The algorithm in the article is actually a cousin of GAs called an evolutionary algorithm. Typically, GAs have crossover (sex) where genetic material from multiple parents is combined. This would be easy to do with the keyboard layout problem, and would probably produce better results.
Often, EAs work better with smaller populations (~100-500) and GAs work better for larger populations, but that's more anecdotal than rigorously shown.
We've had net flix for about 2 months. We've watched about 7 flix, and have 3 ready to watch. Our turnaround is two-three days, because we're in Berkeley, which is nice. The price is right for us, as we're terrible at returning DVDs at blockbuster, so we're saving about 10 bucks a month this way.
Our biggest problem is that it's actually hard to browse their selection. If you know what you want, you can usually find it, but browsing is hard. Their categories aren't as complete as one would like, and the recommendations don't run all that deep (only 10-15 in the indie section, for example). A secondary problem is that we end up with 3 DVDs that we "think" we want to watch someday (e.g. movies like Remains of the Day, Requiem for a Dream) that just sit around because we never quite "feel" like it.
Overall we like the service, but it doesn't replace the impulse movie watching available through our local (non Blockbuster) rental store.
I honestly don't think it matters too much what language is used in teaching intro CS courses, as what must be conveyed in those courses are the basic paradigms that are expressible in all languages, from ML to LISP to Java to C.
Intro to CS courses that use Java could be taught badly, as they would be if they focus on UI issues instead of learning to write the fundamentals. Students should be expected to learn how to write linked lists, trees, and all the basic underlying structures of programming. There should not be too much dependence on the fancy features of the language for doing this, IMHO. Learning to use Java's list class is not the same as writing it yourself, and thus understanding how it's written. Java is fine for testing issues of relative speed and for empirical explorations of computational complexity.
One of the key advantages that java has over C (or especially C++) is that it is relatively clean, and relatively safe. On the other hand, it's not as portable as C or C++, which is a real problem for many schools, since Macs are often used in education.
Scientists report, based on one badly performed and error ridden study, that 1/10 Japanese workers suffers from memory loss, but are considerably more relaxed and happy. A fellow who used to be a salesman was reported as saying "Yeah, I really couldn't remember why I what I was trying to sell anymore, ya know? It sorta had no meaning to me anymore, no depth. I went to a doctor to appease the company (and secure a nice disability check), but man, thank god I'm done with that job".
There are several amazing bits to the movie,
making it well worth watching.
First, the editing is fantastic. It makes the movie work, and makes what would be depressing or gory actually hilarious. The editing at the beginning of the movie, where the characters are introduced, is simply one of the best bits of editing I've seen this year.
Another great bit is Pitt's performance. Although he often has struck me as a bit of a young annoyingly arrogant actor (e.g. I can act b/c I'm cute), he plays the incoherent gypsy who actually is more "with it" than he seems rather well.
Finally, I recommend seeing it in the theatres, or at least with a large group of people, as it's a film that thrives on the audiences humorous reactions, which are self-reinforcing.
Could the whole thing be an elaborate(and twisted) April Fool's joke? Steal the Enigma, then send coded messages back to the authorities regarding its whereabouts?
It strikes me that Dvorak missed many of the most important events and/or people of the year by ignoring the advances in technology and science that took place this year. By focusing nearly entirely on the 'biz' side of things, he has a myopic view, especially for importance. Although I'll be the first to admit that the business side of a company is critical, when one looks at the past and attempts to judge importance, inventions and discoveries seem to rank right up there with savvy business dealings.
What about the first realization of a quantum computer (here)? Or IBM's advances in chip technology? Or any of a number of similar advances that are almost certainly important for the future direction of technology?
For that matter, I think leaving out the continued successful rise and development of cellular phones and the like is quite a mistake. When he puts network PCs and ubiquitous computing on the 'flop' list, he misses the most successful of the network ed appliances, the cell phone. The important future of cell phones (which I already had some good ideas about) was made utterly clear to me when, on Christmas this year, I ordered a book from Amazon on my new Sanyo-4000 using the mini-browser on the phone. Took about 4 minutes (including searching for a few things), and was amazingly easy.
It seems to me that we could all be being led down a goose chase here. The web page reminds me most of the many hoax web sites that one sees, as we don't see anything that really shows the existence of either a prototype or a working product. It reminds me of those who have sites claiming to have solved the P=NP problem, or those who claim to have a unified field theory (discussed several days ago, on/.) I hate to doubt the veracity of someone, but in this case the unprofessional web site and the lack of technical details at least brings it to mind.
Given that these concerns occur to me (not one talented at all at doing due diligence) it strikes me that VCs would also be suspicious. What evidence is being presented to VCs? Perhaps we could help you tune those documents. Although (as in all fields, even slashdot!) there is some nepotism in VC, good, well presented ideas rise to the top, as always.
SSH Version OpenSSH-1.2, protocol version 1.5. Compiled with SSL.
Is that all right? It doesn't mention RSAREF, so I wasn't sure. I suspect it's fine, given that it doesn't directly mention RSAREF, but I thought I'd check.
Although I do believe that there are very real ethical dilemmas surrounding what will happen once HGP is done, it seems to me that JK and most of/. has missed the point, the limitations, and the main dangers of the HGP.
Once the HGP is 100% complete does that mean that the oppression and genetic engineering of humans can start right away? NO! Remember that the HGP is simply the sequencing of the human genome (and only for a subset of humans, obviously). Once we have the sequence and have determined the location of all of the genes, we still have the very difficult task of understanding (1) how the genes form into proteins and (2) the role of these proteins in a developing organism.
In other words, we have yet to solve the protein folding problem and the problem of fully analyzing what proteins do anyway. Although we know a fair bit about what proteins do, we certainly do not fully understand all of their interactions. As for the protein folding problem, anyone who believes we are close to this goal is dreaming. We've only got the 3D structures for about 1000 proteins, and much of the data in that database has flaws and disagreements between researchers. Despite recent large grants for supercomputers for the protein folding problem, being able to accurately simulate the folding of a protein is a long way off.
Without a full understanding of the structure and function of the proteins generated from the genome, we are left with only poor correlational data. In other words -- a gene Y of this type tends to correlate highly with disease or condition Z. But, as we've seen in the breast cancer gene research, the correlations are relatively low, and tend to be of 'enabling' type. Thus, as expected, ones genetic predisposition for something is only part of the story.
It makes some sense to worry about the ethical problems surrounding knowing some correlational information about how gene X affects a persons behavior. If we found a super high correlation between the presence of a gene and violent anti-social behavior, it would be an ethical minefield, for sure. This is the topic that we should be discussing and worrying about, not whether genetic engineering of the perfect child is appropriate or ethical.
In any case, I highly doubt that the population will stand for any significant abuses of this sort of information. We don't allow country of origin (probably the highest correlative factor for tracking down credit card fraud, for example) to be used by the credit card companies, so I doubt that genetic discrimination will be allowed in any but the most clear of cases.
Furthermore, the scariest kinds of discrimination (job discrimination, etc) are the least likely to be highly correlated. The most likely things to be highly correlated are dangerous medical problems (such as sickle cell anemia, for example, for which the correlation is perfect -- with a malformed hemoglobin gene, you get sickle cell anemia b/c the structure of the blood cells is compromised by the malformity -- although it does impart a higher resistance to malaria, fwiw), which can be used by the medical community as indicators to help deal with treatment.
Although I'm for the little guy in this case, the goto.com logo is much interesting and less catching than the go.com design. If I had to choose one, it'd be go.com. But, copyright law protects good and bad design alike. Just because the goto.com logo isn't as effective at evoking the go streetlight symbol as is the go.com logo doesn't mean that it wasn't there first. It's obvious that the two are similar, and goto.com obviously felt like they were being harmed by the similarity. Goto.com is harmed each time a user mistakenly types go.com and doesn't find their search engine. Besides, the publicity of winning the case certainly helps them -- it appears that their site is/.'ed.
What I was surprised by was how tongue in cheek and ironic Gore was in the letter. The "Sonics" bit, the reference to Dubya's failed pop quiz -- I think it's part of a direct effort to seem hip and in touch, to reach the politically cynical and doubting tech crowd. Rather strange for a politician -- a fair bit more subtlety than I expected out of Gore.
Even worse, in the Oakland Tribune, they put the image on their front page, with a caption identifying it as a depiction of an image of the planet from the Berkeley astronomy site. It was not at all mentioned that it was simply an artist's rendition. Claiming that it was "seen" is also a bit odd -- more like it was detected using a visual medium. It's not like a human is sitting there, staring at the star, waiting for a shadow to cross over it. In any case, it's still cool, but once again we see the difficulty of presenting scientific discoveries in the popular press.
>(esp. for 3d gaming, witch is really the only need consumers have for all that FP)
Well, there are other consumer uses... some image processing (although not most), speech recognition, etc. And, if you are doing any research/programming for scientific applications (as are many of the super-computing people), then floating point performance may be the only thing that one is interested in.
Also, it is not always the case that a single chip system will get much higher flops/chip than a multi-chip system -- it depends on how much communication is required in the particular application. If it is something like parallel search, for example, you need hardly any communication between chips and therefore you get good performance. However, for most consumer applications, you don't get as nice speedup.
The list only counts those systems where linpack has been run. I imagine that there are several beowulf (and other self-built)systems that are actually achieving more computational throughput, but aren't going to be noticed. Also, I'm sure the government doesn't release everything it has working on satellite images and what not. Overall, I'd imagine that a large subset of the people who are actually using huge machines for real work (rather than academic research) wouldn't take the time away from their work to even run linpack on their system.
As one example of such a computer, Professor John Koza has a 1000 node (Pentium II 350Mhz) beowulf machine for his Genetic Programming Inc. ( GPI's web site ) research group. He's running genetic programming applied to difficult problems on the machine (such as automatic analog circuit design), and is getting a nearly linear speedup because of the embarrasingly parallel nature of GP.
In addition to the other good comments posted regarding taking this announcement with a grain of salt, I must add that the new system can only recognize a few words -- with only 11 neurons, it couldn't do much else. Without further information, I would guess that training up a net to recognize more words would be quite complicated -- especially given the non-standard training algorithms that were used. It would be great to find a scientific paper written by the researchers on the issue instead of solely press-release material. -dandre
Slashdot Mirror
The algorithm in the article is actually a cousin of GAs called an evolutionary algorithm. Typically, GAs have crossover (sex) where genetic material from multiple parents is combined. This would be easy to do with the keyboard layout problem, and would probably produce better results.
Often, EAs work better with smaller populations (~100-500) and GAs work better for larger populations, but that's more anecdotal than rigorously shown.
We've had net flix for about 2 months. We've watched about 7 flix, and have 3 ready to watch. Our turnaround is two-three days, because we're in Berkeley, which is nice. The price is right for us, as we're terrible at returning DVDs at blockbuster, so we're saving about 10 bucks a month this way.
Our biggest problem is that it's actually hard to browse their selection. If you know what you want, you can usually find it, but browsing is hard. Their categories aren't as complete as one would like, and the recommendations don't run all that deep (only 10-15 in the indie section, for example). A secondary problem is that we end up with 3 DVDs that we "think" we want to watch someday (e.g. movies like Remains of the Day, Requiem for a Dream) that just sit around because we never quite "feel" like it.
Overall we like the service, but it doesn't replace the impulse movie watching available through our local (non Blockbuster) rental store.
Cheers,
David Andre
I honestly don't think it matters too much what language is used in teaching intro CS courses, as what must be conveyed in those courses are the basic paradigms that are expressible in all languages, from ML to LISP to Java to C.
Intro to CS courses that use Java could be taught badly, as they would be if they focus on UI issues instead of learning to write the fundamentals. Students should be expected to learn how to write linked lists, trees, and all the basic underlying structures of programming. There should not be too much dependence on the fancy features of the language for doing this, IMHO. Learning to use Java's list class is not the same as writing it yourself, and thus understanding how it's written. Java is fine for testing issues of relative speed and for empirical explorations of computational complexity.
One of the key advantages that java has over C (or especially C++) is that it is relatively clean, and relatively safe. On the other hand, it's not as portable as C or C++, which is a real problem for many schools, since Macs are often used in education.
David Andre
UC Berkeley
Scientists report, based on one badly performed and error ridden study, that 1/10 Japanese workers suffers from memory loss, but are considerably more relaxed and happy. A fellow who used to be a salesman was reported as saying "Yeah, I really couldn't remember why I what I was trying to sell anymore, ya know? It sorta had no meaning to me anymore, no depth. I went to a doctor to appease the company (and secure a nice disability check), but man, thank god I'm done with that job".
David Andre
dandre@cs.berkeley.edu
There are several amazing bits to the movie, making it well worth watching.
First, the editing is fantastic. It makes the movie work, and makes what would be depressing or gory actually hilarious. The editing at the beginning of the movie, where the characters are introduced, is simply one of the best bits of editing I've seen this year.
Another great bit is Pitt's performance. Although he often has struck me as a bit of a young annoyingly arrogant actor (e.g. I can act b/c I'm cute), he plays the incoherent gypsy who actually is more "with it" than he seems rather well.
Finally, I recommend seeing it in the theatres, or at least with a large group of people, as it's a film that thrives on the audiences humorous reactions, which are self-reinforcing.
-dandre
Could the whole thing be an elaborate(and twisted) April Fool's joke? Steal the Enigma, then send coded messages back to the authorities regarding its whereabouts?
David Andre
I wish I could LOSE -1.4 Billion. Last I checked, that would mean I made money.....
-da
Does anyone have any advice on hooking up xmms to netscape so that it acts as a plugin instead of just as a standalone app?
In other words, can I use spinner.com when I'm running in linux?
Thanks for any advice.
David Andre
It strikes me that Dvorak missed many of the most important events and/or people of the year by ignoring the advances in technology and science that took place this year. By focusing nearly entirely on the 'biz' side of things, he has a myopic view, especially for importance. Although I'll be the first to admit that the business side of a company is critical, when one looks at the past and attempts to judge importance, inventions and discoveries seem to rank right up there with savvy business dealings.
What about the first realization of a quantum computer (here)? Or IBM's advances in chip technology? Or any of a number of similar advances that are almost certainly important for the future direction of technology?
For that matter, I think leaving out the continued successful rise and development of cellular phones and the like is quite a mistake. When he puts network PCs and ubiquitous computing on the 'flop' list, he misses the most successful of the network ed appliances, the cell phone. The important future of cell phones (which I already had some good ideas about) was made utterly clear to me when, on Christmas this year, I ordered a book from Amazon on my new Sanyo-4000 using the mini-browser on the phone. Took about 4 minutes (including searching for a few things), and was amazingly easy.
Cheers,
David Andre
It seems to me that we could all be being led down a goose chase here. The web page reminds me most of the many hoax web sites that one sees, as we don't see anything that really shows the existence of either a prototype or a working product. It reminds me of those who have sites claiming to have solved the P=NP problem, or those who claim to have a unified field theory (discussed several days ago, on /.) I hate to doubt the veracity of someone, but in this case the unprofessional web site and the lack of technical details at least brings it to mind.
Given that these concerns occur to me (not one talented at all at doing due diligence) it strikes me that VCs would also be suspicious. What evidence is being presented to VCs? Perhaps we could help you tune those documents. Although (as in all fields, even slashdot!) there is some nepotism in VC, good, well presented ideas rise to the top, as always.
David Andre
If when I type 'ssh -V' I get:
SSH Version OpenSSH-1.2, protocol version 1.5. Compiled with SSL.
Is that all right? It doesn't mention RSAREF, so I wasn't sure. I suspect it's fine, given that it doesn't directly mention RSAREF, but I thought I'd check.
Thanks,
David Andre
Although I do believe that there are very real ethical dilemmas surrounding what will happen once HGP is done, it seems to me that JK and most of /. has missed the point, the limitations, and the main dangers of the HGP.
Once the HGP is 100% complete does that mean that the oppression and genetic engineering of humans can start right away? NO! Remember that the HGP is simply the sequencing of the human genome (and only for a subset of humans, obviously). Once we have the sequence and have determined the location of all of the genes, we still have the very difficult task of understanding (1) how the genes form into proteins and (2) the role of these proteins in a developing organism.
In other words, we have yet to solve the protein folding problem and the problem of fully analyzing what proteins do anyway. Although we know a fair bit about what proteins do, we certainly do not fully understand all of their interactions. As for the protein folding problem, anyone who believes we are close to this goal is dreaming. We've only got the 3D structures for about 1000 proteins, and much of the data in that database has flaws and disagreements between researchers. Despite recent large grants for supercomputers for the protein folding problem, being able to accurately simulate the folding of a protein is a long way off.
Without a full understanding of the structure and function of the proteins generated from the genome, we are left with only poor correlational data. In other words -- a gene Y of this type tends to correlate highly with disease or condition Z. But, as we've seen in the breast cancer gene research, the correlations are relatively low, and tend to be of 'enabling' type. Thus, as expected, ones genetic predisposition for something is only part of the story.
It makes some sense to worry about the ethical problems surrounding knowing some correlational information about how gene X affects a persons behavior. If we found a super high correlation between the presence of a gene and violent anti-social behavior, it would be an ethical minefield, for sure. This is the topic that we should be discussing and worrying about, not whether genetic engineering of the perfect child is appropriate or ethical.
In any case, I highly doubt that the population will stand for any significant abuses of this sort of information. We don't allow country of origin (probably the highest correlative factor for tracking down credit card fraud, for example) to be used by the credit card companies, so I doubt that genetic discrimination will be allowed in any but the most clear of cases.
Furthermore, the scariest kinds of discrimination (job discrimination, etc) are the least likely to be highly correlated. The most likely things to be highly correlated are dangerous medical problems (such as sickle cell anemia, for example, for which the correlation is perfect -- with a malformed hemoglobin gene, you get sickle cell anemia b/c the structure of the blood cells is compromised by the malformity -- although it does impart a higher resistance to malaria, fwiw), which can be used by the medical community as indicators to help deal with treatment.
Cheers,
David Andre
Although I'm for the little guy in this case, the goto.com logo is much interesting and less catching than the go.com design. If I had to choose one, it'd be go.com. But, copyright law protects good and bad design alike. Just because the goto.com logo isn't as effective at evoking the go streetlight symbol as is the go.com logo doesn't mean that it wasn't there first. It's obvious that the two are similar, and goto.com obviously felt like they were being harmed by the similarity. Goto.com is harmed each time a user mistakenly types go.com and doesn't find their search engine. Besides, the publicity of winning the case certainly helps them -- it appears that their site is /.'ed.
cheers,
dandre
What I was surprised by was how tongue in cheek and ironic Gore was in the letter. The "Sonics" bit, the reference to Dubya's failed pop quiz -- I think it's part of a direct effort to seem hip and in touch, to reach the politically cynical and doubting tech crowd. Rather strange for a politician -- a fair bit more subtlety than I expected out of Gore.
Cheers,
dandre
Even worse, in the Oakland Tribune, they put the image on their front page, with a caption identifying it as a depiction of an image of the planet from the Berkeley astronomy site. It was not at all mentioned that it was simply an artist's rendition. Claiming that it was "seen" is also a bit odd -- more like it was detected using a visual medium. It's not like a human is sitting there, staring at the star, waiting for a shadow to cross over it. In any case, it's still cool, but once again we see the difficulty of presenting scientific discoveries in the popular press.
Cheers,
David Andre
>(esp. for 3d gaming, witch is really the only need consumers have for all that FP)
... some image processing (although not most), speech recognition, etc. And, if you are doing any research/programming for scientific applications (as are many of the super-computing people), then floating point performance may be the only thing that one is interested in.
Well, there are other consumer uses
Also, it is not always the case that a single chip system will get much higher flops/chip than a multi-chip system -- it depends on how much communication is required in the particular application. If it is something like parallel search, for example, you need hardly any communication between chips and therefore you get good performance. However, for most consumer applications, you don't get as nice speedup.
Cheers,
David Andre
The list only counts those systems where linpack has been run. I imagine that there are several beowulf (and other self-built)systems that are actually achieving more computational throughput, but aren't going to be noticed. Also, I'm sure the government doesn't release everything it has working on satellite images and what not. Overall, I'd imagine that a large subset of the people who are actually using huge machines for real work (rather than academic research) wouldn't take the time away from their work to even run linpack on their system.
As one example of such a computer, Professor John Koza has a 1000 node (Pentium II 350Mhz) beowulf machine for his Genetic Programming Inc. ( GPI's web site ) research group. He's running genetic programming applied to difficult problems on the machine (such as automatic analog circuit design), and is getting a nearly linear speedup because of the embarrasingly parallel nature of GP.
Cheers,
David Andre
my web site
disclaimer: I worked with Professor Koza for several years and helped him build some of his previous machines.
In addition to the other good comments posted regarding taking this announcement with a grain of salt, I must add that the new system can only recognize a few words -- with only 11 neurons, it couldn't do much else. Without further information, I would guess that training up a net to recognize more words would be quite complicated -- especially given the non-standard training algorithms that were used. It would be great to find a scientific paper written by the researchers on the issue instead of solely press-release material. -dandre