to be "6" instead of the expected 10%... dunno if they've had any luck though.
Slashdot Mirror
If you don't include yourself in your mental model you would not be able to identify potential risks, as having no self-awareness means not being aware of risks to self. That makes for a very quick "natural de-selection."
I'm not saying it's not important or useful to have a model of yourself. Clearly it is. (Although your risk example isn't complex enough to show that. An amoeba could have a simple rule that says "If you detect a [member of 'dangerous thing' list] then move away", no nontrivial self-modeling necessary. Where having a model of the self becomes more useful is when you want to take into account your own effects on the world, so you have to calculate what those effects are likely to be.)
But that's beside the point. The point is that having a model of yourself doesn't have to be related to consciousness in any way, and probably isn't. It's easy to give a robot a model of its environment, and equally easy to include a representation of the robot itself in the model. But the robot isn't conscious or aware in the usual sense. Modeling yourself is not noticeably different, consciousness-wise, than modeling somebody else and their effects on the world.
Dude - if a character on DS9 asks such a near-rhetorical question
No, that would be a Cardassian question.
Silly people...
Then at some point your "map" has grown so complex that it is only meaningful when you include yourself, your needs and preferences into its model equations. And then you have... consciousness, or a functional equivalent thereof.
Well... it sounds nice, but only if you don't think about it too hard.
The thing is, why should including a model of yourself in the map be so special? It's just another model. You could even argue that it ought to be easier, because the mind has more reliable information about you than about most things it has to model (say, other people).
Consciousness is about awareness, period, not just awareness of the self.
Douglas Hofstadter's book "Gödel, Escher, Bach - an Eternal Golden Braid" has some HEAVY examination of (human and machine) consciousness. My favorite metaphor he uses for consciousness is an ant hill. The ant hill has many layers of emergent, recursive properties.
Careful with this, though. While this is an excellent book, its treatment of mind and consciousness is often a little dubious. (Also, the anthill is more a metaphor for how complexity can arise from simple components rather than consciousness.)
The main point I'd bring up is that there's no particular reason to think that consciousness is related to recursion in any interesting way. It's actually odd how many philosophers define consciousness as "knowing about oneself", as the original poster did. Really, if I know about (have a mental representation of?) my desk, or if I know about you, knowing about me doesn't seem very different.
A better idea about consciousness is one that grows out of what gwernol is talking about: Consciousness is what lets you focus on something, the thing that looks at all the stuff going on in your brain and chooses one item to work on specifically.
The thing about Cyc is, it's all done by a private company, and they don't really release much of anything about their work. So there's no way to know just how much progress they're making, what their strengths and weaknesses are. Which is a shame, because I think they have a promising idea that no one else currently has the resources to execute in the same way: enter into the system, more or less by hand, some ten million or so common-sense facts about the world, together with algorithms for making uncertain, defeasible deductions from these. (Maybe the closest alternatives are systems that try and automatically learn facts from text that's been annotated somehow for the purpose by people. I don't know of any particularly good ones, though.)
This reminded me of Anna Wierzbicka too. I read a paper of hers once. (Oh, look, I actually have it right here because I'm a big dork.) The whole project of reducing language to a few semantic primitives seemed... sort of mildly interesting in an intellectual-challenge kind of way, but overall completely pointless. It's not like brains actually define words in terms of primitives like this; if they did we'd see big processing-time differences for primitives vs. non. So it's just a weird little exercise.
(Also, aren't there words for family relations in there somewhere? Hmm, maybe not.)
(Also also, the paper has three pages of references, 1.5 of which are to herself. I think the rest disagree with her.)
"A computer would never be able to compete with competent chess player, and could certainly never compete with a Grandmaster. Chess requires true human intelligence."
Douglas Hofstadter put forth basically this point of view in his Pulitzer winner "Godel, Escher, Bach" in 1979. He predicted that computers wouldn't be able to beat grandmasters until they'd achieved human-level intelligence in general.
There was a competition for children to design a bit of sculpture for the cathedral. Personally, I think this choice is great: I suspect it's high and out-of-the-way enough, and similar enough to other random gargoyles, that you'll miss it unless you look carefully and think, "Hey, wait a minute..."
I also enjoyed their rationale. Heh. It's even sort of true, if you replace "evil" with "mythical monsters".
"Centuries ago, on our cathedrals, grotesques were intended to symbolize the evil that existed outside the church. Today, Darth Vader is an excellent example of evil in our times."
happiness = 1 / ( 7 - years of marriage )
Ooh... slow growth in happiness for a while, but soon you'll be approaching infinity!
Of course, then you go negative. Tough break, dude.
Wait, are you sure this is marriage you're talking about?
Check out my post elsewhere in the thread for a solution in English.
A very tough problem, involving considerable trial and error and pretty much insolvable in interview-time; I'm guessing the reports that this particular problem was used are probably apocryphal.
The statement of the problem: Given 12 objects, exactly one of which is a different weight than the others, determine a) which object is different and b) whether it's lighter or heavier than the rest. Use only 3 weighings on a balance scale.
It can be solved, but the solution boils down to a bunch of if-then directives.
------------
Weigh 4 balls vs 4 balls.
Equal:
One of the remaining 4 is heavier or ligher.
Weigh 3 "good" balls against 3 of those 4 balls.
Equal:
Weigh the remaining ball against a "good" ball
to see if it's heavier or lighter. []
Different:
Observe whether the three balls were heavier or lighter
than the good balls. If heavier, weigh 1 of the 3
against another of the 3. If these are the same, the
last ball is different and heavier. If they're
different, the heavier ball is the odd one out.
Similarly if the three balls are lighter than the
good ones. []
Unequal:
Take (A) 3 balls from the heavy side and 1 from the light side, and weigh against (B) 3 "good" balls and 1 from the heavy side.
If they're equal, we know one of 3 balls from the light side is lighter than the others. Weigh 2 against each other. []
If side A is heavier, we know one of the heavy-side balls on side A is heavier than the others. Weigh 2 against each other. []
If side B is heavier, we can limit our search to the light-side ball on side A and the heavy-side ball on side B. Weigh one of these against a "good" ball. []
I think there might also be one variation on part of this solution (maybe on the second weighing, if the first turs out unequal).
No... that just means Hotmail receives a lot of spam. So many people use it that a reasonable proportion of possible usernames are taken, and that means spammers can and do use "dictionary" attacks, where they send e-mail to random usernames and then just hang onto the addresses that don't bounce.
I believe that big providers like Hotmail and Yahoo try reasonably hard to prevent people from sending spam from their accounts, as it uses up bandwidth and creates ill will, so they do things like limit number of recipients per message, or recipients per day, that sort of thing. (Can anyone confirm that?)
But a spammer can make their e-mails appear to come from whatever address they want, and if there's a URL in the message they don't need to worry about whether people can reply.
Then he says that rule application can never produce understanding, and tries to make this claim sound obviously correct. (I think it's not, if only because everything above the quantum level, presumably including the brain, seems to be deterministic and in principle describable by rules.)
An intelligent AI would do whatever it was programmed to do. No more, no less. Any goals - or, equivalently, conditions that make the AI "happy" - would have to be installed by its creators. Intelligence in no way implies direction.
I imagine one could also remove mountains to allow remote viewing of approaching territory.
I imagine forgetting to put them back... march march march *splat*
Do these satellites actually detect small changes such as people and tanks? Otherwise, why not just use a map and a laptop?
I agree, this doesn't sound like a terribly useful thing. Certainly the satellites can't detect people through smoke and sandstorms...
4) Put you in my will... by wowbagger
This offer would seem more generous coming from someone with any other username.
I am more amused that she seems to have been using this as an argument for her hiring practices.
"No, you tired old executives, we need to hire more kids! We'd be following the business practices of the most successful chocolate factory in the world! Or do you want to end up like Slugworth?" (And it apparently worked, too.)
Hell, the simplest would be an easy reading comprehension or logic test with a short-answer blank - the computer would never get it, and all humans would.
The only problem with that is that it's nontrivial to make up such tests. And that means that no site will ever have more than a relatively small number of them, in the hundreds or thousands at most. So all an attacker needs is a list of question types and answers.
Part of the reason "read these letters" tests are so widely used is that the number of variations is combinatorially large -- you can automatically vary background, rotation, slant, color or pattern, font, size, baseline, and so forth, individually for each letter.
Amazingly enough, none of these seriously impede people's ability to read the text, amazingly enough. That's some impressive OCR.
Rather, the argument that Spitzer used seemed to be that because the contract was included in the box of a mass market software product and had terms that were anathema to the public interest, those terms should be thrown out (i.e. as in with a contract of adherence).
Um, I don't think it's automatically illegal to do something against the public interest. According to the article:
The argument seems to be that it's a free speech issue, unless the NYT is oversimplifying.
Are we supposed to be impressed with a computer that can serve 8 [actually 11] hits and 4 pages per second?
If that's what a slashdotting consists of, then probably yes.
aimed more at getting more features then having a fast, stable program
As I understand it, "stable" is kind of in opposition to both "fast" and "features". The reviewer's point is that super-optimized code tends to have strange tricks in it that are difficult to read and understand. That makes the code hard to maintain and increases the likelihood of bugs, so performance tweaking isn't a great idea unless speed is really important to an application.
It takes only about a minute with pencil and paper to discover that 2^239 - 1 is a multiple of 479.
If you have a minute, then, how does that work?
Excellent points. I think Brin was pretty much making stuff up as he went along, like the rings representing tools...
>>>This yearning makes sense if you remember that arbitrary lords and chiefs did rule us for 99.44 percent of human existence. It's only been 200 years or so -- an eye blink -- that "scientific enlightenment" began waging its rebellion against the nearly universal pattern called feudalism
>Not to break it to you Einstein, but democracy was invented in ancient Greece. That's not a couple of hundred years, it's a couple of thousand years... just about as old as christianity itself.
Not to mention that feudalism is a lot more recent than Brin seems to think - it requires agriculture (a fact which he mentions and then ignores) that only developed in the last several millenia. Before that - which is the real 99.x% of human history - you pretty much just had small bands of hunter/gatherers, maybe some herders, and you can't be an arbitrary chief when you see everybody you're ruling every day. Maybe we should be yearning for that, huh?
What I meant was, if pi had repeating digits, then you'd be able to show that it was rational. But it isn't rational, therefore it can't have repeating digits without creating a contradiction.
Although nobody's looking for exact repeating patterns, I think some people are on the lookout for digits and patterns that occur with greater probability than chance would predict, like if 12% of the digits of pi seemed to be "6" instead of the expected 10%... dunno if they've had any luck though.