The simple solution is to make "a == b" syntactic sugar for "a.equals(b)", and maybe "a === b" syntactic sugar for "a.referenceEquals(b)" (I know how ugly it looks). In that case, overriding the equals method would always also overrides ==, by definition. Incidentally, this is exactly what Ceylon is doing.
There are some assumptions built into science. Probably the most important one is that science assumes that the laws of the universe do not change over time. If someone observes the universe behaving in a particular way a lot of times, science generally assumes that it will continue to do so.
Interestingly enough, I would say that this assumption is not actually needed. All you need to do is observe that if the laws of the universe change over time, they either do so in a predictable manner, in which case science is still the right tool for the job, or they do so in an unpredictable manner. But see, if you try to predict something in a universe with laws that change unpredictably, well, you might as well predict at random. Your odds of being right will always be the same.
In other words, science is a great system when the laws of the universe do not change. And when the laws of the universe change, well, it so happens that all systems are (equally) bad. So science dominates all other strategies, regardless of whether its "assumption" is correct. Basically, even though we all assume unchanging laws in practice, we don't actually need to - even if you outright reject the assumption, there is no alternative.
You don't need to have "faith" that the laws we have identified will continue to be the way they have been. Think of it this way: supposing that the laws of the universe will suddenly change within a few seconds does not tell you what they will change to. So if you want to, say, optimize your actions, it appears that regardless of what action you choose to do, some law change can reward you and some other law change can punish you. Ergo, if you do not have a model for law changes, all potential actions you could take have the same expected value.
So if you set out to exploit the laws that have worked in the past, you will reap large benefits if they don't change. And if they change, you don't know how in advance, so it's not any better or any worse than anything else you could have done. It's a sort of wager, but it's a good one: you can only prepare for what is predictable.
It's like throwing a rock at a mirror to break it. Maybe the mirror is too solid to be broken by a rock, but then it will stay unbroken regardless of what you do. So even if you don't have "faith" that the mirror can be broken, you might as well throw the rock. Doing science is a bit like that, it's worth doing regardless of whether you have faith in it or not.
The thing is, even if you had one fossil showing up differently than what we would expect, this does not change the fact that we have millions of fossils that do corroborate evolution. If for some reason, in one experiment, one object was found to be falling upwards, you might have to adjust the theory of gravity, but it won't become radically different from what we have now, for the simple reason that there would still be trillions of trillions of objects on this planet that still fall accordingly to the standard theory of gravity. For sure, you won't fall back to "intelligent falling" just because there seem to be a few exceptions to a theory that worked so brilliantly thus far. You will start from what you have, try to find under what conditions there is unexpected behavior, and figure out a theory that's different from the old one whenever these conditions occur, and the same in normal cases. That is exactly what happened with relativity. And you should be aware that the difference between the current theory of evolution and the theory put forth by Darwin over a century ago is greater than relativity is from Newtonian physics. But it's still called evolution, just like gravity is still called gravity, because the vast majority of the observations we have are consistent with the spirit of that nomenclature.
Basically, the more evidence you get in support of a theory, the more evidence you need against that theory in order to reject it. One unexpected observation cannot weigh more than millions of observations that the theory explains properly. Whatever adjustment you make in order to explain this new observation has to ALSO support the others, which means that the new theory will be very similar to the old one. Think about it this way: if you have a theory that works extremely well in 99% of cases, this means that this theory is the limit of the "true theory" in "average conditions". It seems unlikely that something radically different from evolution will look like evolution so much that almost everything we uncover is consistent with it. Whenever surprises happen, we will seek adjustments, we won't just throw away what we have (unless there are so many surprises that it becomes hopeless - but wouldn't this have happened by now?).
Keep in mind that if we reject evolution and replace it by something completely different, we will need to understand how/why it behaves like evolution in the average case. No theory can displace evolution without answering that question (because then it would fail to explain common observations).
It does not matter whether the computer favors a choice more than the other. Imagine that there is maximal favoritism and the computer just always picks rock. If you play randomly, you will win 1/3 of the time (whenever you happen to randomly pick paper), you will lose 1/3 of the time (idem, scissors), and you will tie 1/3 of the time (idem, rock).
Imagine the computer picks rock X% of the time, paper Y% of the time, scissors in all other instances. Whenever it picks rock, you have 1/3 odds of picking paper and winning. Whenever it picks paper, you have 1/3 odds of picking scissors and winning. Whenever it picks scissors, you have 1/3 odds of picking rock and winning. X * 1/3 + Y * 1/3 + (1 - X - Y) * 1/3 = X/3 + Y/3 + 1/3 - X/3 - Y/3 = 1/3, for all X and for all Y. No strategy can expect to win or lose against a random strategy more than 1/3 of the time in the limit of the number of rounds played.
Truly random play has the same expected results against every single strategy.Think about it this way: no matter what the computer thinks you will do, if you play truly randomly, its odds of winning, losing or tying are all 1/3. If it did any better, it would be able to predict randomness, which is by definition impossible, and if it did any worse, then by inverting its strategy it would do better, and the same reasoning holds.
Now, if you try to play randomly, and win in a statistically significant way, you are probably simply acting in a way that feels random to you, but is not random at all, and perhaps this happens to be a counter-strategy to the AI. That is, "playing randomly" for a human often might often involve picking the opposite of what he would have normally picked given the history of throws.
Well, if you watched the show, you would likely have had more exposure to John Locke the character than you ever had to John Locke the philosopher, which will change what you think of first when you see the name. When I see "John Locke", the character pops to mind, then the philosopher a second later, at which point I beat myself up for letting TV shuffle my associative memory around:(
If the name had anything to do with popularity, I'd wager the TV show is to thank. Nobody cares about philosophy, unfortunately.
That's what I thought at first but that's 350k in three months. Still amounts to "only" $485k a year, though, so I figure they didn't do the actual math.
Who gives a fuck if factorization is NP complete or not, that doesn't matter in the slightest.
I might be wrong here but wouldn't a proof that P!=NP only guarantee that NP-complete problems are not in P? That is, if factorization is not NP-complete, then it might actually be in P regardless of whether P=NP or not. So yes, if P=NP it is all moot, but if P!=NP we would much prefer encryption that's NP-complete to break (or provably not in P).
Not really. Intelligent design sees more scorn and obstruction (both justified) from the vast majority of biologists than climate denial does from climate scientists. If there is scientific consensus about something, and well-organized opposition from politicians, industries and/or religions challenges it, scientists will naturally revert to a defensive (and perhaps even paranoid) position - an "us vs them" mentality. These emails showed little more than this defensive position, and creationists would find the same things in private conversations between some evolutionary biologists. Doesn't mean these biologists are crooks - they just feel threatened, and react poorly to it.
You have to keep in mind several things, though. First, OCR would be sufficient, and it's hardly difficult here - this is not handwriting, the area of interest is easy to spot, the video resolution does not have to be capped. I do wonder why they didn't do it, but I cannot see OCR being a significant overhead.
Second, Watson's speed is proportional to the specs of the hardware it runs on. I think it is fair to assume that the algorithms are highly parallel, so by using twice the hardware (or hardware that is twice as fast), Watson should be nearly twice as fast. While it is very well possible that Watson was given an advantage, the alternative is just to give it more computational resources. However, that expense would be difficult to justify: I mean, do you give Watson the clue halfway through the reading of the question "to be fair", or do you give it straight away and cut the specs in half? The results would be the same, but not the (already exorbitant) cost, so it would be folly to pick the former option.
Well, there is another thing about terse clues: they provide less information, so they are inherently harder. Having less time to process them compounds that problem, but five more seconds might not have sufficed anyway.
You know, I'm certain that would be more popular than.info,.name,.biz,.museum and all that stale junk. If I wanted to register a personal domain, I would register a.bacon one, no questions asked.
Why not just open the registration of all TLDs, then? If we are stuck with.com,.org and country TLDs, we might as well promote the x.com domain to the x TLD. If the TLD pool cannot be expanded, then there should be no TLDs to begin with, and all names should reside at the top level. Now, I wouldn't necessarily want that to happen, but it's better than what you are suggesting. Better than both would be to sell TLDs at very high prices with zero approval process, and let the market take care of the rest.
I wager the venture will still have been profitable to them while it lasted. I assume that they will simply settle to drop their incentive system, share profits and probably implement a better system to filter out infringing content. I doubt they will face any real consequences.
The incentive structure does not discourage noninfringing use, it encourages infringing use. I am not saying that's much better, but it's different. I mean, if I want to use the service to upload my own stuff, I don't see what would discourage me.
"Pathetic" is relative. If you expect human-like intelligence, yeah, the state of the art is pretty pathetic. If you compare it to what we had 40 years ago, though, there's been a lot of improvement.
And there's a lot of research in the field. Don't be silly. It might not be glamorous, but it's there.
Well, think about it for a second. If.com,.org and national TLDs are enough, then there is no real difference between registering abc.com and outright registering the abc TLD. In fact, the latter is four keystrokes shorter. If TLDs are locked down in the first place, that's under the assumption that they will become progressively available to meet demand. Else you might as well reserve national TLDs and a few others like.org, and then sell the rest of the TLDs at.com pricing.
Personally, I would just allow anyone to acquire a TLD for a few million bucks and let the market take care of everything else. The high price would ensure TLD scarcity, and we wouldn't be getting all this senseless drama.
Well, let's be fair, here: "Tetris", as a name for a game that uses pieces comprising four blocks, is the equivalent of a car company naming their cars "Caris". If you called your cars a "Carada", the only similarity with "Caris" cars is the part of the name that's a generic description of what the product is.
I mean, come on, this is a game about tetrads or tetrominos, if using tetrads is not an issue, then neither is prefixing the name of the game with "tetr".
You know, that's weird, because I would indeed argue that Windows, Pacman, Mario Bros and Star Trek should all be in the public domain by now, free of copyrights. Now, I would not argue that's because they are commonplace, but rather that the fact that they have been commonplace for quite some time indicates that copyright has already served its purpose for them.
The only purpose of copyright is to provide an incentive for creators to create. I would argue that in the vast majority of cases, 20 years are plenty for a creator to properly exploit their creation, and in some fields, such as software, it should not exceed 5 years. Now, if a work somehow becomes popular 30 years after publication, I'd be sympathetic to extending its protection, but you get the point: you shouldn't be able to milk a blockbuster for 20 years.
The thing is, all ideas are building blocks for further creation, and people naturally internalize everything they come across. Being in contact with Pacman games, and liking Pacman games, will push your thought processes towards using Pacman as part of your own creations. Even if you outright copy it, there is still a chance you will improve it. "You could make original creations" is a red herring: if you are naturally inclined to make a Pacman game, going out of your way to create something "original" will not end up in something better, and it might even demotivate you, because it being Pacman is part of the fun. It's easy not to be sympathetic to "rip-offs", but in theory, a Tetris clone could include some small tweaks that make them better than the original - since Tetris has already paid off the copyright owner, we should encourage such "rip-offs".
In a nutshell, we need copyrights to ensure that creativity is properly rewarded. But they have to end once the reward has been collected, and when something becomes a cultural icon, it's safe to say that's been done.
Instead of trying so hard to optimize JavaScript to do things like that, why can't we just get a wider language choice? Like, a properly sandboxed language with static typing could be handy, and the standard could mandate its presence. Everybody would get a compatible browser eventually, if it allowed them to run web apps that are actually fast.
Slashdot Mirror
The simple solution is to make "a == b" syntactic sugar for "a.equals(b)", and maybe "a === b" syntactic sugar for "a.referenceEquals(b)" (I know how ugly it looks). In that case, overriding the equals method would always also overrides ==, by definition. Incidentally, this is exactly what Ceylon is doing.
There are some assumptions built into science. Probably the most important one is that science assumes that the laws of the universe do not change over time. If someone observes the universe behaving in a particular way a lot of times, science generally assumes that it will continue to do so.
Interestingly enough, I would say that this assumption is not actually needed. All you need to do is observe that if the laws of the universe change over time, they either do so in a predictable manner, in which case science is still the right tool for the job, or they do so in an unpredictable manner. But see, if you try to predict something in a universe with laws that change unpredictably, well, you might as well predict at random. Your odds of being right will always be the same.
In other words, science is a great system when the laws of the universe do not change. And when the laws of the universe change, well, it so happens that all systems are (equally) bad. So science dominates all other strategies, regardless of whether its "assumption" is correct. Basically, even though we all assume unchanging laws in practice, we don't actually need to - even if you outright reject the assumption, there is no alternative.
You don't need to have "faith" that the laws we have identified will continue to be the way they have been. Think of it this way: supposing that the laws of the universe will suddenly change within a few seconds does not tell you what they will change to. So if you want to, say, optimize your actions, it appears that regardless of what action you choose to do, some law change can reward you and some other law change can punish you. Ergo, if you do not have a model for law changes, all potential actions you could take have the same expected value.
So if you set out to exploit the laws that have worked in the past, you will reap large benefits if they don't change. And if they change, you don't know how in advance, so it's not any better or any worse than anything else you could have done. It's a sort of wager, but it's a good one: you can only prepare for what is predictable.
It's like throwing a rock at a mirror to break it. Maybe the mirror is too solid to be broken by a rock, but then it will stay unbroken regardless of what you do. So even if you don't have "faith" that the mirror can be broken, you might as well throw the rock. Doing science is a bit like that, it's worth doing regardless of whether you have faith in it or not.
The thing is, even if you had one fossil showing up differently than what we would expect, this does not change the fact that we have millions of fossils that do corroborate evolution. If for some reason, in one experiment, one object was found to be falling upwards, you might have to adjust the theory of gravity, but it won't become radically different from what we have now, for the simple reason that there would still be trillions of trillions of objects on this planet that still fall accordingly to the standard theory of gravity. For sure, you won't fall back to "intelligent falling" just because there seem to be a few exceptions to a theory that worked so brilliantly thus far. You will start from what you have, try to find under what conditions there is unexpected behavior, and figure out a theory that's different from the old one whenever these conditions occur, and the same in normal cases. That is exactly what happened with relativity. And you should be aware that the difference between the current theory of evolution and the theory put forth by Darwin over a century ago is greater than relativity is from Newtonian physics. But it's still called evolution, just like gravity is still called gravity, because the vast majority of the observations we have are consistent with the spirit of that nomenclature.
Basically, the more evidence you get in support of a theory, the more evidence you need against that theory in order to reject it. One unexpected observation cannot weigh more than millions of observations that the theory explains properly. Whatever adjustment you make in order to explain this new observation has to ALSO support the others, which means that the new theory will be very similar to the old one. Think about it this way: if you have a theory that works extremely well in 99% of cases, this means that this theory is the limit of the "true theory" in "average conditions". It seems unlikely that something radically different from evolution will look like evolution so much that almost everything we uncover is consistent with it. Whenever surprises happen, we will seek adjustments, we won't just throw away what we have (unless there are so many surprises that it becomes hopeless - but wouldn't this have happened by now?).
Keep in mind that if we reject evolution and replace it by something completely different, we will need to understand how/why it behaves like evolution in the average case. No theory can displace evolution without answering that question (because then it would fail to explain common observations).
Excel?!? Please, that's insulting :(
It does not matter whether the computer favors a choice more than the other. Imagine that there is maximal favoritism and the computer just always picks rock. If you play randomly, you will win 1/3 of the time (whenever you happen to randomly pick paper), you will lose 1/3 of the time (idem, scissors), and you will tie 1/3 of the time (idem, rock).
Imagine the computer picks rock X% of the time, paper Y% of the time, scissors in all other instances. Whenever it picks rock, you have 1/3 odds of picking paper and winning. Whenever it picks paper, you have 1/3 odds of picking scissors and winning. Whenever it picks scissors, you have 1/3 odds of picking rock and winning. X * 1/3 + Y * 1/3 + (1 - X - Y) * 1/3 = X/3 + Y/3 + 1/3 - X/3 - Y/3 = 1/3, for all X and for all Y. No strategy can expect to win or lose against a random strategy more than 1/3 of the time in the limit of the number of rounds played.
Truly random play has the same expected results against every single strategy.Think about it this way: no matter what the computer thinks you will do, if you play truly randomly, its odds of winning, losing or tying are all 1/3. If it did any better, it would be able to predict randomness, which is by definition impossible, and if it did any worse, then by inverting its strategy it would do better, and the same reasoning holds.
Now, if you try to play randomly, and win in a statistically significant way, you are probably simply acting in a way that feels random to you, but is not random at all, and perhaps this happens to be a counter-strategy to the AI. That is, "playing randomly" for a human often might often involve picking the opposite of what he would have normally picked given the history of throws.
Well, if you watched the show, you would likely have had more exposure to John Locke the character than you ever had to John Locke the philosopher, which will change what you think of first when you see the name. When I see "John Locke", the character pops to mind, then the philosopher a second later, at which point I beat myself up for letting TV shuffle my associative memory around :(
If the name had anything to do with popularity, I'd wager the TV show is to thank. Nobody cares about philosophy, unfortunately.
The actual figure is (350000 * 0.99 * (365 / 67)) * 0.35 = $660k
Wait, I'm an idiot. Only two months and a week have passed.
That's what I thought at first but that's 350k in three months. Still amounts to "only" $485k a year, though, so I figure they didn't do the actual math.
Who gives a fuck if factorization is NP complete or not, that doesn't matter in the slightest.
I might be wrong here but wouldn't a proof that P!=NP only guarantee that NP-complete problems are not in P? That is, if factorization is not NP-complete, then it might actually be in P regardless of whether P=NP or not. So yes, if P=NP it is all moot, but if P!=NP we would much prefer encryption that's NP-complete to break (or provably not in P).
Not really. Intelligent design sees more scorn and obstruction (both justified) from the vast majority of biologists than climate denial does from climate scientists. If there is scientific consensus about something, and well-organized opposition from politicians, industries and/or religions challenges it, scientists will naturally revert to a defensive (and perhaps even paranoid) position - an "us vs them" mentality. These emails showed little more than this defensive position, and creationists would find the same things in private conversations between some evolutionary biologists. Doesn't mean these biologists are crooks - they just feel threatened, and react poorly to it.
You have to keep in mind several things, though. First, OCR would be sufficient, and it's hardly difficult here - this is not handwriting, the area of interest is easy to spot, the video resolution does not have to be capped. I do wonder why they didn't do it, but I cannot see OCR being a significant overhead.
Second, Watson's speed is proportional to the specs of the hardware it runs on. I think it is fair to assume that the algorithms are highly parallel, so by using twice the hardware (or hardware that is twice as fast), Watson should be nearly twice as fast. While it is very well possible that Watson was given an advantage, the alternative is just to give it more computational resources. However, that expense would be difficult to justify: I mean, do you give Watson the clue halfway through the reading of the question "to be fair", or do you give it straight away and cut the specs in half? The results would be the same, but not the (already exorbitant) cost, so it would be folly to pick the former option.
Well, there is another thing about terse clues: they provide less information, so they are inherently harder. Having less time to process them compounds that problem, but five more seconds might not have sufficed anyway.
You know, I'm certain that would be more popular than .info, .name, .biz, .museum and all that stale junk. If I wanted to register a personal domain, I would register a .bacon one, no questions asked.
But their TLDs disturb me a little.
Why not just open the registration of all TLDs, then? If we are stuck with .com, .org and country TLDs, we might as well promote the x.com domain to the x TLD. If the TLD pool cannot be expanded, then there should be no TLDs to begin with, and all names should reside at the top level. Now, I wouldn't necessarily want that to happen, but it's better than what you are suggesting. Better than both would be to sell TLDs at very high prices with zero approval process, and let the market take care of the rest.
I wager the venture will still have been profitable to them while it lasted. I assume that they will simply settle to drop their incentive system, share profits and probably implement a better system to filter out infringing content. I doubt they will face any real consequences.
The incentive structure does not discourage noninfringing use, it encourages infringing use. I am not saying that's much better, but it's different. I mean, if I want to use the service to upload my own stuff, I don't see what would discourage me.
"Pathetic" is relative. If you expect human-like intelligence, yeah, the state of the art is pretty pathetic. If you compare it to what we had 40 years ago, though, there's been a lot of improvement.
And there's a lot of research in the field. Don't be silly. It might not be glamorous, but it's there.
Well, think about it for a second. If .com, .org and national TLDs are enough, then there is no real difference between registering abc.com and outright registering the abc TLD. In fact, the latter is four keystrokes shorter. If TLDs are locked down in the first place, that's under the assumption that they will become progressively available to meet demand. Else you might as well reserve national TLDs and a few others like .org, and then sell the rest of the TLDs at .com pricing.
Personally, I would just allow anyone to acquire a TLD for a few million bucks and let the market take care of everything else. The high price would ensure TLD scarcity, and we wouldn't be getting all this senseless drama.
Well, let's be fair, here: "Tetris", as a name for a game that uses pieces comprising four blocks, is the equivalent of a car company naming their cars "Caris". If you called your cars a "Carada", the only similarity with "Caris" cars is the part of the name that's a generic description of what the product is.
I mean, come on, this is a game about tetrads or tetrominos, if using tetrads is not an issue, then neither is prefixing the name of the game with "tetr".
You know, that's weird, because I would indeed argue that Windows, Pacman, Mario Bros and Star Trek should all be in the public domain by now, free of copyrights. Now, I would not argue that's because they are commonplace, but rather that the fact that they have been commonplace for quite some time indicates that copyright has already served its purpose for them.
The only purpose of copyright is to provide an incentive for creators to create. I would argue that in the vast majority of cases, 20 years are plenty for a creator to properly exploit their creation, and in some fields, such as software, it should not exceed 5 years. Now, if a work somehow becomes popular 30 years after publication, I'd be sympathetic to extending its protection, but you get the point: you shouldn't be able to milk a blockbuster for 20 years.
The thing is, all ideas are building blocks for further creation, and people naturally internalize everything they come across. Being in contact with Pacman games, and liking Pacman games, will push your thought processes towards using Pacman as part of your own creations. Even if you outright copy it, there is still a chance you will improve it. "You could make original creations" is a red herring: if you are naturally inclined to make a Pacman game, going out of your way to create something "original" will not end up in something better, and it might even demotivate you, because it being Pacman is part of the fun. It's easy not to be sympathetic to "rip-offs", but in theory, a Tetris clone could include some small tweaks that make them better than the original - since Tetris has already paid off the copyright owner, we should encourage such "rip-offs".
In a nutshell, we need copyrights to ensure that creativity is properly rewarded. But they have to end once the reward has been collected, and when something becomes a cultural icon, it's safe to say that's been done.
Instead of trying so hard to optimize JavaScript to do things like that, why can't we just get a wider language choice? Like, a properly sandboxed language with static typing could be handy, and the standard could mandate its presence. Everybody would get a compatible browser eventually, if it allowed them to run web apps that are actually fast.