The first time I encountered work like this was in A.K. Dewdney's column "Simulated Evolution" in Scientific American, May 1989. He presented the program "Palmiter's Protozoa", of which a nice implementation can be found here.
But this is all kid's stuff. Such experiments can be much more interesting nowadays, with the power of computers as we have now. A student of mine studied the evolution of morals in a similar society. His program isn't online yet (will be soon, I guess), but his thesis is.
So assuming that the European Union adopts a more "lenient" patent law, what will happen when American companies start suing europeans because they are "violating" their god like patents?
It depends. Suppose a European firm tries to sell a product in the US that violates some US software patent. Then the firm can be sued in the US. Or, if a European firm has a subsidiary in the US which produces software that violates a US software patent, even if the product is meant for the European market. Again, the firm can be sued in the US. However, as long as the firm is located outside the US, and does not export to the US, it is basically safe (except maybe for a "nukular" threat, but I suppose not even W will go that far).
The funny thing is that a European firm that develops a "new" software concept might get it patented in the US. The net result, without software patents in Europe, is that in Europe many competing businesses might create products based on this "idea", while in the US there will be only one firm that is allowed to sell this product. Imagine what will happen to the price and quality of such a product. I expect that US citizens will get mightly jealous of the great software Europeans are allowed to use for little money, while the US is stuck with one piece of expensive crap. Maybe then the US will get rid of its software patents.
Ah yes, so now the woman has to show in court that astrology really works -- which she can't, of course. Unfortunately, her answer will be that the fact that it doesn't work validates her claim: NASA changed the universe so that she cannot really predict anything anymore. And of course she will bring in a string of witnesses who will claim that her predictions before this date were faultless.
Actually, lawsuits by crackpots are not uncommon. Especially producers of "alternative" medicine are prone to sueing people who state that their product doesn't work. Normally, such a case is judged on the fact whether it can be shown objectively that the defendants statements were false. Science is objective, and such cases are therefore usually resolved by a loss of the crackpot (since alternative medicine is alternative because science has shown that it doesn't really work). I think in this case the same reasoning applies: the question is whether it can be shown objectively that astrology works, and that thus the cook was damaged. As science has shown time and again, astrology is garbage, so the cook will not be able to cash.
Imagine someone invents a new Go algorithm that, when implemented, beats Go masters. I think any CS researcher or Go player would say is a major breakthrough.
Interesting, since I actually am a CS researcher who works in games (not Go, but some of my colleagues work on Go, and I do know a lot about this). I can therefore safely say that, yes, if you invent a new algorithm that significantly improves Go gameplay (even if it doesn't beat Go masters), researchers in this field will consider it a major breakthrough. The question is whether this is (or should be) a patentable breakthrough. I guess that would depend on the algorithm. My personal opinion is that any algorithm is a mathematical construct, and therefore shouldn't be patented, but I admit that we get into a gray area when the algorithm is actually completely novel.
Now say this algorithm is a pattern matcher, a static evaluation function. Given the current board, the algorithm chooses the next move.
This is a bit ambiguous, since any Go evaluation function can be used to choose a next move (when combined with minimax). But lets go on to the main point.
Following your reasoning, this would be a pattern matcher, and so could not be novel. Is this correct?
"A pattern matcher" is too broad. It depends on the kind of pattern matcher. Is it a neural network? Not patentable. Is it a classifier system? Not patentable. Is it... I don't know, something completely new, that is not only applicable to Go, but also to many other problem domains where people haven't found this before... Perhaps.
In my view, if you take something that already exists in some domain, and apply it to another domain, it is silly to allow it to be patented, because there is no advance in the state of the art. Basically, such a patent would allow someone to reap the profits of someone else's work. If, however, you design something completely new, it will ALWAYS have more than one application (at least in computer science), and a patent might be in order. The chance of that happening is very small, though, since in computer science researchers heavily depend on the morcels of knowledge provided to them by their peers.
Most of the AI in CBR is in 1) choosing the right representation, and 2) coming up with the right similarity metric.
In theoretical work, yes. In practice, this is no requirement. And, as I said in another post, they are basically getting a patent on a similarity function.
Instead, they applied statistical techniques to come up with the right similarity metric.
OK, then it's data-mining instead of CBR. Big deal. Data-mining and CBR are closely related anyway.
I don't think CBR circa 1999 was applied to commercial transaction data in order to generate sales recommendations.
And that's just my point. Patents shouldn't be about applying a standard technique to a specific problem. They should be about the standard techniques.
Software patents very often consist of a description of a standard technique, with "on the Internet" attached to it. Such an approach shouldn't be patentable, because of obviousness, and because of prior art.
I recently had to make a list of games that I really enjoyed playing. I found that there were about one or two games per year that I thought were great (since about 1985).
I replay older games. The reason is that there are simply not enough good new games to hold my attention (and I play for about an hour a day, which is not a lot). I rather go back to a classic which I know is good, than that I let myself be bored with the new stuff.
So there is good new stuff, but not enough to occupy all my gaming time. However, the same held when the old stuff was still new. No changes there.
Currently replaying Baldur's Gate, did the Monkey Island games before that. Now if I could only get the original System Shock to run on my new machine...
So, while I still don't think the patent is a good idea, it does seem to be specific enough to be called unique.
Basically, this is a patent on a case-based reasoning application. Its uniqueness is in the specification of the similarity function. However, any case-based reasoning application has to specify a good similarity function, or it won't do its job.
So, did they get a patent on case-based reasoning? That seems ridiculous to me.
Did they then get a patent on a similarity function? What is innovative about that? They have a problem domain and must provide a good function to rate cases in that domain. It is the domain that determines what makes a good function. Give a bunch of programmers and/or mathematicians the job of finding a similarity function for this domain, and what Amazon patented is what they will quickly come up with. So, in this case obviousness should invalidate the patent.
Also, if they got a patent on a similarity function, then anyone who uses a slightly different similarity function should be safe. But, of course, Amazon will claim that they now own the whole approach of case-based reasoning for recommending products.
It is not difficult to make something unique by adding details, such as a specific similarity function. However, uniqueness is no guarantee for non-obviousness.
At the time the patent was filed, it was extremely uncommon for systems to make automatic recommendations based solely on the behavior of users.
That may be true, but in 1999 case-based reasoning (the technique of using past experiences to predict a future outcome) was already old. Basically, this patent covers a specific, very simple, case-based reasoning application. In my opinion, because case-based reasoning exists and is well-documented, any application of case-based reasoning is "obvious", and should not be eligible for patenting. Of course, the nincompoops at the USPTO know bugger-all about computer-science techniques, so for them this probably seems "innovative".
The first main difference between Microsoft and the Internet, and Microsoft and Security, is the fact that IE was a product separate from Windows, while security pervades all of Windows. It is a lot easier to add something than it is to modify it.
The second main difference is that Microsoft could basically BUY themselves the Internet. They just acquired products and companies which went in the direction they wanted to go, slapped a MS logo on their products, and were in business. Security is something they have to solve completely in-house (unless they are willing to release their source code, of course, but that would probably lead to a major DROP in security).
And the third main difference is that adding the Internet does not require a quality increase. Making Windows secure does.
Any poker player will tell you bluffing is where it's at. Without bluffing you play the odds and it just becomes a simple game of chance.
Not (entirely) true. I have followed research in poker programs and I can tell you that programs that bluff play consistently worse than programs that don't. That is, if bluffing is interpreted as "rating your own hand as a lot better than it actually is, in the hope of getting your opponent to fold". The problem is that you are overbetting your hand when you bluff, and if you get called on it, you lose BIG.
The trick is not to just bluff, but to bluff when the hand is worth it, and when you are fairly certain your opponent will fall for it. This means that a good poker program is not one that bluffs, but one that is very good at opponent modelling.
Opponent modelling is in a completely different ballpark than chance bluffing. While there is some research into this, not much has been published. That is why research groups as the one of Jonathan Schaeffer (mentioned in the article) have a far better chance of winning the prize than individuals working on poker programs. I know where I bet my money.
But lesser violations were far more common, including 4.7 percent who admitted to publishing the same data in two or more publications
Only 4.7 percent? Ridiculous. Every scientist publishes the same data in two or more publications. Not only do they have to do that to meet preset quantities of publications (which are ridiculously high), but it is also a way to get more attention for their work. Furthermore, I have found that journals and conferences often approach me to ask for a derivative of an article I wrote somewhere else. What could possibly be wrong about writing two aricles on the same data?
It will be easier in the next game, because it will have toilets, which the creature will recognise and use. Instead, you should now teach it to flush afterwards, and not leave the seat up (unless you play Evil, in which case you SHOULD leave the seat up).
Hogwash. Splinter Cell is completely different. I love Thief. I hate Splinter Cell. Why? Spinter Cell requires complex button mashing to execute exactly the right moves, which, for me, means that I have to save before nearly every action because I will fall to my death with more than 50% probability. Furthermore, Splinter Cell is annoyingly linear, while Thief is Freedom Itself.
We know M$ gets aways with all sort of things, so EVEN if this guy is taking advantage of software patents and is intentions are bad, he WON against the proverbial 800-pound gorilla. That's something to CELEBRATE. It tells all those little guys somewhere, that there's hope on their processes. Yes, you CAN beat a giant at their own game.
I don't think Microsoft really lost here.
Think with me: if Microsoft really wanted to win this case, they would just appeal. They've got the money, they've got the lawyers, there is NO WAY Joe Smallpotatoes would win in the end. Especially not this ridiculous patent, which should be easy to overthrow on the grounds of obviousness.
So, I can only conclude that Microsoft is actually happy to lose this one. And why would that be? My guess is that they simply have lots of these obvious patents themselves, which they hope to apply tactically in the near future to bring down small entrepreneurs. Since they now lost this case, in the future, when someone they sue tries to tell the judge that a Microsoft patent is obvious, Microsoft can reply by pointing out the historic case in which a judge upheld a similar patent, which is therefore non-obvious.
This is a tactical loss for Microsoft. And I see a bleak future.
Slashdot Mirror
But this is all kid's stuff. Such experiments can be much more interesting nowadays, with the power of computers as we have now. A student of mine studied the evolution of morals in a similar society. His program isn't online yet (will be soon, I guess), but his thesis is.
What, with the cynical slashdot crowd?
It depends. Suppose a European firm tries to sell a product in the US that violates some US software patent. Then the firm can be sued in the US. Or, if a European firm has a subsidiary in the US which produces software that violates a US software patent, even if the product is meant for the European market. Again, the firm can be sued in the US. However, as long as the firm is located outside the US, and does not export to the US, it is basically safe (except maybe for a "nukular" threat, but I suppose not even W will go that far).
The funny thing is that a European firm that develops a "new" software concept might get it patented in the US. The net result, without software patents in Europe, is that in Europe many competing businesses might create products based on this "idea", while in the US there will be only one firm that is allowed to sell this product. Imagine what will happen to the price and quality of such a product. I expect that US citizens will get mightly jealous of the great software Europeans are allowed to use for little money, while the US is stuck with one piece of expensive crap. Maybe then the US will get rid of its software patents.
Dammit! Not being a native English speaker catches up with me once again!
Ah yes, so now the woman has to show in court that astrology really works -- which she can't, of course. Unfortunately, her answer will be that the fact that it doesn't work validates her claim: NASA changed the universe so that she cannot really predict anything anymore. And of course she will bring in a string of witnesses who will claim that her predictions before this date were faultless.
Actually, lawsuits by crackpots are not uncommon. Especially producers of "alternative" medicine are prone to sueing people who state that their product doesn't work. Normally, such a case is judged on the fact whether it can be shown objectively that the defendants statements were false. Science is objective, and such cases are therefore usually resolved by a loss of the crackpot (since alternative medicine is alternative because science has shown that it doesn't really work). I think in this case the same reasoning applies: the question is whether it can be shown objectively that astrology works, and that thus the cook was damaged. As science has shown time and again, astrology is garbage, so the cook will not be able to cash.
Yes, and when a tree has been struck by lightning once, it has less chance to be struck by lightning again.
By this reasoning, people who never make a mistake have a greater chance to make a mistake than people who make mistakes all the time.
Interesting, since I actually am a CS researcher who works in games (not Go, but some of my colleagues work on Go, and I do know a lot about this). I can therefore safely say that, yes, if you invent a new algorithm that significantly improves Go gameplay (even if it doesn't beat Go masters), researchers in this field will consider it a major breakthrough. The question is whether this is (or should be) a patentable breakthrough. I guess that would depend on the algorithm. My personal opinion is that any algorithm is a mathematical construct, and therefore shouldn't be patented, but I admit that we get into a gray area when the algorithm is actually completely novel.
Now say this algorithm is a pattern matcher, a static evaluation function. Given the current board, the algorithm chooses the next move.
This is a bit ambiguous, since any Go evaluation function can be used to choose a next move (when combined with minimax). But lets go on to the main point.
Following your reasoning, this would be a pattern matcher, and so could not be novel. Is this correct?
"A pattern matcher" is too broad. It depends on the kind of pattern matcher. Is it a neural network? Not patentable. Is it a classifier system? Not patentable. Is it... I don't know, something completely new, that is not only applicable to Go, but also to many other problem domains where people haven't found this before... Perhaps.
In my view, if you take something that already exists in some domain, and apply it to another domain, it is silly to allow it to be patented, because there is no advance in the state of the art. Basically, such a patent would allow someone to reap the profits of someone else's work. If, however, you design something completely new, it will ALWAYS have more than one application (at least in computer science), and a patent might be in order. The chance of that happening is very small, though, since in computer science researchers heavily depend on the morcels of knowledge provided to them by their peers.
In theoretical work, yes. In practice, this is no requirement. And, as I said in another post, they are basically getting a patent on a similarity function.
Instead, they applied statistical techniques to come up with the right similarity metric.
OK, then it's data-mining instead of CBR. Big deal. Data-mining and CBR are closely related anyway.
I don't think CBR circa 1999 was applied to commercial transaction data in order to generate sales recommendations.
And that's just my point. Patents shouldn't be about applying a standard technique to a specific problem. They should be about the standard techniques.
Software patents very often consist of a description of a standard technique, with "on the Internet" attached to it. Such an approach shouldn't be patentable, because of obviousness, and because of prior art.
I replay older games. The reason is that there are simply not enough good new games to hold my attention (and I play for about an hour a day, which is not a lot). I rather go back to a classic which I know is good, than that I let myself be bored with the new stuff.
So there is good new stuff, but not enough to occupy all my gaming time. However, the same held when the old stuff was still new. No changes there.
Currently replaying Baldur's Gate, did the Monkey Island games before that. Now if I could only get the original System Shock to run on my new machine...
Basically, this is a patent on a case-based reasoning application. Its uniqueness is in the specification of the similarity function. However, any case-based reasoning application has to specify a good similarity function, or it won't do its job.
So, did they get a patent on case-based reasoning? That seems ridiculous to me.
Did they then get a patent on a similarity function? What is innovative about that? They have a problem domain and must provide a good function to rate cases in that domain. It is the domain that determines what makes a good function. Give a bunch of programmers and/or mathematicians the job of finding a similarity function for this domain, and what Amazon patented is what they will quickly come up with. So, in this case obviousness should invalidate the patent.
Also, if they got a patent on a similarity function, then anyone who uses a slightly different similarity function should be safe. But, of course, Amazon will claim that they now own the whole approach of case-based reasoning for recommending products.
It is not difficult to make something unique by adding details, such as a specific similarity function. However, uniqueness is no guarantee for non-obviousness.
That may be true, but in 1999 case-based reasoning (the technique of using past experiences to predict a future outcome) was already old. Basically, this patent covers a specific, very simple, case-based reasoning application. In my opinion, because case-based reasoning exists and is well-documented, any application of case-based reasoning is "obvious", and should not be eligible for patenting. Of course, the nincompoops at the USPTO know bugger-all about computer-science techniques, so for them this probably seems "innovative".
Why not Scotland Yard? It's IDEAL to be played in the centre of London!
Real men do NOT ask for directions!
Real men rely on technology.
The first main difference between Microsoft and the Internet, and Microsoft and Security, is the fact that IE was a product separate from Windows, while security pervades all of Windows. It is a lot easier to add something than it is to modify it.
The second main difference is that Microsoft could basically BUY themselves the Internet. They just acquired products and companies which went in the direction they wanted to go, slapped a MS logo on their products, and were in business. Security is something they have to solve completely in-house (unless they are willing to release their source code, of course, but that would probably lead to a major DROP in security).
And the third main difference is that adding the Internet does not require a quality increase. Making Windows secure does.
Maybe because it would give slightly unethical programmers a reason to deliberately put bugs in the code?
Not (entirely) true. I have followed research in poker programs and I can tell you that programs that bluff play consistently worse than programs that don't. That is, if bluffing is interpreted as "rating your own hand as a lot better than it actually is, in the hope of getting your opponent to fold". The problem is that you are overbetting your hand when you bluff, and if you get called on it, you lose BIG.
The trick is not to just bluff, but to bluff when the hand is worth it, and when you are fairly certain your opponent will fall for it. This means that a good poker program is not one that bluffs, but one that is very good at opponent modelling.
Opponent modelling is in a completely different ballpark than chance bluffing. While there is some research into this, not much has been published. That is why research groups as the one of Jonathan Schaeffer (mentioned in the article) have a far better chance of winning the prize than individuals working on poker programs. I know where I bet my money.
But lesser violations were far more common, including 4.7 percent who admitted to publishing the same data in two or more publications
Only 4.7 percent? Ridiculous. Every scientist publishes the same data in two or more publications. Not only do they have to do that to meet preset quantities of publications (which are ridiculously high), but it is also a way to get more attention for their work. Furthermore, I have found that journals and conferences often approach me to ask for a derivative of an article I wrote somewhere else. What could possibly be wrong about writing two aricles on the same data?
It will be easier in the next game, because it will have toilets, which the creature will recognise and use. Instead, you should now teach it to flush afterwards, and not leave the seat up (unless you play Evil, in which case you SHOULD leave the seat up).
Hardcore = People who play wolfenstein and study tactics and play the same map 40 times.
Casual Hardcore = People who play wolfenstein, enjoy it, try a new game, for at least 40 hours a week.
Casual = People who play a game when they have the time, which is usually not a lot, because they have a life.
To sell games, make good games, do not try to invent a game that appeals to an arbitrarily construed focus group.
But I assume, to be a columnist, you have to write at least 500 words per article.
Hogwash. Splinter Cell is completely different. I love Thief. I hate Splinter Cell. Why? Spinter Cell requires complex button mashing to execute exactly the right moves, which, for me, means that I have to save before nearly every action because I will fall to my death with more than 50% probability. Furthermore, Splinter Cell is annoyingly linear, while Thief is Freedom Itself.
I don't think Microsoft really lost here.
Think with me: if Microsoft really wanted to win this case, they would just appeal. They've got the money, they've got the lawyers, there is NO WAY Joe Smallpotatoes would win in the end. Especially not this ridiculous patent, which should be easy to overthrow on the grounds of obviousness.
So, I can only conclude that Microsoft is actually happy to lose this one. And why would that be? My guess is that they simply have lots of these obvious patents themselves, which they hope to apply tactically in the near future to bring down small entrepreneurs. Since they now lost this case, in the future, when someone they sue tries to tell the judge that a Microsoft patent is obvious, Microsoft can reply by pointing out the historic case in which a judge upheld a similar patent, which is therefore non-obvious.
This is a tactical loss for Microsoft. And I see a bleak future.
Not particularly, but I do think back in horror to the voice of the French git in Ultima IX.
Porn for nerds. Stuff that splatters.