I read your comment with a sense of awe that you know both so much and so little simultaneously.
When establishing an SSL session, the first thing that happens is the receipt of the certificate of the destination party (e.g. I am the client, I request the certificate of the server).
The certificate is signed by a certificate authority, such as Verisign, who has presumably done some basic research on the company - the idea is that Verisign has verified that the target company is real. Inside of the certificate is the URL of the destination server and the public key of that same server.
At this point, I (the client), know three things for sure that could not have been faked (unless Verisign was tricked into creating a bad certificate): I know the URL of the destination server, I know the public key of the destination server and I know the identity of the controlling party of that server.
When establishing an encrypted session, I use the public key of the destination server to encrypt my own communications to establish a session key. The address I connect to is the URL found in the certificate. There is no opportunity in this exchange for a man in the middle attack, unless that man in the middle has the private key of the destination server.
I believe the digital fingerprinting in this case may be more of an individualized watermark that uniquely identifies an individual purchaser. So, for instance, if you buy a copy of Star Wars online, it will watermark the movie with a unique identifier that links in some database to you. This watermarking may be noticed by network sniffers, but the more important aspect of this is that if a copy of Star Wars with your id in it shows up all over the place, the MPAA knows who to sue.
If all this were was a tagging on movies to tell they were movies, it would be a PHB joke because it would take the movie swapping public exactly 10 seconds to add encryption. I'm sure the MPAA knows this.
I am glad the MPAA is doing this and I hope legislation is passed forcing ISPs to add sniffers to their networks.
Why? Because the amount of encrypted traffic on the Internet will explode and we will get closer to a truly private Internet network (as apposed to our very PUBLIC Internet today).
The ISPs will be legally required to do man in the middle attacks. When you start up an SSL connection they will accept it as if they were the destination and then make a request to the destination for a connection. They will then pipe all info between the two connections through their fingerprinting program, and then pipe the approved data to you and to them. None of this will ever happen.
This will never happen, for many reasons, not the least of which being that it wouldn't work (the destination URL and encryption keys would not match - the ISP cannot intercept the communication). Politically, you must remember, the MPAA would have to wage a war against the banks to get this through. And I don't care how tough the MPAA is, when they go up against financial institutions they will loose.
(3) Creationism. This is not a serious option anywhere in the Western world, but a large percentage of smart Americans still think that evolution is doubtful and that creationism is a real competitor.
Indeed, and in many circles in America that takes the form of looking upon God as an advanced extraterrestrial species and mankind as a genetically engineered (or uplifted) species.
It is good to have guns, because it is good for the government to be scared of the people. POINT (and in America, we often wonder why the rest of the world doesn't get this)
The second issue is more complicated and there are several sources to this image of America not caring about the environment. But first, you must realize that the liberal left-wing in America are the most die-hard extremists imaginable when it comes to the environment. Nobody in the rest of the world seems to notice our extremists that are pro-environment. However, their zeal induces an equal and opposite passion from our right wing.
The net result? You don't see our left wing, because their alignment with you makes them invisible. But our right wing, with their obnoxious stance on the environment sticks out like a tennis-ball sized wart on our faces. Part of their ability to act that way is as the parent poster pointed out, due to their blind faith in what their religious leader, George Bush, tells them.
But biometrics are important. Eventually, security will converge to encryption keys / certificates stored on physical keys (tokens), accessed by some biometric (such as a fingerprint) and a password.
As a previous poster put it, something you are, something you own and something you know. This provides the greatest degree of security. For this system to be compromised, the theif must steal your biometric data, your physical token and your knowledge. For the user, it feels no different than using a password with the possible exception of slipping in your token.
The major problem, of course, is that of loosing your key (token). Personally, I like the idea of a 3-way raided token. Leave one with your computer, put one in a safe and keep one with you. Anytime you access your accounts you put two of the three tokens from the stripe together and you can access the data in the stripe.
As someone who has both worked in Internet security and had serious physical and electronic fraud committed against him, I can tell you these issues are not a joke and something indeed does need to be done.
Is it just an accident (pun intended) that Slashdot ran this story at the same time as the current one on including random processes on-die?
Kinda spooky (or is it the global collective consciousness expressing it's desire to us, the technical geeky guys, to do it's will)...
Real Estate and the age of InkJet Construction
on
Machine-Grown Housing
·
· Score: 1
This and other robotic construction techniques are inevitable. The idea of what essentially amounts to InkJet Construction is quite interesting and is certain to take off and evolve.
What is truly interesting to consider, however, are the economic ramifications of this change. Make no mistake, this is as drastic an advancement for construction, architecture and in fact, civilization as can be.
While the switch-over to these techniques may occur slowly at first, once the kinks are worked out of the systems and economies of scale brings the equipment prices down to a reasonable level, automated construction will overwhelm manual construction. It will occur because the resultant structures will be drastically cheaper to build, stronger, more attractive and in all other ways better than hand built structures.
How many people will this put out of work? Consider it for a moment. What percentage of the workforce is construction?
How about real-estate prices? The new structures will be in all ways superior to hand-build structures and can be built in days for the cost of the materials (such as plastic and concrete). Existing structures will loose their values unless they have some historical or other significance, they will be "antiques". Real-Estate prices will evaporate, approaching the price of the lot only (for all one has to do is purchase a lot and hire out one of these machines for a couple of days as competition).
In residential real-estate, people will knock down their own homes and hire out machines to build a better, bigger, custom made one for them.
Anyone else have thoughts of battery powering a Mac Mini, equipped with airport and an iSight for applications like this?
Seems to me you could plop a Mini and an iSight on just about any remote controlled vehicle and have quite a bit of fun with it. You could use iChat to get high-resolution high-framerate remote video and audio.
Hell, the mini could be the robotic brain for the vehicle as well. This sounds like a great new project - the "Mac Mini robotics project!"
This is the first time I have run across this phrase: "But that's 20th century tech".
It marks a threshold in my mind. Interesting. I will now predict that this phrase "20th century tech" will become part of popular culture, describing shoddy, low-tech, incompatible, bug-prone, poorly designed gadgets that have no style.
The reason that the U.S. is not innovating in the area of energy production has to do with politics. America controls (directly and in the case of Saudi, indirectly) world oil production and therefore world energy. Alternate energy sources, especially those that free nations from the oil addiction reduce dependence on America and therefore reduce America's power.
China, knowing this, is actively persuing alternate energy policy including nuclear, hydrogen and more novel approaches. They want to detach themselves from the oil addiction so that they have independence from the U.S. and U.S. controled energy interests.
Again, politics.
But, the results are inevitable: As a result of these politics, the Chinese will inevitably control more advanced and more important energy technologies (both economicaly and ecologically). So the conclusion to this will be exactly the opposite of that desired by the status quo (America controlled energy). However, the administration doesn't care because they will be retired, rich, fat and happy (or dead of old age) when China turns it all around on America and effectively takes control of world energy production.
Your words are cutting and probably accurate... However, there is a twist:
I agree, the US is over as the preeminent power, its just going to take time for people to realize it. Around 2030 when China and the US face off probably in the Middle East for the remaining easy oil, Americans will get a rare taste of what war is like from the losing side.
I think this is quite well understood in the power-circles of the U.S. I believe the current military actions are being taken specifically to force a worldwide energy monopoly to be in the hands of the U.S. to offset the trade and deficit imbalances. These will probably continue during the Bush presidency.
I also believe that the U.S. will fight before it loses the ability to win, if it looks like it is slipping from it's position of power in the world.
For the sake of our children, world peace and everything holy, let us pray that the U.S. is not slipping out of world-power. It may in fact spell the end of the world as our desperate leaders lash out at the world to keep their power. America would become like Nazi Germany (who by the standards of the time had just as significant a technology and military advantage as the U.S. does today).
Given how poorly the western world handles these issues, i can't imagine how well it'll be done elsewhere...
Why is it that we tend to assume that we (in America - which is what I assume you mean by the Western world) always do things better? So just because we do a poor job of handling our waste, that means China will automatically do a worse job?
Remember, this is a country with over 3000 years of continuous existence, compared to our 200 years. I would suggest to you that they may know more about maintaining their environment and preserving for the future than we do.
Seriously though, what is the incentive for robots to reproduce?
One word: Motivation
Truly intelligent robots will not be deterministic, intelligence does not function like that. Intelligence is an emergent property. Since it cannot be programmed, per sey, providing some "purpose" to the AI's existence becomes an important problem. Without some *hunger* or *drive* the AI would just sit there doing nothing.
Why would this be a post of a "Linux" user switching? This, like many articles on slashdot since Ballmer beat his chest about attacking the OSS community where it lives (uhum... here?) looks to me like a psychologically tuned meme designed to undermine Linux users' pride in their choices...
Let's stick to windows users switching to OS X. We all know that Linux users love it too, but it is highly unlikely that they would "switch" - just buy a mini or a laptop to augment their collection of hardware.
Many have posted objections regarding uplifting other terrestrial species with human DNA. The objections seem rational, as it appears the uplift would be done in order to exploit the results.
What about the reverse, uplifting humans by grafting in superior dna found in the animal kingdom? Better eyes? Superior strength? whatever...
For that matter, he could be padding a nest for Carmack and crew to incubate their eggs in.
Nice thought, but this is a war of egos. Carmack has a name that resonates with a generation of gamers. His name has more power than Bezos' already. Do you think Bezos would let himself be overshadowed by the genius of Carmack?
Much more likely that Bezos thinks he can just throw money at the problem until it is solved, contrasted to Carmack's approach of throwing his own proven genius at that same problem.
Personally, my money is on Carmack.
But Stephenson does spin a nice tale, huh? Claiming to be a minor character in a heinlein novel is akin to suggesting that Bezos has a breakthrough in propulsion that will cause a paradigm shift in space technology.... If he wasn't a novelist (and therefore a sensationalist by nature), I might take that claim more seriously (after all, who doesn't like Stephenson?)...
Anyway, the ones you don't hear about (directly) are businessmen. They could give a fuck less about the secret service or NSA or any other organization unless it was easy to fence the data. Thus most people have no reason to fear them.
You appear to be glamorizing the life of a common thief. I hate to break this to you, but anyone who spends their time hacking computers for money is not only without decency, but also without brains.
These guys do eventually get caught, or worse, end up slaves to some crime syndicate. These people DO HURT others through identity theft, credit-card fraud, banking fraud and other malicious activities.
And I dare say, the ones that don't get caught probably don't get caught because of the fierce terrorizing power of the crime-syndicate that they are a part of (e.g. anyone gets close, they are scared away), rather than because of some super-talent. The super-talented guys can make it in the real world.
I concluded that it isnt practical for general use, it took too much time to compress an image (alright it was five years ago and so today it probably wouldnt matter), but most importantly there is no easy general compression solution for all images (for instance one that compresses tree pics well wont do faces well and vice versa).
I did some basic expirementation with Genetic Algorithms and fractal compression and I can tell you, GA does solve the problem. Not only solve it, but obliterates it. With GA, fractal compression can achieve compression ratios and quality that are unheard of with other techniques.
Of course, this is to be expected, after all - it is what nature does with us. Our genecode is the compressed image, our bodies are the uncompressed results.
"If the GA starts with 20 predetermined tuning parameter sets and mates and mutates those, then all we have is a fancy hill-climber."
This isn't true. The mutation alone differentiates it from a hill-climber.
Mutation in Genetic Algorithms are supposed to act as hill-climbers *most of the time*. Mutations are not supposed to make drastic changes - that is the job of random individulas inserted into populations and recombination (mating). Mutation (of the bit flipping variety) is mostly there to provide the opportunity make good solutions better.
The problem of hill-climbers is that they get stuck at local peaks. Genetic algorithms don't.
Genetic Algorithms get stuck at local maxima all the time, especially in populations with low population count and low genetic diversity.
"In order to find novel Maxima in the search-space, the GA should not be constrained - it must have the ability to exhaustively search, even though it does not (of course) do such a thing."
I don't understand what you consider to be the constraint in this situation. The genetic algorithm does have the ability to find any part of the search space as far as I can tell.
I need to take a look at the code, but my understanding from reading the article was that the GA starts with predefined tuning solutions and uses recombination and mutation to change them. In order to *effectively* search the search-space, more randomness needs to be present (e.g. generate a couple of random solutions each generation). Otherwise, the GA will pretty quickly draw to a local maxima and stick there (even with mutation, altough mutation *can* kick it out of a local maxima eventually - assuming there is higher peak to be found within the range of the mutation function).
I don't see how a fitness function could determine whether or not it had found "the hill".
Why does it have to? The point is, don't use the GA *all the time*. Use it until the system stabilizes and then hill climb for a while. The results will be better for a couple of reasons: (1) You are not running all the bad solutions at the bottom of your population all the time while hill climbing (only the best solution) and (2) Your best solution will be getting deterministically better through hill-climbing. A GA is a bad hill-climber but good at finding hills. Augmenting a GA with a hill climber will improve solutions (often dramatically).
Writing a GA is simple. Writing a GA that delivers results, now that is a different beast alltogether. My experience has shown me that there are some rather more complicated things that need to be taken into account such as:
Remembering previous good solutions instead of breeding them out of the genepool. In this scenario, it is likely that there are multiple eigenloads that a machine finds itself in. Having to re-generate the correct tuning parameters each time the load switches between these *magic* states is cumbersome. But if the genes that worked in the other eigenstate are still in the genepool, it becomes trivial. Often, this is achievable by augmenting the fitness function to take into account the previous aggregate fitness of a solution or somesuch.
Keep the right amount of randomness flowing. If the GA does not have a good source of randomness to draw on, it will operate much like a hill-climber and will oscillate between a few solutions. It will get stuck.
Keep the population correctly sized
Have multiple populations that occassionally interact
Statistical mating is important
In a case like this, the best current solution should get the most timeslices. Period. And this should be like 10x the total number of timeslices for the entire population (e.g. population=20, each generation is 220 timeslices - top solution gets 200, each other gets 1)
Tuning mutation and crossover is extremely important
These statements apply to all search algorithms for parameterized problem domains. No search algorithm is very effective if it is bottled off from significant portions of the domain.
Thank you. We agree. That was exactly my point - the implications are that in a scenario like this, if you allow the search to progress as it should, you will constantly be testing very sub-optimal solutions which will result in negative side-effects to system performance. For that reason, it would be necessary to place heavy restrictions on the percentage of time that the GA is testing new solutions and use other techniques such as hill-climbing to augment the GA.
A GA does exhaustively search a search space, given enough time. It just so happens that the solutions it finds while searching are generally very useful. These solutions may be used immediately and in some cases, you can stop searching further. But if you let the algorithm continue long enough, it WILL SEARCH THE ENTIRE SPACE.
If the GA is not able to reach the entire space or if it is biased in such a way as to make portions of the space hard to reach, then it is not an effective GA. If I wanted to state this in one sentence, I could say "A GA is only truly effective if you let it exhaustively search the search space." This was the exact statement that you took issue to. Please note that nowhere did I state that you had to let the seach finish before the results were useful.
I thought long how the "fittness" part of that GA might work. Ultimately the only "fittness" the kernel can with high precision measure is "load" isn't it?
Any better ideas?
I suppose the only way would be to have some "tuning" hooks that applications could use to announce their own performance metrics. The GA could optimize around load when these metrics are not available and take them into account if they are.
Slashdot Mirror
I read your comment with a sense of awe that you know both so much and so little simultaneously.
When establishing an SSL session, the first thing that happens is the receipt of the certificate of the destination party (e.g. I am the client, I request the certificate of the server).
The certificate is signed by a certificate authority, such as Verisign, who has presumably done some basic research on the company - the idea is that Verisign has verified that the target company is real. Inside of the certificate is the URL of the destination server and the public key of that same server.
At this point, I (the client), know three things for sure that could not have been faked (unless Verisign was tricked into creating a bad certificate): I know the URL of the destination server, I know the public key of the destination server and I know the identity of the controlling party of that server.
When establishing an encrypted session, I use the public key of the destination server to encrypt my own communications to establish a session key. The address I connect to is the URL found in the certificate. There is no opportunity in this exchange for a man in the middle attack, unless that man in the middle has the private key of the destination server.
So, sorry, you are just wrong.
I believe the digital fingerprinting in this case may be more of an individualized watermark that uniquely identifies an individual purchaser. So, for instance, if you buy a copy of Star Wars online, it will watermark the movie with a unique identifier that links in some database to you. This watermarking may be noticed by network sniffers, but the more important aspect of this is that if a copy of Star Wars with your id in it shows up all over the place, the MPAA knows who to sue.
If all this were was a tagging on movies to tell they were movies, it would be a PHB joke because it would take the movie swapping public exactly 10 seconds to add encryption. I'm sure the MPAA knows this.
I am glad the MPAA is doing this and I hope legislation is passed forcing ISPs to add sniffers to their networks.
Why? Because the amount of encrypted traffic on the Internet will explode and we will get closer to a truly private Internet network (as apposed to our very PUBLIC Internet today).
The ISPs will be legally required to do man in the middle attacks. When you start up an SSL connection they will accept it as if they were the destination and then make a request to the destination for a connection. They will then pipe all info between the two connections through their fingerprinting program, and then pipe the approved data to you and to them. None of this will ever happen.
This will never happen, for many reasons, not the least of which being that it wouldn't work (the destination URL and encryption keys would not match - the ISP cannot intercept the communication). Politically, you must remember, the MPAA would have to wage a war against the banks to get this through. And I don't care how tough the MPAA is, when they go up against financial institutions they will loose.
(3) Creationism. This is not a serious option anywhere in the Western world, but a large percentage of smart Americans still think that evolution is doubtful and that creationism is a real competitor.
;-)
Indeed, and in many circles in America that takes the form of looking upon God as an advanced extraterrestrial species and mankind as a genetically engineered (or uplifted) species.
So what? It could be true...
It is good to have guns, because it is good for the government to be scared of the people. POINT (and in America, we often wonder why the rest of the world doesn't get this)
The second issue is more complicated and there are several sources to this image of America not caring about the environment. But first, you must realize that the liberal left-wing in America are the most die-hard extremists imaginable when it comes to the environment. Nobody in the rest of the world seems to notice our extremists that are pro-environment. However, their zeal induces an equal and opposite passion from our right wing.
The net result? You don't see our left wing, because their alignment with you makes them invisible. But our right wing, with their obnoxious stance on the environment sticks out like a tennis-ball sized wart on our faces. Part of their ability to act that way is as the parent poster pointed out, due to their blind faith in what their religious leader, George Bush, tells them.
But biometrics are important. Eventually, security will converge to encryption keys / certificates stored on physical keys (tokens), accessed by some biometric (such as a fingerprint) and a password.
As a previous poster put it, something you are, something you own and something you know. This provides the greatest degree of security. For this system to be compromised, the theif must steal your biometric data, your physical token and your knowledge. For the user, it feels no different than using a password with the possible exception of slipping in your token.
The major problem, of course, is that of loosing your key (token). Personally, I like the idea of a 3-way raided token. Leave one with your computer, put one in a safe and keep one with you. Anytime you access your accounts you put two of the three tokens from the stripe together and you can access the data in the stripe.
As someone who has both worked in Internet security and had serious physical and electronic fraud committed against him, I can tell you these issues are not a joke and something indeed does need to be done.
Is it just an accident (pun intended) that Slashdot ran this story at the same time as the current one on including random processes on-die?
Kinda spooky (or is it the global collective consciousness expressing it's desire to us, the technical geeky guys, to do it's will)...
This and other robotic construction techniques are inevitable. The idea of what essentially amounts to InkJet Construction is quite interesting and is certain to take off and evolve.
What is truly interesting to consider, however, are the economic ramifications of this change. Make no mistake, this is as drastic an advancement for construction, architecture and in fact, civilization as can be.
While the switch-over to these techniques may occur slowly at first, once the kinks are worked out of the systems and economies of scale brings the equipment prices down to a reasonable level, automated construction will overwhelm manual construction. It will occur because the resultant structures will be drastically cheaper to build, stronger, more attractive and in all other ways better than hand built structures.
How many people will this put out of work? Consider it for a moment. What percentage of the workforce is construction?
How about real-estate prices? The new structures will be in all ways superior to hand-build structures and can be built in days for the cost of the materials (such as plastic and concrete). Existing structures will loose their values unless they have some historical or other significance, they will be "antiques". Real-Estate prices will evaporate, approaching the price of the lot only (for all one has to do is purchase a lot and hire out one of these machines for a couple of days as competition).
In residential real-estate, people will knock down their own homes and hire out machines to build a better, bigger, custom made one for them.
The implications are astounding.
Anyone else have thoughts of battery powering a Mac Mini, equipped with airport and an iSight for applications like this?
Seems to me you could plop a Mini and an iSight on just about any remote controlled vehicle and have quite a bit of fun with it. You could use iChat to get high-resolution high-framerate remote video and audio.
Hell, the mini could be the robotic brain for the vehicle as well. This sounds like a great new project - the "Mac Mini robotics project!"
Gotta get me one of those now...
This is the first time I have run across this phrase: "But that's 20th century tech".
It marks a threshold in my mind. Interesting. I will now predict that this phrase "20th century tech" will become part of popular culture, describing shoddy, low-tech, incompatible, bug-prone, poorly designed gadgets that have no style.
The reason that the U.S. is not innovating in the area of energy production has to do with politics. America controls (directly and in the case of Saudi, indirectly) world oil production and therefore world energy. Alternate energy sources, especially those that free nations from the oil addiction reduce dependence on America and therefore reduce America's power.
China, knowing this, is actively persuing alternate energy policy including nuclear, hydrogen and more novel approaches. They want to detach themselves from the oil addiction so that they have independence from the U.S. and U.S. controled energy interests.
Again, politics.
But, the results are inevitable: As a result of these politics, the Chinese will inevitably control more advanced and more important energy technologies (both economicaly and ecologically). So the conclusion to this will be exactly the opposite of that desired by the status quo (America controlled energy). However, the administration doesn't care because they will be retired, rich, fat and happy (or dead of old age) when China turns it all around on America and effectively takes control of world energy production.
Your words are cutting and probably accurate... However, there is a twist:
I agree, the US is over as the preeminent power, its just going to take time for people to realize it. Around 2030 when China and the US face off probably in the Middle East for the remaining easy oil, Americans will get a rare taste of what war is like from the losing side.
I think this is quite well understood in the power-circles of the U.S. I believe the current military actions are being taken specifically to force a worldwide energy monopoly to be in the hands of the U.S. to offset the trade and deficit imbalances. These will probably continue during the Bush presidency.
I also believe that the U.S. will fight before it loses the ability to win, if it looks like it is slipping from it's position of power in the world.
For the sake of our children, world peace and everything holy, let us pray that the U.S. is not slipping out of world-power. It may in fact spell the end of the world as our desperate leaders lash out at the world to keep their power. America would become like Nazi Germany (who by the standards of the time had just as significant a technology and military advantage as the U.S. does today).
Given how poorly the western world handles these issues, i can't imagine how well it'll be done elsewhere...
Why is it that we tend to assume that we (in America - which is what I assume you mean by the Western world) always do things better? So just because we do a poor job of handling our waste, that means China will automatically do a worse job?
Remember, this is a country with over 3000 years of continuous existence, compared to our 200 years. I would suggest to you that they may know more about maintaining their environment and preserving for the future than we do.
Seriously though, what is the incentive for robots to reproduce?
One word: Motivation
Truly intelligent robots will not be deterministic, intelligence does not function like that. Intelligence is an emergent property. Since it cannot be programmed, per sey, providing some "purpose" to the AI's existence becomes an important problem. Without some *hunger* or *drive* the AI would just sit there doing nothing.
Why would this be a post of a "Linux" user switching? This, like many articles on slashdot since Ballmer beat his chest about attacking the OSS community where it lives (uhum... here?) looks to me like a psychologically tuned meme designed to undermine Linux users' pride in their choices...
Let's stick to windows users switching to OS X. We all know that Linux users love it too, but it is highly unlikely that they would "switch" - just buy a mini or a laptop to augment their collection of hardware.
Many have posted objections regarding uplifting other terrestrial species with human DNA. The objections seem rational, as it appears the uplift would be done in order to exploit the results.
What about the reverse, uplifting humans by grafting in superior dna found in the animal kingdom? Better eyes? Superior strength? whatever...
I am curious about your view on this.
For that matter, he could be padding a nest for Carmack and crew to incubate their eggs in.
Nice thought, but this is a war of egos. Carmack has a name that resonates with a generation of gamers. His name has more power than Bezos' already. Do you think Bezos would let himself be overshadowed by the genius of Carmack?
Much more likely that Bezos thinks he can just throw money at the problem until it is solved, contrasted to Carmack's approach of throwing his own proven genius at that same problem.
Personally, my money is on Carmack.
But Stephenson does spin a nice tale, huh? Claiming to be a minor character in a heinlein novel is akin to suggesting that Bezos has a breakthrough in propulsion that will cause a paradigm shift in space technology.... If he wasn't a novelist (and therefore a sensationalist by nature), I might take that claim more seriously (after all, who doesn't like Stephenson?)...
Anyway, the ones you don't hear about (directly) are businessmen. They could give a fuck less about the secret service or NSA or any other organization unless it was easy to fence the data. Thus most people have no reason to fear them.
You appear to be glamorizing the life of a common thief. I hate to break this to you, but anyone who spends their time hacking computers for money is not only without decency, but also without brains.
These guys do eventually get caught, or worse, end up slaves to some crime syndicate. These people DO HURT others through identity theft, credit-card fraud, banking fraud and other malicious activities.
And I dare say, the ones that don't get caught probably don't get caught because of the fierce terrorizing power of the crime-syndicate that they are a part of (e.g. anyone gets close, they are scared away), rather than because of some super-talent. The super-talented guys can make it in the real world.
I have some experience with this, see http://slashdot.org/comments.pl?sid=135763&cid=113 33754
I concluded that it isnt practical for general use, it took too much time to compress an image (alright it was five years ago and so today it probably wouldnt matter), but most importantly there is no easy general compression solution for all images (for instance one that compresses tree pics well wont do faces well and vice versa).
I did some basic expirementation with Genetic Algorithms and fractal compression and I can tell you, GA does solve the problem. Not only solve it, but obliterates it. With GA, fractal compression can achieve compression ratios and quality that are unheard of with other techniques.
Of course, this is to be expected, after all - it is what nature does with us. Our genecode is the compressed image, our bodies are the uncompressed results.
Interesting thought food, huh...
Writing a GA is simple. Writing a GA that delivers results, now that is a different beast alltogether. My experience has shown me that there are some rather more complicated things that need to be taken into account such as:
These statements apply to all search algorithms for parameterized problem domains. No search algorithm is very effective if it is bottled off from significant portions of the domain.
Thank you. We agree. That was exactly my point - the implications are that in a scenario like this, if you allow the search to progress as it should, you will constantly be testing very sub-optimal solutions which will result in negative side-effects to system performance. For that reason, it would be necessary to place heavy restrictions on the percentage of time that the GA is testing new solutions and use other techniques such as hill-climbing to augment the GA.
A GA does exhaustively search a search space, given enough time. It just so happens that the solutions it finds while searching are generally very useful. These solutions may be used immediately and in some cases, you can stop searching further. But if you let the algorithm continue long enough, it WILL SEARCH THE ENTIRE SPACE.
If the GA is not able to reach the entire space or if it is biased in such a way as to make portions of the space hard to reach, then it is not an effective GA. If I wanted to state this in one sentence, I could say "A GA is only truly effective if you let it exhaustively search the search space." This was the exact statement that you took issue to. Please note that nowhere did I state that you had to let the seach finish before the results were useful.
Do you still not understand?
I thought long how the "fittness" part of that GA might work. Ultimately the only "fittness" the kernel can with high precision measure is "load" isn't it?
Any better ideas? I suppose the only way would be to have some "tuning" hooks that applications could use to announce their own performance metrics. The GA could optimize around load when these metrics are not available and take them into account if they are.