Quote :

"All programs that mimic a human saying something are still crap to the rest of us."

Sorry, but you can't speak for the entire species. Some folks are enamored with current voice-recognition technology. I remember watching one of my coworkers getting his laptop to accept verbal command. Lots of problems, some success. As you note, companies already make us deal with difficult communication. Human don't answer many phones anymore. You have to fight through the recording first. I suspect it'll be the same with early conversational interfaces.

[Edited on May 27, 2008 at 2:34 PM. Reason : h00mahnz]

5/27/2008 2:33:25 PM

screw voice recognition (at least for now.. it's shamefully bad at the moment)

emails and text messages is where communication is at.. and if you have a decent translation program to screen those emails and messages you aren't gonna be slowed down.

5/27/2008 3:20:01 PM

Speech-to-text programs work reasonably well. I know my mother has used them to write papers and such.

5/27/2008 3:42:15 PM

Quote :

"good thing we don't consider Africa a real place...because they generally speak french as a second language."

I think some of the more functional African states are English-speaking, Commonwealth nations. The bigger players such as South Africa, Nigeria, Kenya, and Zimbabwe fall under this category. If anything a lot of the nations in Africa use English or French as a common national language to prevent perceived favoritism by selecting one of the tribal languages instead.

5/27/2008 4:17:26 PM

Quote :

"Pinyin is probably the latest attempt, it was originally developed to replace Chinese characters by the CCP back in the 1950s"

Not exactly -- it was adopted as a teaching method then and has been used as such ever since. The situation with the standard script is much more complex -- Mao decided to "simplify" the writing system so there's two writing systems in use, one for the mainland ("Simplified") and another for Taiwan ("Traditional"). Of course Taiwan didn't adopt Mao's system of writing Chinese characters and likely never will. In addition to that, there is a cursive script that is in common use.

There's some discussion at the grassroots (blog) level about replacing characters with pinyin. But the situation is very complex in that there are so few sounds in the language. Having only about 400 unique sounds (not counting tones), everything written in a romanized form basically looks the same. Characters do help to disambiguate -- and it's this usage which has often kept them around in Mandarin and other sound-poor languages.

I don't see characters going anywhere; what I do see happening is a lot of Westerners becoming semi-literate in Mandarin (ie just learning spoken and a few thousand characters) and a lot of Chinese becoming semi-literate in English (per the Li Yang Crazy English program ) and meeting in the middle somewhere.

Until, of course, we reverse engineer the human brain, all futurism comes true, etc. etc.

5/27/2008 9:00:38 PM

Quote :

"I'll begin with a question: Why do you write about linear increases in processing power? Moore's Law and company are explicitly exponential. That's a basic principle of Kurzweil-style future studies. Indeed, evidence shows that even the rate of doubling is increasing. This rate of growth is absolutely key. Linear progress would take forever to get us anywhere, but that's not what we have in information technology."

Well, the main issue here is that I'm not writing about linear increases in processing power. I am talking about linear increases in technological capability.

Moore's Law is about the number of transistors on a chip, and that chip is used to power a VN machine. Again, Computer Science -- just because the processing power (however you define it) of a VN machine increases exponentially, does not -- and I repeat, DOES NOT -- increase exponentially the class of problems that a VN machine can solve.

In other words: digital computers aren't all-purpose devices for solving all problems. They are practical implementations of a theoretical construct for computing _certain_ functions. There are functions that they cannot compute. Just as a car cannot fly, a digital computer cannot, by itself, properly solve NLP because it is not in the class of problems to which it is suited. The computer that does solve NLP will likely be something else like analog, neural, quantum, choose your favorite buzzword. But most likely not the VN machine sitting on your desk.

What a computer (or a network of computers) CAN do, is it can approximate solutions to NLP. The depth to which it can reasonably approximate NLP is questionable, and "moron" and I have been discussing that. Please feel free to add your input with substantial detail to back it up (beyond predictions, actual details on technology).

As to the rest of your post --

OK, fine, I get it -- Google is improving NLP. Duh. Someone should tell your linkee what an "asymptotic function" is when he's considering notions of "progress." The brain stuff is nice but not really appropriate to this discussion.

5/27/2008 9:25:17 PM

Quote :

"I am talking about linear increases in technological capability."

Could you explain what you mean by this? How do you measure technological capability? (The way futurists measure it, it doesn't increase linearly.)

Quote :

"What a computer (or a network of computers) CAN do, is it can approximate solutions to NLP. The depth to which it can reasonably approximate NLP is questionable, and "moron" and I have been discussing that."

And it's been a far cry from you earlier attempt to equate NLP with perpetual motion machines. I'm happy to see you backed away from that assertion. As noted by John Smart, simple statistical techniques are improving NLP. This undermines that idea that it's impossibly hard to program.

Quote :

"Please feel free to add your input with substantial detail to back it up (beyond predictions, actual details on technology)."

I'll contribute however I please, thank you. If I feel predictions are relevant, I'll provide predictions. You're free to dismiss whatever I post, but don't expect me to acquiesce to your attempt to control the debate. As we're talking about the future, the opinions and arguments of futurists seem appropriate.

Quote :

"Someone should tell your linkee what an "asymptotic function" is when he's considering notions of "progress.""

I suspect John Smart is aware of asymptotic functions. If you honesty don't think so, feel free to drop him an email. I'm sure he'd appreciate your insight. He's quite fond of criticism. For example, see here:

http://www.accelerationwatch.com/critiques.html

Quote :

"The brain stuff is nice but not really appropriate to this discussion."

It's appropriate even if I only judge by your posts in this thread. You compared NLP to solving the AI problem in general. Brain scanning and modeling could do exactly that.

By the way, IBM researchers explicitly say real-time machine translation is on its way. See this page for more information:

http://www.research.ibm.com/compsci/spotlight/nlp/

[Edited on May 27, 2008 at 10:10 PM. Reason : IBM added]

5/27/2008 9:52:02 PM

Quote :

"From what I gather about the Japanese language, the written language would be nearly incomprehensible without the use of Chinese characters (pure Hiragana with no Kanji) given the extremely large number of homonyms."

Yeah that's bullcrap. Japanese could get along just fine without kanji. But you have a bunch of traditionalists who use that homophone argument as a justification of the kanji. It's all about context clues

Technology is doing a pretty good job of marginalizing the importance of kanji though. Most of my kids can't remember how to write all the kanji they're supposed to know by high school so they'll cheat and use their cell phones to look them up. But sometimes they screw up and write the wrong character I wouldn't be surprised if some time in the future they shorten the list of required kanji.

5/28/2008 1:12:16 AM

Quote :

"One of my basic problems with context as an issue -- aside from the processing details you discussed -- is that as a theoretical concept, we don't really know what context is. That to me is the big intractable problem that really makes NLP hard on an other-worldly level. It might be said to be world knowledge -- but it's really not, sometimes it's simply a mental representation in someone's mind of what world knowledge is. e.g., one's own assumptions about the world versus facts."

I was thinking about bringing this up, but it's too complex of an issue for me to want to type about at 2AM. But, this is a big issue that would have to be dealt with. In general though, any 2 people speaking with each other rarely are viewing things from the same context, they may just not realize it until there's an actual physical misunderstanding. But, this is an inherent problem with any communication of this nature, so it may be something that we have to accept. A human can't translate something with 100% accuracy, even a team of humans, we shouldn't expect an algorithm to be able to get it perfect either. Essentially, we do the best we can, and it should be "good enough" as with any translation.

But, this is why a think an emergent/genetic algorithm would be best, because it allows for fuzziness.

Quote :

"This aspect actually interests me a lot because, as English speakers, we're rather addicted to the idea that words have strictly objective meanings underlying them. We have this vague, "common sense" notion of the "literal sense" of words that is actually a mental formulation and not a well-defined (theoretically) concept."

I think people think they think this (sic), but in reality this is not the case. If you just think of all the conversations on a weekly basis, or the things people say to you, you would realize there's a lot of things that aren't clear what someone means, you either just charge ahead anyway, or take the risk that you're doing the wrong thing. But because we filter out things so effectively, we become desensitized to the flaws in language.

In essence, context is a fleeting idea. It changes with society. So the fact that it can't be nailed down is not a problem, if you system is based around it being reflective of societal memes, which is one of the strengths of latent semantic analysis. The mappings can change as the inputs change, but they still represent the same things they always represented.

Quote :

"I kind of believe that, as we approach true "solutions" to NLP, the problem converges into the basic Turing test. i.e. if computers really can 'understand' natural language, then basically they must also be able to pass a Turing test since many of the same basic skills involved in deriving meanings from words are involved. I'm sure this is a fairly obvious idea but it's worth mentioning."

What i haven't been saying, or have been saying implicitly, is that truly good NLP would literally require a robust AI system. Or at least, that's what I would call something that gathers visual, audio, chemical, and textual inputs, continuously forms contextual relationships on those inputs, and formulates responses to queries based solely on those relationships. It would start off with something simple at first that could be used for translations, but would extend with some effort in to something that can pass the turing test.

I think we'll see something that can at least pass the turing test in my lifetime, that may have the effective intelligence of a young child that's REALLY good at basic math.

5/28/2008 2:10:38 AM

Quote :

"Not exactly -- it was adopted as a teaching method then and has been used as such ever since. The situation with the standard script is much more complex -- Mao decided to "simplify" the writing system so there's two writing systems in use, one for the mainland ("Simplified") and another for Taiwan ("Traditional"). Of course Taiwan didn't adopt Mao's system of writing Chinese characters and likely never will. In addition to that, there is a cursive script that is in common use."

I was under the impression that pinyin was originally developed with the goal of trying to replace classical characters entirely but when it proved too much, they shifted to the simplified version. I swore I read it somewhere while doing research on the "golden era" of the CCP, but ah well. Personally, I find their simplified character set a complete mess and a butchery of the language; I think they could have done a better and more intuitive set, but what's done is done.

Quote :

"There's some discussion at the grassroots (blog) level about replacing characters with pinyin. But the situation is very complex in that there are so few sounds in the language. Having only about 400 unique sounds (not counting tones), everything written in a romanized form basically looks the same. Characters do help to disambiguate -- and it's this usage which has often kept them around in Mandarin and other sound-poor languages."

That doesn't surprise me at all for the same reason the Japanese can't get rid of their characters either. It reminds me of something my parents said once, about how one advantage of characters was that it allowed the point to be portrayed quickly and precisely (for example, newspaper headlines) than when written out phonetically in Korean.

5/28/2008 2:14:08 AM

Quote :

"That doesn't surprise me at all for the same reason the Japanese can't get rid of their characters either. "

But Japan can get rid of kanji and has slowly been on its way to doing that. Sometimes I'll include kanji on a worksheet for a word and my students have no clue what it says, even though Japanese is their native tongue. Or what I really love is when I go to a museum or something with a Japanese friend, and I have to explain what something is because they can't read the kanji

5/28/2008 2:47:24 AM

^^

That's interesting, if you know the source of that I'd love to read it; the history of this stuff is fascinating. As far as I know, pinyin was never meant to replace hanzi. In fact Mao himself was very much into Chinese calligraphy so it's hard to imagine he'd want to get rid of characters.

Quote :

"That doesn't surprise me at all for the same reason the Japanese can't get rid of their characters either."

Mind you I think it's an incredibly bad reason. The writing system is insane -- it's really hard to write the hanzi properly, they are very hard to remember (even when you get radicals and stroke order and all that), and they convey very little meaningful information in their form.

5/28/2008 5:15:11 AM

Quote :

" I'll contribute however I please, thank you. If I feel predictions are relevant, I'll provide predictions. You're free to dismiss whatever I post, but don't expect me to acquiesce to your attempt to control the debate. As we're talking about the future, the opinions and arguments of futurists seem appropriate."

Well, I'm really not trying to "control the debate." In point of fact, I'm telling you what I personally am willing to debate about. I do get to choose what I post here, would you not agree?

You can feel free to post whatever you want and contribute. Cheers, I say. But at this point I don't really care what you have to say about futurism -- I mean, it's basically just religious proselytizing.

5/28/2008 5:27:12 AM

Quote :

"A human can't translate something with 100% accuracy, even a team of humans, we shouldn't expect an algorithm to be able to get it perfect either. Essentially, we do the best we can, and it should be "good enough" as with any translation."

Exactly. In fact, I'd say you can't ever translate anything with perfect accuracy. English ain't Mandarin, and Mandarin ain't English. They won't ever be completely equivalent. In practice, this doesn't much matter. The whole process of communicating with language is a messy business.

Quote :

"But at this point I don't really care what you have to say about futurism -- I mean, it's basically just religious proselytizing."

Because all religions were founded on careful and detailed analysis of technological trends, right? You're parroting a common and dubious attack on Kurzweil and company. Interesting and useful ways to question the Singularity exist. This tired emotional appeal isn't one of them.

In any case, the issue of functional real-time translation can be separated from the rapture of the nerds. As I've shown, mainstream researchers say such translation programs are coming. They don't necessarily plan on uploading their brains or anything similar. They're primarily trying to make a buck and help people across the world communicate. You apparently think this impossible or unlikely. They don't, and won't stop working on it any time soon.

5/28/2008 10:58:01 AM

English is the international language

5/28/2008 11:36:09 AM

Quote :

"Sorry, but you can't speak for the entire species. Some folks are enamored with current voice-recognition technology. I remember watching one of my coworkers getting his laptop to accept verbal command. Lots of problems, some success. As you note, companies already make us deal with difficult communication. Human don't answer many phones anymore. You have to fight through the recording first. I suspect it'll be the same with early conversational interfaces."

okay, we'll split humanity into two. The ones who think the voice recognition and production programs are the greatest thing since sliced break, and the ones they force the technology upon.

Since the beginning of automation we started a war between people who desire human interaction for what they do and a system that would rather not pay for it. What you propose is not a solution to this, it would make it worse. I can very well imagine a distant future where half of the people walking on the street are convincing androids out to gather your personal information. The more you implement such systems, the more cynical most people will get of using them. There is some reasonable expectation of being dealt with by a human simply because oneself if a human as well.

On the quite different subject of technological singularity you keep blabbing about,

Quote :

"Interesting and useful ways to question the Singularity exist. This tired emotional appeal isn't one of them. "

Sheah, I'd say so. Moore's law had the bottom fall out on it. We're not reaching this singularity with a linear growth in processing power (I think someone mentioned this). We now must move to a computing environment where processing is done by a large community of autonomous processors instead of brute force that we're more used to. And we'll realize just how much a far cry our artificial processors are from the system of neurons nature developed.

Eventually we'll be able to recreate a human brain from the ground up. Not in the next 10 years. More like 100 or 1000 or 10000 years. Futurists giving overly optimistic estimates is not unprecedented.

More important than moore's law type arguments, I think projecting when computers will match humans is a grossly limited way of thinking. They are two completely different beasts - imagining that artificial life will advance so much to fully encompass the abilities of natural life is stupid and conceded. Instead, both will advance. Isn't this one of the futurist arguments? That biotechnology will get good enough to apply itself in other fields never before thought possible.

Our ability to manipulate life will advance just as computer technology continues to advance, and the abilities of one will never engulf the abilities of the other. The idea that computers will soon be able to do "human" things is the exact same trap people were falling into when the computer revolution happened. It's always been predicted, never happened, and we have less reason to believe it now than ever.

The monstrosities created by these overoptimistic predictions will probably be so horrendous people will long for the days of press 1 for ...

5/28/2008 11:44:08 AM

Quote :

"Sheah, I'd say so. Moore's law had the bottom fall out on it."

How so? I'm not aware of this happening. Intel's Pat Gelsinger sees no end in sight.

http://java.sys-con.com/read/557154.htm

Quote :

"We're not reaching this singularity with a linear growth in processing power (I think someone mentioned this)."

What is this mystical linear growth y'all keep referring to?

Quote :

"Eventually we'll be able to recreate a human brain from the ground up. Not in the next 10 years. More like 100 or 1000 or 10000 years."

Maybe a hundred, but current estimates of brain processing and computer improvement suggest otherwise. Your other numbers are wildly speculative. One only needs basic knowledge of history to conclude that the world be a wholly different place in ten thousand years. We've had significant scientific advancement for a few hundred years at most. For it to take so long to reverse engineer the brain would mean either an unprecedented retardation of progress or the discovery that the brain is many orders of magnitude more complex than we now believe. Possible, yes, but not likely.

Quote :

"Futurists giving overly optimistic estimates is not unprecedented."

Neither is excessive skepticism.

Quote :

"More important than moore's law type arguments, I think projecting when computers will match humans is a grossly limited way of thinking. They are two completely different beasts - imagining that artificial life will advance so much to fully encompass the abilities of natural life is stupid and conceded."

Well, I think exactly the opposite. Giving special powers to the meat is stupid and conceded. In this area, it's a philosophical and aesthetic debate as much as scientific one.

Quote :

"Instead, both will advance. Isn't this one of the futurist arguments? That biotechnology will get good enough to apply itself in other fields never before thought possible."

John Smart has a interesting article on the limits of biotechnology. You might want to read it. I'm not sure one way or the other, but I favor Smart's view.

Quote :

"Our ability to manipulate life will advance just as computer technology continues to advance, and the abilities of one will never engulf the abilities of the other."

Why not? Why are the two separate but apparently equal? I say it's all clumps of matter, and living organism aren't anywhere close to optimized. We can do better.

Quote :

"It's always been predicted, never happened, and we have less reason to believe it now than ever."

Why? That's a strange argument. As computers grow more and more potent, we have less reason to believe they'll match humans? By the way, it's been predicted for an incredibly short amount of time, not always. And because a predicted event doesn't happen when expected in no way suggest it'll never happen. Human were dreaming of traveling to the moon in ancient times. In the 17th century, The Man in the Moone had Spaniard flying there in a fantastic engine. You could consider this very early science fiction. Science fiction about lunar travel then proliferated. This didn't prevent us from actually setting foot on the moon in 1969. Those sort of predictions and fantasies have a funny way of coming true.

5/28/2008 2:08:58 PM

The two of us don't inherently differ, I have my optimism about the future, but it looks completely different from yours.

The shortest way I can put my stance:
You, like many in your camp, have a simple problem of telling the future what to be like. If we're looking at even 30 years in the future, the form of advancement will be amazingly different from what we can predict.


At the point we landed on the moon, it seemed perfectly reasonable that we would be on Mars in a decade or two, right? Given any particular time, take the fastest growing sector, project it out at the current rates, and that's the keystone of what the future looks like. Problem is, that's wrong. Industrialization changed the world, but limitations are eventually exerted downward forces on the massive engineering projects, and many of the biggest of something we built in the 70s haven't grown much since (tallest/largest building for example).

Just look at the Spruce Goose, or the Titanic. The people from those times would be awfully confused as to how our technology is a great leap forward - because it takes more than what meets the eye.

You right now can't go to the store and buy a processor that is 2^3=8 times as fast as what you could get 6 years ago. But it's much more common to get 4 linked together, and the same speed of processor may be cheaper, hugely energy efficient, and better designed. Plus, processor speed isn't the only spec of the world computer network. I see much more cause for optimism in solid-state storage technology, display technology, portability, (fingers crossed) network speeds, and integration into society. Again, a hard sell for someone from our time to fall in love with, who may be used to a bit different kind of progression.

You won't see a sudden halt in advancement of processors as molecular limits are hit, it will be a gradual slowing, and yes, that's not even guaranteed, as we have hope for the memristor or various kinds of new technology. There are already lots of things computers do better than the human brain.

Quote :

"Why not? Why are the two separate but apparently equal? I say it's all clumps of matter, and living organism aren't anywhere close to optimized. We can do better."

They're not equal in any sense, that's the point. The human brain was 100% evolved and programed from experience, whereas computers are the result of some advanced etching tool in a factory which was physically created in a very short amount of time. The abilities don't come close to overlapping for most problems. Never mind the fact that computers don't solve problems on their own.

Why the hell did evolution decide to keep over 1,000,000 species of animals on earth? By your logic of processors "equaling and surpassing" organic brains, we don't need but one kind of thing for problem solving right? Problem solving can't require more diversity than the biosphere - or can it?

No, the future holds much more diversification rather than consolidation. The software aspect of computer is already evolving to be more like organic brains. I'm about to tear my hair out from reading so many papers about genetic algorithms and artificial neural networks. Evolution got it right, and more and more of human advancement will come from understanding exactly what evolution did. The methods and construction that nature came up with will never be thrown out the window.

What John Smart said would be more relevant to this topic
http:///message_topic.aspx?topic=526546

The fact evolution isn't doing much anymore, and selective breading is of limited usefulness has little bearing on the use of naturally evolved structures from the last 4 billion years. He advocates the use of our ground-up designed technology more, but isn't that just because we can't yet understand the natural structures in question. As nanotechnology improves, what we make will look more and more like what nature looks like.

If one anticipates an eventual fusion of natural and artificial nanotechnology, then what John Smart says implies that ground-up engineered technologies have much further to go. And that natural structures are a hard to crack. Yeah, that sounds about right.

5/28/2008 3:45:18 PM

Quote :

"You, like many in your camp, have a simple problem of telling the future what to be like."

Kurzweil has a surprisingly decent record so far.

Quote :

"If we're looking at even 30 years in the future, the form of advancement will be amazingly different from what we can predict."

Maybe, maybe not. That remains to be seen. More critically, the future don't merely unfold around us. We make it. Folks like Marvin Minsky and Ben Goertzel aren't just speculating about AI, they're actively pursuing the technology. If you suspect a certain future is possible, you can work toward the goal. That's part of the explanation for the advance predicted by Moore's Law. Chip manufacturers strive for that rate of increase.

Quote :

"I see much more cause for optimism in solid-state storage technology, display technology, portability, (fingers crossed) network speeds, and integration into society."

Oh, I'm excited by those thing as well. Futurists now invoke Moore's Law as an example of exponential increase in technology.

Quote :

"There are already lots of things computers do better than the human brain."

Absolutely. I love Moravec's imagery of intelligence as a landscape. I'll quote:

Quote :

"Computers are universal machines, their potential extends uniformly over a boundless expanse of tasks. Human potentials, on the other hand, are strong in areas long important for survival, but weak in things far removed. Imagine a "landscape of human competence," having lowlands with labels like "arithmetic" and "rote memorization", foothills like "theorem proving" and "chess playing," and high mountain peaks labeled "locomotion," "hand-eye coordination" and "social interaction." We all live in the solid mountaintops, but it takes great effort to reach the rest of the terrain, and only a few of us work each patch. Advancing computer performance is like water slowly flooding the landscape. A half century ago it began to drown the lowlands, driving out human calculators and record clerks, but leaving most of us dry. Now the flood has reached the foothills, and our outposts there are contemplating retreat. We feel safe on our peaks, but, at the present rate, those too will be submerged within another half century. I propose (Moravec 1998) that we build Arks as that day nears, and adopt a seafaring life! For now, though, we must rely on our representatives in the lowlands to tell us what water is really like."

http://www.transhumanist.com/volume1/moravec.htm

Quote :

"By your logic of processors "equaling and surpassing" organic brains, we don't need but one kind of thing for problem solving right?"

I wouldn't that far. I think we can improve on nature in all areas. As Drexler writes, nanotech mimics with the efficiency of current solar cells could out-compete real plants. Natural selection is a tedious and haphazard way to design anything.

Quote :

"As nanotechnology improves, what we make will look more and more like what nature looks like."

To some extent, yes. But it'll be far more capable. Molecular nanotechnology will bring replication to an entirely different level.

5/28/2008 6:53:18 PM

Quote :

"You're parroting a common and dubious attack on Kurzweil and company."

I wasn't attacking Kurzweil, I was attacking you. I'm sure I could have quite an interesting conversation with Kurzweil about this stuff and learn a lot from him. But the difference is, he understands the basics, and you don't.

Or do you consider yourself to be in "the company" of Kurzweil? That's a laugh.

[Edited on May 29, 2008 at 5:24 AM. Reason : foo]

5/29/2008 5:23:20 AM

Uh, it's most commonly an attack on the movement in general. Oh dear, they see Kurzweil as a prophet, uploading as heaven, etc. I'll admit your apparent respect for the man breaks the mold. Most who use the religion argument dismiss him as a cult leader.

Quote :

"But the difference is, he understands the basics, and you don't."

You've yet to demonstrate this. I'm quite capable of discussing technological possibilities. You're angry for unclear reasons. You enthusiastically hurl insults. That's fun but unenlightening. Feel free to explain yourself. I think we've have an interesting debate in this thread. I don't know why you're so pissed.

5/29/2008 1:44:08 PM

By the way, as far as basic knowledge goes, I'm still confused by your attempt to equate NLP and/or strong AI with perpetual motion machines, infinite free energy, and teleportation. You have yet to cite any physical laws to support this strange claim. On the reality of artificial intelligence, I'll quote Ben Goertzel:

Quote :

"Very few contemporary scientific researchers—in AI, computer science, neuroscience or any other field— believe AGI is impossible. The philosophy literature contains a variety of arguments against the possibility of generally intelligent software, but none are very convincing. Perhaps the strongest counterargument is the Penrose/Hameroff speculation that human intelligence is based on human consciousness which in turn is based on unspecified quantum gravity based dynamics operating within brain dynamics; but evidence in favor of this is nonexistent. But this is totally unsupported by evidence and almost nobody believes it. I've never seen a survey, but my strong impression is that nearly all contemporary scientists believe that AGI at the human level and beyond is possible in principle. In other words, they nearly all believe that AGI is not a matter of if, it's a matter of when."

http://www.kurzweilai.net/meme/frame.html?m=3

Note that Goertzel happens to be a pretty righteous dude. Observe:



Check out that hair.

5/29/2008 2:20:24 PM

http://arstechnica.com/news.ars/post/20080529-the-semantic-web-gets-a-boost-from-functional-mris.html

Here's some moderately interesting new research. The conclusions reached in the article are of the "duh" variety, but the usage of biometric technology to enhance computing algorithms is only going to grow.

Quote :

""But at this point I don't really care what you have to say about futurism -- I mean, it's basically just religious proselytizing." Because all religions were founded on careful and detailed analysis of technological trends, right? You're parroting a common and dubious attack on Kurzweil and company. Interesting and useful ways to question the Singularity exist. This tired emotional appeal isn't one of them. "

Calling it a religion might be a little too far, but you certainly seem to be responding to it like a religion. The basis of your argument here is that "it's predicted by futurism" and your support is just reporting the words of your prophets/scientists. If their arguments were so strong, why not post their data instead of their assertions? And the whole idea of a technological singularity is almost like a biblical rapture. Some great event at some unspecified point in the future will happen, in response to a vague set of incidents that will trigger the event.

I'm sure the future will happen, but thats its job. But pointing out that it will happen just-because has no bearing on asking what will make it happen and why. AI is not going to happen just by people saying its going to happen. People have to discuss it and find ways to make it happen.

[Edited on May 30, 2008 at 12:01 AM. Reason : ]

5/29/2008 11:55:45 PM

So, what's the going rate for a futurist? 5 moonbeams and 3 unicorn horns per hour?

5/30/2008 1:41:14 AM

Quote :

"You're angry for unclear reasons. You enthusiastically hurl insults."

I'm not insulting you nor am I angry. You're ignorant. That's not an insult -- it's a fact. If you feel hurt by the fact that you don't honestly know something, then that's your issue. Doesn't change anything.

Here are the facts, ma'am:

I do technology for a living. I have a degree in Computer Science. I make money because I understand the limitations of what technology can and cannot do. I've been successful at this by most standards. If you want to characterize my job as just "programming" then by all means, do so, but you really just don't know the first thing.

You, on the other hand, read some books with some nice stories and now you think you're qualified to talk about Natural Language Processing. You're not.

I don't read books about Constitutional law and presume to argue with professional lawyers about how the Supreme Court is going to be ruling on specific cases. I really don't see why you think you have any credibility in a discussion about NLP. You just don't.

[Edited on May 30, 2008 at 5:38 AM. Reason : fo]

5/30/2008 5:38:00 AM

Quote :

"But Japan can get rid of kanji and has slowly been on its way to doing that. Sometimes I'll include kanji on a worksheet for a word and my students have no clue what it says, even though Japanese is their native tongue. Or what I really love is when I go to a museum or something with a Japanese friend, and I have to explain what something is because they can't read the kanji "

Curious:
What character did a Japanese person not know at a museum?

You can't type, maybe just look up the code
http://www.csse.monash.edu.au/~jwb/cgi-bin/wwwjdic.cgi?1B

5/30/2008 8:47:39 AM

Quote :

"Calling it a religion might be a little too far, but you certainly seem to be responding to it like a religion. The basis of your argument here is that "it's predicted by futurism" and your support is just reporting the words of your prophets/scientists."

It's not only religion that cite its sources, you know. In fact, it's an academic habit. Something NC State trained me to do. (Honestly, we work more dubious documents in the field of history.) I quote the leading thinkers because their comments are relevant and insightful.

Quote :

"If their arguments were so strong, why not post their data instead of their assertions?"

What type of data would you like? I haven't gotten much response from my attempts to discuss the character of progress. Using Moore's Law as an example, I say it's exponential. For unspecified reasons, others say it's linear. The data appears all around us. Look what's being researched at this very moment.

Quote :

"And the whole idea of a technological singularity is almost like a biblical rapture. Some great event at some unspecified point in the future will happen, in response to a vague set of incidents that will trigger the event."

I agree, but I find your description lacking. The progression of the Singularity is far clearer than the second quoted sentence. In some way, yes, the transhumanism movement resembles a religion. In other ways, not at all. The similarity to religion does not invalidate the theory.

Quote :

"AI is not going to happen just by people saying its going to happen. People have to discuss it and find ways to make it happen."

Absolutely. We can see various paths toward strong AI. I'm most familiar with two. One would be duplicating the brain in electronics, the approach described by Kurzweil and pursued by various scientists. The other would be Goertzel's effort to mimic brain function with more traditional computer science techniques and current neurological knowledge.

On the other hand, you can lose the big picture if you worry too much about details. The approach used for solving immediate problems won't work as well for prediction. This may be why certain folks with practical experience in a field make wildly pessimistic future estimates. History overflows with examples of seemingly insurmountable problems being quickly taken care of. You have to consider this. You can't limit your projections to strictly what you know in the present. I don't see why this thread should be restricted to purely technical discussion.

Quote :

"I'm not insulting you nor am I angry."

Putting "foo" in every edit box doesn't exactly convey calm. And you been explicitly insulting from early in this debate. Why deny it? I understand the appeal. Anyone can easily read the thread. Denial won't work.

Quote :

"You, on the other hand, read some books with some nice stories and now you think you're qualified to talk about Natural Language Processing. You're not."

I'm not convinced by your credentials, especially considering that I don't know what you really do, as you've noted. You haven't provide this information in a verifiable form. More importantly, I'm willing to argue with any person, regardless of their field. If I'm wrong, show me. I'll again ask you to support your assertion that NLP and/or strong AI is as impossible as perpetual motion machines, infinite free energy, and teleportation. When doing so, please cite a natural law for support.

And I've yet to see an argument from you why functional (not perfect) real-time translation should be so unlikely. Statistical methods seem to be advancing translation presently. Companies doing the research, such as IBM, expect these developments. They're not talking about AI or the Rapture for nerds, only solid machine translation. The technology already technically exists. You can find real-time translation programs for text. We mainly need to refine it.

Quote :

"I don't read books about Constitutional law and presume to argue with professional lawyers about how the Supreme Court is going to be ruling on specific cases."

You probably should.

Quote :

"I really don't see why you think you have any credibility in a discussion about NLP. You just don't."

I need no extraordinary credibility to make arguments, note evidence, and provide expert opinion. While you may be a genius from your point of view, I naturally see things differently. You're a handle to me. Think about that and you might better understand my position. Statements from prominent researchers and futurists carry more weight than one dude on TWW. Even under the traditional theory of credentials and credibility, you've got far less than Kurzweil. I could potentially cite him in an academic work. I couldn't cite Smoker4. Claims of expertise on internet message boards aren't the high-water mark.

[Edited on May 30, 2008 at 10:50 AM. Reason : history]

5/30/2008 10:24:45 AM

http://www.cbc.ca/world/story/2008/05/30/brazil-tribe.html

A tribe was found in Brazil that has never had contact with the modern world (hilarious picture there).

What is relevant to this article, I think, is that just because we're humans doesn't mean we'll advance. It takes specific circumstances to cause technology to be created, it just doesn't happen because of past trends.

5/30/2008 11:15:42 AM

Oh, I agree. Under different conditions, we wouldn't advance. Luckily for us, we're not isolated in an Amazonian jungle. A catastrophic event could theoretically hurl the species back to that state. I don't see that as likely, but it could happen. Barring such change, though, the suggests evidence change will continue at an accelerating pace. We can choose to hinder or bolster this process. I want to encourage it.

I question the methodology of basing predictions on current technical limitations. If you only foresee what you know, you'll always be wrong. Extrapolation has a better, though imperfect, record. Hard physical restrictions, on the other hand, I can accept. This relates to Smoker4's assertion. He equates NLP/AI with infinite free energy and accuses me of not knowing the basics. I don't get it. We have working models of language processing all around us. No such models of infinite energy or perpetual motion machines exist.

[Edited on May 30, 2008 at 12:39 PM. Reason : haste makes waste]

5/30/2008 12:37:28 PM

Quote :

"What type of data would you like? I haven't gotten much response from my attempts to discuss the character of progress. Using Moore's Law as an example, I say it's exponential. For unspecified reasons, others say it's linear. The data appears all around us. Look what's being researched at this very moment. "

Okay, someone mentioned it was linear once. The number of transistors per given area (or inverse of size of transistor) has exhibited exponential growth over a certain time period with a certain kind of transistor technology, thus exhibiting the behavior predicted by Moore's law.



You are looking to advance your idea by gross oversimplification. Not all technologies there can be said to exhibit "Moore's law", and those that can, do not have the same doubling time.

Atoms exist. There, I proved that there exists no technology for which density of electronic components can follow Moore's law for an infinite amount of time. In reality, manufacturing limitations has already began to 'flatten out' this generation's Moore's law. Here's one way; the price of a single production line increases as processor advance through time, you can't make a line that costs more than the value of the computer industry - that's a practical limitation. I know what you want to do here - you want to say that innovation will dictate that new technologies and human ingenuity will allow the more general metric of 'capabilities of computers' will follow the same trend.

You have no basis for saying that. Your treatment of the issue makes no allowance for different technologies. A car made of steel may never be able to achieve certain things doable by a car made of carbon-fiber. Vice versa may also apply.

Quote :

"A catastrophic event could theoretically hurl the species back to that state. I don't see that as likely, but it could happen."

Right, because tribal humans are the most adaptable form of humans.

5/30/2008 3:33:30 PM

Quote :

"Not all technologies there can be said to exhibit "Moore's law", and those that can, do not have the same doubling time."

Sure. But the exponential trend applies to various technologies, not merely transistors. For a quick example, look to the progress of the human genome project. It neither applies to everything nor to only to computer chips.

Quote :

"Atoms exist. There, I proved that there exists no technology for which density of electronic components can follow Moore's law for an infinite amount of time."

I readily accept such arguments. As you know, I'd point to quantum computing and company to continue the trend.

Quote :

"I know what you want to do here - you want to say that innovation will dictate that new technologies and human ingenuity will allow the more general metric of 'capabilities of computers' will follow the same trend."

Exactly.

Quote :

"You have no basis for saying that."

None save modern history.

Quote :

"Your treatment of the issue makes no allowance for different technologies."

I don't think the broadest view can or needs to. When we get into specifics, they can go either way. They're as likely to increase the rate of progress as to decrease it. Your criticism is quite legitimate. Trends stop, trends change. But we have to work with the historical data. What's your alternative? Pure uncertainly? That's the most accurate prediction, but it's not useful.

Some of the more radical members of the community have criticized Kurzweil on this point, saying that he sticks too rigidly to this graphs. It's appropriate to his approach. These numbers provide the strongest foundation for speculation. Attempt to estimate the exact effect of each potential technology becomes nightmarishly complex. I encourage such speculation, but I like Kurzweil's conservative method as well.

Quote :

"Right, because tribal humans are the most adaptable form of humans."

Could you expand on this? I can't tell if you're serious or it's an subtle flame I'm too stupid to see.

[Edited on May 30, 2008 at 4:45 PM. Reason : a bit dull are we?]

5/30/2008 4:42:37 PM

I don't really have a response, but I must give props on the handling of the obligatory cynicism.

5/30/2008 5:39:13 PM

Quote :

" A tribe was found in Brazil that has never had contact with the modern world (hilarious picture there)."

Wow, that almost belongs in the Onion ...

Quote :

" You probably should."

No, because I'm not a pseudo-intellectual fool. I also don't argue with my tax advisor over how to compute AMT, with my dentist over how to identify and fill cavities, etc. etc.

As to this "free energy" business, I already cited a "natural law" (being a mathematical concept you aren't familiar with) several times, and you've handily ignored it because you a) don't know what it is, and b) don't know anything but how to cite books and URLs. The analogy was not perfect but for all intents and purposes it was good enough; that you don't see that, is because you just don't understand what I was talking about to begin with. And you aren't willing to listen.

Both "moron" and "mrfrog" had perfectly reasonable discussions with me on the subject; and in fact they pointedly disagreed. They also knew what I was talking about and addressed my arguments directly. I'm willing to accept and discuss differences of opinion. What I'm not willing to accept here, is a basically religious argument interjected into a worthwhile topic.

I think if you look back at the methods of the more unsavory posters on this board (on topics of social stratification, shall we say), they use basically the same methods as you to get their points across. No real discussion of "facts on the ground," lots of deference to arbitrary (internet) authority, no real credentials/authority/expertise to back up the position, no data, etc. etc. In other words, you're really just a troll here.

As to the "identity" problem -- well, let's compare. On the one hand, I made some fairly straightforward arguments from established Computer Science about the nature of NLP. I used my expertise only to make the point that I actually am capable of discussing this stuff further and I understand the basics. Anybody on this thread who understands the basic theory (like mrfrog and moron) was capable of seeing what I was saying and speaking directly to it.

In your case, you represent yourself as well-versed in futurism; you represent yourself as a proxy for the views of Ray Kurzweil and company, but you aren't capable of representing their opinions in the context of my argument (or ANY reasonable argument based in theory, no matter who posts it). Because, as I said, you don't understand what I was talking about and it's just easier to post links and block quotes.

As to the "foo" thing -- considering I've used it in almost every single edit I've made in posting here for eight years, well ... let's just say you look really foolish right now. Not just because of that, but that (again) you are so clueless you don't understand the connotation behind it ...

5/30/2008 7:50:40 PM

lol, before i searched for NLP i was thinking it was a part of the



thingamabob.

I always hated that stuff. Abstract as fuck. I don't think it even says much about NLP. I not sure it belongs in any of those circles (not to exclude the possibility of overlap).

5/30/2008 10:43:41 PM

I think it's a pretty safe bet that the optimal solution to NLP is NP complete.

5/30/2008 10:49:51 PM

^ that may be true, but the optimal solution would surpass what the smartest of the smart human would be able to determine, and that level of precision is not required.

5/30/2008 11:54:47 PM

but given that language problems don't have an 'optimal solution' at all I would question the applicability of this categorization scheme.

5/31/2008 12:05:42 AM

^ i think if someone/something knew the entire vocabulary for a set of languages, and all slang and figurative phrases and their origins, they would be able to find a translation that would be better in some way than one even a top expert could come up with.

Here's an amusing site: http://www.blahblahfish.com/?selection=top_rated

You put in a phrase, pick a language, and it converts your phrase to that language, then back again.

[Edited on May 31, 2008 at 12:30 AM. Reason : ]

5/31/2008 12:16:15 AM

Quote :

"and that level of precision is not required"

What level of precision are we talking about here exactly?

5/31/2008 7:02:34 AM

The precision of using a theoretical NP-complete algorithm in order to find the solution.

5/31/2008 11:10:45 AM

Holy circular reasoning

5/31/2008 5:27:27 PM

Quote :

"Smoker4: No, because I'm not a pseudo-intellectual fool."

Well, I am. As I'm experienced in the field, I assume you'll defer to my opinion here.

Quote :

"I also don't argue with my tax advisor over how to compute AMT, with my dentist over how to identify and fill cavities, etc. etc."

I'm not paying you provide any service for me. Even assuming you predict NLP/AI futures for living, the analogy fails.

Quote :

"As to this "free energy" business, I already cited a "natural law" (being a mathematical concept you aren't familiar with) several times, and you've handily ignored it because you a) don't know what it is, and b) don't know anything but how to cite books and URLs."

Please support or withdraw these assertions.

Quote :

"The analogy was not perfect but for all intents and purposes it was good enough; that you don't see that, is because you just don't understand what I was talking about to begin with. And you aren't willing to listen."

No, it's not enough. The difficulty in programming NLP is no way equivalent to the conservation of energy. If you don't believe me, try running your claim by a physicist. You can't design your way past the conservation of energy. We know no means of evading this fundamental physical principle. On the other hand, we know for a fact that piles of atoms can process language. Our argument alone demonstrates this. To create a language processor, we need only look to ourselves. We have six billion working models.

You originally seemed to understand this. In your third post in the thread, you reasonably characterized NLP as a tough problem. You didn't initially suggest that a solution couldn't possibly be found. I'm not sure why you made this leap in your fourth post. It's no coincidence that few knowledgeable observers claim human-level AI is impossible, while most note perpetual motion machines violate an unbreakable law. I've never before seen anyone assert that NLP, a mere subset of AI, couldn't be done.

Quote :

"They also knew what I was talking about and addressed my arguments directly. I'm willing to accept and discuss differences of opinion."

Not all differences of opinion, apparently. Specifically, not ones posted by GoldenViper. Despite your protests, I have addressed the issue. I suspect we'll get functional NLP by refining the current statistical methods. Regardless of how that goes, human-level language processing will come once we reverse-engineer the brain. I'll grant I approach the problem from farther away. I'm interested in the grand trends, you're interested in the current details. This difference doesn't have to put us at one another's throats. I'm sadden that we haven't been able to make this dialogue less hostile.

Quote :

"What I'm not willing to accept here, is a basically religious argument interjected into a worthwhile topic."

Then don't invent one.

Quote :

"I think if you look back at the methods of the more unsavory posters on this board (on topics of social stratification, shall we say), they use basically the same methods as you to get their points across."

Ha. Between transhumanism and anarchism, I'm might be the new Kris. I'm no salisburyboy.

Quote :

"No real discussion of "facts on the ground," lots of deference to arbitrary (internet) authority, no real credentials/authority/expertise to back up the position, no data, etc. etc."

I reject your characterization of my debating style in this thread. I've discussed the facts, just not in the way you want. My sources are hardly arbitrary; they're futurists and industry insiders. I don't see why you're opposed to topical links, though I notice you haven't provided any yourself. Instead, you expect me to respect your credentials, which you haven't even established.

Quote :

"On the one hand, I made some fairly straightforward arguments from established Computer Science about the nature of NLP."

You also equated solving NLP with infinite free energy, an assertion you maintain.

Quote :

"Because, as I said, you don't understand what I was talking about and it's just easier to post links and block quotes."

By the same logic, may I conclude that you don't understand future studies? That would explain why you attempt to dismiss me as religious zealot rather than argue on the evidence, theory, and methodology.

Quote :

"As to the "foo" thing -- considering I've used it in almost every single edit I've made in posting here for eight years, well ... let's just say you look really foolish right now. Not just because of that, but that (again) you are so clueless you don't understand the connotation behind it ..."

Out curiosity, do you believe this paragraph makes you appear less angry? Does it support the claim that you don't hurl insults? The fact that you've been doing the same for years doesn't necessarily rob it of meaning.

5/31/2008 8:35:58 PM

Quote :

"The precision of using a theoretical NP-complete algorithm in order to find the solution."

My question was about the goal of NLP, not the algorithm itself. Let's see if we can narrow this down.

A "good" NLP solution, in my view, allows someone to listen to speech in _real time_ and understand it in context. Also it allows someone to speak and be understood, in real time. This is what humans do every day with language. And that is, in my view, an NP-Complete level of complexity -- not some ideal understanding that is above and beyond what humans do in ordinary communication.

This idea that statistical models can do the above with "enough" precision is, frankly, bollocks. I do believe that interpretation will be (and nearly is) "good enough" for rudimentary work like translating documents quickly -- given enough manual intervention to take some poorly translated text and put it into appropriate context. I do believe statistical models are helpful for this particular application.

But speaking and understanding language is not a matter of statistics, it's just black and white truth for most people. Only in specific cases of real ambiguity does the human interpretation boil down to a matter of "pick and choose." Not only that, but I would wager that for real communication, basic miscommunication is extremely taboo, a source of true embarrassment, and a computer algorithm that is "right" 95% of the time still won't be "good enough" if it fouls up even just a few instances of basic day-to-day communication.

Language is a sensitive thing. Arabic script was not printed with printing presses for over 300 years after the invention thereof because type-setting algorithms couldn't produce the calligraphic script with any degree of fidelity. It's not that people couldn't read it; the culture could surely have adapted.

The underlying question in this discussion is the goal of NLP technology. If it's to make language differences obsolete, I still believe that is wholly impossible, a pipe dream, a meltdown waiting to happen if it is tried. If it's to improve communication through the use of rudimentary but useful tools, then fine. We're already there and progressing along a well-defined curve.

As such I suspect the source of "optimism" around NLP expressed here is true naivetie about the role of language and the importance of software that "gets it right" every time before it's considered trustworthy and adopted in the mainstream of human society.

6/1/2008 6:31:04 AM

Quote :

" Out curiosity, do you believe this paragraph makes you appear less angry? Does it support the claim that you don't hurl insults? The fact that you've been doing the same for years doesn't necessarily rob it of meaning."

:sigh:

Here's your goddamned topical link.

http://en.wikipedia.org/wiki/Foo

Quote :

"Foo is a metasyntactic variable used heavily in computer science to represent concepts abstractly and can be used to represent any part of a complicated system or idea including the data, variables, functions, and commands. Foo is commonly used with the metasyntactic variables bar and foobar. The word foo itself has no meaning and is merely a commonly used logical representation that is used much in the way that the letters x and y are used in algebra to represent a number. In computer programming metasyntactic variables such as foo are used as a variable to represent the name of a subroutine, variable, or any other programmer named part of a program."

True definition of a troll: spews BS post after post, but is completely unable to research anything himself.

[Edited on June 1, 2008 at 6:39 AM. Reason : foo]

6/1/2008 6:39:31 AM

In the programming world, "foo" is commonly used as a placeholder variable for anything (commonly juxtaposed with "bar"-- foobar, get it?). It's kind of like 'lorem ipsum" if you're familiar with that. He wasn't trying to say like a Mr. T "i pity da foo'" kind of thing.

6/1/2008 9:10:59 PM

Quote :

"But speaking and understanding language is not a matter of statistics, it's just black and white truth for most people. Only in specific cases of real ambiguity does the human interpretation boil down to a matter of "pick and choose.""

There's a lot of ambiguity in language. As y'all have discussed, the context is key. We handle structural and lexical ambiguity regularly. Occasionally, we get confused and ask for clarification. I think I see where you're going here, but describing language understanding as black and white truth doesn't sit well with me.

Quote :

"Not only that, but I would wager that for real communication, basic miscommunication is extremely taboo, a source of true embarrassment, and a computer algorithm that is "right" 95% of the time still won't be "good enough" if it fouls up even just a few instances of basic day-to-day communication."

How do you figure? Miscommunication happens to me daily, even in simple interactions. The sky doesn't fall. The world doesn't end. I think translation errors would be accepted in many applications. Folks communicate despite low proficiency all across the world, especially in English. This relates to the notion of English as an international language. Yes, but a limited form of English. Many of the world's English speaker are far from what Americans would consider fluent. They still speak it.

Quote :

"The underlying question in this discussion is the goal of NLP technology. If it's to make language differences obsolete, I still believe that is wholly impossible, a pipe dream, a meltdown waiting to happen if it is tried."

And for the longterm, I continue to disagree. Strong AI is coming, whether in thirty years or a hundred. Eventually, the posthumans will create superior forms of communication. Nothing like current languages could be considered the pinnacle, though I imagine unaltered humans will speak something equivalent for as long as they exist.

Quote :

"If it's to improve communication through the use of rudimentary but useful tools, then fine. We're already there and progressing along a well-defined curve."

Yeah. Note that I included improved ways of learning new language in my initial post. I wasn't only talking about translation

Quote :

"Here's your goddamned topical link."

That does clarify things. Thank you for the information.

Quote :

"True definition of a troll: spews BS post after post, but is completely unable to research anything himself."

I assure I could have Googled the term. I would have if I hadn't I assumed a different meaning. Why this fills you with rage I don't know. By the way, what BS have I spewed?

[Edited on June 1, 2008 at 10:56 PM. Reason : posthumans]

6/1/2008 10:27:37 PM

I remain somewhat confused by the defense of linguistic diversity and opposition to international tongue. While I agree establishing a universal language would be difficult under current circumstances, it's quite conceivable. Cultural dominance would be the most likely method. After all, people didn't used to speak English from Florida to California. Do y'all also desire linguistic fragmentation within the United States, or are you content to maintaining the current level of diversity? Would it be a good thing if I lost the ability to communicate easily with people from Texas?

I'm sure developing a unique North Carolinian language incomprehensible to outsiders would have certain cultural and mental advantages. To some extent, this already exists. Various communities have radically different dialects. Yet most English speaker can move closer to the standard when needed, facilitating communication. Would it be better to widen our differences? If so, why? If not, then why wouldn't an international language be desirable?

6/1/2008 11:09:40 PM

^

You're confused by linguistic diversity? What's so confusing about it?

What makes economies tick isn't homogeneity, it's the ability of people to express preference. And therefore the existence of it.

We'd have a very efficient free market, for example, if everyone only wanted white bread, and nobody wanted wheat bread (or artisan honey-crusted bread with sesame seeds). Marketing would be easy. Just make white bread.

Did you ever stop and think that people speak different languages, and adhere to different cultures, because that's what they want? Take Mandarin for example. Although it has not been strictly the same over time, it is effectively one of the oldest continuously spoken languages on Earth. It expresses concepts in ways that English doesn't, using sounds we don't have. I believe the Chinese people are rightfully proud of their language.

This discussion of linguistic fragmentation in the U.S. is a horrendous strawman argument. In China there are many different dialects which are all mutually exclusive to each other. Cantonese is not Mandarin, and they are as different as English and German. The whole world is not the United States. But do you think Hong Kong or Shanghai wants to be exactly the same as Beijing?

Looking at economics and world culture purely from the standpoint of efficiency is silly. Economics / Business / etc. is only about efficiency once preference is taken into account. You don't even start drawing the lines until you understand what they represent. You don't start Hello Kitty airlines until you realize it's an international craze. Then you try to optimize route pricing.

Likewise in the previous post, you asked me why miscommunication is such a big deal. You say that people communicate despite low proficiency in a given language. And yet, I say to you -- language is a major point of social stratification. Do you think people can make six figure incomes doing enterprise sales using a translator that causes them to mispronounce "chicken" as "kitchen?" Or that doesn't convey subtlety during a sensitive negotiation process? How about dating? How about languages like Japanese that have extensive notions of social hierarchy built into them, to be breached at your own peril?

When you don't speak the language, properly, you're not part of the culture. Not all cultures assimilate people the way the U.S. does. It's profoundly arrogant to believe they will begin to just because some people can sometimes communicate through a translator.

Language and culture are both overt statements of preference. Preference is just as valid in economics and global business as efficiency. By beginning to learn a language in earnest, you are expressing a preference for that culture. By deferring to an imperfect and (occasionally) silly machine, you are saying -- your culture is a burden, I only want to interact on a 'good enough' basis. I don't share your preference.

Not exactly a strong foundation for the future of human interaction in a globalized marketplace, IMO. NLP has its place like all tools, and that's that. Downplaying linguistic and therefore cultural diversity is completely inane.

6/2/2008 1:30:05 AM

Quote :

"You're confused by linguistic diversity? What's so confusing about it?"

No, that's not what I typed. I'm confused by the adamant defense of linguistic diversity I've seen in this thread.

Quote :

"Did you ever stop and think that people speak different languages, and adhere to different cultures, because that's what they want?"

I'm well aware of that. However, language isn't a simple matter of preference. It's not equivalent to bread. You can't yet go into the store and purchase fluency. If you could, I'd speak all major languages and many minor ones.

Quote :

"This discussion of linguistic fragmentation in the U.S. is a horrendous strawman argument."

How so? I'm trying to understand your ideal system. In what context is linguistic diversity desirable? You apparently favor it enough to oppose a universal auxiliary language. If it didn't already exist, would you want to invent it?

Note that a universal auxiliary language wouldn't erase other language, only threaten them. Many humans know and use more than one language on a regular basis. You could preserve consider linguistic diversity under such a system. Yet y'all don't like this. You value the diversity much more than I do. I'm trying to understand this.

Quote :

"Looking at economics and world culture purely from the standpoint of efficiency is silly."

I'll reiterate that the language barrier is real and harms people daily. No amount explaining, romanticizing, and justifying will change this fundamental reality. Attempting to solve an obvious problem isn't silly.

Quote :

"And yet, I say to you -- language is a major point of social stratification. Do you think people can make six figure incomes doing enterprise sales using a translator that causes them to mispronounce "chicken" as "kitchen?" Or that doesn't convey subtlety during a sensitive negotiation process? How about dating?"

That's all true. I foresee hilarity ensuing from various translation errors. I'm sure the elites will prefer expert human translators for a while, but that won't prevent the rest of us from using decent machine translation. Until we smash hierarchy, I agree language will remain a point of social stratification.

Quote :

"Not all cultures assimilate people the way the U.S. does. It's profoundly arrogant to believe they will begin to just because some people can sometimes communicate through a translator."

I don't remember claiming anything about cultural acceptance.

Quote :

"By deferring to an imperfect and (occasionally) silly machine, you are saying -- your culture is a burden, I only want to interact on a 'good enough' basis. I don't share your preference."

People already rely on human translators, who also happen to make mistakes. Is that similarly insulting?

6/2/2008 1:56:58 AM

^

is it really too hard for you to just write a reply instead of re-posting everything line by line? We're not in kindergarten here.

First of all, you obviously don't get the point that in the U.S. we're rather willing to accept people who don't speak English absolutely fluently. That is us. The rest of the world is not us. So in point of fact, you WERE claiming something about cultural acceptance, you just didn't realize it because your viewpoint is entirely America-centric. And we definitely have limits.

Having said that -- the notion that a language barrier "harms" people is so absurd, it's borderline psychotic. What's considered "harmful" in any reasonable discussion is relative to what's "normal." Your idea of normalcy is apparently a utopian world where everyone communicates equally well with each other, and anything other than that is a "harm." Pardon me if I don't exactly go along on your trip through the clouds. The air is thin up there.

Back here in reality, where I live, there's no opportunity cost to the language barrier. The simple reason is that the language barrier exists because people want it. If some culture wants to exclude others by virtue of speaking a different language, then that's their choice and your religious notion that everyone should cross-communicate equally doesn't factor into the equation. Because that's _your_ viewpoint; it's not necessary in this world that everyone share your personal preferences. Hard to believe, huh?

The fact is, I am sure that as of today that many countries _can_ "buy" fluency. There's nothing stopping national governments from instituting rigorous education from K (or equivalent) and up in some "world language" like English. China in particular could; they already tell their citizens what words to use for daily speech.

But -- imagine this -- they don't want that. And in my mind, people being free to express their preferences in terms of language and culture is a "good" thing. People self-determining "who they are" is their first basic right. That's as about as desirable as it gets. If you need help with why that's good, well ... I won't go there.

6/2/2008 4:54:22 AM