That's more of an adblock-equivalent, and I found it lacking when I first tried it (I don't remember my specific complaints, and it may have gotten better). The more direct noscript-equivalent is "NotScripts", which is quite simple and works pretty nicely.
Indeed, I was shooting from the hip a bit. I didn't mean to argue that integer factorization *cannot* be NP-complete, in the same way that I wouldn't argue that NP != PH or P != NP. However, among experts in the field, it's generally expected (not known, not proven) that all those things are true. When I said it would be "surprising", I meant that many well-informed people would be surprised. I like using words to mean what they mean.
Basically, yes, it is possible that P = NP = CoNP = PH, but it's reasonable in many contexts to assume otherwise.
As for NP=CoNP implying NP=PH, that was actually out of my notes for a computational complexity course I took. There's a proof in these lecture notes under Theorem 3 (not the course I took, and I haven't checked the proof in detail, but it looks about right).
If they get quantum computers to work at a useful scale, they'll be near useless for communication (both because they're so expensive and because of the networking problems you mentioned), but will be great for breaking all the encryption that everyone else uses around the world. In short, we need a classical cryptography scheme that's still secure with quantum computing.
Yeah, that's true.
Note: factoring isn't NP-complete! So far there's no reason to believe it's not an "easy" problem, except that we haven't figured out how to do it.
Much like people work under the assumption that factoring is hard, you are working under the assumption that factoring is not NP-Complete. Nobody has proven this either...
That's true, but it's a pretty safe assumption. Integer factorization has been proven to be in both NP and CoNP, so if it's NP-complete that would mean that NP=CoNP. This, in turn, would imply that NP=PH. This would be, suffice it to say, very surprising.
A minor nit: any "hard" problem that's harder one way than the other will ultimately be attackable via quantum methods.
Can you point me toward more information on this? I haven't heard anything like that before -- all arguments I've seen that say quantum computing breaks cryptographic schemes are just based on Shor's algorithm, which I didn't think had such broad implications. (I didn't know it breaks ECC, too.)
Yeah, I didn't know about ECC also being vulnerable (I learned something, too!). The problem with using QKD is that it requires all involved parties to be on a network of quantum computers. The biggest danger I see is when a few people (like the NSA) have quantum computers, but no one else does. If there aren't classical public-key schemes that can stand up to quantum computing, then security as we know it is basically broken, and anyone who wants a real guarantee of privacy will have to resort to one-time pads.
Wait, who didn't know this already? The title is misleading, but the fact that quantum computing breaks RSA is pretty standard knowledge (among people who have heard of quantum computing at all, I guess). Of course, there are other encryption schemes that seem to work just fine (e.g. Elliptic curve cryptography) with quantum computing, and there's not much evidence that algorithms other than RSA are broken. Note: factoring isn't NP-complete! So far there's no reason to believe it's not an "easy" problem, except that we haven't figured out how to do it. More intersetingly, there's a lot of research being done on quantum cryptography, which is really quite cool. In total, quantum computing should probably give us more security than it breaks, except for the idiots who keep using outdated algorithms long after they're broken, but they'd be screwed anyway.
So, the sky is falling! Oh wait, no, that's just the weather changing.
I would say if Watson's programmers are fairly confident he can get more than 50% of the questions correct, all they have to do is program him to buzz in on every question. Could make for a boring Jeopardy game.
...have you ever watched Jeopardy? If Watson buzzes in first on every question and gets 50% of them right, it'll end up somewhere around $0. If it gets a question wrong, it loses money and the other two contestants have a chance to buzz in to answer the question. If the programmers are confident it can get enough questions right for "buzz in on every question" to be a winning strategy, then yeah, it'll win. But if it's that accurate, well, mission accomplished.
You realize we (as in the human race) are really far away from fully emulating ourselves, right? Nobody involved in this project or anything related to the field is fooling themselves into thinking we've developed human-level AI.
In short: human-level AI is really really hard. That the state of the art isn't there yet doesn't mean making improvements isn't a challenge, or that it isn't interesting research.
From what the article describes, all this IS, is a test of the AI's database mining and parsing abilities.
[...]
There's even more that goes into the game. But this won't be a demonstration of AI vs. computer at Jeopardy!, it will be a demonstration of an AI database mining vs. a human, using Jeopardy! style questions and format as a framework.
Um... yeah? That's been pretty clear from the beginning. This is a feat of natural language processing (within pretty well-defined constraints) and information retrieval more than anything else. Who said otherwise?
Depends. Is the computer allowed to use wikipedia (during the show, or somewhere in the past)?
Otherwise, the computer knows only as much as the programmers have taught it.
Asking whether it's allowed to use an archived (or, more likely, well-indexed) copy of wikipedia is like asking whether the human contestants are allowed to remember something they read on wikipedia. There's no question that computers can store more information than humans; that's not what this is testing, and it's probably a fair guess that "Watson" will have the answer to most every question asked. The hard part, however, is parsing the clues and understanding what they're looking for with a reasonable degree of accuracy, and doing so faster than the human contestants. Humans are great at this sort of thing, and it's really hard to write a program that does it at all well.
So wait... how exactly can I view, let alone export, all the personal data that Google has collected on me over the years? What if I want to switch to a different search engine but don't want to lose all the behind the scenes tweaking that can be done with a good decades worth of search history?
I don't know if they provide an export feature, but all the searches you made while logged in are there. I found out I've done 11307 searches! That's actually fewer than I thought... It's pretty interesting to look at what I was searching for 4 years ago.
Yeah, my last comment was basically to say that I don't have a dog in this fight, so I don't particularly care what Ubuntu does. I'll probably continue using Ubuntu on my laptop, as long as it does what I want and there's nothing else that's demonstrably better.
Fortunately, X is not Linux, and Linux is not X, which might help clear up some of your confusion. When someone says "I use Linux!", it means exactly that, and you can't really assume a priori what else they've got. Of course, most Ubuntu users will say "I use Ubuntu!", which should make things easier.
Personally, I rarely do anything that really depends on X being X, so my reaction is essentially "huh, I wonder how that'll work out."
I was assuming we'd just bombard them at range... you know, use the railgun as a railgun. I have to imagine it wouldn't be *that* hard to destroy a star (I have to imagine it because it would just be so awesome).
Or, combining the two ideas (and having more fun with the less realistic scale), have a ton of magnets around the solar system to form a maglev loop! You can start near the sun, then spiral out as you get faster, and then you have essentially a railgun (as the tail of the 6) to aim you out of the solar system.
Of course, we could probably just do the railgun the size of the solar system... when it's that large, you don't need to accelerate very quickly, since you have plenty of time to get to top speed. Also, then we could totally declare war on Alpha Centauri.
If they have CCTV footage of an anonymous person on the street spraying them and running away, it doesn't matter what's in the bottle. They can still successfully use the "crazy guy sprayed me" defense, assuming there's not other strong evidence linking them to the crime.
I'd bet this would be used (well, in a reasonable world) in conjunction with witness reports or camera footage to gather suspects, or to determine the owner of recovered goods. I am curious, though... if you get sprayed accidentally while holding your camera, then your camera gets stolen, will it be returned to the owner of the spray without question?
1. Buy your own unique DNA spray 2. Steal shit 3. Tag the stolen goods with your spray, and maybe scratch off serial numbers or other identifying information 4. Leave them with some homeless guy and report them stolen (maybe spray the homeless guy too) 5. The police notice the homeless guy with professional photography gear matching your description, confiscate it and find the DNA tag 6. The police helpfully deliver your stolen goods to your door 7. Profit!
Probably this isn't nearly worth anyone's time, and there are so many ways it could go wrong, but it seems like it could work...
They claim to be able to encode over 10 billion unique values, so I doubt they plan ever to reuse codes -- they'll improve the product or go out of business long before they get 10 billion sales. The fee is probably to pay for using their lab to check for matches (also, it's far more lucrative than a one-time fee). So... I'm also curious what they do if they find a match that's not being paid for. Ethically, it would be nice if they told the police. From a business perspective, they don't want customers to know they'll still tell police even if you don't pay anymore, so they should keep it quiet. My guess is they remove the code from their database when you stop paying, so they honestly don't know one way or the other.
When will people learn to stop acting suspiciously after they do something they aren't supposed to do?
A large subset of thieves (and many other types of criminals) are also stupid, or have low self-control. If you can control yourself and are reasonably smart, you can probably profit more through various less risky legal means.
Slashdot Mirror
There are more details here.
Personally, I'm just gonna sit back and watch this unfold *grabs popcorn*
Quite honestly, I'm surprised this didn't cause some sort of panic...
That's more of an adblock-equivalent, and I found it lacking when I first tried it (I don't remember my specific complaints, and it may have gotten better). The more direct noscript-equivalent is "NotScripts", which is quite simple and works pretty nicely.
Indeed, I was shooting from the hip a bit. I didn't mean to argue that integer factorization *cannot* be NP-complete, in the same way that I wouldn't argue that NP != PH or P != NP. However, among experts in the field, it's generally expected (not known, not proven) that all those things are true. When I said it would be "surprising", I meant that many well-informed people would be surprised. I like using words to mean what they mean.
Basically, yes, it is possible that P = NP = CoNP = PH, but it's reasonable in many contexts to assume otherwise.
As for NP=CoNP implying NP=PH, that was actually out of my notes for a computational complexity course I took. There's a proof in these lecture notes under Theorem 3 (not the course I took, and I haven't checked the proof in detail, but it looks about right).
That reinforces my point, though :)
If they get quantum computers to work at a useful scale, they'll be near useless for communication (both because they're so expensive and because of the networking problems you mentioned), but will be great for breaking all the encryption that everyone else uses around the world. In short, we need a classical cryptography scheme that's still secure with quantum computing.
Yeah, that's true. Note: factoring isn't NP-complete! So far there's no reason to believe it's not an "easy" problem, except that we haven't figured out how to do it.
Much like people work under the assumption that factoring is hard, you are working under the assumption that factoring is not NP-Complete. Nobody has proven this either...
That's true, but it's a pretty safe assumption. Integer factorization has been proven to be in both NP and CoNP, so if it's NP-complete that would mean that NP=CoNP. This, in turn, would imply that NP=PH. This would be, suffice it to say, very surprising.
A minor nit: any "hard" problem that's harder one way than the other will ultimately be attackable via quantum methods.
Can you point me toward more information on this? I haven't heard anything like that before -- all arguments I've seen that say quantum computing breaks cryptographic schemes are just based on Shor's algorithm, which I didn't think had such broad implications. (I didn't know it breaks ECC, too.)
Yeah, I didn't know about ECC also being vulnerable (I learned something, too!). The problem with using QKD is that it requires all involved parties to be on a network of quantum computers. The biggest danger I see is when a few people (like the NSA) have quantum computers, but no one else does. If there aren't classical public-key schemes that can stand up to quantum computing, then security as we know it is basically broken, and anyone who wants a real guarantee of privacy will have to resort to one-time pads.
Yeah, that's true.
Wait, who didn't know this already? The title is misleading, but the fact that quantum computing breaks RSA is pretty standard knowledge (among people who have heard of quantum computing at all, I guess). Of course, there are other encryption schemes that seem to work just fine (e.g. Elliptic curve cryptography) with quantum computing, and there's not much evidence that algorithms other than RSA are broken. Note: factoring isn't NP-complete! So far there's no reason to believe it's not an "easy" problem, except that we haven't figured out how to do it. More intersetingly, there's a lot of research being done on quantum cryptography, which is really quite cool. In total, quantum computing should probably give us more security than it breaks, except for the idiots who keep using outdated algorithms long after they're broken, but they'd be screwed anyway.
So, the sky is falling! Oh wait, no, that's just the weather changing.
Apparently someone at google labs is a fan of the whole "pirates prevent global warming" joke: http://ngrams.googlelabs.com/graph?content=pirates%2Cninjas&year_start=1800&year_end=2008&corpus=0&smoothing=3
I would say if Watson's programmers are fairly confident he can get more than 50% of the questions correct, all they have to do is program him to buzz in on every question. Could make for a boring Jeopardy game.
...have you ever watched Jeopardy? If Watson buzzes in first on every question and gets 50% of them right, it'll end up somewhere around $0. If it gets a question wrong, it loses money and the other two contestants have a chance to buzz in to answer the question. If the programmers are confident it can get enough questions right for "buzz in on every question" to be a winning strategy, then yeah, it'll win. But if it's that accurate, well, mission accomplished.
You realize we (as in the human race) are really far away from fully emulating ourselves, right? Nobody involved in this project or anything related to the field is fooling themselves into thinking we've developed human-level AI.
In short: human-level AI is really really hard. That the state of the art isn't there yet doesn't mean making improvements isn't a challenge, or that it isn't interesting research.
From what the article describes, all this IS, is a test of the AI's database mining and parsing abilities.
[...]
There's even more that goes into the game. But this won't be a demonstration of AI vs. computer at Jeopardy!, it will be a demonstration of an AI database mining vs. a human, using Jeopardy! style questions and format as a framework.
Um... yeah? That's been pretty clear from the beginning. This is a feat of natural language processing (within pretty well-defined constraints) and information retrieval more than anything else. Who said otherwise?
Watson has some advantages, the humans have some advantages. Why is this a problem?
Depends. Is the computer allowed to use wikipedia (during the show, or somewhere in the past)?
Otherwise, the computer knows only as much as the programmers have taught it.
Asking whether it's allowed to use an archived (or, more likely, well-indexed) copy of wikipedia is like asking whether the human contestants are allowed to remember something they read on wikipedia. There's no question that computers can store more information than humans; that's not what this is testing, and it's probably a fair guess that "Watson" will have the answer to most every question asked. The hard part, however, is parsing the clues and understanding what they're looking for with a reasonable degree of accuracy, and doing so faster than the human contestants. Humans are great at this sort of thing, and it's really hard to write a program that does it at all well.
So wait... how exactly can I view, let alone export, all the personal data that Google has collected on me over the years? What if I want to switch to a different search engine but don't want to lose all the behind the scenes tweaking that can be done with a good decades worth of search history?
https://www.google.com/history/
I don't know if they provide an export feature, but all the searches you made while logged in are there. I found out I've done 11307 searches! That's actually fewer than I thought... It's pretty interesting to look at what I was searching for 4 years ago.
1. Linux is not Unix
2. X is neither part of Unix nor required for it.
Anything else you'd like to add to this discussion?
Nice troll! You managed to choose a topic that is probably as complex and volatile as Kirk vs. Picard, but yet is not as familiar.
Nah, it's pretty well known and accepted that Linux is not Unix. Linux is certainly and undeniably Unix-like, but it's not Unix.
Not really complex. Not really volatile. Not a troll.
Yeah, my last comment was basically to say that I don't have a dog in this fight, so I don't particularly care what Ubuntu does. I'll probably continue using Ubuntu on my laptop, as long as it does what I want and there's nothing else that's demonstrably better.
Fortunately, X is not Linux, and Linux is not X, which might help clear up some of your confusion. When someone says "I use Linux!", it means exactly that, and you can't really assume a priori what else they've got. Of course, most Ubuntu users will say "I use Ubuntu!", which should make things easier.
Personally, I rarely do anything that really depends on X being X, so my reaction is essentially "huh, I wonder how that'll work out."
I was assuming we'd just bombard them at range... you know, use the railgun as a railgun. I have to imagine it wouldn't be *that* hard to destroy a star (I have to imagine it because it would just be so awesome).
Or, combining the two ideas (and having more fun with the less realistic scale), have a ton of magnets around the solar system to form a maglev loop! You can start near the sun, then spiral out as you get faster, and then you have essentially a railgun (as the tail of the 6) to aim you out of the solar system.
Of course, we could probably just do the railgun the size of the solar system... when it's that large, you don't need to accelerate very quickly, since you have plenty of time to get to top speed. Also, then we could totally declare war on Alpha Centauri.
If they have CCTV footage of an anonymous person on the street spraying them and running away, it doesn't matter what's in the bottle. They can still successfully use the "crazy guy sprayed me" defense, assuming there's not other strong evidence linking them to the crime.
I'd bet this would be used (well, in a reasonable world) in conjunction with witness reports or camera footage to gather suspects, or to determine the owner of recovered goods. I am curious, though... if you get sprayed accidentally while holding your camera, then your camera gets stolen, will it be returned to the owner of the spray without question?
1. Buy your own unique DNA spray
2. Steal shit
3. Tag the stolen goods with your spray, and maybe scratch off serial numbers or other identifying information
4. Leave them with some homeless guy and report them stolen (maybe spray the homeless guy too)
5. The police notice the homeless guy with professional photography gear matching your description, confiscate it and find the DNA tag
6. The police helpfully deliver your stolen goods to your door
7. Profit!
Probably this isn't nearly worth anyone's time, and there are so many ways it could go wrong, but it seems like it could work...
They claim to be able to encode over 10 billion unique values, so I doubt they plan ever to reuse codes -- they'll improve the product or go out of business long before they get 10 billion sales. The fee is probably to pay for using their lab to check for matches (also, it's far more lucrative than a one-time fee). So... I'm also curious what they do if they find a match that's not being paid for. Ethically, it would be nice if they told the police. From a business perspective, they don't want customers to know they'll still tell police even if you don't pay anymore, so they should keep it quiet. My guess is they remove the code from their database when you stop paying, so they honestly don't know one way or the other.
When will people learn to stop acting suspiciously after they do something they aren't supposed to do?
A large subset of thieves (and many other types of criminals) are also stupid, or have low self-control. If you can control yourself and are reasonably smart, you can probably profit more through various less risky legal means.