The AQA is the exam board that resulted from the merge of the JMB into a number of other exam boards. The JMB universities (Manchester, Sheffield and Lancaster are the three I recall) who ran the original JMB exam board have exceptionally strong reputations in computer science and software engineering. This doesn't mean much for how it is today, only that historically that was always a good place to look for that kind of skill.
However, for the purposes of the OP's question, the main considerations are that they have a clear syllabus and several years of past exam papers to practice with. In home education, the quality of the instructor is a total unknown, as is the quality of the "IT facility" (read: home computer). Without this information, the best syllabus in the world could be a disaster but then the worst in the world could also be a great starting-point for amazing success. But with known exam papers where the percentage of takers with any given grade is also known, it should be quite easy to get a feel for where a person in in relation to their peers.
My thought: It doesn't matter where you learn or how you learn, the fundamentals are universal.
AQA offers a suggested schooling curriculum and past papers for the exams they set. Sure it's UK not US, but C is C, HTML is HTML, MS Office is MS Office and small furry creatures from alpha centauri make great soup if you put them in the blender for long enough.
But by being the fattest, they have twice the heart disease levels and heart attacks of Britain (which, in turn, is obese compared to much of Europe). This costs money. If you look at the total cost of those burgers, rather than the cost at the till, it stops looking so good. It's just that you're paying it via assorted taxes, insurance charges, co-pay fees, etc, so it's all invisible. Well, it's a bit worse. Junk food can be iron-deficient, damaging your concentration and eventually your brain. Reduced job market value is also a cost. Junk food also limits the immune system's ability to cope, which means even normal sicknesses are more common. Absenteeism, especially for contractors, is a huge cost. But again, it's all deferred and spread over many different things with no easy way for someone to find out exactly what those costs were.
If you could somehow add up all the hidden costs and add them to the burger, it would stop being $1 and start being damn scary.
Of course, the law of diminishing returns says that you'll somewhere hit a point that isn't great but is still very respectable whilst also being close enough to cheap. I have no idea what that point would be or what life would be like if that were the standard of eating that was considered ok for those on welfare or minimum wage. Bet it would be a good deal more interesting than life is today.
Jamie Oliver demonstrated by switching a school's menu that a poor diet causes the masses to become brainless. (The improved diet, once accepted, caused exam scores to skyrocket and absence to plumet. After that, both media and schools started taking his views a bit more seriously - except in LA, where he was banned.)
It follows that you've a self-perpetuating cycle. People on heavily-processed, factory-farmed diets will, in general, be too stupid - as a direct result of those diets - to change.
If the bucket averages out at $3 but the medical bills afterwards average out at $30 (because of superbugs, toxicity from avian meds, or whatever), then $30 per bucket with no medical bills afterwards is a save. It doesn't mean that this is the case, what it does mean is that there are cases when $30 for a KFC bucket really IS a win for the consumer, no matter what it feels like at the register.
No more so than Sugar is. One of the great things about heterogeneous environments is that you tend to see specialization and optimization rather than outright competition (which just drains resources from everyone). Everyone gets a perfect whatever-they-want rather than half-baked one-size-fits-all solutions (which, by trying to compete with everyone, will always lose against everyone).
GUIs are overhead in a lot of areas. Graphics can be done via framebuffers, for example, and if you're simply wanting to flip through a photo album you're far better off doing it that way than to have the overheads of X in there. On the other hand, I'd hate to do video editing in a console app.
It's like arguing whether axes are better than arrows. Even with a 170 lb. longbow, it'll take time to cut down a tree, but an archer will still beat an axe-wielding maniac.
I use WindowMaker, Enlightenment, AfterSTEP, FVWM2, OpenLook, KDE, Gnome and the one that's an Archimedes GUI ripoff. Not usually at the same time, though that's happened, but whichever one happens to fit my mood and/or fits with whatever I'm doing better. I can't understand those who live in a single GUI. It's like trying to live in a single country.
Oh, and yes, to answer the other question, it does. Every component in a crypto system can be attacked, so having a sifter/sieve algorithm plus N proxies adds N+2 components that are potential targets for attack. Since the proxies decrypt and re-encrypt, if the proxies are broken then the plaintext is readable. If the sieve or sifter are broken, the combinatorial explosion is of no significance since the attacker will look only at the channels of interest.
No, I'm proposing that the network topology can essentially add bits to an encryption key when you're already at the upper limit of what the algorithm itself can take without harming security. If the problem of decryption can be made exponentially harder for each node added, then one node equals some fixed number of bits (probably 1). If you can make the problem increase in severity faster than that, then every node added adds more bits to the key than the last.
And that's the key question. If you can only do exponential growth at the very best, and quantum computers can beat you now, then quantum computers will always keep pace. It's only when you get into the hyperexponential territory that you can guarantee, flat-out, that you can make RSA-type problems worse than can be solved by quantum computers.
It's not really security through obscurity, since it isn't the algorithms that are secret, it's the seed information (which is just a different kind of shared key and shared keys aren't considered as security through obscurity).
A mitm would need to have N private keys (one to decrypt each proxy, since each uses a different key pair to talk to the recipient). But how to identify the keys? The problem with brute-forcing a key is that you've got to know what the message's general form has to be versus what a failed decode looks like. With a one-time pad, all messages of the maximum length or less are equally possible and you can't tell which one is the real one. In order to make RSA act like a one-time pad, you have to have a way to make it impossible to distinguish a correct key from an incorrect key.
How to do this? Well, I'm leaning heavily on the so-called Byzantine General's Problem. In short, I'm treating each channel as a node in the BGP that is acting on behalf of the attacker by carrying information to them and then making the majority of those channels "traitors". The practical upshot is that the attacker can't be guaranteed able to identify real keys when false messages might yield apparently real keys and real messages (since they're compressed) might not.
The idea is to swamp the qc with too many possibilities, too many avenues to go down. If you can't swamp with key length, drown in options.
Please! This is Slashdot. They're channeling William Shakespere (who wrote all of Shakespere's plays under the pseudonym of William Shakespere, to deprive Francis Bacon of recognition as the author).
One way you -might- be able to securely exchange keys using an RSA-style algorithm would be to chop the keys up. What I'm thinking is this:
N proxy servers, of which M receive part of the key using RSA or something similar. The rest get something random, but it's also encrypted using the same method. Each server then transmits what they get to the intended recipient the same way. You'd need some algorithm to generate what combination is the correct one (which then becomes vulnerable to attack itself) and some covert method of exchanging the seed information (less data than a OTP but still a serious problem) or the method is useless.
For the actual data, you have the data block also split into M fragments, with the N-M remaining proxies getting random data.
Since you cannot distinguish between a failed decrypt of a valid assemblage of packets and a valid decrypt of an invalid assemblage of packets, the number of possible combinations to try goes up factorially. For a large enough N and a small enough M, it should be possible to keep such a system ahead of quantum computers as it grows faster than exponentially.
The other problem is species. It wasn't that long ago that genetic research forced the reclassification of northern right whales into two distinct species (essentially slashing the population size). I don't know how much research has been done on pilot whales - not much, if it's as data-deficient as it appears to be - and it may well prove that there's only one distinct species there. On the other hand, if there are N+1 species or subspecies then there's a maximum of 1/(2^N) of the estimated numbers in the smallest-numbered group.
That the northern right whale split was such a surprise (it was only discovered around 2000, long after the discovery that dorsal fins uniquely identify cetaceans and certainly after the invention of photography) tells me that the assumptions being made aren't always being checked very carefully.
The numbers probably are sustainable, provided N=0 or 1 (ie: 1 or 2 species), not convinced the numbers are as good if N is any larger.
The ethical issues get a bit trickier. It's observed that bottlenose dolphins and orcas can manufacture tools, work with intermediate languages when pods gather, and can even operate touchscreen panels with purpose (well, orcas might well use porpoise sometimes, they're like that). That starts getting into cloudy issues, especially as you can't really do similar studies on whales very easily. What sized screen would you use for a blue whale, anyway? The Japanese argument that this makes cetaceans no smarter than a dog just doesn't hold up. Dogs don't manufacture tools and the only thing they're likely to see touchscreens as good for is as a chewtoy. However, we can't compare species directly, research is way too primitive to draw any meaningful conclusions, and even if we could, we've no framework for defining when something shouldn't be a food item for intellectual reasons, only environmental ones.
The last estimate the IWC did for pilot whales was 1989 and they calculated between 440,000 - 1,370,000. Which, in my books, translates to "we haven't a clue". MarineBio.org lists their population as "unknown but not considered endangered". Which also translates to "we haven't a clue".
Whale populations are extremely difficult to estimate, though the Japanese method of hauling them ashore and counting them one at a time is probably not the best.
Occam still exists. KROC supports the latest version of Occam (Occam-Pi), which supports mobile processes, generics and other concepts acquired from the developments in serial programming and computer clusters.
I consider it to be one of the finest languages out there for learning parallel programming and consider that most of the modern failures in the field are a result of people not knowing it and therefore not knowing the fundamentals. You can't run if you can't walk.
The transputer was the ultimate in computer cluster technology - a serial line is all it took. No switches, no routers, no expensive hardware, just raw mesh topology. It was a very sad day when it was sold to ST Electronics. It still exists, but is used only in multimedia appliances these days. A bit of a comedown for a chip that threatened to crush Cray.
Slashdot Mirror
If Alpha Centuran vegetables are unavailable, you can substitute those from Proxima Centauri. Do not, however, use root vegetables from Cygnus X1.
The AQA is the exam board that resulted from the merge of the JMB into a number of other exam boards. The JMB universities (Manchester, Sheffield and Lancaster are the three I recall) who ran the original JMB exam board have exceptionally strong reputations in computer science and software engineering. This doesn't mean much for how it is today, only that historically that was always a good place to look for that kind of skill.
However, for the purposes of the OP's question, the main considerations are that they have a clear syllabus and several years of past exam papers to practice with. In home education, the quality of the instructor is a total unknown, as is the quality of the "IT facility" (read: home computer). Without this information, the best syllabus in the world could be a disaster but then the worst in the world could also be a great starting-point for amazing success. But with known exam papers where the percentage of takers with any given grade is also known, it should be quite easy to get a feel for where a person in in relation to their peers.
A GATTAling gun - used to shoot holes in genetics theories.
Good advice.
My thought: It doesn't matter where you learn or how you learn, the fundamentals are universal.
AQA offers a suggested schooling curriculum and past papers for the exams they set. Sure it's UK not US, but C is C, HTML is HTML, MS Office is MS Office and small furry creatures from alpha centauri make great soup if you put them in the blender for long enough.
What about the Nukem part? :)
But by being the fattest, they have twice the heart disease levels and heart attacks of Britain (which, in turn, is obese compared to much of Europe). This costs money. If you look at the total cost of those burgers, rather than the cost at the till, it stops looking so good. It's just that you're paying it via assorted taxes, insurance charges, co-pay fees, etc, so it's all invisible. Well, it's a bit worse. Junk food can be iron-deficient, damaging your concentration and eventually your brain. Reduced job market value is also a cost. Junk food also limits the immune system's ability to cope, which means even normal sicknesses are more common. Absenteeism, especially for contractors, is a huge cost. But again, it's all deferred and spread over many different things with no easy way for someone to find out exactly what those costs were.
If you could somehow add up all the hidden costs and add them to the burger, it would stop being $1 and start being damn scary.
Of course, the law of diminishing returns says that you'll somewhere hit a point that isn't great but is still very respectable whilst also being close enough to cheap. I have no idea what that point would be or what life would be like if that were the standard of eating that was considered ok for those on welfare or minimum wage. Bet it would be a good deal more interesting than life is today.
Jamie Oliver demonstrated by switching a school's menu that a poor diet causes the masses to become brainless. (The improved diet, once accepted, caused exam scores to skyrocket and absence to plumet. After that, both media and schools started taking his views a bit more seriously - except in LA, where he was banned.)
It follows that you've a self-perpetuating cycle. People on heavily-processed, factory-farmed diets will, in general, be too stupid - as a direct result of those diets - to change.
If the bucket averages out at $3 but the medical bills afterwards average out at $30 (because of superbugs, toxicity from avian meds, or whatever), then $30 per bucket with no medical bills afterwards is a save. It doesn't mean that this is the case, what it does mean is that there are cases when $30 for a KFC bucket really IS a win for the consumer, no matter what it feels like at the register.
Can I use the usual customers in McDonalds' as proof that something brain-damaging gets into the food chain through these farms?
Given that recent lawsuits against cigarette makers have been lost, there seem plenty of people willing to believe they're harmless, even today.
No more so than Sugar is. One of the great things about heterogeneous environments is that you tend to see specialization and optimization rather than outright competition (which just drains resources from everyone). Everyone gets a perfect whatever-they-want rather than half-baked one-size-fits-all solutions (which, by trying to compete with everyone, will always lose against everyone).
GUIs are overhead in a lot of areas. Graphics can be done via framebuffers, for example, and if you're simply wanting to flip through a photo album you're far better off doing it that way than to have the overheads of X in there. On the other hand, I'd hate to do video editing in a console app.
It's like arguing whether axes are better than arrows. Even with a 170 lb. longbow, it'll take time to cut down a tree, but an archer will still beat an axe-wielding maniac.
I use WindowMaker, Enlightenment, AfterSTEP, FVWM2, OpenLook, KDE, Gnome and the one that's an Archimedes GUI ripoff. Not usually at the same time, though that's happened, but whichever one happens to fit my mood and/or fits with whatever I'm doing better. I can't understand those who live in a single GUI. It's like trying to live in a single country.
Oh, and yes, to answer the other question, it does. Every component in a crypto system can be attacked, so having a sifter/sieve algorithm plus N proxies adds N+2 components that are potential targets for attack. Since the proxies decrypt and re-encrypt, if the proxies are broken then the plaintext is readable. If the sieve or sifter are broken, the combinatorial explosion is of no significance since the attacker will look only at the channels of interest.
No, I'm proposing that the network topology can essentially add bits to an encryption key when you're already at the upper limit of what the algorithm itself can take without harming security. If the problem of decryption can be made exponentially harder for each node added, then one node equals some fixed number of bits (probably 1). If you can make the problem increase in severity faster than that, then every node added adds more bits to the key than the last.
And that's the key question. If you can only do exponential growth at the very best, and quantum computers can beat you now, then quantum computers will always keep pace. It's only when you get into the hyperexponential territory that you can guarantee, flat-out, that you can make RSA-type problems worse than can be solved by quantum computers.
It's not really security through obscurity, since it isn't the algorithms that are secret, it's the seed information (which is just a different kind of shared key and shared keys aren't considered as security through obscurity).
A mitm would need to have N private keys (one to decrypt each proxy, since each uses a different key pair to talk to the recipient). But how to identify the keys? The problem with brute-forcing a key is that you've got to know what the message's general form has to be versus what a failed decode looks like. With a one-time pad, all messages of the maximum length or less are equally possible and you can't tell which one is the real one. In order to make RSA act like a one-time pad, you have to have a way to make it impossible to distinguish a correct key from an incorrect key.
How to do this? Well, I'm leaning heavily on the so-called Byzantine General's Problem. In short, I'm treating each channel as a node in the BGP that is acting on behalf of the attacker by carrying information to them and then making the majority of those channels "traitors". The practical upshot is that the attacker can't be guaranteed able to identify real keys when false messages might yield apparently real keys and real messages (since they're compressed) might not.
The idea is to swamp the qc with too many possibilities, too many avenues to go down. If you can't swamp with key length, drown in options.
This was answered in I think it was the fourth season of the revamped Doctor Who.
Please! This is Slashdot. They're channeling William Shakespere (who wrote all of Shakespere's plays under the pseudonym of William Shakespere, to deprive Francis Bacon of recognition as the author).
One way you -might- be able to securely exchange keys using an RSA-style algorithm would be to chop the keys up. What I'm thinking is this:
N proxy servers, of which M receive part of the key using RSA or something similar. The rest get something random, but it's also encrypted using the same method. Each server then transmits what they get to the intended recipient the same way. You'd need some algorithm to generate what combination is the correct one (which then becomes vulnerable to attack itself) and some covert method of exchanging the seed information (less data than a OTP but still a serious problem) or the method is useless.
For the actual data, you have the data block also split into M fragments, with the N-M remaining proxies getting random data.
Since you cannot distinguish between a failed decrypt of a valid assemblage of packets and a valid decrypt of an invalid assemblage of packets, the number of possible combinations to try goes up factorially. For a large enough N and a small enough M, it should be possible to keep such a system ahead of quantum computers as it grows faster than exponentially.
It would be better if you use TLS, though, as it overcomes some of the problems with SSL.
The other problem is species. It wasn't that long ago that genetic research forced the reclassification of northern right whales into two distinct species (essentially slashing the population size). I don't know how much research has been done on pilot whales - not much, if it's as data-deficient as it appears to be - and it may well prove that there's only one distinct species there. On the other hand, if there are N+1 species or subspecies then there's a maximum of 1/(2^N) of the estimated numbers in the smallest-numbered group.
That the northern right whale split was such a surprise (it was only discovered around 2000, long after the discovery that dorsal fins uniquely identify cetaceans and certainly after the invention of photography) tells me that the assumptions being made aren't always being checked very carefully.
The numbers probably are sustainable, provided N=0 or 1 (ie: 1 or 2 species), not convinced the numbers are as good if N is any larger.
The ethical issues get a bit trickier. It's observed that bottlenose dolphins and orcas can manufacture tools, work with intermediate languages when pods gather, and can even operate touchscreen panels with purpose (well, orcas might well use porpoise sometimes, they're like that). That starts getting into cloudy issues, especially as you can't really do similar studies on whales very easily. What sized screen would you use for a blue whale, anyway? The Japanese argument that this makes cetaceans no smarter than a dog just doesn't hold up. Dogs don't manufacture tools and the only thing they're likely to see touchscreens as good for is as a chewtoy. However, we can't compare species directly, research is way too primitive to draw any meaningful conclusions, and even if we could, we've no framework for defining when something shouldn't be a food item for intellectual reasons, only environmental ones.
So it's like fuzzy logic, only they got tired of having muppets run the IT department?
No, since you can't crack non-quantum one-time pad encryption without the encryption pad.
Spellcheckers don't usually help with grammar.
The last estimate the IWC did for pilot whales was 1989 and they calculated between 440,000 - 1,370,000. Which, in my books, translates to "we haven't a clue". MarineBio.org lists their population as "unknown but not considered endangered". Which also translates to "we haven't a clue".
Whale populations are extremely difficult to estimate, though the Japanese method of hauling them ashore and counting them one at a time is probably not the best.
Occam still exists. KROC supports the latest version of Occam (Occam-Pi), which supports mobile processes, generics and other concepts acquired from the developments in serial programming and computer clusters.
I consider it to be one of the finest languages out there for learning parallel programming and consider that most of the modern failures in the field are a result of people not knowing it and therefore not knowing the fundamentals. You can't run if you can't walk.
The transputer was the ultimate in computer cluster technology - a serial line is all it took. No switches, no routers, no expensive hardware, just raw mesh topology. It was a very sad day when it was sold to ST Electronics. It still exists, but is used only in multimedia appliances these days. A bit of a comedown for a chip that threatened to crush Cray.