From what I remember (and it has been a while), some of the problems that this approach has been applied to in the past included the traveling salesman problem (or at least the observation that some of the molecules gravitated into a low-energy state that happened to also solve TSP), which is why I brought up NP-complete in the first place. This certainly doesn't make NP-complete go away; they are just doing a near-exhaustive search in a small amount of time.
The error rate that they cited (98% or whatever) probably gets much worse as they increase the problem size, since the probability of randomly discovering an answer of the increased complexity would drop exponentially (which, as you point out, is one of the reason NP-complete problems are so nasty).
It's my understanding that all they are doing is allowing molecules to combine into a tremendous number of configurations, then filtering out the ones that don't have the characteristics they'd expect from a solution to a particular problem. Then they just verify the shape of the structure of the remaining molecules. It's only slightly more sophisticated than having a trillion monkeys typing on a trillian keyboards (except in this case, they know when a monkey is close to the answer they want).
It might be possible to solve NP-complete problems in this fashion (i.e. is there a hamiltonian circuit containing N vertices in this molecule's structure), but the amount of time and effort needed to set up the system and filter out the results does not seem worthwhile. Further, this requires that they already know what kind of structure they expect as an answer (in order to filter it out from the rest), so it will only work on problems where they already have a good guess about the answer. Not something you can expect to see as a general problem-solver.
In otherwords, I don't expect to see Apache running on this anytime, ever. Might be interesting for conjecture, but my money's on quantum computing for this kind of problem solving (at least q-bits have a chance of being interfaced with existing computer hardware).
so please don't get disheartened, just use a more realistic compression setting.
I just come from a long tradition of gzip -9 and kill -9 -- I like my programs to have maximum effect. But I suppose I should have tried an easier setting.
Even so, the 118MB file only came down to 70-something megs, for about 35% compression, which just isn't quite enough for it to be worthwhile for me. I'd like to be able to archive my CD collection at lossless quality, but being able to backup the archive in a cost-effective manner would require that I can at least reliably cut the size in half. Even then, backing up 80 CDs onto 40 CDs does not cut it for me (at that point, trying to save the archive from a fire is almost as hard as trying to save the original discs).
But I will try FLAC again, right after I pick up a few 100GB drives to put into a RAID, so backup becomes automatic. Or when DVD burners come down in price.
On your recommendation, I went out, grabbed the FLAC source, and ripped the Sanctuary Remix of Delerium's "Silence", off of the 2CD version of "Karma".
The track is typical of what I listen to in terms of content and genre, and when doing a VBR compression using notlame (32-320kbps, 44.1Khz), it averages 209kbps, and takes up 17.6MB (compared
to the CDDA image, which is 118MB). The compression to MP3 took less than 10 minutes using notlame, even though I have all the features turned on.
FLAC is still working on the same track. I made
it as aggressive as it would go, since a one-time CPU cost is worth a long-term storage savings (especially if I burn the.flac files back to CD; I would not want to have to split albums across CD's, so I want as close to 2:1 as I can get). On my Athlon 600, it has taken 39 minutes to do 20% of the 11 minute track, with only a 43% reduction in file size to that point.
So, in order to do my entire CD collection (currently 7.5GB of MP3s), it would take about 58 days of CPU time (for a program that does not seem to multithread, so an SMP system would only help if flac were run repeatedly by hand). This is for a collection that can be played in its entirety
in under 80 hours.
It just seems a bit impractical. I mean, by the
time I finish compressing all of those files, hard drive prices will have dropped to the point that it would be just as economical to keep the files in their raw CDDA format, and I wouldn't have paid for the electricity to keep my processor pegged for two months.
I'm all for lossless, but I can't tell the difference between the real thing and mp3 files with carefully chosen encoding parameters. I think I'd rather wait for the lossless compression codecs to mature a bit.
And it's still encoding. 30% after 58 minutes of CPU time at 600Mhz, estimated 43% savings, if I let it go another two hours.
This may come off sounding redundant, but there is
a lot to be said about using the correct settings for your encoding. I went about building a dedicated MP3 player/server which plays nonstop to both my stereo and an icecast stream, and also serves Samba shares. Before encoding my entire music collection (only 84 albums, but I don't want to do it twice) I did some research on the subject.
The r3mix comparison shows that LAME (and by extension, notlame) is probably the best way to go for mp3 encoding. Some
discussion is also provided regarding the advantages of using VBR for encoding in general (click on the "encoding" tab).
As for my own observations, I found that using this set of command line arguments works best:
The results are files that will go as high as 320Kbps if needed (on average, they tend to float around 200Kbps) at 44.1Khz. There is no loss of quality that I can hear; the parts of the music that need a full 320kbps get it, and the silent parts drop to 32kbps.
And the players I've tried all seem to work fine; winamp and xmms (using mpg123) have no trouble with the VBR files. I don't know about portable players. For streaming, I use liveice to resample the files to a particular CBR (128Kbps for my apartment, 64Kbps for sending out over the network); so I can listen to the files at whatever quality I want without sacrificing the quality of my archival files.
So, if I did own a Rio or a CD player that would read mp3 CD's, I could just resample the files I wanted to a particular bitrate (pretty much on the fly).
I see no reason why anybody would only sample their files at 160kbps or 256kbps; every file I've done has needed 320kbps at some point to get the best fit. And the size isn't unreasonable; a typical album is between 70-80MB, with some extreme cases taking up to 110MB. Sure, it's big,
but I've only used 25% of the 30GB drive I bought for the purpose (which right now costs very little), including the RedHat install to run
the machine.
As for using Ogg, I looked into it. If I could be convinced that the quality would be good enough to archive my entire collection, I would have used it. But it just didn't seem mature enough at the time I began all of this (such as no support for joint stereo), so I chose mp3.
Of course, with storage at its current price, I am tempted to give up compression altogether and just store entire CDDA images. A typical album would then take around 700MB to store, but a $200 drive could then be used to keep over 100 CDs with absolutely no loss in quality.
Regarding speaker/mic: some of the Inuktuns (including the MicroTracs) have a microphone and speakers, with a headset that the operator can wear.
Regarding tethered/wireless: the Inuktuns are tethered. This limits their range, but makes it possible to use them continuously, and there is really no risk of losing communications (unless the tether is physically broken). Further, if needed, the robot can be retrieved by following the tether.
The iRobot/RWI robots are wireless, and would normally be capable of running with full autonomy (i.e. they have onboard processing and a sensor suite). They are limited by their battery life, and if the wireless communications are lost (due to interference or otherwise), they may be difficult to recover. For this reason, the Inuktuns were used primarily.
Check out
the USF USAR page (not yet updated with WTC
information) for more.
One problem with this is that the material in unpublished papers is considered very sensitive. If somebody were to use such a system to read a paper, then steal the results of that research to file a patent, the original researcher could get screwed.
I just don't see those kinds of protections in the Slashdot/kuro5hin models.
Okay, now that I'm properly awake, here are some examples of "military technology" that have been put to "peaceful purposes":
Encryption. One of the earlier known forms of encryption, the so-called Caesar cipher, was used to transfer information to generals in the field. Encryption has been studied, and used, by the military ever since. However, at the same time, you can use SSL or SSH or whatever to securely send your CC numbers or email.
Computers. The first examples of modern computers (electronic, not Babbage's) were a result of military research. Bletchley Park developed what is arguably the first modern computer to break encryption, and ENIAC was built to compute ballistics tables for artillery. Yet, your post to Slashdot uses computer technology peacefully.
The Space Program. The V1 and V2 rockets designed by the Nazis to hurl high explosives at London were a terror weapon during WWII. However, the same technology that enabled that (as well as the ICBM's to follow) also created the space program, which has even been used recently for tourism (which is a peaceful enterprise).
GPS. Among other things that have been boosted into orbit, the GPS constellation was initially a military venture. Nowadays, you can get a GPS receiver for your Palm.
The Internet. Or shall I say ARPA Net? The protocols behind the Internet were developed by and for DARPA. It just turns out that the rest of the world had something else to use them for.
At the risk of running late for work, I'll stop here. But in closing: technology research is something you should be applauding, regardless. Technology does not exist in a vacuum; you cannot develop it for one application (say, military) without finding benefits elsewhere (say, civilian).
So-called "military technology" will have peaceful applications, just as many things that start out on the "peaceful" side of the fence are adopted for military purposes.
Supposedly, Archimedes of Syracuse designed machines to ward off Roman invaders based on levers and pulleys, some of the most basic machines that exist. It isn't so much that these are "military technology", but that they are technology that can be put to, among other things, military uses.
Similarly, any implement you can grab, from a phone receiver to a frozen ham, can be used as a lethal weapon.
So even if its initial purpose is military, the results of such research are likely to be useful for far more. "Seek and destroy" becomes "search and rescue", and those heavy bombers become your commercial airliners. Same tech, different app.
As far as collaborative, web-based artwork, the
Internet Movie Project has been going on for a while now. There is also the IRTCGroup Image Project which is a collaborative, web-based art endeavor.
Of course, in both of these cases, the projects are not being done blindly; everybody who contributes has some idea of the bigger picture (pun may or may be intended; it's early yet).
"...but why would humans make disobedient robots?"
Because sometimes, free will (or at least an independent line of reasoning) is the only way to find an answer. Attributed to Patton:
"Never tell people how to do things. Tell them what to do and they will surprise you with their ingenuity."
If an artificial intelligence was not capable of being disobedient, would you really consider it intelligent? In other words, if all it were capable of doing was that which its (possibly myopic) creator/controller had said, then how useful will it be?
Asimov's laws of robotics were an interesting idea, but I would not expect to see them in the real thing. They are too simple as stated; for a robot (or AI or whatever) to understand when it might be breaking those laws, it would probably already be smart enough to disregard them. The answer? A bright, red, easily accessible kill switch. And a remote kill switch in case the thing gets away.
I found the following quote in the St. Pete article attached to the story:
"That's awesome," said Souders, 35, a caterer. "If you don't have anything to worry about, it won't bother you. As far as any
invasion of rights -- if you're breaking the law, your rights are kind of dissolved."
If the argument "if you don't have anything to worry about, it shouldn't bother you" makes sense to you, then you have no respect for privacy (your own or that of anybody else). It's about information, and how little information is needed to separate the "good" people from the "bad" people. It is due to a logical fallacy, which
only stands if people do not recognize it as such.
This is the same kind of argument that marginalizes encryption. In other words, if encryption is only used by people who have something to hide, then anybody using encryption is suspect. Same goes for the paper shredder I recently bought: if I only shred sensitive documents, then it can be assumed that anything I shred is important.
Examples:
"If only people who are trying to hide something
complain about public surveillance, then simply investigate those who complain." This kind of thinking will destroy privacy with amazing efficiency. If people fight the surveillance system for the sake of privacy in general, then it remains impossible to isolate people based on their reaction to the new technology.
"If only people who are hiding something use encryption, then simply investigate those who use encryption." One way destroy this kind of thinking is to use encryption as much as possible to add noise to the system.
"If only important documents are shredded, then it may be worthwhile to reconstruct everything that is shredded." Shred everything. Doing so lowers any perceived value of whatever is shredded. It also increases the volume of things to reconstruct, and makes it difficult to determine how much or how often something important is shredded (i.e. traffic analysis).
I realize that this all sounds somewhat paranoid, but consider how little information it may require to go smashing down your privacy. Just because a technology makes life more difficult for criminals does not mean it is good for everybody else. In other words: I have nothing to hide, but you are just going to have to take my word on that, unless (through due process) you have real evidence to the contrary. Without this attitude, there can be no privacy.
Sorry for the insinuation on the original post (as this drifts offtopic). After shelling out for LW, I take a low view of pirates. Mine is the educational edition as well.
The L[6] to L[6.5] upgrade can be had at www.lightwave6.com as well as a feature list. The new features include better volumetrics, some plugins, and faster renders.
Very cool software. The docs are there as well,
which (as I understand) are better written than
the original L[6] docs.
The OpenGL textured previews are not rendered in realtime, however. They are rendered (slowly) into memory,
then played back from there. You can run an
animation in near-real-time if you are content
with wireframe only; but if you are using plugins
to generate motion, it will still be well short of
real-time, as it is a CPU burden rather than the
video card.
To render a real frame, with quality cranked up, can still take hours, but this is true of any
rendering package.
Incidentally, the upgrade from Lightwave 6 to 6.5 is free, so
unless you've got a cracked copy, you really ought
to be using 6.5. I've been using Lightwave since 5.6 and 6.5 is incredible by comparison.
In reality, we see the world by combining the images formed by our left and right eye and get our perception of depth by viewing objects from two points separated by a few inches. Our brain takes care of the rest.
Two nitpicks: first, if my eyes were a "few inches" apart, I'd look quite a bit different -- the distances between eyes (on humans at least) is
only a couple of inches, max.
Second nitpick: studies in cognitive psychology have shown that the perception of depth does not benefit all that much from stereo vision. Yes, humans get depth from stereo. However, humans also get depth from occlusion, optic flow, expectation (how large you think something should be impacts your perception of how far away it is), and other visual clues like shadowing, perspective, context (how big things around an object are), and depth of focus (i.e. your eyes provide feedback about focal depth). There is also depth from convergence (your eyes don't look in parallel; you become more cross-eyed as you study things directly in front of your nose). Studies have shown that of these, depth from stereo contributes very little to the overall sensation of depth.
Not to say that a 3D display isn't cool. But much of the illusion of depth is already illustrated in 2D images, from motion, occlusion, and lighting cues.
The only other thing to do would be to get real-time random-dot stereograms working on your computer monitor.
I wanted to point out that perhaps the reason that the animation (which did use a lot of matte backdrops) followed in the footsteps of Antz more than Toy Story 2 is because Dreamworks did Antz and Shrek. The next Pixar flick (Monsters, Inc.) will feel more like the Toy Story series.
The state of the art in CGI is roughly as follows:
in Toy Story, everything looked plastic...but that was OK, because all of the characters were supposed to be plastic. This goes not only for the surfaces, but the modeling and motion as well. Antz and Bug's Life improved on these quite a bit; still using the convenient exoskeleton to keep from having to animate much skin, but apart from more detail and more natural scenes (like the tree in Bug's Life), there were still some things lacking: good rendering of hair and fur, good surface textures, and realistic motion on some characters.
Stewart Little and Dinosaur began to push the state of the art; the creators of Stuart Little had to create not only realistic fur for the lead, but also cloth. In the process, they contributed to the power of Alias|Wavefront's Maya package (some of their hair/cloth code made it into Maya).
It is getting better, and Shrek illustrates this (as will Monsters, Inc.). Detail is incredible, surface textures are bright and realistic, and motion is getting better (some characters were smooth, others were lacking (a bit like the skipping motions used in Toy Story 2, in both the "When somebody loves you" bit and the airport)). Fur is looking pretty good, and hair is OK (it renders well, but despite how the
various Chris Landreth Maya animations, including "The End" and Bingo have turned out, it doesn't blow in the wind much).
It is getting better, though. The faces on the humans were incredibly detailed, and there is a tremendous attention to detail, both in character animation and in the scenery. I greatly anticipate the (not-so-distant) future of this style of CGI animation.
I wish they had listed the software/hardware they used at the end (even just a kudos to SGI or Sun or somebody). It is nice to know what was used.
UC Berkeley offers a service
(Moss: Measure of Software Similarity) that allows
programs written in many languages (C/C++, ML, LISP, ADA, Pascal, Scheme) to be automatically examined and the results mailed back. It actually parses out the code and looks for matching parse trees, so regardless of changes in indenting or variable names, people who cheat show up with a high correlation. It also nicely formats all of the results (color-coded, even) to allow for lines of similar code to be compared head-to-head. It offers features such as being able to specify a base file that others were derived from, and particular patterns to expect to see often that should not be flagged.
This is not enough to accuse people of cheating, but it is certainly a good way to find papers that should be examined more closely. When I have graded papers, I usually catch everything that Moss does, but it is nice to know I haven't missed anything.
Programs like this will only become more common. I, for one, am glad; if people who are not willing to do work are barred from getting a BS or MS degree, then it makes my own degrees mean a little more.
There are already numerous pieces of software out
there that require a regular payment to keep it
active. Alias|Wavefront Maya, Matlab, and
Allegro Common Lisp all require regular upkeep
(at least for the licenses I've seen). Newtek's
Lightwave may begin offering an academic license
that is renewed anually as well.
While I personally steer clear of software with this kind of license (I tend to save up all of my pennies for a large one-time purchase) this is certainly nothing new.
Apart from the representation in inodes versus
FAT tables versus everything else, don't many filesystems also have a minimum disk block size? If he split the file into 218 pieces, then each one would take up 14430 bytes (on avereage, assuming uniformly random data), which is not evenly divisible by any significant power of two...meaning that unused space in each disk block at the end of each file would occupy far more than the savings gained by splitting the file in the first place.
As an aside, I find it amusing that one of the people he quoted regarding the outcome of the contest was one David Scott, who sounds a lot like someone who also posts to sci.crypt, and gives./ trolls a bad name. The SCOTT16U.ZIP controversy, in which he personally blasted Bruce Schneier, was particularly amusing.
I just retired a couple of characters on MUME recently...one had 27 days of play time, and the other around 20. Those were two, and at one time I had seven.
I know that Everquest is addictive, which is why I will never play it. But for sheer mind-sucking addiction, free of charge, MUDs are still ranked pretty high in my book. I probably spent six months doing nothing but playing one game, over the course of four years.
Bingo is probably one of the best short animations I've ever seen. The link will lead to a 24MB download; well worth it.
405 is also very well done; it was somebody's personal project that turned out to be extremely succssful.
They also provide materials that could be useful in a teaching environment.
Flay is a Lightwave-specific site, but they post animations (including 405, above) as they are discovered.
This may be a good thing, especially for everybody who is concerned about privacy.
I'm serious, and on-topic.
If you read Schneier's "Secrets and Lies", he talks about the need for a trusted path within a computer, which can be used to implement access control. That is, people who are supposed to be able to see files can see them, and nobody else can -- in such a way that trojans, viruses, and the like cannot violate.
If we have technology to enforce access control on media files built into every new computer that comes off the line, then it will be that much easier to improve the security of those computers. By adding more authentication and access control, it will actually become easier to lock down a computer and protect its data.
As far as I see it, losing the ability to copy protected works without limit also comes with the protection from the unlimited copying or alteration of critical system files. This could be a good thing. And even if the DMCA gets thrown out and the RIAA told to fsck themselves, the technology that everybody is getting upset over right now will still be there, and may be of very great value.
Slashdot Mirror
From what I remember (and it has been a while), some of the problems that this approach has been applied to in the past included the traveling salesman problem (or at least the observation that some of the molecules gravitated into a low-energy state that happened to also solve TSP), which is why I brought up NP-complete in the first place. This certainly doesn't make NP-complete go away; they are just doing a near-exhaustive search in a small amount of time.
The error rate that they cited (98% or whatever) probably gets much worse as they increase the problem size, since the probability of randomly discovering an answer of the increased complexity would drop exponentially (which, as you point out, is one of the reason NP-complete problems are so nasty).
It's my understanding that all they are doing is allowing molecules to combine into a tremendous number of configurations, then filtering out the ones that don't have the characteristics they'd expect from a solution to a particular problem. Then they just verify the shape of the structure of the remaining molecules. It's only slightly more sophisticated than having a trillion monkeys typing on a trillian keyboards (except in this case, they know when a monkey is close to the answer they want).
It might be possible to solve NP-complete problems in this fashion (i.e. is there a hamiltonian circuit containing N vertices in this molecule's structure), but the amount of time and effort needed to set up the system and filter out the results does not seem worthwhile. Further, this requires that they already know what kind of structure they expect as an answer (in order to filter it out from the rest), so it will only work on problems where they already have a good guess about the answer. Not something you can expect to see as a general problem-solver.
In otherwords, I don't expect to see Apache running on this anytime, ever. Might be interesting for conjecture, but my money's on quantum computing for this kind of problem solving (at least q-bits have a chance of being interfaced with existing computer hardware).
I just come from a long tradition of gzip -9 and kill -9 -- I like my programs to have maximum effect. But I suppose I should have tried an easier setting.
Even so, the 118MB file only came down to 70-something megs, for about 35% compression, which just isn't quite enough for it to be worthwhile for me. I'd like to be able to archive my CD collection at lossless quality, but being able to backup the archive in a cost-effective manner would require that I can at least reliably cut the size in half. Even then, backing up 80 CDs onto 40 CDs does not cut it for me (at that point, trying to save the archive from a fire is almost as hard as trying to save the original discs).
But I will try FLAC again, right after I pick up a few 100GB drives to put into a RAID, so backup becomes automatic. Or when DVD burners come down in price.
The track is typical of what I listen to in terms of content and genre, and when doing a VBR compression using notlame (32-320kbps, 44.1Khz), it averages 209kbps, and takes up 17.6MB (compared to the CDDA image, which is 118MB). The compression to MP3 took less than 10 minutes using notlame, even though I have all the features turned on.
FLAC is still working on the same track. I made it as aggressive as it would go, since a one-time CPU cost is worth a long-term storage savings (especially if I burn the .flac files back to CD; I would not want to have to split albums across CD's, so I want as close to 2:1 as I can get). On my Athlon 600, it has taken 39 minutes to do 20% of the 11 minute track, with only a 43% reduction in file size to that point.
So, in order to do my entire CD collection (currently 7.5GB of MP3s), it would take about 58 days of CPU time (for a program that does not seem to multithread, so an SMP system would only help if flac were run repeatedly by hand). This is for a collection that can be played in its entirety in under 80 hours.
It just seems a bit impractical. I mean, by the time I finish compressing all of those files, hard drive prices will have dropped to the point that it would be just as economical to keep the files in their raw CDDA format, and I wouldn't have paid for the electricity to keep my processor pegged for two months.
I'm all for lossless, but I can't tell the difference between the real thing and mp3 files with carefully chosen encoding parameters. I think I'd rather wait for the lossless compression codecs to mature a bit.
And it's still encoding. 30% after 58 minutes of CPU time at 600Mhz, estimated 43% savings, if I let it go another two hours.
The r3mix comparison shows that LAME (and by extension, notlame) is probably the best way to go for mp3 encoding. Some discussion is also provided regarding the advantages of using VBR for encoding in general (click on the "encoding" tab).
As for my own observations, I found that using this set of command line arguments works best:
notlame -v -V 1 -B 320 -p tempcdda.wav [outputfile]
The results are files that will go as high as 320Kbps if needed (on average, they tend to float around 200Kbps) at 44.1Khz. There is no loss of quality that I can hear; the parts of the music that need a full 320kbps get it, and the silent parts drop to 32kbps.
And the players I've tried all seem to work fine; winamp and xmms (using mpg123) have no trouble with the VBR files. I don't know about portable players. For streaming, I use liveice to resample the files to a particular CBR (128Kbps for my apartment, 64Kbps for sending out over the network); so I can listen to the files at whatever quality I want without sacrificing the quality of my archival files.
So, if I did own a Rio or a CD player that would read mp3 CD's, I could just resample the files I wanted to a particular bitrate (pretty much on the fly).
I see no reason why anybody would only sample their files at 160kbps or 256kbps; every file I've done has needed 320kbps at some point to get the best fit. And the size isn't unreasonable; a typical album is between 70-80MB, with some extreme cases taking up to 110MB. Sure, it's big, but I've only used 25% of the 30GB drive I bought for the purpose (which right now costs very little), including the RedHat install to run the machine.
As for using Ogg, I looked into it. If I could be convinced that the quality would be good enough to archive my entire collection, I would have used it. But it just didn't seem mature enough at the time I began all of this (such as no support for joint stereo), so I chose mp3.
Of course, with storage at its current price, I am tempted to give up compression altogether and just store entire CDDA images. A typical album would then take around 700MB to store, but a $200 drive could then be used to keep over 100 CDs with absolutely no loss in quality.
Regarding tethered/wireless: the Inuktuns are tethered. This limits their range, but makes it possible to use them continuously, and there is really no risk of losing communications (unless the tether is physically broken). Further, if needed, the robot can be retrieved by following the tether.
The iRobot/RWI robots are wireless, and would normally be capable of running with full autonomy (i.e. they have onboard processing and a sensor suite). They are limited by their battery life, and if the wireless communications are lost (due to interference or otherwise), they may be difficult to recover. For this reason, the Inuktuns were used primarily.
Check out
the USF USAR page (not yet updated with WTC
information) for more.
There's slightly more information available (also confirming this) at Excite news.
Your nuke will be delivered in 30 minutes or less, or it's free.
Yes, I'm pretty sure this has military potential.
I just don't see those kinds of protections in the Slashdot/kuro5hin models.
Encryption. One of the earlier known forms of encryption, the so-called Caesar cipher, was used to transfer information to generals in the field. Encryption has been studied, and used, by the military ever since. However, at the same time, you can use SSL or SSH or whatever to securely send your CC numbers or email.
Computers. The first examples of modern computers (electronic, not Babbage's) were a result of military research. Bletchley Park developed what is arguably the first modern computer to break encryption, and ENIAC was built to compute ballistics tables for artillery. Yet, your post to Slashdot uses computer technology peacefully.
The Space Program. The V1 and V2 rockets designed by the Nazis to hurl high explosives at London were a terror weapon during WWII. However, the same technology that enabled that (as well as the ICBM's to follow) also created the space program, which has even been used recently for tourism (which is a peaceful enterprise).
GPS. Among other things that have been boosted into orbit, the GPS constellation was initially a military venture. Nowadays, you can get a GPS receiver for your Palm.
The Internet. Or shall I say ARPA Net? The protocols behind the Internet were developed by and for DARPA. It just turns out that the rest of the world had something else to use them for.
At the risk of running late for work, I'll stop here. But in closing: technology research is something you should be applauding, regardless. Technology does not exist in a vacuum; you cannot develop it for one application (say, military) without finding benefits elsewhere (say, civilian).
Supposedly, Archimedes of Syracuse designed machines to ward off Roman invaders based on levers and pulleys, some of the most basic machines that exist. It isn't so much that these are "military technology", but that they are technology that can be put to, among other things, military uses.
Similarly, any implement you can grab, from a phone receiver to a frozen ham, can be used as a lethal weapon.
So even if its initial purpose is military, the results of such research are likely to be useful for far more. "Seek and destroy" becomes "search and rescue", and those heavy bombers become your commercial airliners. Same tech, different app.
Of course, in both of these cases, the projects are not being done blindly; everybody who contributes has some idea of the bigger picture (pun may or may be intended; it's early yet).
Because sometimes, free will (or at least an independent line of reasoning) is the only way to find an answer. Attributed to Patton: "Never tell people how to do things. Tell them what to do and they will surprise you with their ingenuity."
If an artificial intelligence was not capable of being disobedient, would you really consider it intelligent? In other words, if all it were capable of doing was that which its (possibly myopic) creator/controller had said, then how useful will it be?
Asimov's laws of robotics were an interesting idea, but I would not expect to see them in the real thing. They are too simple as stated; for a robot (or AI or whatever) to understand when it might be breaking those laws, it would probably already be smart enough to disregard them. The answer? A bright, red, easily accessible kill switch. And a remote kill switch in case the thing gets away.
"That's awesome," said Souders, 35, a caterer. "If you don't have anything to worry about, it won't bother you. As far as any invasion of rights -- if you're breaking the law, your rights are kind of dissolved."
If the argument "if you don't have anything to worry about, it shouldn't bother you" makes sense to you, then you have no respect for privacy (your own or that of anybody else). It's about information, and how little information is needed to separate the "good" people from the "bad" people. It is due to a logical fallacy, which only stands if people do not recognize it as such.
This is the same kind of argument that marginalizes encryption. In other words, if encryption is only used by people who have something to hide, then anybody using encryption is suspect. Same goes for the paper shredder I recently bought: if I only shred sensitive documents, then it can be assumed that anything I shred is important.
Examples:
"If only people who are trying to hide something complain about public surveillance, then simply investigate those who complain." This kind of thinking will destroy privacy with amazing efficiency. If people fight the surveillance system for the sake of privacy in general, then it remains impossible to isolate people based on their reaction to the new technology.
"If only people who are hiding something use encryption, then simply investigate those who use encryption." One way destroy this kind of thinking is to use encryption as much as possible to add noise to the system.
"If only important documents are shredded, then it may be worthwhile to reconstruct everything that is shredded." Shred everything. Doing so lowers any perceived value of whatever is shredded. It also increases the volume of things to reconstruct, and makes it difficult to determine how much or how often something important is shredded (i.e. traffic analysis).
I realize that this all sounds somewhat paranoid, but consider how little information it may require to go smashing down your privacy. Just because a technology makes life more difficult for criminals does not mean it is good for everybody else. In other words: I have nothing to hide, but you are just going to have to take my word on that, unless (through due process) you have real evidence to the contrary. Without this attitude, there can be no privacy.
The L[6] to L[6.5] upgrade can be had at www.lightwave6.com as well as a feature list. The new features include better volumetrics, some plugins, and faster renders.
Very cool software. The docs are there as well, which (as I understand) are better written than the original L[6] docs.
Incidentally, the upgrade from Lightwave 6 to 6.5 is free, so unless you've got a cracked copy, you really ought to be using 6.5. I've been using Lightwave since 5.6 and 6.5 is incredible by comparison.
In reality, we see the world by combining the images formed by our left and right eye and get our perception of depth by viewing objects from two points separated by a few inches. Our brain takes care of the rest.
Two nitpicks: first, if my eyes were a "few inches" apart, I'd look quite a bit different -- the distances between eyes (on humans at least) is only a couple of inches, max.
Second nitpick: studies in cognitive psychology have shown that the perception of depth does not benefit all that much from stereo vision. Yes, humans get depth from stereo. However, humans also get depth from occlusion, optic flow, expectation (how large you think something should be impacts your perception of how far away it is), and other visual clues like shadowing, perspective, context (how big things around an object are), and depth of focus (i.e. your eyes provide feedback about focal depth). There is also depth from convergence (your eyes don't look in parallel; you become more cross-eyed as you study things directly in front of your nose). Studies have shown that of these, depth from stereo contributes very little to the overall sensation of depth.
Not to say that a 3D display isn't cool. But much of the illusion of depth is already illustrated in 2D images, from motion, occlusion, and lighting cues.
The only other thing to do would be to get real-time random-dot stereograms working on your computer monitor.
Incidentally, download link of Bingo can be found at http://scifi.ign.com/movies/1677.html. Good stuff.
The state of the art in CGI is roughly as follows: in Toy Story, everything looked plastic...but that was OK, because all of the characters were supposed to be plastic. This goes not only for the surfaces, but the modeling and motion as well. Antz and Bug's Life improved on these quite a bit; still using the convenient exoskeleton to keep from having to animate much skin, but apart from more detail and more natural scenes (like the tree in Bug's Life), there were still some things lacking: good rendering of hair and fur, good surface textures, and realistic motion on some characters.
Stewart Little and Dinosaur began to push the state of the art; the creators of Stuart Little had to create not only realistic fur for the lead, but also cloth. In the process, they contributed to the power of Alias|Wavefront's Maya package (some of their hair/cloth code made it into Maya).
It is getting better, and Shrek illustrates this (as will Monsters, Inc.). Detail is incredible, surface textures are bright and realistic, and motion is getting better (some characters were smooth, others were lacking (a bit like the skipping motions used in Toy Story 2, in both the "When somebody loves you" bit and the airport)). Fur is looking pretty good, and hair is OK (it renders well, but despite how the various Chris Landreth Maya animations, including "The End" and Bingo have turned out, it doesn't blow in the wind much).
It is getting better, though. The faces on the humans were incredibly detailed, and there is a tremendous attention to detail, both in character animation and in the scenery. I greatly anticipate the (not-so-distant) future of this style of CGI animation.
I wish they had listed the software/hardware they used at the end (even just a kudos to SGI or Sun or somebody). It is nice to know what was used.
This is not enough to accuse people of cheating, but it is certainly a good way to find papers that should be examined more closely. When I have graded papers, I usually catch everything that Moss does, but it is nice to know I haven't missed anything.
Programs like this will only become more common. I, for one, am glad; if people who are not willing to do work are barred from getting a BS or MS degree, then it makes my own degrees mean a little more.
While I personally steer clear of software with this kind of license (I tend to save up all of my pennies for a large one-time purchase) this is certainly nothing new.
As an aside, I find it amusing that one of the people he quoted regarding the outcome of the contest was one David Scott, who sounds a lot like someone who also posts to sci.crypt, and gives ./ trolls a bad name. The SCOTT16U.ZIP controversy, in which he personally blasted Bruce Schneier, was particularly amusing.
I just retired a couple of characters on MUME recently...one had 27 days of play time, and the other around 20. Those were two, and at one time I had seven. I know that Everquest is addictive, which is why I will never play it. But for sheer mind-sucking addiction, free of charge, MUDs are still ranked pretty high in my book. I probably spent six months doing nothing but playing one game, over the course of four years.
405 is also very well done; it was somebody's personal project that turned out to be extremely succssful. They also provide materials that could be useful in a teaching environment.
Flay is a Lightwave-specific site, but they post animations (including 405, above) as they are discovered.
The Internet Raytracing Competition has numerous animations...some of them appeal to you.
I'm serious, and on-topic.
If you read Schneier's "Secrets and Lies", he talks about the need for a trusted path within a computer, which can be used to implement access control. That is, people who are supposed to be able to see files can see them, and nobody else can -- in such a way that trojans, viruses, and the like cannot violate.
If we have technology to enforce access control on media files built into every new computer that comes off the line, then it will be that much easier to improve the security of those computers. By adding more authentication and access control, it will actually become easier to lock down a computer and protect its data.
As far as I see it, losing the ability to copy protected works without limit also comes with the protection from the unlimited copying or alteration of critical system files. This could be a good thing. And even if the DMCA gets thrown out and the RIAA told to fsck themselves, the technology that everybody is getting upset over right now will still be there, and may be of very great value.