Has anyone seen it? I hope not. Well, I guess it's not much worse than other 60s sci-fi (except the good ones), but it's pretty bad. I mean, did you check out those rocket jets? And the end...charles dickens...ugh!
But like all ridiculously bad movies, it can be pretty funny to watch:)
Imaging a shell script called ILOVEYOU.txt arrives as an attachment to a newbie Linux user (that guy's mother, for example). Wouldn't it still be natural to open it, with the same catastrophic results?
No. You "open" files by using open(2) or fopen(3). The newbie's mother would probably view the the file in ed, which would in turn open the file for read and write. The script would never get executed, unless the mother entered    %./ILOVEYOU.txt This is sane behavior.
-----------------------------------------------
Random.com is minimal...so why must they...
on
Hump Day Quickies
·
· Score: 1
not support text browsers?!? They're gonna have to change this soon...
Could the size of the permutation be increased every once in a while and still maintain the properties of LCGs? With an infinite tape, a counter could be incremented with each generated number. When a multiple of a certain number (the period of the current permutation?) is reached, a larger permutation is used.
And what about other generators? There's a funny one called Cliff Random Number Generator that produces random floats, taking logs of numbers. With an infinite tape, you could take out these log calculations to arbitrary precision (again, decided by a counter) to get a sequence that never repeats itself. If you want to make the sequence truly unpredicatable, you could also reverse the bit order before generating each number, so that least significant bits become most significant. I don't know the math behind sequences of chopped-logged-swapped numbers, but I don't see why in principle a non-periodic sequence that appears random could be generated.
Just some random noise to think about until newscientist sets MaxClients > 0...if these questions/observations have obvious/accepted answers (or not), please reply (flame away):
Pseudorandom number sequences always have some period because the algorithms that generate them run on a finite amount of tape. If your computer has an infinite tape, it can generate pseudorandom sequeunces with no period. Quantum noise is perfectly random and so has no period. Could quantum noise be generated on a computer with infinite tape? Another way to put this: could a purely random sequence be computed given some seed? (i.e., impossible to predict--without knowing the seed)
If above question answered positively, what are the implications for quantum computers (such as the universe)? Would they be real Turing Machines, as opposed to the mediocre, wannabe, we-shall-always-remain-finite-state-machines that are currently most popular?
wow! pure quantum-mystical-pseudo-cosmo-cpu-babble! and i haven't even read penrose! (though permutation city was almost as hokey (still a cool book, though...))
From experience, I tend to agree with the results of the experiment (and like most of you I've had a lot of experience with sleep deprivation:-() It was definitely a bad thing(lgpl) taking a proof-oriented math midterm after an hour or so of sleep. (Then again, it was pretty bad taking a take-home math final with 10 hours sleep...on the keyboard before the deadline;) But on the other hand, I don't really have a problem with BS'ing (i.e., exercising my creative writing powers) under sleep deprivation. Indeed, it seems to encourage my ranting ability. Too bad I got a full 6 hours of sleep today or else I'd be able to continue...
The two properties of DNA most important to molecular computing are:
Massive parallelism
Base pair matching (A-T, G-C)
The second property is what gives it the capability of computing things. Adleman's original solution to the Hamiltonian path problem worked in the following way:
Give each of the n (=7) cities a unique encoding (string of DNA bases) c_i = t_i f_i. That is, each city encoding is divided into two parts: "to" and "from".
If the graph contains an edge from city i to city j, create the edge e_ij = f'_i t'_j, where f'_i and t'_j are the (DNA base) complements of f'_i and t'_j, respectively.
Dump all the c_i and e_ij into a clean test tube and wait a while for them to match up.
Remove all paths (i.e., double strands) of length != n. Then remove all paths that don't start with the first city and end with the last one. Then remove all paths that don't contain each city. (All of these filtering processes can be done by well-developed biochemical/mechanical techniques used regularly in normal DNA labs.)
If there are any paths left, you got yourself a Hamiltonian path!
I'd love to draw diagrams, but it's kinda hard using lynx... One shortcoming of Adleman's experiment was that the method only works on this particular problem (kind of a DNA hack). But since then, several more general models have been created. One developed by Richard Lipton encodes binary numbers by representing them as graphs, and then using Adleman's method. Specifically, he has a vertex v_i for each bit position, and connects it to v_i+1 by way of "zero" and "one" vertices 0_i and 1_i (so there are edges from (v_i,0_i), (v_i, 1_i), (0_i, v_i+1), (1_i, v_i+1)). A particular number is just a path along this graph. So, you can dump all the the dna into a test tube, perform the operations needed your problem, and see if any of the remaining double strands are paths (not hamiltonian) from the first "bit" to the last, and thus solutions to your problem...
But as some people have noted, DNA computing is slow...and while speed*parallelism makes up for some of it, it's still slower than a pc, and much much slower than distributed computing. There are different ways it could compete, though:
3D nature better exploited. That is, the techniques used up to now don't really lend themselves to scaling in huge-ass vats. If they could, maybe you could fill your house up with DNA instead of building that Beowulf.
Solving more "natural" problems. Protein folding?
Nanotechnology. Instead of computing things, we could use the attachment properties of DNA to build nice little things (or make crystals to build nice little things). DNA is easily broken, though, so you have to be careful.
People working on molecular computing have looked into this. Some literature:
Adleman, Rothemund, Roweiss, Winfree. "On applying molecular computation to the Data Encryption Standard." From 2nd DIMACS workshop on DNA Computers, 1996.
Boneh, Dunworth, Lipton. "Breaking DES using a molecular computer." From first DIMACS workshop on DNA Computers, 1995.
I think the algorithms basically do the same thing you described. Adleman et al's approach would require about 1 gram of DNA. To do the "hard part", it builds up from a set of primitive operations (merge, separate, set, and clear). These can be controlled robotically, but will still take a bit of time since it takes a while before all the molecules settle into place after each operation. IIRC, bottom line a year or two ago was several months in theory. I'm not sure where current technology stands.
(this overview of DES cracking as well as more information about D/RNA computing in general can be found in DNA Computing by Paun, Rozenberg, Salomaa)
Do you think some trader wrote a perl script to search through news for the word "linux" and buy? I can't think of any other explanation for the following (note: this is a bulleting board traded stock)...
ISOL.OB 2:19PM 2 5/16 +1 5/16 +131.25% 441,100
Fri Dec 10 ISOL.OB 1MAGE Software Announces Document Imaging for Linux - Business Wire
Hmm...small point...wouldn't having 10 weeks of backups each with only a single byte (or even a few) of randomness be pretty easy to recover? Assuming the backed up files don't change too much, you simply take the most frequent byte value as the correct one. Since the backups do change, it requires a bit more trickiness, but still, it doesn't seem too hard. Anyone with experience in this area?
Hmm...that sounded kind of harsh. I don't want to come off as bashing Knuth in any way. For one thing, he could probably come up with a better way to get a nicely typeset message using lynx. I guess my problem with the lecture was that I came in expecting to see a fearless code warrior(tm) but instead heard a pretty old guy say things like (don't hold me to wording) "I recommend [doing group translations of bible verses]. It's very educational" and "Working on 3:16 was great because I learned so many new things about the Bible."
Knuth's lecture series is subtitled (something like) "Things that a Computer Scientist doesn't talk about." I think that's the problem. When people try to go out of their field, they usually end up sounding unprofessional. I got the same impression from reading some of Douglas Hofstadter's new book, Le Ton Bon de Marot, which is also about translation. I didn't read the whole book, so maybe my opinion is worthless, but the first few chapters weren't that great and gave a really amateurish impression.
The moral: computer scientists should give talks about computer science. Go ahead and study the Bible, read French poetry, smoke pot, etc., on your own free time. But don't tell us about it.
Actually, I just came back from Knuth's 3rd lecture (well...had dinner first). It was on translating Bible verses without knowledge of Hebrew or Greek. Yes, I came in saying to everyone around that "Knuth is God", etc. But the lecture was horrible. Well...at least it was extremely uninteresting, IMHO. Maybe some of the other lectures in the series will be better (maybe first two were), but that looks doubtful. Art of CP, TeX, good math are all well and supernatural, but 3:16 (what he talked about today) is not. He sounded more like a preacher and less like a diety; that and he can't really give a presentation. Well...I'm going to go back to my altar now...
But try searching for altavista on lycos. A whole bunch of results pop up. Granted, the first result is "AltaVista Firewall 98", but see what it has to say about the search engine: Renowned search engine also sports news provided by ABCNews.com, free e-mail, services, and Usenet search.
BTW, the link is to the old altavista.digital.com
I'm not sure what this all means...secret alliances between web portals?
One-time pad through quantum encryption prevents Eve from listening into the conversation across the line or changing the signal. But it does nothing, as indeed it can't, to prevent her from capturing the unencrypted information on either Alice or Bob's end. Eventually, all cypher must be converted to plaintext, where it is susceptible to any number of attacks. Of course, Alice and Bob could be locked inside steel vaults, with no Internet connection and a room full of M-16s...
Slashdot Mirror
Sorry, this is false.
-----------------------------------------------
But like all ridiculously bad movies, it can be pretty funny to watch :)
-----------------------------------------------
No. You "open" files by using open(2) or fopen(3). The newbie's mother would probably view the the file in ed, which would in turn open the file for read and write. The script would never get executed, unless the mother entered ./ILOVEYOU.txt
   %
This is sane behavior.
-----------------------------------------------
-----------------------------------------------
Could the size of the permutation be increased every once in a while and still maintain the properties of LCGs? With an infinite tape, a counter could be incremented with each generated number. When a multiple of a certain number (the period of the current permutation?) is reached, a larger permutation is used.
And what about other generators? There's a funny one called Cliff Random Number Generator that produces random floats, taking logs of numbers. With an infinite tape, you could take out these log calculations to arbitrary precision (again, decided by a counter) to get a sequence that never repeats itself. If you want to make the sequence truly unpredicatable, you could also reverse the bit order before generating each number, so that least significant bits become most significant. I don't know the math behind sequences of chopped-logged-swapped numbers, but I don't see why in principle a non-periodic sequence that appears random could be generated.
-----------------------------------------------
wow! pure quantum-mystical-pseudo-cosmo-cpu-babble! and i haven't even read penrose! (though permutation city was almost as hokey (still a cool book, though...))
-----------------------------------------------
-----------------------------------------------
The second property is what gives it the capability of computing things.
Adleman's original solution to the Hamiltonian path problem worked in the following way:
I'd love to draw diagrams, but it's kinda hard using lynx... One shortcoming of Adleman's experiment was that the method only works on this particular problem (kind of a DNA hack). But since then, several more general models have been created. One developed by Richard Lipton encodes binary numbers by representing them as graphs, and then using Adleman's method. Specifically, he has a vertex v_i for each bit position, and connects it to v_i+1 by way of "zero" and "one" vertices 0_i and 1_i (so there are edges from (v_i,0_i), (v_i, 1_i), (0_i, v_i+1), (1_i, v_i+1)). A particular number is just a path along this graph. So, you can dump all the the dna into a test tube, perform the operations needed your problem, and see if any of the remaining double strands are paths (not hamiltonian) from the first "bit" to the last, and thus solutions to your problem...
But as some people have noted, DNA computing is slow...and while speed*parallelism makes up for some of it, it's still slower than a pc, and much much slower than distributed computing. There are different ways it could compete, though:
-----------------------------------------------
I think the algorithms basically do the same thing you described. Adleman et al's approach would require about 1 gram of DNA. To do the "hard part", it builds up from a set of primitive operations (merge, separate, set, and clear). These can be controlled robotically, but will still take a bit of time since it takes a while before all the molecules settle into place after each operation. IIRC, bottom line a year or two ago was several months in theory. I'm not sure where current technology stands.
(this overview of DES cracking as well as more information about D/RNA computing in general can be found in DNA Computing by Paun, Rozenberg, Salomaa)
-----------------------------------------------
ISOL.OB 2:19PM 2 5/16 +1 5/16 +131.25% 441,100
Fri Dec 10 ISOL.OB 1MAGE Software Announces Document Imaging for Linux - Business Wire
Hmm...small point...wouldn't having 10 weeks of backups each with only a single byte (or even a few) of randomness be pretty easy to recover? Assuming the backed up files don't change too much, you simply take the most frequent byte value as the correct one. Since the backups do change, it requires a bit more trickiness, but still, it doesn't seem too hard. Anyone with experience in this area?
Encrypt the string 00000000 and you'll get a pleasant surprise. Now, do I get "Insightful", "Informative", or "Troll"?
Knuth's lecture series is subtitled (something like) "Things that a Computer Scientist doesn't talk about." I think that's the problem. When people try to go out of their field, they usually end up sounding unprofessional. I got the same impression from reading some of Douglas Hofstadter's new book, Le Ton Bon de Marot, which is also about translation. I didn't read the whole book, so maybe my opinion is worthless, but the first few chapters weren't that great and gave a really amateurish impression.
The moral: computer scientists should give talks about computer science. Go ahead and study the Bible, read French poetry, smoke pot, etc., on your own free time. But don't tell us about it.
Actually, I just came back from Knuth's 3rd
lecture (well...had dinner first). It was on
translating Bible verses without knowledge of
Hebrew or Greek. Yes, I came in saying to everyone
around that "Knuth is God", etc. But the lecture
was horrible. Well...at least it was extremely
uninteresting, IMHO. Maybe some of the other
lectures in the series will be better (maybe first two were), but
that looks doubtful. Art of CP, TeX, good math are all well and supernatural, but
3:16 (what he talked about today) is not. He sounded more like a preacher and less like a diety; that and he can't really give a presentation. Well...I'm going to go back to my altar now...
But try searching for altavista on lycos.
A whole bunch of results pop up. Granted, the
first result is "AltaVista Firewall 98", but see
what it has to say about the search engine:
Renowned search engine also sports news provided by
ABCNews.com, free e-mail, services, and Usenet search.
BTW, the link is to the old altavista.digital.com
I'm not sure what this all means...secret alliances between web portals?
One-time pad through quantum encryption prevents Eve from listening into the conversation across the line or changing the signal. But it does nothing, as indeed it can't, to prevent her from capturing the unencrypted information on either Alice or Bob's end. Eventually, all cypher must be converted to plaintext, where it is susceptible to any number of attacks. Of course, Alice and Bob could be locked inside steel vaults, with no Internet connection and a room full of M-16s...