First things first: this is fringe science, making its way into Scientific American. FTFA:
Says Barry Smith, a distinguished professor of bioinformatics and ontology at the State University of New York at Buffalo who is familiar with Bringsjord's work: "He's known as someone on the fringe of philosophy and computer science."
Bringsjord sounds a bit naive when considering the threat such programs could pose in virtual environments like Second Life.
"I wouldn't release E or anything like it, even in purely virtual environments, without engineered safeguards," Bringsjord says. These safeguards would be a set of ethics written into the software, something akin to author Isaac Asimov's "Three Laws of Robotics" that prevent a robot from harming humans, requires a robot to obey humans, and instructs a robot to protect itselfâ"as long as that does not violate either or both of the first two laws. "Because I have a lot of faith in this approach," he says, "E will be controlled."
Really? May be E will be controllable. But what about E'? You know, the one with the first 2 of the Three Laws edited in the source to be the boolean negation (!) of the original..
Fortunately for the researchers though, there's always someone out there willing to pay for this kind of stuff:
Bringsjord and Smith both have an interest in finding ways to better understand human behavior, and their work has attracted the attention of the intelligence community, which is seeking ways to successfully analyze the information they gather on potential terrorists. "To solve problems in intelligence analysis, you need more accurate representations of people," Smith says. "Selmer is trying to build really good representations of human beings in all of their subtlety."
Maybe that explains E's dark hair and five o'clock shadow.
Maybe not. Taking the claim at face value, then we'll never be quite dead: there will be always a copy of our brain somewhere ready to be loaded into a VM by some system admin.
Exactly. It's been done since time immemorial, a few cases of which are documented in Chomsky's Manufacturing Consent. (Nice tag, whoever did it.)
The difference between now and when Chomsky wrote his book is the web, of course. Kudos to this blogger Michael Geist for helping expose a farcical consensus. Hopefully we'll see more of this kind of analysis for other lobby groups as well.
Compared to my screen, my keyboard is awfully dirty. It's shiny with grime, even though I clean it about once month. I'm thinking if I had a touch screen, I'd have to wipe that screen clean about once an hour, at least.
Sort of. About 40 seconds into the
demo video cited by the story, you see a so so virtual keyboard overlayed on top of a dimmed background.
It's a shame that the virtual keyboard is so small that the user must type with one finger. There's plenty of screen real estate there to have a very nice large (virtual) keyboard. (I imagine this single finger text input interface might be deliberate--e.g. multiple fingers touching the screen might confuse the touch-screen hardware?)
This is perhaps offtopic, but cloud computing probably wont be relevant forever. It is hard to imagine, however, how open source will not remain relevant. More than just code, open source is a part of a collective heritage that assimilates the new and builds on the old. It's a tall argument to say that one's heritage may some day become irrelevant.
Below some (offtopic) thoughts on why cloud computing may not stay relevant (if it is, right now) into the future.
Geometric growth rate of hardware performance at the ends of the network may meet most application needs. To wit, recall the ephemerally named classic Managing Gigabytes (although, of course, the content of that book is even more relevant to today's data sizes).
Open source focus may shift away to the ends of the network. Think p2p, or the desktop.
Cloud computing may get commoditized. The drivers behind this may be portability (businesses may not want to commit to a single cloud provider), and yes, open source. Open source is often behind the feature curve when compared to closed source, but it's stronger in assimilating new features and ideas that come into vogue.
I think most would agree that to say "FOSS usability sucks" is an over-generalization. A lot of free software is not targeted at an end-user audience anyway. In my experience, FOSS usability shines in that non-end-user space. So if we broke down the user space like this..
Programmers
System admininstrators
End users
I think, FOSS usability is good, if not often superior to commercial software, for the first two categories of users.
So (again, an over-generalization) why does FOSS suck for end-users? I don't imagine it's because the end-users aren't geeky enough. It's not like only uber-geeks post to user mailing lists of successful FOSS projects: you'll find a big mix of expertise (programmers and non-programmers) in users posting to the Apache mailing lists, for example. Rather, I guess, the reason why end-user-facing FOSS projects often fall short on usability is that the user-feedback loop is somehow broken.
Here are some ideas off the top of my head on how to improve the situation:
Educate the user on how the FOSS process works. This could be done, hopefully discretely, in a GUI's startup splash screen, for example.
Provide an easy-to-use interface to the product's user mailing list.
Communicate bug and feature-request status information effectively to end-users.
Allow end users to escalate issues and somehow recognize them as contributors.
Maybe come up with usability standards on how end-users communicate with a FOSS projects.
Any other way to involve the end user.
If those are good/ok suggestions, then the good news is that much of the infrastructure for this feedback mechanism already exists in off-the-shelf FOSS components.
* *
On a tangential note, I don't quite buy in to the argument that FOSS usability suffers because it's boring. First, it's not boring to everyone, and second, depending on the perspective, there are lots of boring tasks involved in maintaining a FOSS project, and third, most UI projects want to be as user-friendly as possible--so there is no lack of motivation.
So if I have an algorithm that can verify an integer factorization quickly, it means there must be an algorithm that can factor any integer quickly? How would that work?
The anonymous poster makes a good counter argument against the idea that the algorithm must be easily defeatible: just because you have an algorithm that detects human behavior does not imply you have an algorithm that emulates the human behavior detected by the original algorithm.
In fact, there are many, so-called, one-way (correct terminology?) algorithms. So, for example, for a given file, it's easy to compute its MD5; harder to compute a file for a given MD5 (though doable). And of course, the AC's better example which is impossibly hard in reverse for composite numbers made from very large prime factors.
So no. Labeling the idea flawedbydesign is jumping the gun--logically, speaking.
When repressive governments set up offices and instruments to have citizens spy on each other, what usually happens is that it just becomes a tool for parties to private disputes to hassle each other. It becomes easy to set up a bunch of false witnesses and turn in your adversary to the authorities.
And the government employees who run this racket, soon discover this abuse. And in an effort to separate the "good" reports from the "bad", they become gatekeepers. So now if you want to turn your adversary in, you must spend a bit of money.
I thought this might just be a joke, but then I checked the calendar and noticed it isn't April yet.
A project team member is quoted as comparing to the distributed system to the borg. To drive the point, he adds
So far, there is little danger of the cybots getting out of control.
Then after some nonsense about how they don't have the resources to test this in the wild, the article closes ominously
He said there is some urgency in developing UNTAME. "We know we can do this," he said. "That means other people can do it." U.S. government officials assume that other countries are working on cyber warfare capabilities. "If we don't deploy this to defend the enterprise, someone else could turn this around and use it as an offensive weapon."
Short on details and long on threats that if you don't fund me bad things will happen. Sound familiar?
Parent must be right.. And I suspect this sort of chip design necessitates the same sort of rethinking about algorithms that near future multi-core processors will demand. There, the emphasis is on parallelization; here, the issue is randomness and probability.
Probabilistic (or randomized) algorithms have already become mainstream. (I don't know if Knuth has come out with his 4th volume, yet, but I believe he's promised to devote an entire section to the subject there.) The now classic example of a probabilistic algo is the
Skip List (1990, Pugh). So it seems natural to extend the domain of "the randomized model of computation" down to the hardware, especially if this randomness comes for free.
Now, in order to have any faith in a probabilistic algorithm, you need to know something about the probability distribution of the random variables. If I had to guess, an important requirement of this chip design must be that the randomness must not be a function of input (else, that would open an algo to pathological behavior on certain inputs).
And how does Murdoch figure in all of this? I wonder. Besides being the new owner of the WSJ, he apparently has his fingers in the music industry also.
I'll go out on a limb, put on my clairvoyant glasses, and describe the scene today in the past tense:
As the decade grew to a close, it must have been clear to Microsoft, the first and most enduring oligarch of the digital age, that it would soon secede leadership to a band of toolsmiths loosely organized under a cloak of Linux, a disdain for the oligarch's controlling and sometimes capricious ways, and a nascent humanist infomatic ideology . The war had been fought on many battle lines: from servers, routers, cell phones and other specialized hardware, to virtualized environments and even the desktop. On every front, Linux had pushed the oligarch back and was gaining ground. It was only a question of degree: the trend was already clear.
This was unlike any threat the oligarch had faced in its existence: this was guerilla warfare. An almost faceless enemy, a militia that was said to be capable of subsisting on just pizza and soda. And because of an almost monastic devotion to contributing the fruits of their labor to what they called the free bazaar, they enjoyed both the support of the citizenry and an emerging business nobility that had grown sour of the oligarch's onerous taxes.
And while the writing must have clearly been on the wall inside the Redmond offices, the pundits of the time, still bickered over so-called Linux adoption rates, ignoring the fact that the in the end the oligarch had lost most battles it had chosen to fight and was incapable of defending its steadily shrinking territory.
We have to have quite a bit of hubris to think that our ways of ordering things are the only ways of doing so.
Actually, there are very precise, non-anthropic definitions of entropy for us computer types. For example, using Chaitin's defintion (
google query), the entropy of a given byte sequence is defined to be the [byte] length of the shortest computer program that can generate that sequence.
There are an insane amount of "ordered" sequences (c.f. http://www.research.att.com/~njas/sequences/index.html), each one no more "random" than any other, given the appropriate context. Just because humans like x*2, and can pick it out easily, that doesn't mean that an alien species wouldn't find x^2 "more aesthetic" (or the Fibonacci sequence, or the digits of pi base 23...)
Which perhaps hints at the fact that the cards dealt in the parent's example are not as ordered (low entropy) as suggested. Why? Because the Chaitin entropy of the sequence includes the definition of
f(x) = (2*x) mod 13 and g(x) = (x^2) mod 13, respectively, x in [14,19].
(which btw, is not the shortest program that can generate the sequence, given how short the sequence itself is: the shortest program for *that* sequence is probably a "hard-coded" list. But I digress.)
So to put it another way, the algorithmic entropy of a byte sequence is governed by how compressible it is. But the method the parent proposes proposes to compress the card sequence would have to include, at the very least, numbers representing the choices of f(x) and g(x), and also the domain on which they are to be applied. So no, the cards dealt in the example in fact have quite high entropy.
There are many alternative power sources to petroleum. Whatever alternative power sources become economical in the future, I imagine it would be sensible to standardize on the way that power is delivered to the consumer. Electric power seems like a good candidate; hydrogen, for example, (hoax?) is not.
So as long as this grid is not hard-wired for a specific type of power source (I can't think how it could be), then I think this investment by the Australian gov should pay off.
Offtopic observation: posts in this page, and other slashdot stories concerning children are a good statistical sample for tracking and/or estimating slashdot user demographics concerning age and number of children in household.
markets begin to break down when certain features of our world have to start obeying the laws of geometry vs some idealized form of socio-economic system.
Someone once wrote (I think it was Nassim Taleb), that the magic of capitalism lies in its capacity for creative destruction; it's claim to efficiency (in the engineering sense) is really a farce. The free market, I'd venture to say, is only asymptotically efficient. That is, only the grotesquely inefficient die of natural causes; the rest are mostly victims of a circumstantial jungle.
I haven't tried snowl, but I think this type of effort has promise. I believe we have already surrendered too much control to the convenience of web-based interfaces for social networking. (Think web-based email.) Snowl's approach is more end-to-end.
Every web property owner will, at the end of the day, protect their own turf--at the possible expense of the user. So, for example, I can't expect facebook to play nice with say a competitor like Plaxo. I'd like them to--because I have accounts at both, and find duplicating my personal information at both places, among other things, really tedious. And it's only getting worse with time.
The solution to this is not yet another web-based aggregator. No, all that does is set up yet another middleman whose business model will be to eventually screw you. Much better to put all the smart at your end. That'll keep us, the end users, in control.
The U.S. National Aeronautics and Space Administration confirmed Thursday that it built a special chip used in a disputed demonstration of quantum computing in February.
NASA engineers used their experience with sub-micrometer dimensions and ultra-low temperatures to build a quantum processor for Canadian startup D-Wave Systems Inc., said Alan Kleinsasser, principal investigator in the quantum chip program at NASA's Jet Propulsion Laboratory in Pasadena, California.
To judge the veracity of that article, I
googled nasa + d-wave, and found a lot of articles corroborating the opposite conclusion: it appears a good number of people in the QC field take these guys seriously.
According to the AP article, Obama did vote for an ammendment that would have stripped the telecom immunity provision. But I guess it's the end result that really matters. And Obama too has failed us.
Please educate me. If the process is transparent, then why doesn't the reporter, instead of using the passive construction, say something like ".. Congressman McFinger slipped an election transaction reporting provision into the bill."
One nice thing about a content management system is that it's easy to see last minute changes--or changes to the revision to the Congressman's staff last read.
Slashdot Mirror
First things first: this is fringe science, making its way into Scientific American. FTFA:
Says Barry Smith, a distinguished professor of bioinformatics and ontology at the State University of New York at Buffalo who is familiar with Bringsjord's work: "He's known as someone on the fringe of philosophy and computer science."
Bringsjord sounds a bit naive when considering the threat such programs could pose in virtual environments like Second Life.
"I wouldn't release E or anything like it, even in purely virtual environments, without engineered safeguards," Bringsjord says. These safeguards would be a set of ethics written into the software, something akin to author Isaac Asimov's "Three Laws of Robotics" that prevent a robot from harming humans, requires a robot to obey humans, and instructs a robot to protect itselfâ"as long as that does not violate either or both of the first two laws. "Because I have a lot of faith in this approach," he says, "E will be controlled."
Really? May be E will be controllable. But what about E'? You know, the one with the first 2 of the Three Laws edited in the source to be the boolean negation (!) of the original..
Fortunately for the researchers though, there's always someone out there willing to pay for this kind of stuff:
Bringsjord and Smith both have an interest in finding ways to better understand human behavior, and their work has attracted the attention of the intelligence community, which is seeking ways to successfully analyze the information they gather on potential terrorists. "To solve problems in intelligence analysis, you need more accurate representations of people," Smith says. "Selmer is trying to build really good representations of human beings in all of their subtlety."
Maybe that explains E's dark hair and five o'clock shadow.
I have read and heard here and there that Rupert was naturalized by an act of Congress. If true, can anyone cite the act?
--
Tangential discussions tend to go offtopic. So what.
Aw, shit. We're all gonna die..
Maybe not. Taking the claim at face value, then we'll never be quite dead: there will be always a copy of our brain somewhere ready to be loaded into a VM by some system admin.
Exactly. It's been done since time immemorial, a few cases of which are documented in Chomsky's Manufacturing Consent. (Nice tag, whoever did it.)
The difference between now and when Chomsky wrote his book is the web, of course. Kudos to this blogger Michael Geist for helping expose a farcical consensus. Hopefully we'll see more of this kind of analysis for other lobby groups as well.
Compared to my screen, my keyboard is awfully dirty. It's shiny with grime, even though I clean it about once month. I'm thinking if I had a touch screen, I'd have to wipe that screen clean about once an hour, at least.
Either that, or stop licking my fingers.
Will it have a virtual keyboard though?
Sort of. About 40 seconds into the demo video cited by the story, you see a so so virtual keyboard overlayed on top of a dimmed background.
It's a shame that the virtual keyboard is so small that the user must type with one finger. There's plenty of screen real estate there to have a very nice large (virtual) keyboard. (I imagine this single finger text input interface might be deliberate--e.g. multiple fingers touching the screen might confuse the touch-screen hardware?)
Below some (offtopic) thoughts on why cloud computing may not stay relevant (if it is, right now) into the future.
I think most would agree that to say "FOSS usability sucks" is an over-generalization. A lot of free software is not targeted at an end-user audience anyway. In my experience, FOSS usability shines in that non-end-user space. So if we broke down the user space like this..
I think, FOSS usability is good, if not often superior to commercial software, for the first two categories of users.
So (again, an over-generalization) why does FOSS suck for end-users? I don't imagine it's because the end-users aren't geeky enough. It's not like only uber-geeks post to user mailing lists of successful FOSS projects: you'll find a big mix of expertise (programmers and non-programmers) in users posting to the Apache mailing lists, for example. Rather, I guess, the reason why end-user-facing FOSS projects often fall short on usability is that the user-feedback loop is somehow broken.
Here are some ideas off the top of my head on how to improve the situation:
If those are good/ok suggestions, then the good news is that much of the infrastructure for this feedback mechanism already exists in off-the-shelf FOSS components.
* *
On a tangential note, I don't quite buy in to the argument that FOSS usability suffers because it's boring. First, it's not boring to everyone, and second, depending on the perspective, there are lots of boring tasks involved in maintaining a FOSS project, and third, most UI projects want to be as user-friendly as possible--so there is no lack of motivation.
I disagree. I don't think there's anything terribly un-mimicable about the way humans interact with web pages.
Maybe, maybe not. The point was that claiming
it should indeed be defeatable by another algorithm
is not a logical slam-dunk.
So if I have an algorithm that can verify an integer factorization quickly, it means there must be an algorithm that can factor any integer quickly? How would that work?
The anonymous poster makes a good counter argument against the idea that the algorithm must be easily defeatible: just because you have an algorithm that detects human behavior does not imply you have an algorithm that emulates the human behavior detected by the original algorithm.
In fact, there are many, so-called, one-way (correct terminology?) algorithms. So, for example, for a given file, it's easy to compute its MD5; harder to compute a file for a given MD5 (though doable). And of course, the AC's better example which is impossibly hard in reverse for composite numbers made from very large prime factors.
So no. Labeling the idea flawedbydesign is jumping the gun--logically, speaking.
When repressive governments set up offices and instruments to have citizens spy on each other, what usually happens is that it just becomes a tool for parties to private disputes to hassle each other. It becomes easy to set up a bunch of false witnesses and turn in your adversary to the authorities.
And the government employees who run this racket, soon discover this abuse. And in an effort to separate the "good" reports from the "bad", they become gatekeepers. So now if you want to turn your adversary in, you must spend a bit of money.
And so a mob forms.. me thinks.
A choice video demo-ing the original technology can be found here.
A project team member is quoted as comparing to the distributed system to the borg. To drive the point, he adds
So far, there is little danger of the cybots getting out of control.
Then after some nonsense about how they don't have the resources to test this in the wild, the article closes ominously
He said there is some urgency in developing UNTAME. "We know we can do this," he said. "That means other people can do it." U.S. government officials assume that other countries are working on cyber warfare capabilities. "If we don't deploy this to defend the enterprise, someone else could turn this around and use it as an offensive weapon."
Short on details and long on threats that if you don't fund me bad things will happen. Sound familiar?
Parent must be right.. And I suspect this sort of chip design necessitates the same sort of rethinking about algorithms that near future multi-core processors will demand. There, the emphasis is on parallelization; here, the issue is randomness and probability.
Probabilistic (or randomized) algorithms have already become mainstream. (I don't know if Knuth has come out with his 4th volume, yet, but I believe he's promised to devote an entire section to the subject there.) The now classic example of a probabilistic algo is the Skip List (1990, Pugh). So it seems natural to extend the domain of "the randomized model of computation" down to the hardware, especially if this randomness comes for free.
Now, in order to have any faith in a probabilistic algorithm, you need to know something about the probability distribution of the random variables. If I had to guess, an important requirement of this chip design must be that the randomness must not be a function of input (else, that would open an algo to pathological behavior on certain inputs).
And how does Murdoch figure in all of this? I wonder. Besides being the new owner of the WSJ, he apparently has his fingers in the music industry also.
I'll go out on a limb, put on my clairvoyant glasses, and describe the scene today in the past tense:
As the decade grew to a close, it must have been clear to Microsoft, the first and most enduring oligarch of the digital age, that it would soon secede leadership to a band of toolsmiths loosely organized under a cloak of Linux, a disdain for the oligarch's controlling and sometimes capricious ways, and a nascent humanist infomatic ideology . The war had been fought on many battle lines: from servers, routers, cell phones and other specialized hardware, to virtualized environments and even the desktop. On every front, Linux had pushed the oligarch back and was gaining ground. It was only a question of degree: the trend was already clear.
This was unlike any threat the oligarch had faced in its existence: this was guerilla warfare. An almost faceless enemy, a militia that was said to be capable of subsisting on just pizza and soda. And because of an almost monastic devotion to contributing the fruits of their labor to what they called the free bazaar, they enjoyed both the support of the citizenry and an emerging business nobility that had grown sour of the oligarch's onerous taxes.
And while the writing must have clearly been on the wall inside the Redmond offices, the pundits of the time, still bickered over so-called Linux adoption rates, ignoring the fact that the in the end the oligarch had lost most battles it had chosen to fight and was incapable of defending its steadily shrinking territory.
mod parent up, that was a good read..
We have to have quite a bit of hubris to think that our ways of ordering things are the only ways of doing so.
Actually, there are very precise, non-anthropic definitions of entropy for us computer types. For example, using Chaitin's defintion ( google query), the entropy of a given byte sequence is defined to be the [byte] length of the shortest computer program that can generate that sequence.
There are an insane amount of "ordered" sequences (c.f. http://www.research.att.com/~njas/sequences/index.html), each one no more "random" than any other, given the appropriate context. Just because humans like x*2, and can pick it out easily, that doesn't mean that an alien species wouldn't find x^2 "more aesthetic" (or the Fibonacci sequence, or the digits of pi base 23 ...)
Which perhaps hints at the fact that the cards dealt in the parent's example are not as ordered (low entropy) as suggested. Why? Because the Chaitin entropy of the sequence includes the definition of
f(x) = (2*x) mod 13 and g(x) = (x^2) mod 13, respectively, x in [14,19].
(which btw, is not the shortest program that can generate the sequence, given how short the sequence itself is: the shortest program for *that* sequence is probably a "hard-coded" list. But I digress.)
So to put it another way, the algorithmic entropy of a byte sequence is governed by how compressible it is. But the method the parent proposes proposes to compress the card sequence would have to include, at the very least, numbers representing the choices of f(x) and g(x), and also the domain on which they are to be applied. So no, the cards dealt in the example in fact have quite high entropy.
There are many alternative power sources to petroleum. Whatever alternative power sources become economical in the future, I imagine it would be sensible to standardize on the way that power is delivered to the consumer. Electric power seems like a good candidate; hydrogen, for example, (hoax?) is not.
So as long as this grid is not hard-wired for a specific type of power source (I can't think how it could be), then I think this investment by the Australian gov should pay off.
Offtopic observation: posts in this page, and other slashdot stories concerning children are a good statistical sample for tracking and/or estimating slashdot user demographics concerning age and number of children in household.
markets begin to break down when certain features of our world have to start obeying the laws of geometry vs some idealized form of socio-economic system.
Someone once wrote (I think it was Nassim Taleb), that the magic of capitalism lies in its capacity for creative destruction; it's claim to efficiency (in the engineering sense) is really a farce. The free market, I'd venture to say, is only asymptotically efficient. That is, only the grotesquely inefficient die of natural causes; the rest are mostly victims of a circumstantial jungle.
I haven't tried snowl, but I think this type of effort has promise. I believe we have already surrendered too much control to the convenience of web-based interfaces for social networking. (Think web-based email.) Snowl's approach is more end-to-end.
Every web property owner will, at the end of the day, protect their own turf--at the possible expense of the user. So, for example, I can't expect facebook to play nice with say a competitor like Plaxo. I'd like them to--because I have accounts at both, and find duplicating my personal information at both places, among other things, really tedious. And it's only getting worse with time.
The solution to this is not yet another web-based aggregator. No, all that does is set up yet another middleman whose business model will be to eventually screw you. Much better to put all the smart at your end. That'll keep us, the end users, in control.
--
Have USB, Will Travel
Though I'd google for citations and ran into this PCWorld article. Quoting the article
The U.S. National Aeronautics and Space Administration confirmed Thursday that it built a special chip used in a disputed demonstration of quantum computing in February.
NASA engineers used their experience with sub-micrometer dimensions and ultra-low temperatures to build a quantum processor for Canadian startup D-Wave Systems Inc., said Alan Kleinsasser, principal investigator in the quantum chip program at NASA's Jet Propulsion Laboratory in Pasadena, California.
To judge the veracity of that article, I googled nasa + d-wave, and found a lot of articles corroborating the opposite conclusion: it appears a good number of people in the QC field take these guys seriously.
Yeh, he's been disappointing me recently, also.
According to the AP article, Obama did vote for an ammendment that would have stripped the telecom immunity provision. But I guess it's the end result that really matters. And Obama too has failed us.
Please educate me. If the process is transparent, then why doesn't the reporter, instead of using the passive construction, say something like ".. Congressman McFinger slipped an election transaction reporting provision into the bill."
One nice thing about a content management system is that it's easy to see last minute changes--or changes to the revision to the Congressman's staff last read.