The idea with most hashes is that you can go forward, but you can't (reasonably) go backwards (i.e. you can't "decode" a hash): i.e. I can safely transmit a hash of my password because you can't magically "decode" it back to my password. Of course an infinite number of data streams would lead to the same hash (i.e. this is the entropy thing again : An SHA1 hash is only 160 bits long, so obviously there will be collisions [though it's designed to avoid that, of course it will happen given an infinite amount of input]), so obviously you can't just say well "give me the string that leads to that hash!" because there's an infinite number of strings that would lead to the same hash.
The problem with the millions of half-thought out compression ideas is exactly that: They're half thought out. It's easy to think "Well I'll just search for sequences in pi because pi is infinitely long and if you had an infinite number of monkeys....". It all comes down to entropy or the variability in a particular bit space (yup I yabbled about this in another post), and the simple cold hard reality is that 1 byte can represent no more than 256 different "realities", versus 2 bytes which can hold 65536 different values : 2 bytes cannot POSSIBLY represent every combination of values that could be contained by even 3 bytes (which has 16777216 different combinations): There is no trick or slight of hand to get around this basic mathematical fact.
That basic fact immediately discounts and proves impossible any compressibility of random data, absolutely and non-refutably. Imagine up ways to store floating point numbers, or to "3d encapsulate the space-time continuum", but it all comes down to entropy, and the limitation of it when trying to represent X amount of data in
$ make a lot of things happen...
on
The Drone War
·
· Score: 3
That goes both ways though: Military contractors make a lot of money when there are conflicts (for instance apparently the military is ordering cruise missiles faster than they can be built), so you get a dangerous situation where there are elements that have no personal risk to themselves so they encourage the government (explicitly and subvertly) to engage in conflicts. Behind almost everything there is the almighty dollar.
I have never seen a web poll that isn't pure bullshit. On the one hand you have polls that tend to cater to a certain crowd that is more motivated (for instance if there's a "favourite operating system" poll what kind of people are going to go on there and answer it? Bob, the guy checking his email who runs "Windows something", or Joe the Linux Geek. How many times here on SLASHDOT have they encouraged people to run and vote on behalf of Apache/Linux/etc.? On the other hand you have poll rigging, which EVERY camp has people who try to abuse polls (again the prisoner's dilemma: Most people justify it by presuming that everyone else is doing it anyways). In the end this petty little poll garnered a measly 1415 votes, meaning that the massive Microsoft rigging at most counted for 1057: Big frickin' deal.
This is so ridiculously NOT news. Did the poll say "Employees and family of Sun or Microsoft are excluded". Furthermore did it state "The fanatical are excluded as well". We have no idea if Java jumped in the front due to the much more devious web stuffing by Sun.
The bottom line is that web polls are absolutely, positively useless. The only purpose they hold is to give the converted a chubby ("Oh gosh look I'm gonna blow! Linux is the #1 server OS based on the poll on Slashdot!"), or in this case to get millions of hits based on a rather dubious decision to pronounce some great fraud because your bogo web poll was otherwise that inane.
Re:Prevailing market conditions...
on
No Solaris 9 for x86
·
· Score: 2, Insightful
Yes, in terms of actual revenue Solaris on Intel is a complete failure. In fact, it is farcical. But I dont think Sun ever considered it to be a primary business venture - more of a 'loss leader'.
Indeed, I've read many complaints arguing that Sun does a horrible job of optimizing on the Intel platform (hence "Slowaris") intentionally to make their own hardware look that much better. However that's a catch-22: It makes their hardware look better, but it makes their software look worse.
Prevailing market conditions...
on
No Solaris 9 for x86
·
· Score: 5, Insightful
The market conditions are that Solaris on Intel machines is a total failure. As another poster in another argument mentioned: The only people who Solaris on Intel machines seem to be just taking it for a test run, and then they go back to their real OS (be it Windows, Linux, FreeBSD, etc.).
Just thought that was a little more honest than claiming it's the recession or Sept. 11th fallout.
During the "Lunchbox PC" discussion a couple of days ago the soundcard was brought up as an important factor, and several people mentioned getting an external USB D/A converter. I was intrigued and went looking to find that there are quite a fewoptions out there right now already, and this Creative entry is just yet another addition to the selection list. Note that Creative would probably be the last company most musicians would look to as a leader in the high fidelity audio field.
While many people have rightly mentioned that USB has ample bandwidth for 2-channel audio (though even that at 96Khz/24bits hits 4.6Mbps - 96000 * 24 * 2): Note that this device claims Dolby Digital 5.1 -> Now I don't know if they do the decoding in the external box, or if they actually send 6 streams, but if they send six separate streams that's about 13.8Mbps (and of course USB is limited to 11Mbps). Just something to consider. It is a fair statement to say that Firewire or USB2 (it is just USB1 isn't it?) would make me feel a lot more comfortable about the unrestrained capabilities.
Totally unrelated, and I imply no correlation and similarity in facts, however I came across this old story which is always an interesting read: http://www.thestandard.com/article/0,1902,16368,00.html?body_page=11. Again it's totally unrelated to the ventures of these data compression people, but just is an interesting read for Slashdot folks.
True, and that disclaimer itself basically discounts this being anything better than "another variant of LZW". Perhaps what they're really selling is technology (or the dream at least) that can find patterns that aren't readily apparent.
Of course if you want to really liberally interpret "practically random", I could say that the string AAAAAAAAAAAA(repeat 100 times. I originally did to get the Slashcode error "that's an awful long string of letters there":-)) is `practically' random (meaning that it's equally as possible in a random sequence as any other 100 character string), therefore my amazing compression can represent that as 100A! Please send checks care of...
Pi does not qualify as random as a whole, though, because it's a known quantity. To have enough possible variations to use such a random representation method for 1000 different "random" streams you would need 1000 different representations. Therein lies the quandry of trying to compress something other than pi.
Speaking of that there's the brilliant Dilbert where Dilbert is led around I believe accounting and is shown their "random number generator" which is a guy constantly saying "7". Dilbert asks if they're sure it works, to which his host replies "That's the problem with random numbers: We'll never know". Anyone know where that is in graphical form? It was from October I believe and the Dilbert site pulled it, and everyone else seems to have linked to the Dilbert site.
At least PGP uses various timing values for random data as well (the timing of typing in addition to some other timing sources I believe). If it was just typing then that would be scary: How many "random" keystrokes seem to always have "asdf"? There is nothing random typing at a keyboard.
Well firstly I'd say the press release gives a pretty clear picture of the reality of their technology: It has such an overuse of supposedly TM'd (anyone want to double check the filings? I'm going to guess that there are none) "technoterms" like "TunerAccelerator" and "BinaryAccelerator" that it just is screaming hoax (or creative deception), not to mention a use of Flash that makes you want to punch something. Note that they give themselves huge openings such as always saying "practically random" data: What the hell does that mean?
I think one way to understand it (Because all of us at some point or another have thought up some half-assed, ridiculous way of compressing any data down to 1/10th -> "Maybe I'll find a denominator and store that with a floating point representation of..."), and I'm saying this as not a mathematician or compression expert : Let's say for instance that this compression ratio is 10 to 1 on random data, and I have every possible random document 100 bytes long -> That means I have 6.6680144328798542740798517907213e+240 different random documents (256^100). So I compress them all into 10 byte documents, but the maximum variations of a 10 byte documents is 1208925819614629174706176 : There isn't the entropy in a 10-byte document to store 6.6680144328798542740798517907213e+240 different possibilities (it is simply impossible, no matter how many QuantumStreamTM HyperTechTM TechoBabbleTM TermsTM) : You end up needed, tada, 100 bytes to have the entropy to possibly store all variants of a 100 byte document, but of course most compression routines put in various logic codes and actually increase the size of the document. In the case of the ZeoSync claim though they're apparently claiming that somehow you'll represent 6.6680144328798542740798517907213e+240 different variations in a single byte : So somehow 64 tells you "Oh yeah, that's variation 5.5958572359823958293589253e+236!". Maybe they're using SubSpatialQuantumBitsTM.
For lossless compression simply saying "There were 5 red shirts and 7 blue shirts" isn't enough: You'd have to also store information on exactly where those 5 red shirts and 7 shirts were in the sample to be able to recreate the situation exactly as it was. Because of this it has been found to be impossible to "compress" truly random data without actually increasing the size of the file.
Of course if you're talking lossy then everything changes: Who cares where the shirts are just tell em how many there was. Unfortunately lossy is only relevant for images and sounds.
The dot COM bust I'm sure has been a nightmare for them (in the same way as it's been for F5, Cisco, Nortel, etc.): Suddenly instead of the smaller market buying new, they're all buying used from.COM shops that went under (and, of course, Herman Miller doesn't include a "non-transferrable" license with their chairs...:-)).
Maybe I have a bad unit or something (though that would mean that many others with the same observation have as well), however in lighting in a house at nighttime it is unusable literally right underneath a 60W light. The only place where I've found it to be usable is in direct sunlight.
Well my gaming comment was moreso directed to the flock of seagulls that likewise claim that a "GeForce 2 is more than anyone needs" (the exact same argument and technique as the CPU argument). Having said that: Games with greater AI and modelling of the real world (i.e. not a FPS where it's 96% graphics/4% computations), such as the example given (Operation Flashpoint), do beg for faster processors: In OpFlash the scope of a battle is heavily limited not by your video card, but rather by the CPU in your system. Falcon 4 from a couple of years ago simulated full-scale combat and as such still is begging for a faster processor on the fastest systems available.
The GBA is completely unusable (including with the aftermarket add on front lights) in anything but very bright conditions, which is something that I think they should be a little more truthful about (i.e. in the commercial where the guy is playing in a church: Fat chance).
Why is it that every story about ever increasing CPU speeds is met by about 50% postings claiming that such speed isn't necessary unless you "want 500 fps in Quake"? I've seen these sorts of posts since back in the BBS days when the new 486 came out, immediately to be met by 50 posts (usually by people who feel a need to justify whatever they own) claiming that "a 386/33 is more power than anyone needs anyways!". Bah.
There are countless benefits to the increased speed (and of course like always: Once you use a higher speed system for a while suddenly you notice, clear as a sunny day, that yes there IS a very noticable difference, and suddenly that previously adequate machine seems pokey), and if you don't realize what they are then continue using whatever it is you use, but save the "500 fps in Quake" rhetoric (here I am with what would have been a cutting edge machine one year ago and Operation Flashpoint runs with frame rates in the single digits, yet even still it isn't a fraction as complex of a "world" as it could be if more powerful systems were prevalent).
What proof is there that overclocking lowers the life of the chip beyond its likely gross obsolescence anyways? I have a Celeron "450a" (300 overclocked 50%) that ran fabulously for a couple of years, including increasing the voltage 50%, and it's still working 100% as well as a test machine (despite the fact that the market value of the chip now is in the very low 2 digits).
That does make sense. I'm always shocked when I actually convert to rotations per second to discover how slow it really is (i.e. 120 rotations per second on a 7200 RPM drive).
Having said that this has me curious: Are there drives with mutiple heads at different locations on the drive (i.e. One immediate design I could imagine is opposite heads on the same mechanism [assuring accuracy]), basically so with two heads at opposite points the rotational seek time is halved and the throughput could be doubled with some creative buffering and concatenation.
For seek times I can see that RPMs would matter (though the majority of the time for seeking is physically moving the head radially on the disk), however my point is that sure, changing the RPMs for a given density and head count matters, but that comparing RPMs across different densities is comparing based upon partial info. For the sake of example, imagine that on one drive one revolution contains 100KB of data, whereas on another drive it contains 50KB of data, but drive one runs at a slow 5400 RPM versus the "speedy" 7200 RPM of drive two.
However, the solution is not to remove raw sockets, it's to (a) forcibly educate the users, such as making them run through a tutorial on first- boot from a PC, or (b) lock down the system, instead of leaving it open, like MS typically does.
The real solution is at the ISP levels: All ISPs should be mandated (at risk of being held accountable due to negligence for financial loss held by others) to filter out IP spoofing -> If someone is sending out packets from the address a.b.c.d and you know that they're z.y.x.w then shut down their connection or at lest filter the packets out. The whole idea of IP spoofing is absurd. The same holds true for any other manner of malformed packets: They should be dropped at the first router they hit.
Slashdot Mirror
The idea with most hashes is that you can go forward, but you can't (reasonably) go backwards (i.e. you can't "decode" a hash): i.e. I can safely transmit a hash of my password because you can't magically "decode" it back to my password. Of course an infinite number of data streams would lead to the same hash (i.e. this is the entropy thing again : An SHA1 hash is only 160 bits long, so obviously there will be collisions [though it's designed to avoid that, of course it will happen given an infinite amount of input]), so obviously you can't just say well "give me the string that leads to that hash!" because there's an infinite number of strings that would lead to the same hash.
The problem with the millions of half-thought out compression ideas is exactly that: They're half thought out. It's easy to think "Well I'll just search for sequences in pi because pi is infinitely long and if you had an infinite number of monkeys....". It all comes down to entropy or the variability in a particular bit space (yup I yabbled about this in another post), and the simple cold hard reality is that 1 byte can represent no more than 256 different "realities", versus 2 bytes which can hold 65536 different values : 2 bytes cannot POSSIBLY represent every combination of values that could be contained by even 3 bytes (which has 16777216 different combinations): There is no trick or slight of hand to get around this basic mathematical fact.
That basic fact immediately discounts and proves impossible any compressibility of random data, absolutely and non-refutably. Imagine up ways to store floating point numbers, or to "3d encapsulate the space-time continuum", but it all comes down to entropy, and the limitation of it when trying to represent X amount of data in
That goes both ways though: Military contractors make a lot of money when there are conflicts (for instance apparently the military is ordering cruise missiles faster than they can be built), so you get a dangerous situation where there are elements that have no personal risk to themselves so they encourage the government (explicitly and subvertly) to engage in conflicts. Behind almost everything there is the almighty dollar.
Amazing, Linux is apparently going to jump from ~1% marketshare to a stunning 80%+ during 2002! Geez these web polls are magical tea leaves pronouncing the future!
I have never seen a web poll that isn't pure bullshit. On the one hand you have polls that tend to cater to a certain crowd that is more motivated (for instance if there's a "favourite operating system" poll what kind of people are going to go on there and answer it? Bob, the guy checking his email who runs "Windows something", or Joe the Linux Geek. How many times here on SLASHDOT have they encouraged people to run and vote on behalf of Apache/Linux/etc.? On the other hand you have poll rigging, which EVERY camp has people who try to abuse polls (again the prisoner's dilemma: Most people justify it by presuming that everyone else is doing it anyways). In the end this petty little poll garnered a measly 1415 votes, meaning that the massive Microsoft rigging at most counted for 1057: Big frickin' deal.
This is so ridiculously NOT news. Did the poll say "Employees and family of Sun or Microsoft are excluded". Furthermore did it state "The fanatical are excluded as well". We have no idea if Java jumped in the front due to the much more devious web stuffing by Sun.
The bottom line is that web polls are absolutely, positively useless. The only purpose they hold is to give the converted a chubby ("Oh gosh look I'm gonna blow! Linux is the #1 server OS based on the poll on Slashdot!"), or in this case to get millions of hits based on a rather dubious decision to pronounce some great fraud because your bogo web poll was otherwise that inane.
Yes, in terms of actual revenue Solaris on Intel is a complete failure. In fact, it is farcical. But I dont think Sun ever considered it to be a primary business venture - more of a 'loss leader'.
Indeed, I've read many complaints arguing that Sun does a horrible job of optimizing on the Intel platform (hence "Slowaris") intentionally to make their own hardware look that much better. However that's a catch-22: It makes their hardware look better, but it makes their software look worse.
The market conditions are that Solaris on Intel machines is a total failure. As another poster in another argument mentioned: The only people who Solaris on Intel machines seem to be just taking it for a test run, and then they go back to their real OS (be it Windows, Linux, FreeBSD, etc.).
Just thought that was a little more honest than claiming it's the recession or Sept. 11th fallout.
During the "Lunchbox PC" discussion a couple of days ago the soundcard was brought up as an important factor, and several people mentioned getting an external USB D/A converter. I was intrigued and went looking to find that there are quite a few options out there right now already, and this Creative entry is just yet another addition to the selection list. Note that Creative would probably be the last company most musicians would look to as a leader in the high fidelity audio field.
While many people have rightly mentioned that USB has ample bandwidth for 2-channel audio (though even that at 96Khz/24bits hits 4.6Mbps - 96000 * 24 * 2): Note that this device claims Dolby Digital 5.1 -> Now I don't know if they do the decoding in the external box, or if they actually send 6 streams, but if they send six separate streams that's about 13.8Mbps (and of course USB is limited to 11Mbps). Just something to consider. It is a fair statement to say that Firewire or USB2 (it is just USB1 isn't it?) would make me feel a lot more comfortable about the unrestrained capabilities.
Totally unrelated, and I imply no correlation and similarity in facts, however I came across this old story which is always an interesting read: http://www.thestandard.com/article/0,1902,16368,00 .html?body_page=11. Again it's totally unrelated to the ventures of these data compression people, but just is an interesting read for Slashdot folks.
True, and that disclaimer itself basically discounts this being anything better than "another variant of LZW". Perhaps what they're really selling is technology (or the dream at least) that can find patterns that aren't readily apparent.
Of course if you want to really liberally interpret "practically random", I could say that the string AAAAAAAAAAAA(repeat 100 times. I originally did to get the Slashcode error "that's an awful long string of letters there" :-)) is `practically' random (meaning that it's equally as possible in a random sequence as any other 100 character string), therefore my amazing compression can represent that as 100A! Please send checks care of...
Pi does not qualify as random as a whole, though, because it's a known quantity. To have enough possible variations to use such a random representation method for 1000 different "random" streams you would need 1000 different representations. Therein lies the quandry of trying to compress something other than pi.
Speaking of that there's the brilliant Dilbert where Dilbert is led around I believe accounting and is shown their "random number generator" which is a guy constantly saying "7". Dilbert asks if they're sure it works, to which his host replies "That's the problem with random numbers: We'll never know". Anyone know where that is in graphical form? It was from October I believe and the Dilbert site pulled it, and everyone else seems to have linked to the Dilbert site.
At least PGP uses various timing values for random data as well (the timing of typing in addition to some other timing sources I believe). If it was just typing then that would be scary: How many "random" keystrokes seem to always have "asdf"? There is nothing random typing at a keyboard.
Well firstly I'd say the press release gives a pretty clear picture of the reality of their technology: It has such an overuse of supposedly TM'd (anyone want to double check the filings? I'm going to guess that there are none) "technoterms" like "TunerAccelerator" and "BinaryAccelerator" that it just is screaming hoax (or creative deception), not to mention a use of Flash that makes you want to punch something. Note that they give themselves huge openings such as always saying "practically random" data: What the hell does that mean?
I think one way to understand it (Because all of us at some point or another have thought up some half-assed, ridiculous way of compressing any data down to 1/10th -> "Maybe I'll find a denominator and store that with a floating point representation of..."), and I'm saying this as not a mathematician or compression expert : Let's say for instance that this compression ratio is 10 to 1 on random data, and I have every possible random document 100 bytes long -> That means I have 6.6680144328798542740798517907213e+240 different random documents (256^100). So I compress them all into 10 byte documents, but the maximum variations of a 10 byte documents is 1208925819614629174706176 : There isn't the entropy in a 10-byte document to store 6.6680144328798542740798517907213e+240 different possibilities (it is simply impossible, no matter how many QuantumStreamTM HyperTechTM TechoBabbleTM TermsTM) : You end up needed, tada, 100 bytes to have the entropy to possibly store all variants of a 100 byte document, but of course most compression routines put in various logic codes and actually increase the size of the document. In the case of the ZeoSync claim though they're apparently claiming that somehow you'll represent 6.6680144328798542740798517907213e+240 different variations in a single byte : So somehow 64 tells you "Oh yeah, that's variation 5.5958572359823958293589253e+236!". Maybe they're using SubSpatialQuantumBitsTM.
For lossless compression simply saying "There were 5 red shirts and 7 blue shirts" isn't enough: You'd have to also store information on exactly where those 5 red shirts and 7 shirts were in the sample to be able to recreate the situation exactly as it was. Because of this it has been found to be impossible to "compress" truly random data without actually increasing the size of the file.
Of course if you're talking lossy then everything changes: Who cares where the shirts are just tell em how many there was. Unfortunately lossy is only relevant for images and sounds.
The dot COM bust I'm sure has been a nightmare for them (in the same way as it's been for F5, Cisco, Nortel, etc.): Suddenly instead of the smaller market buying new, they're all buying used from .COM shops that went under (and, of course, Herman Miller doesn't include a "non-transferrable" license with their chairs... :-)).
Maybe I have a bad unit or something (though that would mean that many others with the same observation have as well), however in lighting in a house at nighttime it is unusable literally right underneath a 60W light. The only place where I've found it to be usable is in direct sunlight.
Well my gaming comment was moreso directed to the flock of seagulls that likewise claim that a "GeForce 2 is more than anyone needs" (the exact same argument and technique as the CPU argument). Having said that: Games with greater AI and modelling of the real world (i.e. not a FPS where it's 96% graphics/4% computations), such as the example given (Operation Flashpoint), do beg for faster processors: In OpFlash the scope of a battle is heavily limited not by your video card, but rather by the CPU in your system. Falcon 4 from a couple of years ago simulated full-scale combat and as such still is begging for a faster processor on the fastest systems available.
The GBA is completely unusable (including with the aftermarket add on front lights) in anything but very bright conditions, which is something that I think they should be a little more truthful about (i.e. in the commercial where the guy is playing in a church: Fat chance).
Why is it that every story about ever increasing CPU speeds is met by about 50% postings claiming that such speed isn't necessary unless you "want 500 fps in Quake"? I've seen these sorts of posts since back in the BBS days when the new 486 came out, immediately to be met by 50 posts (usually by people who feel a need to justify whatever they own) claiming that "a 386/33 is more power than anyone needs anyways!". Bah.
There are countless benefits to the increased speed (and of course like always: Once you use a higher speed system for a while suddenly you notice, clear as a sunny day, that yes there IS a very noticable difference, and suddenly that previously adequate machine seems pokey), and if you don't realize what they are then continue using whatever it is you use, but save the "500 fps in Quake" rhetoric (here I am with what would have been a cutting edge machine one year ago and Operation Flashpoint runs with frame rates in the single digits, yet even still it isn't a fraction as complex of a "world" as it could be if more powerful systems were prevalent).
What proof is there that overclocking lowers the life of the chip beyond its likely gross obsolescence anyways? I have a Celeron "450a" (300 overclocked 50%) that ran fabulously for a couple of years, including increasing the voltage 50%, and it's still working 100% as well as a test machine (despite the fact that the market value of the chip now is in the very low 2 digits).
Your roommate should get a CF reader for his PC. I gave a link, but I've seen dozens of them on the market.
That does make sense. I'm always shocked when I actually convert to rotations per second to discover how slow it really is (i.e. 120 rotations per second on a 7200 RPM drive).
Having said that this has me curious: Are there drives with mutiple heads at different locations on the drive (i.e. One immediate design I could imagine is opposite heads on the same mechanism [assuring accuracy]), basically so with two heads at opposite points the rotational seek time is halved and the throughput could be doubled with some creative buffering and concatenation.
For seek times I can see that RPMs would matter (though the majority of the time for seeking is physically moving the head radially on the disk), however my point is that sure, changing the RPMs for a given density and head count matters, but that comparing RPMs across different densities is comparing based upon partial info. For the sake of example, imagine that on one drive one revolution contains 100KB of data, whereas on another drive it contains 50KB of data, but drive one runs at a slow 5400 RPM versus the "speedy" 7200 RPM of drive two.
However, the solution is not to remove raw sockets, it's to (a) forcibly educate the users, such as making them run through a tutorial on first- boot from a PC, or (b) lock down the system, instead of leaving it open, like MS typically does.
The real solution is at the ISP levels: All ISPs should be mandated (at risk of being held accountable due to negligence for financial loss held by others) to filter out IP spoofing -> If someone is sending out packets from the address a.b.c.d and you know that they're z.y.x.w then shut down their connection or at lest filter the packets out. The whole idea of IP spoofing is absurd. The same holds true for any other manner of malformed packets: They should be dropped at the first router they hit.