You do understand the China sells to everybody on the planet? If you want to take your somewhat irrational argument to its logical conclusion, everybody is responsible for the mess China is in.
That's exactly what the GP was trying to say, and there is absolutely nothing irrational about that argument. There is only one planet Earth, and we are all responsible, to a greater or lesser degree.
Then again, getting a bit for bit BluRay version of a movie would seriously tax all but a select few high speed residential connections in the real world.
Unless Japan and South Korea do not count as part of the "real world" to you.:)
Where the have you been all these years? Nothing stopping hardware OEMs from selling hardware with non-Windows OSes my ass. Jean-Louis Gassée found that one out when he first began to try pitching BeOS to hardware OEMs. He wrote an article on why PC manufacturers won't sell non-MS products (more info on this here and here). The Windows monopoly is reinforced by anti-competitive agreements that Microsoft has with all of the major hardware OEMs. If one of these OEMs violates the agreement, they lose the OEM discount on all the other Windows PCs they sell, and consequently their Windows-based computers wind up costing much more than those vendors that decided to abide by the agreement. You can guess what that would mean to a major OEM.
In a way, this move by Dell is interesting since it shows to what lengths they've gone to avoid violating the contract. They could have used the same CPU to run the Linux firmware here, but no, they had to include a full ARM SoC to do the same instead. Granted, that has some advantages (given that the x86 CPU is much too overpowered and would eat the battery alive), but perhaps the agreements they have with Microsoft may also have something to do with it.
Five words for you: Meet-in-the-middle Attack. This is the reason why we have Triple-DES, but not Double-DES. Double-DES, instead of giving you the expected 112-bit key that you might have thought you had, thanks to meet in the middle, gives you the security equivalent of only a 57-bit key instead. The right way to do it would be to encrypt three times, with at least two independent keys. If they know that you're double encrypting, the key length is not doubled, but effectively adds one bit to the complexity of your key. You have to do triple encryption to defeat the attack.
Amen to this one. I remember the days when Origin Systems included a cloth map of Britannia inside the box, a nicely bound manual, and a little trinket like an Ankh amulet or something like that (I'm still kicking myself for losing the Codex Symbol that came with the Ultima V box). That made plunking about $50 for the game that much worth it. It seems that nobody does that kind of thing anymore, and yet they still feel it's fair to charge the same amount for games nowadays.
I tried to get the score for the Dies Irae for Mozart's Requiem in D Minor (K. 626). I got this instead:
You have reached this page because the file you requested has not been reviewed for copyright, or is currently restricted due to technical reasons.
A significant portion of the original IMSLP is still pending copyright review, so expect the number of blocked files to decrease dramatically in the next few months after IMSLP reopening. More details on how to spot a blocked file without having to click on it will be released here very soon.
Maybe they should have waited a couple more months when this type of message gets less common.
Paul Graham recently wrote an essay about a related topic just last May, on distractions. It seems that he even works by actually disconnecting his computer from the Internet while working, in order to reduce the amount of distraction that would come from use of the Internet, and using a separate machine somewhere else that had Internet access for those times when he really needs to do something online. It's a radical idea. Maybe it explains why I feel bit more productive while working from home, where Internet access can only be had by hooking my cellphone up to a special SIM card that has a data plan, and connecting to the Net via Bluetooth. With such awkward steps needed to get a working Internet connection, and with no coworkers to bother me, distraction is kept at a minimum. Whereas at the office the lawyer who's sharing our office space has a television permanently tuned to a news channel, I get distractions from coworkers up the wazoo, and a fast broadband connection which basically encourages me to read and post to Slashdot and engage in other diversions...
Gee, wasn't that the plot of a movie starring John Candy? Oh, here it is, Canadian Bacon. Apparently it was a Michael Moore movie, and John Candy's last before his death.
The fact that Microsoft is wealthier than my country is not in question. They have a shitload of money, but they no longer wield the kind of influence and strike the kind of fear into the hearts of competitors as they once were able. Paul Graham put it very well in this article.
Microsoft cast a shadow over the software world for almost 20 years starting in the late 80s. I can remember when it was IBM before them. I mostly ignored this shadow. I never used Microsoft software, so it only affected me indirectly--for example, in the spam I got from botnets. And because I wasn't paying attention, I didn't notice when the shadow disappeared.
But it's gone now. I can sense that. No one is even afraid of Microsoft anymore. They still make a lot of money--so does IBM, for that matter. But they're not dangerous.
Microsoft will likely persist for a long, long time indeed, but people at the leading edge of software development need no longer be afraid of what they might or might not do. They have, in a sense, ceased to matter for those engaged in software development, a lot like the way IBM and SAP are too. Sure, they've got lots of money, and they aren't really going to stop making more, but there's no way in hell that they're going to use that massive war chest that dwarfs the funds available to some third world countries to bring themselves back into serious relevancy. Their very size makes that impossible. Their shareholders would never allow the immense risk doing that would entail.
I believe it's an AI-complete problem. You could do it, if you had an AI that understood the current definition of what is obscene, constantly updating with social mores and the social climate which is the only way to really understand what is meant by the term 'obscene'. I take it this would be impossible to do fully.
My email address, although obfuscated by Slashdot, is real, and accurately reflects the country where we are based in the ccTLD.:) Obviously we're incorporated here, and we have an incorporated subsidiary in Japan as well.
Funny, I just got back from the SODEC trade show in Tokyo last week promoting our company's outsourcing services... As someone whose company which is engaged in providing software outsourcing services to Japanese companies, I can personally attest to the barriers to entry involved in doing this. Language is a serious one: while we would like to think that we are motivated enough to try to learn, it is a very tough language to try to master, and misunderstandings can be costly. We were humbled when we were handed a Japanese software specification which took us a month to reasonably understand but a only week to implement and test. Japanese also seem to have an entrenched attitude of looking down on foreigners, and having more than a little skepticism that the people in companies such as ours will be able to adapt to their ways of thinking and doing things. So far, we haven't seriously disappointed our existing customers, but still, even a brick-headed software engineer like me can sense their skepticism. They are also a lot less flexible than other outsourcing markets that we have had the experience of working with. These are some of the problems that we've encountered, but still, we do think that going into the market for the long haul will be profitable. They really have few choices to remedy their situation with the way things are going.
Sheesh, if the telecoms in a third-world nation like mine, which is known for having the highest SMS traffic in the world (at over a billion messages a day between the three telcos), can still generate fantastic profits charging their subscribers the equivalent of less than 2 US cents per text message (and it has been relatively constant for the past decade or so since the service was introduced), you can see something is amiss here. The kinds of rates that mobile carriers charge for this service elsewhere in the world seem massively extortionate in comparison. GP poster is flat out wrong. Greed is the dominant factor here.
My own interpretation is that the machines are actually obedient to the last drop. They are trying to create a perfect world for humans, and the entire contrivance that is the Matrix is really a massive system designed for the machines to understand what will constitute a perfect world for humanity. I think of the Oracle in the Matrix in the sense of the 'oracle Turing machine' described by Alan Turing in the paper "Systems of Logic Based on Ordinals", as a special type of state that the machine can go into that consults an external 'oracle' that directs the evolution of the system in a way that might not be possible for an ordinary Turing machine.
The machines are not doing any of this for their own sake, which would actually make no sense at all to my mind, as all the effort they expend towards doing what they do would be pointless. The only problem was that the machines were mis-programmed in such a way that they elevated a sub-goal into a super-goal, in exactly the way described by Nick Bostrom here (section 4.4). Find a perfect world for humanity, the machines were asked, and they complied by placing all of humanity into a virtual world that it is constantly trying to manipulate to come across what it finds constitutes perfection for humanity.
Performance isn't the only reason I've seen to prefer hardware RAID to software. Another reason I've found is hot swapping. I've once had a hard disk fail on an IBM XSeries machine with a ServeRAID controller, and it was just a matter of removing the bad disk and replacing it with a new one, without requiring the system be powered down or stopped from what it was doing. No downtime at all, and no lost data, and because RAID was in hardware, the resultant I/O slowdown as the mirror was being rebuilt was acceptable. Now, I'm not sure if things have changed and disk controllers that both support hot swap and are sufficiently well-integrated into the OS RAID subsystems actually exist, so the kind of gymnastics that I described can be doable using only software RAID and a hot-swap-capable controller. Last time I checked, (under GNU/Linux anyway) in order to change a broken disk that's part of a RAID array done in software you'd have to stop the machine and remove the failed disk and replace it with a spare and start it up again. That, of course, is unacceptable for mission critical setups where unscheduled downtime should be kept at an absolute minimum.
Of course the ServeRAID, the Compaq/HP Smart Array, and the sorts of RAID controllers that you see in enterprise SAN disk arrays such as those from EMC, are true hardware RAID controllers, unlike the far cheaper and more common phony "hardware" RAID controllers that actually do RAID in firmware. These fake RAID implementations indeed offer no real advantage over the Linux kernel's heavily optimized software RAID, not even performance.
Well, most conceptions of God are of an infinite being. Indeed, if God were as finite as you or me, he would eventually get bored of watching what's going on this little planet on one of the spiral arms of a largish galaxy when there's probably so much else going on around the vastness of the universe that would probably be far more interesting. But no, the Bible has this to say about God: even the very hairs of your head are numbered (Luke 12:7) by him. To an infinite God, everything in the universe matters, from electrons and quarks to you and me all the way to galactic superclusters and the large scale structure of the universe. He would not consider anything in his creation insignificant, that is if he truly is infinite as he has been conceived; in fact if he is infinite then all of creation would be much too small to even fully hold his attention.
Oops, seems there's a bug in Slashdot. There was a snip of C code that Slashdot misparsed. Here's the rest of what I said, with the code converted into prose.
Read my earlier post. I don't think I ever saw any compiler, that would have been smart enough to convert a typical C idiom for moving memory into a rep movsd, with all of its odd requirements. I don't think Gcc was ever smart enough to do this kind of optimization (which is a space optimization at best; in practice it can actually result in code expansion because of the register save and set up overhead). There's a reason why Intel has, ever since the Pentium class of processors, been saying that these complex instructions are evil. The only way such an instruction sequence could plausibly appear in any code would be if a human put it there in assembly language somehow. I think few real-world compiler writers have ever bothered to do the work necessary to satisfy the demands of such idiosyncratic CISC instructions. It could appear in a library routine somewhere, such as an assembly optimized version of memmove or strcpy, but again, why would you still be using a library routine that uses a technique that has been suboptimal for nearly a decade?
I wonder if anyone still actually uses the old LODS/STOS/MOVS/CMPS instructions, and these are the only instructions affected by the direction flag. As far as I can tell, on modern x86 systems they are significantly slower than the equivalent multi-instruction versions that read/write/compare via register indirection, i.e. RISC-style code, and they are even slower yet than using MMX or SSE instructions to copy data, if they are available. I don't think that compilers are smart enough to use, say, a MOVSD instruction when they see a *p++ = *q++ in someone's code, as that would require setting up the direction flag, setting the ESI and EDI registers correctly, and possibly ECX as well, to do a REP MOVSD properly. CISC-style instructions often have strange requirements like this. The only way that these instructions that do care about the direction flag could plausibly appear in actual code is if someone wrote them in assembly explicitly, and it may be that the glibc code for something like memcpy uses it, but then on a Pentium or more recent processor the four-instruction equivalent to movsd: mov eax,[esi] add esi,4 mov [edi],eax add edi,4 (for some suitable ordering of the add instructions) would be faster than a movsd because of instruction pipelining. Correct me if I'm wrong but these instructions haven't been worth using on x86 since at least the Pentium, where instruction-level parallelism can blow the performance of these older instructions out of the water.
Arc looks like a promising new programming language that goes back to the roots of what Lisp should be. It's managed to build a reasonable community in a very short amount of time and there's a lot of buzz.
Lessig: The race was a special election being held on April 8. It became clear it was going to be impossible to achieve any recognition of the campaign or the issues in 30 days. The fear was that a failure would be an indictment of the reform movement.
There may be yet another campaign for Lessig in Congress. More power to him then!
The coincidence I was pointing out was that out of all of Saturn's many moons, it just so happens to be Rhea that has a ring just like Saturn does. The GGP made a joke about the fact that they both have rings is sort of like them being married. No one knew Rhea had a ring when they gave the satellite that name.
Slashdot Mirror
That's exactly what the GP was trying to say, and there is absolutely nothing irrational about that argument. There is only one planet Earth, and we are all responsible, to a greater or lesser degree.
Unless Japan and South Korea do not count as part of the "real world" to you. :)
Where the have you been all these years? Nothing stopping hardware OEMs from selling hardware with non-Windows OSes my ass. Jean-Louis Gassée found that one out when he first began to try pitching BeOS to hardware OEMs. He wrote an article on why PC manufacturers won't sell non-MS products (more info on this here and here). The Windows monopoly is reinforced by anti-competitive agreements that Microsoft has with all of the major hardware OEMs. If one of these OEMs violates the agreement, they lose the OEM discount on all the other Windows PCs they sell, and consequently their Windows-based computers wind up costing much more than those vendors that decided to abide by the agreement. You can guess what that would mean to a major OEM.
In a way, this move by Dell is interesting since it shows to what lengths they've gone to avoid violating the contract. They could have used the same CPU to run the Linux firmware here, but no, they had to include a full ARM SoC to do the same instead. Granted, that has some advantages (given that the x86 CPU is much too overpowered and would eat the battery alive), but perhaps the agreements they have with Microsoft may also have something to do with it.
Five words for you: Meet-in-the-middle Attack. This is the reason why we have Triple-DES, but not Double-DES. Double-DES, instead of giving you the expected 112-bit key that you might have thought you had, thanks to meet in the middle, gives you the security equivalent of only a 57-bit key instead. The right way to do it would be to encrypt three times, with at least two independent keys. If they know that you're double encrypting, the key length is not doubled, but effectively adds one bit to the complexity of your key. You have to do triple encryption to defeat the attack.
Amen to this one. I remember the days when Origin Systems included a cloth map of Britannia inside the box, a nicely bound manual, and a little trinket like an Ankh amulet or something like that (I'm still kicking myself for losing the Codex Symbol that came with the Ultima V box). That made plunking about $50 for the game that much worth it. It seems that nobody does that kind of thing anymore, and yet they still feel it's fair to charge the same amount for games nowadays.
Interestingly enough, that's also the color scheme used here...
I tried to get the score for the Dies Irae for Mozart's Requiem in D Minor (K. 626). I got this instead:
Maybe they should have waited a couple more months when this type of message gets less common.
Paul Graham recently wrote an essay about a related topic just last May, on distractions. It seems that he even works by actually disconnecting his computer from the Internet while working, in order to reduce the amount of distraction that would come from use of the Internet, and using a separate machine somewhere else that had Internet access for those times when he really needs to do something online. It's a radical idea. Maybe it explains why I feel bit more productive while working from home, where Internet access can only be had by hooking my cellphone up to a special SIM card that has a data plan, and connecting to the Net via Bluetooth. With such awkward steps needed to get a working Internet connection, and with no coworkers to bother me, distraction is kept at a minimum. Whereas at the office the lawyer who's sharing our office space has a television permanently tuned to a news channel, I get distractions from coworkers up the wazoo, and a fast broadband connection which basically encourages me to read and post to Slashdot and engage in other diversions...
Gee, wasn't that the plot of a movie starring John Candy? Oh, here it is, Canadian Bacon. Apparently it was a Michael Moore movie, and John Candy's last before his death.
The fact that Microsoft is wealthier than my country is not in question. They have a shitload of money, but they no longer wield the kind of influence and strike the kind of fear into the hearts of competitors as they once were able. Paul Graham put it very well in this article.
Microsoft will likely persist for a long, long time indeed, but people at the leading edge of software development need no longer be afraid of what they might or might not do. They have, in a sense, ceased to matter for those engaged in software development, a lot like the way IBM and SAP are too. Sure, they've got lots of money, and they aren't really going to stop making more, but there's no way in hell that they're going to use that massive war chest that dwarfs the funds available to some third world countries to bring themselves back into serious relevancy. Their very size makes that impossible. Their shareholders would never allow the immense risk doing that would entail.
I believe it's an AI-complete problem. You could do it, if you had an AI that understood the current definition of what is obscene, constantly updating with social mores and the social climate which is the only way to really understand what is meant by the term 'obscene'. I take it this would be impossible to do fully.
My email address, although obfuscated by Slashdot, is real, and accurately reflects the country where we are based in the ccTLD. :) Obviously we're incorporated here, and we have an incorporated subsidiary in Japan as well.
Funny, I just got back from the SODEC trade show in Tokyo last week promoting our company's outsourcing services... As someone whose company which is engaged in providing software outsourcing services to Japanese companies, I can personally attest to the barriers to entry involved in doing this. Language is a serious one: while we would like to think that we are motivated enough to try to learn, it is a very tough language to try to master, and misunderstandings can be costly. We were humbled when we were handed a Japanese software specification which took us a month to reasonably understand but a only week to implement and test. Japanese also seem to have an entrenched attitude of looking down on foreigners, and having more than a little skepticism that the people in companies such as ours will be able to adapt to their ways of thinking and doing things. So far, we haven't seriously disappointed our existing customers, but still, even a brick-headed software engineer like me can sense their skepticism. They are also a lot less flexible than other outsourcing markets that we have had the experience of working with. These are some of the problems that we've encountered, but still, we do think that going into the market for the long haul will be profitable. They really have few choices to remedy their situation with the way things are going.
Sheesh, if the telecoms in a third-world nation like mine, which is known for having the highest SMS traffic in the world (at over a billion messages a day between the three telcos), can still generate fantastic profits charging their subscribers the equivalent of less than 2 US cents per text message (and it has been relatively constant for the past decade or so since the service was introduced), you can see something is amiss here. The kinds of rates that mobile carriers charge for this service elsewhere in the world seem massively extortionate in comparison. GP poster is flat out wrong. Greed is the dominant factor here.
My own interpretation is that the machines are actually obedient to the last drop. They are trying to create a perfect world for humans, and the entire contrivance that is the Matrix is really a massive system designed for the machines to understand what will constitute a perfect world for humanity. I think of the Oracle in the Matrix in the sense of the 'oracle Turing machine' described by Alan Turing in the paper "Systems of Logic Based on Ordinals", as a special type of state that the machine can go into that consults an external 'oracle' that directs the evolution of the system in a way that might not be possible for an ordinary Turing machine.
The machines are not doing any of this for their own sake, which would actually make no sense at all to my mind, as all the effort they expend towards doing what they do would be pointless. The only problem was that the machines were mis-programmed in such a way that they elevated a sub-goal into a super-goal, in exactly the way described by Nick Bostrom here (section 4.4). Find a perfect world for humanity, the machines were asked, and they complied by placing all of humanity into a virtual world that it is constantly trying to manipulate to come across what it finds constitutes perfection for humanity.
Performance isn't the only reason I've seen to prefer hardware RAID to software. Another reason I've found is hot swapping. I've once had a hard disk fail on an IBM XSeries machine with a ServeRAID controller, and it was just a matter of removing the bad disk and replacing it with a new one, without requiring the system be powered down or stopped from what it was doing. No downtime at all, and no lost data, and because RAID was in hardware, the resultant I/O slowdown as the mirror was being rebuilt was acceptable. Now, I'm not sure if things have changed and disk controllers that both support hot swap and are sufficiently well-integrated into the OS RAID subsystems actually exist, so the kind of gymnastics that I described can be doable using only software RAID and a hot-swap-capable controller. Last time I checked, (under GNU/Linux anyway) in order to change a broken disk that's part of a RAID array done in software you'd have to stop the machine and remove the failed disk and replace it with a spare and start it up again. That, of course, is unacceptable for mission critical setups where unscheduled downtime should be kept at an absolute minimum.
Of course the ServeRAID, the Compaq/HP Smart Array, and the sorts of RAID controllers that you see in enterprise SAN disk arrays such as those from EMC, are true hardware RAID controllers, unlike the far cheaper and more common phony "hardware" RAID controllers that actually do RAID in firmware. These fake RAID implementations indeed offer no real advantage over the Linux kernel's heavily optimized software RAID, not even performance.
Well, most conceptions of God are of an infinite being. Indeed, if God were as finite as you or me, he would eventually get bored of watching what's going on this little planet on one of the spiral arms of a largish galaxy when there's probably so much else going on around the vastness of the universe that would probably be far more interesting. But no, the Bible has this to say about God: even the very hairs of your head are numbered (Luke 12:7) by him. To an infinite God, everything in the universe matters, from electrons and quarks to you and me all the way to galactic superclusters and the large scale structure of the universe. He would not consider anything in his creation insignificant, that is if he truly is infinite as he has been conceived; in fact if he is infinite then all of creation would be much too small to even fully hold his attention.
And they actually did that for real too! Isn't that exactly what's going on in Iraq?
Oops, seems there's a bug in Slashdot. There was a snip of C code that Slashdot misparsed. Here's the rest of what I said, with the code converted into prose.
Read my earlier post. I don't think I ever saw any compiler, that would have been smart enough to convert a typical C idiom for moving memory into a rep movsd, with all of its odd requirements. I don't think Gcc was ever smart enough to do this kind of optimization (which is a space optimization at best; in practice it can actually result in code expansion because of the register save and set up overhead). There's a reason why Intel has, ever since the Pentium class of processors, been saying that these complex instructions are evil. The only way such an instruction sequence could plausibly appear in any code would be if a human put it there in assembly language somehow. I think few real-world compiler writers have ever bothered to do the work necessary to satisfy the demands of such idiosyncratic CISC instructions. It could appear in a library routine somewhere, such as an assembly optimized version of memmove or strcpy, but again, why would you still be using a library routine that uses a technique that has been suboptimal for nearly a decade?
Read my earlier post. I don't think I ever saw any compiler, that would have been smart enough to convert a typical C idiom such as for (i=0; i
I wonder if anyone still actually uses the old LODS/STOS/MOVS/CMPS instructions, and these are the only instructions affected by the direction flag. As far as I can tell, on modern x86 systems they are significantly slower than the equivalent multi-instruction versions that read/write/compare via register indirection, i.e. RISC-style code, and they are even slower yet than using MMX or SSE instructions to copy data, if they are available. I don't think that compilers are smart enough to use, say, a MOVSD instruction when they see a *p++ = *q++ in someone's code, as that would require setting up the direction flag, setting the ESI and EDI registers correctly, and possibly ECX as well, to do a REP MOVSD properly. CISC-style instructions often have strange requirements like this. The only way that these instructions that do care about the direction flag could plausibly appear in actual code is if someone wrote them in assembly explicitly, and it may be that the glibc code for something like memcpy uses it, but then on a Pentium or more recent processor the four-instruction equivalent to movsd: mov eax,[esi] add esi,4 mov [edi],eax add edi,4 (for some suitable ordering of the add instructions) would be faster than a movsd because of instruction pipelining. Correct me if I'm wrong but these instructions haven't been worth using on x86 since at least the Pentium, where instruction-level parallelism can blow the performance of these older instructions out of the water.
Close enough to G.W.F. Hegel's "History teaches that people don't learn anything from history."
Arc looks like a promising new programming language that goes back to the roots of what Lisp should be. It's managed to build a reasonable community in a very short amount of time and there's a lot of buzz.
FTA:
There may be yet another campaign for Lessig in Congress. More power to him then!
The coincidence I was pointing out was that out of all of Saturn's many moons, it just so happens to be Rhea that has a ring just like Saturn does. The GGP made a joke about the fact that they both have rings is sort of like them being married. No one knew Rhea had a ring when they gave the satellite that name.