I really hate physicists' obsession with drawing power laws, and the fact that science journalists always eat it up as important science. This article embodies all these annoying things. So I'd like to agree with you. BUT I have to note that Clauset was the co-author of a conceptually very sound, important paper about fitting distributions to empirical data: Clauset-Shalizi-Newman - Power-law distributions in empirical data. I don't think he would make the very mistake he identifies in this paper.
Well reasoned argument, but you leave out one important possibility. I don't think that an alien civilization will colonize Earth because Earth is some special, valuable, interesting place. They will colonize Earth because they will colonize all of their light cone. Let me try to explain how and why would they do that.
The why question: The relevant buzzwords here are computronium and negentropy. Basically, for any sufficiently advanced civilization, there is only one valuable resource: computation. (If this sounds implausible, just try to imagine how much of our existence will be spent in virtual spaces in 50 years from now, and then extrapolate another 100 million years from there. Upload is another relevant buzzword.) So on a conceptual level, and strictly from an outside view, you can imagine an advanced civilization as a process to build the largest possible computer. And that requires a huge amount of space.
The how question: Neumann probes are already a well-known possible method for totally terraforming a region of space, but I think much more effective methods are possible, too. We don't know any laws of physics that would be violated by an expansion speed converging to the speed of light. I imagine this as a turbo-charged version of the terraforming method described in "His Master's Voice", with teleportation instead of the gentle catalysis.
In fact, I believe in a 0-1 law for the expansion speed of civilizations. I argue that there is only a very short timeframe in the life of a civilization when their sphere of influence is already expanding, but not yet expanding with the speed of light. Note that this idea is an (admittedly somewhat contrived) solution to the Fermi Paradox: If they are before this short phase transition, they can’t be observed with current human technology. After the phase transition they can’t be observed at all.
Nice to hear that number. I was a bit worried that economic growth will suffer while things inevitably converge to their copyright-free state. But no, even the perfect demolition of a 4% GDP industry would not ruin world economy.
I have seen hundreds of posts on Slashdot about incompetence and abuse of power by Wikipedia editors. NONE of these posts contained reference to the events described. Citation needed, indeed.
It is quite possible that the distribution is lognormal. I don't have a citation right now, but I know that some of these guessing distributions are indeed lognormal. That would mean that the geometric average is the correct formula, and not the mean or median.
This is a theoretic, not immediately applicable result, but it shows that it is possible to set up such a distributed system so that it is immune to poisoning attempts.
You are the clueless one here. See some of the other replies to your post, most of them do have a clue. It is a shame that currently you are modded to the top, and they were not even moderated. New Scientist is prone to sensationalism, and this field is often misrepresented in the press. So I was actually quite surprised to see an informative and accurate piece such as this.
You are right. But it is a bit more complicated than that. For n less than 100, these are the values where the greedy method fails: 41,52,59,60,66,74,79,87,88,94,98. So, supposedly these are the values where backtracking helps. Unfortunately, in practice, it does not help at all: For these, the backtracking version of the algorithm does not terminate in a reasonable amount of time.
An interesting question: The OP's implementation does not really care about how to break ties when it chooses the move with the best value. My list above depends on the implementation details. Wolfram MathWorld talks about n less than 76, not n less than 41, so maybe they know about a different implementation. Is there a local tie-breaking rule that leads to a successful greedy algorithm for all values of n?
There is something about the OP's algorithm that I'm not sure he realized: It never backtracks. The 'choose the loneliest' method works even when applied in a greedy fashion. I did the following patch to artificially make the backtrack codepath run:
Changed the key=lambda c: reduce( part to key=lambda c: random.random()*3+reduce(
This means that the 'choose the loneliest' rule is broken sometimes. Backtracking is able to correct these mistakes sometimes, mostly if they did not happen too early in the search.
No, they can't do this because it violates MY patent on patenting patent trolling!
They can do this, if they pay you a licensing fee. Lucky for them, this is a one-time fee. On the other hand, they can get a fee from EACH patent troll after this. Their patent is smarter than yours.
I always thought that an LHC black hole accident wouuld not be good material for an Armaggedon-style Bruckheimer movie, because Earth's destruction would be instant. Now that I know that this takes longer time, I can envision it: Terrifying earthquakes. The world's leaders argue about course of action. Moscow disappeaers in a 100 mile hole. Brave scientists send nuclear warheads to center of Earth. NY's disappearance in a 1000 mile hole is just prevented.
Important caveat: After one year into the competition, sometime last autumn maybe, the leading team had to publish all their methods to claim the $50,000 yearly prize. If this guy did nothing else but correctly reimplement this published algorithm, he would be at this score.
I didn't check your calculation, but if it is correct, that would mean that you need a 4800mx4800m area to permanently provide energy for one 60W lightbulb. (sqrt(60.0/0.045*365*24*2))
Isn't this image of Carmack partly due to his willingness to talk about his bonehead mistakes? I don't know anything about the alt.space community's standards, but Carmack-the-game-programmer is well-known for his openness. Is Carmack's very detailed report the norm in the alt.space community, or is it more like "We crashed. The GPS went wrong. Regards."?
Let's change every single mention of "book" to "web page" in your post, and adjust the last sentences accordingly:
"I have built a web page X recently. They have an entire copy of my web page sitting on their servers. (It may in fact be an index and hence a derivative work from which the complete original can be constructed, but that is still subject to copyright.) They are using their complete copy of X to make profit. I don't see a penny of this except maybe occasionally they link to my web page. And there's certain no law that compels me to accept web traffic in recompense for abuse of my copyright."
My non-rhetorical question: is (classic) Google Search a fully opt-in system? I didn't read any relevant legal documents, so I honestly don't know. Maybe it is a two-level opt-in system, where individual content owners (say, bloggers) give explicit permission for the site owner to give permission for google to index and cache their content? Exactly what rights are granted to Google this way?
I think it is obvious that the Wayback Machine archives and redistributes copyrighted material. It is also obvious that Google and http proxies also do this.
I'm sure the lawyers at Google and the Wayback Machine already figured out why this activity is legal, but I have never heard an explanation before. Anyone here?
One method is the automatic detection of bilingual websites.
Our reasearch team used a manual method. We have built a large (2 million sentences) Hungarian-English parallel text database using Project Gutenberg, open source software documentation, EU legal texts and other online resources.
I'm not very familiar with filesystem terminology. But we are talking about two completely different innovations. You are talking about filesystem-level metadata. I didn't say that was invented by Apple. No way. Was I was talking about was file TYPE metadata. Basically, Apple built an extensible type hierarchy with multiple inheritance. It's not like xattrs at all. It's like MIME types, but much-much better. Read the article page that I referenced.
I'm not sure who invented this in academia. I'm sure it's not new. After all, this is the proper way of doing it. But in mass production, I think it's quite new.
Re:Not a cron replacement, a init replacement
on
Does launchd Beat cron?
·
· Score: 3, Informative
So I think what Apple did rules. They didn't ask requirements from anyone, they had a flash of inspiration, created something awesome that works good for them, and then said 'hey, if you want to use this, go ahead!'
They did something similar with an even more important technology: file type metadata. Read the very good overview in the referenced Ars Technica article:
Also, the fact that quantum computers can factor large integers efficiently necessarily implies that they can do other NP-complete problems efficiently, such as the traveling salesman problem.
IAWAUGTOQC? No way. You know nothing about the theory of quantum computation. Factoring integers is NOT an NP-complete problem. Quantum computers are NOT believed to be able to solve NP-complete problems. Moderate parent DOWN.
Now that it has been released, what 64-bit compilers are available for the operating system? The last time I looked, Microsoft was planning to use an ugly data model (LLP64) where only "long long" variables and pointers would be 64-bits. To me, that's a chicken-shit decision, broken code should be fixed or rewritten, not accommodated by crippling the compiler.
No. Code that relies on long being 32-bit is not broken. All right, not very broken. But very-very common. It should recompile and work out of the box without porting.
Suppose someone lends you a car, and you drive a 1000 miles in a month. That someone shows up and takes the car away because he suddenly stops liking you. Do you say, "Damn, I knew it! We should have kept walking" or "Oh well, at least we made good progress for a month?"
This analogy works best if the guy takes back his car in the middle of a huge desert.
Slashdot Mirror
I really hate physicists' obsession with drawing power laws, and the fact that science journalists always eat it up as important science. This article embodies all these annoying things. So I'd like to agree with you. BUT I have to note that Clauset was the co-author of a conceptually very sound, important paper about fitting distributions to empirical data: Clauset-Shalizi-Newman - Power-law distributions in empirical data. I don't think he would make the very mistake he identifies in this paper.
Well reasoned argument, but you leave out one important possibility. I don't think that an alien civilization will colonize Earth because Earth is some special, valuable, interesting place. They will colonize Earth because they will colonize all of their light cone. Let me try to explain how and why would they do that.
The why question: The relevant buzzwords here are computronium and negentropy. Basically, for any sufficiently advanced civilization, there is only one valuable resource: computation. (If this sounds implausible, just try to imagine how much of our existence will be spent in virtual spaces in 50 years from now, and then extrapolate another 100 million years from there. Upload is another relevant buzzword.) So on a conceptual level, and strictly from an outside view, you can imagine an advanced civilization as a process to build the largest possible computer. And that requires a huge amount of space.
The how question: Neumann probes are already a well-known possible method for totally terraforming a region of space, but I think much more effective methods are possible, too. We don't know any laws of physics that would be violated by an expansion speed converging to the speed of light. I imagine this as a turbo-charged version of the terraforming method described in "His Master's Voice", with teleportation instead of the gentle catalysis.
In fact, I believe in a 0-1 law for the expansion speed of civilizations. I argue that there is only a very short timeframe in the life of a civilization when their sphere of influence is already expanding, but not yet expanding with the speed of light. Note that this idea is an (admittedly somewhat contrived) solution to the Fermi Paradox: If they are before this short phase transition, they can’t be observed with current human technology. After the phase transition they can’t be observed at all.
Supersonic ballistic whoosh.
Nice to hear that number. I was a bit worried that economic growth will suffer while things inevitably converge to their copyright-free state. But no, even the perfect demolition of a 4% GDP industry would not ruin world economy.
I have seen hundreds of posts on Slashdot about incompetence and abuse of power by Wikipedia editors. NONE of these posts contained reference to the events described. Citation needed, indeed.
It is quite possible that the distribution is lognormal. I don't have a citation right now, but I know that some of these guessing distributions are indeed lognormal. That would mean that the geometric average is the correct formula, and not the mean or median.
Goldwasser,Kalai,Rothblum - Delegating Computation: Interactive Proofs for Muggles
http://www.mit.edu/~rothblum/papers/del.pdf
This is a theoretic, not immediately applicable result, but it shows that it is possible to set up such a distributed system so that it is immune to poisoning attempts.
You are the clueless one here. See some of the other replies to your post, most of them do have a clue. It is a shame that currently you are modded to the top, and they were not even moderated. New Scientist is prone to sensationalism, and this field is often misrepresented in the press. So I was actually quite surprised to see an informative and accurate piece such as this.
You are right. But it is a bit more complicated than that. For n less than 100, these are the values where the greedy method fails: 41,52,59,60,66,74,79,87,88,94,98. So, supposedly these are the values where backtracking helps. Unfortunately, in practice, it does not help at all: For these, the backtracking version of the algorithm does not terminate in a reasonable amount of time.
An interesting question: The OP's implementation does not really care about how to break ties when it chooses the move with the best value. My list above depends on the implementation details. Wolfram MathWorld talks about n less than 76, not n less than 41, so maybe they know about a different implementation. Is there a local tie-breaking rule that leads to a successful greedy algorithm for all values of n?
There is something about the OP's algorithm that I'm not sure he realized: It never backtracks. The 'choose the loneliest' method works even when applied in a greedy fashion. I did the following patch to artificially make the backtrack codepath run:
Changed the
key=lambda c: reduce(
part to
key=lambda c: random.random()*3+reduce(
This means that the 'choose the loneliest' rule is broken sometimes. Backtracking is able to correct these mistakes sometimes, mostly if they did not happen too early in the search.
No, they can't do this because it violates MY patent on patenting patent trolling!
They can do this, if they pay you a licensing fee. Lucky for them, this is a one-time fee. On the other hand, they can get a fee from EACH patent troll after this. Their patent is smarter than yours.
I always thought that an LHC black hole accident wouuld not be good material for an Armaggedon-style Bruckheimer movie, because Earth's destruction would be instant. Now that I know that this takes longer time, I can envision it: Terrifying earthquakes. The world's leaders argue about course of action. Moscow disappeaers in a 100 mile hole. Brave scientists send nuclear warheads to center of Earth. NY's disappearance in a 1000 mile hole is just prevented.
Important caveat: After one year into the competition, sometime last autumn maybe, the leading team had to publish all their methods to claim the $50,000 yearly prize. If this guy did nothing else but correctly reimplement this published algorithm, he would be at this score.
I didn't check your calculation, but if it is correct, that would mean that you need a 4800mx4800m area to permanently provide energy for one 60W lightbulb. (sqrt(60.0/0.045*365*24*2))
Isn't this image of Carmack partly due to his willingness to talk about his bonehead mistakes? I don't know anything about the alt.space community's standards, but Carmack-the-game-programmer is well-known for his openness. Is Carmack's very detailed report the norm in the alt.space community, or is it more like "We crashed. The GPS went wrong. Regards."?
http://weblog.fortnow.com/2004/09/blue-state-of-sc ience.html
Let's change every single mention of "book" to "web page" in your post, and adjust the last sentences accordingly:
"I have built a web page X recently. They have an entire copy of my web page sitting on their servers. (It may in fact be an index and hence a derivative work from which the complete original can be constructed, but that is still subject to copyright.) They are using their complete copy of X to make profit. I don't see a penny of this except maybe occasionally they link to my web page. And there's certain no law that compels me to accept web traffic in recompense for abuse of my copyright."
My non-rhetorical question: is (classic) Google Search a fully opt-in system? I didn't read any relevant legal documents, so I honestly don't know. Maybe it is a two-level opt-in system, where individual content owners (say, bloggers) give explicit permission for the site owner to give permission for google to index and cache their content? Exactly what rights are granted to Google this way?
I think it is obvious that the Wayback Machine archives and redistributes copyrighted material. It is also obvious that Google and http proxies also do this.
I'm sure the lawyers at Google and the Wayback Machine already figured out why this activity is legal, but I have never heard an explanation before. Anyone here?
One method is the automatic detection of bilingual websites.
Our reasearch team used a manual method. We have built a large (2 million sentences) Hungarian-English parallel text database using Project Gutenberg, open source software documentation, EU legal texts and other online resources.
It has a simple web query interface at
http://szotar.mokk.bme.hu/hunglish/corpus
The search is still completely unoptimized, and the user interface is spartan, but it works.
In the near future, we'll all be paying a monthly fee for having a memory, too!
Combine this comment with some joke about the original post being a dupe. Profit.
I'm not very familiar with filesystem terminology. But we are talking about two completely different innovations. You are talking about filesystem-level metadata. I didn't say that was invented by Apple. No way. Was I was talking about was file TYPE metadata. Basically, Apple built an extensible type hierarchy with multiple inheritance. It's not like xattrs at all. It's like MIME types, but much-much better. Read the article page that I referenced.
I'm not sure who invented this in academia. I'm sure it's not new. After all, this is the proper way of doing it. But in mass production, I think it's quite new.
So I think what Apple did rules. They didn't ask requirements from anyone, they had a flash of inspiration, created something awesome that works good for them, and then said 'hey, if you want to use this, go ahead!'
/ 11
They did something similar with an even more important technology: file type metadata. Read the very good overview in the referenced Ars Technica article:
http://arstechnica.com/reviews/os/macosx-10.4.ars
Also, the fact that quantum computers can factor large integers efficiently necessarily implies that they can do other NP-complete problems efficiently, such as the traveling salesman problem.
IAWAUGTOQC? No way. You know nothing about the theory of quantum computation. Factoring integers is NOT an NP-complete problem. Quantum computers are NOT believed to be able to solve NP-complete problems. Moderate parent DOWN.
Now that it has been released, what 64-bit compilers are available for the operating system? The last time I looked, Microsoft was planning to use an ugly data model (LLP64) where only "long long" variables and pointers would be 64-bits. To me, that's a chicken-shit decision, broken code should be fixed or rewritten, not accommodated by crippling the compiler.
No. Code that relies on long being 32-bit is not broken. All right, not very broken. But very-very common. It should recompile and work out of the box without porting.
Read The Old New Thing. Search for Raymond Chen's replies.
Suppose someone lends you a car, and you drive a 1000 miles in a month. That someone shows up and takes the car away because he suddenly stops liking you. Do you say, "Damn, I knew it! We should have kept walking" or "Oh well, at least we made good progress for a month?"
This analogy works best if the guy takes back his car in the middle of a huge desert.