I've seen several presentations this year from bleeding-edge researchers from different disciplines (operating systems, programming languages, hardware) and everybody agrees that the only way we're going to keep up with Moore's Law in the next decade is by adding more cores. The hardware guys are nearing the limit of just how many transistors they can jam into a single core (which is why clock speeds have stopped increasing), so the new trend is to add more cores to a chip.
Windows and Linux do a nice job of dispatching threads to your 2 or 4-way SMP system; since we're usually running a few programs simultaneously, adding a single core will give us a speed boost even when all of our programs are single-threaded. But those days will end: We're not running thousands of programs, and people ARE talking about thousands of cores. In order to take advantage of those cores, programming tools are going to need to change dramatically; it's just too hard to effectively leverage that sort of parallelization by hand, especially with heterogeneous cores, and it's pretty much impossible to automatically infer very much parallelism from the usual imperative languages.
There's a HUGE distinction between duplication and forgery when it comes to cryptographically signed data.
Forgery involves fabrication of information: changing names, photographs, birthdates... This is computationally infeasible without the government's private key used to sign the information.
The issue with RFID passports would be if they could be/forged/... it doesn't matter if they can be duplicated.
Sure, there's a minor privacy issue if the passport can be read by proximity (how close do you need to be? ten inches?), but really... this is blown out of proportion.
but a completely open-source business? If that's not novel, it's certainly unusual. I think it's pretty cool. Hopefully we get a follow-up post in six months to see how well the concept works.
What would be really interesting, though, would be to have some more "standard" open source businesses... laundromat, convenience store, teriyaki joint, etc. The kinds of shops that don't take immense resources to start up.
I suppose the franchise concept provides a closed-source equivalent: pay money for a howto
It won't be zero, but it also can't be very high or else it wouldn't be cost effective for the police. Assuming that it takes a non-trivial amount of human time to process each positive, a high false-to-true positive ratio would be a show-stopper.
If I told you that there was a "best" strategy for Tic-Tac-Toe, and that any other strategy was idiotic, would you say the designers of Tic-Tac-Toe are rolling in their graves? Would you insist on playing losing strategies just to prove an idealistic point?
The Founding Fathers were smart guys; they would agree with me.
Voting your conscience is a naive (read: bad) strategy, unless your conscience happens to be a candidate with a chance of winning. I'm sorry to sound cynical, but it is mathematical fact.
If the Libertarians and the Greens actually understood this, they would spend their effort trying to CHANGE THE VOTING SYSTEM instead of contiually running candidates doomed to failure*.
* worse than failure, actually.. as with Perot/92 and Nader/2000
From a game theory standpoint, your best* strategy in our voting system is to cast a vote AGAINST a candidate:
1. Identify your least favorite candidate with a chance of winning
2. Vote for the guy who is most likely to beat him
Now, before you Libertarian/Green folks reply with idealistic "Vote your conscience!" idiocy, do everyone a favor and read up on:
Tactical voting
Gibbard-Satterthwaite Theorem
Duverger's Law
* best = most likely to obtain the outcome that most closely approximates your actual preferences
Yes. And when Page and Brin die, their embalmbed bodies will be displayed in glass cases with softly blinking LED lights. Pilgrims to the garage will leave offerings of Mountain Dew.
Seriously, though, I wonder what kind of security they had? Was it designed by someone competent using well-known strong encryption, or was it designed by the same clowns who brought us DVD-CSS and Diebold?
I seem to recall that the DoD software for (classified) security clearance applications still uses 56-bit AES1 --- e.g., someone with significant (but not unobtainable) resources could crack the encryption. And if that's what the Department of Defense considers secure, it's not likely the Census bureau is doing any better...
After much research, I have reverse engineered John Dvorak.
1. Write sensationalist article about Microsoft's imminent doom
2. Lure tens of thousands of Slashdot eyeballs to web site
3. Profit!
Seriously, though, why does every utterance from this guy make it to the Slashdot front page? This particular article is a little less ridiculous than the last one (some kind of fan-boy fantasy about how Apple is going to take over the world), but it's still absurd to think that MS is adrift. Dvorak's claims about how Vista and Office don't offer compelling reasons to upgrade are the exact same things people said about Windows 2000 and Windows XP (and their Office contemporaries), but those seem to have worked out pretty well for MS. The.NET CLR is a solid idea, and will be a major threat to OSS if MS succeeds in getting everyone to write managed code (and I say that as a Linux programmer), but it will take some time for that to happen. MSN sucks, and has always sucked, so that's hardly news.
The only real threat to MS is Google, a competition which Dvorak calls a "preoccupation". Has anyone else tried out the new Google calendar? It's good. Adios, Outlook.
The cell phone industry is a nice counter-example for anyone who insists that a free market is always good for the consumer, unless you redefine "consumer" as "wireless provider". NOBODY wants to carry around a phone that does what this article describes. Even those who aren't concerned about the privacy implications are going to be nonplussed by the fact that their batteries suddenly only last half as long because their phones are so busy processing and transmitting this marketing trash.
And to broaden my rant: Who are these people who think that playing TV programs and games on a phone is a great idea? Where are these people? I would love to see all of the marketing and R&D dollars poured into these stupid, stupid features go instead into producing smaller phones that have increased range, longer battery life and a user interface not designed by a team of raccoons. Is that so ridiculous?
I'll grant you that the underpinnings of cars, airplanes, washing machines, and light-bulbs predate the twentieth century, but the rate of increase of knowledge has never been greater than it is today. Just because the fruits of this knowlege don't manifest as objects the size of Volkswagens is no basis for claiming that western civilization is on the slide.
The twentieth century redefined physics. It invented computers and computer science -- and to dismiss those under the umbrella of the integrated circuit or the transistor does them a tremendous injustice.
Now, I'm not saying we're NOT on the decline --- everytime I have the misfortune of watching more than five minutes of prime-time TV it makes me want to start hoarding spices because they'll be valuable once civilization collapses and we revert to a barter economy --- I'm just saying that the advances we're making today are a little less obvious but no less significant than those of a century ago.
It seems to me that if your data is important enough to back up, then it's important enough to transfer to new media every few years. Media is/cheap/. For the last six years, I've kept data backed up on redundant hard drives in different locations, and I fully expect to replace the hard drives every two or three years. I never have to worry about having important data stored on obsolete technology, and unlike a collection of CDs, I can copy the entire archive in one operation and search for things efficiently.
This scheme might break down if you're some kind of media-fiend who consistently wants to backup really large amounts of data (e.g. exceeding the capacity of a few state-of-the-art hard drives) --- but these people stopped using CDs long ago.
Okay, so they've programmed a robot to understand mirrors... that's hardly the same as "self-awareness" in the sense of sentience or consciousness.
The article isn't very descriptive, but it sounds like stupid pseudo-science:
"This so-called mirror image cognition is based on artificial nerve cell groups built into the robot's computer brain that give it the ability to recognize itself and acknowledge others."... what does that even mean?
The real question is: was this robot programmed to recognize itself in a mirror, or did it come to the realization through observation and experimentation? If the latter, that's really impressive. If the former (more likely), this is no more "Artificial Intelligence" than that horrid chat-bot thing and it doesn't warrant any mention from anyone.
And while I'm ranting--- The term Artificial Intelligence makes me cringe. One-third of AI can be better described as computational statistics (pattern recognition), another third as an exercise in ontology (expert systems), and the other third is the territory of pseudo-scientific hacks who like say things like "this robot's computer brain has artificial nerve cell groups" when they really mean "our robot is a wheeled computer with some sensors attached"
Okay, okay, I exaggerated. Slackware has evolved. I was just trying to make the point that newer, sexier, more ricey distributions are eating up its market share.
It's all about package management. Slackware has long since been replaced as the hobbyist distro-of-choice by Debian and Gentoo, and the casual users who simply want a "basic distribution that works" have turned to Ubuntu, Fedora, Knoppix, etc. Slackware has not evolved.
My roommate, I suspect, is the typical Slackware user... using it because it's what he was using in the 90's, when he chose Slackware for its laid-back, non-commercial attitude and for its geek-chic. More power to him, but I think that the newbie-hobbyists of today are not choosing Slackware as their first distribution crush... it simply doesn't offer the compelling technical advantages of Gentoo or Debian.
(note that the author of the linked article doesn't even mention Gentoo... -1 credibility...)
It seems that user-defined types are limited to a type name and a series of attributes ("details"). Attribute types are limited to primitives like "Web URL", "text string", "location".
Hopefully they do expand the type system to allow for richer types (generalized algebraic data types, anyone?)
I don't see anything about an API either, but I think it's a fair bet that Google will supply one at some point. They've been very good about APIs for their other services.
Larry: I've been trying to infer structure from web pages all day, and boy are my arms tired!
Sergey: Hey Larry, what if we could trick our users into supplying structure along with their data?
Larry: Why, we would rule the world!
[Larry and Sergey burst into song...]
~
But seriously, this is way cool. Here's where Google Base could go from here:
1. Allow more complex data types (the AJAX interface is beautiful and simple, but really limiting... maybe allow more complex types to be built through an API?) I'm thinking of structures and some kind of instantiation system.
2. Publish a really rich library of types. What they've got right now is a good start, but more common types will prevent people from creating redundant and not-quite-identical types. For instance, if there had been a "Slashdot geek" category, we wouldn't all have had to create our own when we entered ourselves into the 'Base after reading this headline.
3. Allow data to be restricted to groups of users. Google already has concepts of users and groups, so this is a no-brainer... a few customized viewing pages, and it could replace Flickr, Blogs,... everything. Just by the example details you can tell it's poised to kill craigslist's classifieds.
(about the subject title... my cat wrote it just now when he jumped onto my laptop, and it seemed as good as anything I was about to come up with)
Slashdot Mirror
I've seen several presentations this year from bleeding-edge researchers from different disciplines (operating systems, programming languages, hardware) and everybody agrees that the only way we're going to keep up with Moore's Law in the next decade is by adding more cores. The hardware guys are nearing the limit of just how many transistors they can jam into a single core (which is why clock speeds have stopped increasing), so the new trend is to add more cores to a chip.
Windows and Linux do a nice job of dispatching threads to your 2 or 4-way SMP system; since we're usually running a few programs simultaneously, adding a single core will give us a speed boost even when all of our programs are single-threaded. But those days will end: We're not running thousands of programs, and people ARE talking about thousands of cores. In order to take advantage of those cores, programming tools are going to need to change dramatically; it's just too hard to effectively leverage that sort of parallelization by hand, especially with heterogeneous cores, and it's pretty much impossible to automatically infer very much parallelism from the usual imperative languages.
I missed that one. You're right. Amended:
It's the second worst analogy of the day!
That's absurd.
There's a HUGE distinction between duplication and forgery when it comes to cryptographically signed data.
Forgery involves fabrication of information: changing names, photographs, birthdates...
This is computationally infeasible without the government's private key used to sign the information.
ha, that's exactly what I came here to post =)
It's the worst analogy of the day!
Is this really a big deal?
/forged/... it doesn't matter if they can be duplicated.
The issue with RFID passports would be if they could be
Sure, there's a minor privacy issue if the passport can be read by proximity (how close do you need to be? ten inches?), but really... this is blown out of proportion.
Who cares if a passport can be cloned?
Doesn't the real problem occur if information can be forged?
I haven't seen anything that suggests this is possible.
The laser-etching part isn't unique...
but a completely open-source business? If that's not novel, it's certainly unusual. I think it's pretty cool. Hopefully we get a follow-up post in six months to see how well the concept works.
What would be really interesting, though, would be to have some more "standard" open source businesses... laundromat, convenience store, teriyaki joint, etc. The kinds of shops that don't take immense resources to start up.
I suppose the franchise concept provides a closed-source equivalent: pay money for a howto
It won't be zero, but it also can't be very high or else it wouldn't be cost effective for the police. Assuming that it takes a non-trivial amount of human time to process each positive, a high false-to-true positive ratio would be a show-stopper.
Ah, a different sort of tactical voting =)
If I told you that there was a "best" strategy for Tic-Tac-Toe, and that any other strategy was idiotic, would you say the designers of Tic-Tac-Toe are rolling in their graves? Would you insist on playing losing strategies just to prove an idealistic point?
The Founding Fathers were smart guys; they would agree with me.
Voting your conscience is a naive (read: bad) strategy, unless your conscience happens to be a candidate with a chance of winning. I'm sorry to sound cynical, but it is mathematical fact.
If the Libertarians and the Greens actually understood this, they would spend their effort trying to CHANGE THE VOTING SYSTEM instead of contiually running candidates doomed to failure*.
* worse than failure, actually.. as with Perot/92 and Nader/2000
Quite right.
From a game theory standpoint, your best* strategy in our voting system is to cast a vote AGAINST a candidate:
1. Identify your least favorite candidate with a chance of winning
2. Vote for the guy who is most likely to beat him
Now, before you Libertarian/Green folks reply with idealistic "Vote your conscience!" idiocy, do everyone a favor and read up on:
Tactical voting
Gibbard-Satterthwaite Theorem
Duverger's Law
* best = most likely to obtain the outcome that most closely approximates your actual preferences
Yes. And when Page and Brin die, their embalmbed bodies will be displayed in glass cases with softly blinking LED lights. Pilgrims to the garage will leave offerings of Mountain Dew.
How many laptops lost per household is that?
~
Seriously, though, I wonder what kind of security they had?
Was it designed by someone competent using well-known strong encryption, or was it designed by the same clowns who brought us DVD-CSS and Diebold?
I seem to recall that the DoD software for (classified) security clearance applications still uses 56-bit AES1 --- e.g., someone with significant (but not unobtainable) resources could crack the encryption. And if that's what the Department of Defense considers secure, it's not likely the Census bureau is doing any better...
After much research, I have reverse engineered John Dvorak.
.NET CLR is a solid idea, and will be a major threat to OSS if MS succeeds in getting everyone to write managed code (and I say that as a Linux programmer), but it will take some time for that to happen. MSN sucks, and has always sucked, so that's hardly news.
1. Write sensationalist article about Microsoft's imminent doom
2. Lure tens of thousands of Slashdot eyeballs to web site
3. Profit!
Seriously, though, why does every utterance from this guy make it to the Slashdot front page? This particular article is a little less ridiculous than the last one (some kind of fan-boy fantasy about how Apple is going to take over the world), but it's still absurd to think that MS is adrift. Dvorak's claims about how Vista and Office don't offer compelling reasons to upgrade are the exact same things people said about Windows 2000 and Windows XP (and their Office contemporaries), but those seem to have worked out pretty well for MS. The
The only real threat to MS is Google, a competition which Dvorak calls a "preoccupation". Has anyone else tried out the new Google calendar? It's good. Adios, Outlook.
Too much coffee? Didn't RTFA?
Both! But come on, this is Slashdot. What do you expect?
The cell phone industry is a nice counter-example for anyone who insists that a free market is always good for the consumer, unless you redefine "consumer" as "wireless provider". NOBODY wants to carry around a phone that does what this article describes. Even those who aren't concerned about the privacy implications are going to be nonplussed by the fact that their batteries suddenly only last half as long because their phones are so busy processing and transmitting this marketing trash.
And to broaden my rant: Who are these people who think that playing TV programs and games on a phone is a great idea? Where are these people? I would love to see all of the marketing and R&D dollars poured into these stupid, stupid features go instead into producing smaller phones that have increased range, longer battery life and a user interface not designed by a team of raccoons. Is that so ridiculous?
This is inane.
I'll grant you that the underpinnings of cars, airplanes, washing machines, and light-bulbs predate the twentieth century, but the rate of increase of knowledge has never been greater than it is today. Just because the fruits of this knowlege don't manifest as objects the size of Volkswagens is no basis for claiming that western civilization is on the slide.
The twentieth century redefined physics. It invented computers and computer science -- and to dismiss those under the umbrella of the integrated circuit or the transistor does them a tremendous injustice.
Now, I'm not saying we're NOT on the decline --- everytime I have the misfortune of watching more than five minutes of prime-time TV it makes me want to start hoarding spices because they'll be valuable once civilization collapses and we revert to a barter economy --- I'm just saying that the advances we're making today are a little less obvious but no less significant than those of a century ago.
Search Amazon for "Fluency with Information Technology : Skills, Concepts, and Capabilities" by Lawrence Snyder.
Exactly the comment I was about to make ;-)
I would have expected an editor at the Economist to have caught the misuse of deus ex machina, but, well, here we are.
Maybe it's an intentional misuse of the phrase gone awry? Who knows.
Using CDs for backup is /so/ 1996.
/cheap/. For the last six years, I've kept data backed up on redundant hard drives in different locations, and I fully expect to replace the hard drives every two or three years. I never have to worry about having important data stored on obsolete technology, and unlike a collection of CDs, I can copy the entire archive in one operation and search for things efficiently.
It seems to me that if your data is important enough to back up, then it's important enough to transfer to new media every few years. Media is
This scheme might break down if you're some kind of media-fiend who consistently wants to backup really large amounts of data (e.g. exceeding the capacity of a few state-of-the-art hard drives) --- but these people stopped using CDs long ago.
Okay, so they've programmed a robot to understand mirrors... that's hardly the same as "self-awareness" in the sense of sentience or consciousness.
... what does that even mean?
The article isn't very descriptive, but it sounds like stupid pseudo-science:
"This so-called mirror image cognition is based on artificial nerve cell groups built into the robot's computer brain
that give it the ability to recognize itself and acknowledge others."
The real question is: was this robot programmed to recognize itself in a mirror, or did it come to the realization through observation and experimentation? If the latter, that's really impressive. If the former (more likely), this is no more "Artificial Intelligence" than that horrid chat-bot thing and it doesn't warrant any mention from anyone.
And while I'm ranting--- The term Artificial Intelligence makes me cringe. One-third of AI can be better described as computational statistics (pattern recognition), another third as an exercise in ontology (expert systems), and the other third is the territory of pseudo-scientific hacks who like say things like "this robot's computer brain has artificial nerve cell groups" when they really mean "our robot is a wheeled computer with some sensors attached"
Okay, okay, I exaggerated. Slackware has evolved. I was just trying to make the point that newer, sexier, more ricey distributions are eating up its market share.
It's all about package management. Slackware has long since been replaced as the hobbyist distro-of-choice by Debian and Gentoo, and the casual users who simply want a "basic distribution that works" have turned to Ubuntu, Fedora, Knoppix, etc. Slackware has not evolved.
My roommate, I suspect, is the typical Slackware user... using it because it's what he was using in the 90's, when he chose Slackware for its laid-back, non-commercial attitude and for its geek-chic. More power to him, but I think that the newbie-hobbyists of today are not choosing Slackware as their first distribution crush... it simply doesn't offer the compelling technical advantages of Gentoo or Debian.
(note that the author of the linked article doesn't even mention Gentoo... -1 credibility...)
It seems that user-defined types are limited to a type name and a series of attributes ("details"). Attribute types are limited to primitives like "Web URL", "text string", "location".
Hopefully they do expand the type system to allow for richer types (generalized algebraic data types, anyone?)
I don't see anything about an API either, but I think it's a fair bet that Google will supply one at some point. They've been very good about APIs for their other services.
Larry: I've been trying to infer structure from web pages all day, and boy are my arms tired!
... everything. Just by the example details you can tell it's poised to kill craigslist's classifieds.
Sergey: Hey Larry, what if we could trick our users into supplying structure along with their data?
Larry: Why, we would rule the world!
[Larry and Sergey burst into song...]
~
But seriously, this is way cool. Here's where Google Base could go from here:
1. Allow more complex data types (the AJAX interface is beautiful and simple, but really limiting... maybe allow more complex types to be built through an API?) I'm thinking of structures and some kind of instantiation system.
2. Publish a really rich library of types. What they've got right now is a good start, but more common types will prevent people from creating redundant and not-quite-identical types. For instance, if there had been a "Slashdot geek" category, we wouldn't all have had to create our own when we entered ourselves into the 'Base after reading this headline.
3. Allow data to be restricted to groups of users. Google already has concepts of users and groups, so this is a no-brainer... a few customized viewing pages, and it could replace Flickr, Blogs,
(about the subject title... my cat wrote it just now when he jumped onto my laptop, and it seemed as good as anything I was about to come up with)