I guess I still don't see that happening anytime soon. I think humans will continue for a long time to be pushed into jobs that can't be done effectively or economically by machines.
But you never know... Numbers 1 and 3 would solve a lot of problems permenantly (in the worst sense of the phrase). And number 2 would probably be just as bad -- work is a very important part of our psychology. Any of your 3 choices would probably be the end of humanity.
If option 4 (people always have a place) doesn't exist, we're in trouble in the long term.
Like, duh. Half of the jobs available today will be taken by robots. This is called "industrialization," and it has been happening for hundreds of years. As machines become more capable, they take over in jobs that used to require humans. This has several consequences.
The global workforce is essentially bigger, since a machine was added, but the people it replaced were not removed. This surplus can cause a number of different results: unemployment, fewer working hours, or growth of production.
Fast change leads to unemployment. The situation changes more rapidly than the economy can adjust for. The human psychology seems to prefer working a full day, so while we really could probably be working only 1 hour a day and still maintaining sustenance levels of production, we are still working 8 (or more!) hours for a nice lifestyle. While we work less than most people did 100 years ago, the trend seems to be more for the surplus to move into growth of production.
Since we can only eat so much food, the production has to find more creative outlets. Computer Games, Themed Desktops, screen savers with flying toasters -- these are all technically unnecessary, yet somebody gets paid to make them.
So that is what happened to the jobs that were stolen by the tractor -- they were pushed out of the plow driver market and into the Video Game Industry. It took time, but that is where technology has gotten us.
So I'm not too worried. Yeah, I've got to keep sharp so that some AI doesn't steal my job. But I'm not too worried -- I'll hopefully have found something more fun to do when a computer takes over my current position.
Only if the point is to sell the advertised product. Not all spam need be so direct.
For example: CALL 1-800-SOMEBODY-THE-SPAMMER-HATES AND WE WILL GIVE YOU FREE MONEY!!!!!!!!!!
So people start calling some random business's 1-800 number demanding their free money or complaining about the spam. Phone bill goes through the roof, legit calls get DOS'ed, and the spammer might actually be able to put some small company out of business.
I'm sure the more creative among you could come up with even more fun scenarios than this. But let's not give anybody any ideas, eh?
But you and the rest of the Slashdot crowd don't get to decide what those rules are. That is for judges who know the law to decide. This is what drives me nuts about people on Slashdot who post things like "Microsoft is a convicted criminal, and therefore..." Microsoft is a monopoly, which means that they operate under restrictions that other companies do not have. But those restrictions are decided by judges, not by Slashbots.
Funny you should mention it, but I spent the last few evenings undoing that mistake.
Split your requirements into two computers. One should be a devoted gateway. It should do any routing, NATing (if you do that sort of thing), firewalling, traffic shaping, DNS, DHCP, etc., everything associated with "getting the Internet into my residence". Maybe email fits in as well. If you want, you can even put Squid on it. I got a box that is perfect for this for about $105 (after shipping). Just add a second NIC (everything else is included). Maybe a bit of overkill (400 MHz, 128 MB RAM, 6 GB HDD), but hey, it was cheap. Not as small as the IWill, but since I tend to not mess with my router much, it doesn't really need to be any smaller than it is.
Then have second machine -- perhaps the one referenced in this article -- for everything else. Forward port 80 to it so your web sites appear to come from the main system. Forward port 2222 to 22 so that you can SSH into it from the outside world. Put samba, MySQL, XFree, etc. on it.
The point of this is, put the stuff that you simply set up and forget on one machine. Make it dedicated to its purpose of handling the network needs for your residence. Then for everything else, use one or more other machines. Then, if you crash or screw up or whatever the one machine, you haven't taken down your whole local network. You don't have to worry about accidentally messing up your firewall config, since it is on a box that you never touch. You still get a web (and any other) server that appears to be directly exposed, but you can mess with the web server configuration without risking messing with your whole network. Finally, you can disconnect and take off with your secondary server without messing anything up. Before I split the job over two computers, I wanted to take my computer somewhere else for a few days. But I couldn't -- taking it would take the network down. And to make my routing computer work properly in the new environment, I would have to reconfigure it. And then I would have to configure it again to get it back up and running once I put it back to work as a router. Now, I can just unplug the secondary computer. It is already set up to use DHCP, and it doesn't expect to act as a router, so no reconfiguration. And the network at home stays up -- the web server is gone, but that is all.
Anyway, my suggestion is to give permanent services a permanent, stable home. Give the dynamic services and the stuff you will play with a separate box. Kind of like the kernel-space, userland separation that makes an OS stable...
While clock speed and power usage are correlated, they aren't the same thing. "Given the same performance, power usage is a valid factor in deciding on a processor" is not the same as "given the same performance, clock speed is a valid factor in deciding on a processor." One is true, the other is not.
True. Power consumption is a valid criterion for processor selection. However, that doesn't invalidate my point: low GHz/Flops is not a valid criterion.
Some of the compiler optimizations can make a huge difference, especially on a raw benchmark like this. Using SSE2 or not using SSE2 could make a LOT of difference. Turning on SSE2 flags for GCC doesn't necessarily mean that SSE2 will be used in all of the places that it should be used, while in the real world, the computation intensive stuff will probably make good use of SSE2.
GCC is almost certainly the compiler that PPC apps will use. It is the one that Macintosh uses, distributes, and supports. But it isn't necessarily the one that x86 apps will use -- x86 developers also use MSVC and Intel, among others.
I'm guessing you're trying to imply that the Xeon machine was somehow unstable. That could be true, but we don't have enough information.
Why would the Xeon machine crash? Perhaps:
* The Mac version of Mathematica got recompiled for a 64 bit architecture, so it could handle 64 bit memory space. The Xeon machine didn't have a 64 bit version of Mathematica, and therefore couldn't handle it.
* The Xeon version didn't handle PAE properly and had bugs.
It only can offer that power if it can overcome the megahertz disparity.
You can't tell me that a 2 GHz G5 is more powerful than a 3 GHz P4 just because it has higher Flops/GHz. If the performance of the two chips is approximately equal, then they are equal. Case closed. You can't say "It's better because it does it with fewer GHz," since I don't care how many GHz it took. I care how fast it got the job done and how much it cost me.
If IBM is able to bring the clock speed of the G5 up to be on par with the P4 while maintaining its CPI count, then you'll have something to say. Until then, the comparison is pretty meaningless.
That's meaningless. The benchmarks measure performance. If the Apple chips get their performance through RISC, and the Pentium chips don't, who cares? If Apple has the better underlying technology, they should win. If Pentium comes out on top, it means that they have better technology, RISC or no RISC.
They were expecting the article to sit in queue for a day or two before being posted. By some fluke, the article got posted while still relevant. What is SlashDot coming to???
Anything new added to the Windows CE lineup would have to go through a pretty long process before it got PocketPC support, so you won't see any new processor flavors anytime soon.
Actually, in all seriousness, investigating the sources of a story can lead to "Unfair" metamods. This has happened to me twice now:
1. Somebody posts a comment that on the surface looks reasonable, but is actually completely off base and wrong. You can't tell that the comment is wrong without doing some research or following the links.
2. Moderator actually discovers that the poster is blatantly lying, and mods post as Troll.
3. Metamoderator sees that a perfectly reasonable looking post got marked as Troll, so marks that moderation as Unfair.
That is why the metamoderation system is flawed...
I would agree with the idea that Halo was rushed. The later levels lack the depth and the "WOW" factor that I felt in the first few levels. They definitely get more repetitive (think Library, as well as others).
It was definitely good enough, and it was probably the "correct" decision at the time to ship it instead of wait. At that point, XBox really needed a few good titles to get it off the ground, and Halo was needed. On the other hand, now there are enough titles out there that Halo 2 isn't needed to keep the XBox going. Shipping later will probably cost them sales, as Christmas is a great time for it to come out, but it doesn't affect the strategic positioning of the XBox any -- if anything, it improves the strategic positioning because the XBox market will not have another unfinished game. By not rushing Halo 2, Microsoft is stating "we think we would lose more sales by shipping an unfinished product than we would lose by not shipping in time for Christmas."
Competetive means that if you raise your price, the customer will choose the competition.
Non competetive means that if you raise your price, the customer will choose to not purchase the product at all (more accurately, your competition is literally "nothing").
Non competetive means higher prices, but not infinite prices.
Everybody is different, and I suppose I wander in more conservative circles. But nearly everybody I've ever met who has a tattoo regrets it. Several of them are saving for laser tattoo removal. Remember that while the tattoo may express who you are today, you will be a different person 20 years from now. The tattoo might not express who you will be, but it will still be there expressing itself.
Perhaps there are better ways of expressing yourself. I mean, what does dying your hair or getting a piercing say? Everybody is going to interpret it differently, and many of them will get the wrong meaning. If the idea is to make yourself look a certain way, then I suppose it works. But if the idea is to communicate something, you probably could find a more effective and accurate way to do it.
Remember one thing. While we would all like to think that professional sports are about playing games and showing who has the best team, in reality it is about making money. Think about that for a minute.
If your objective is to make the most money, then the ref's job is to keep the fans happy, not to make fair calls. Obviously, the fans will be disappointed if blatently unfair calls are made, so that helps keep the calls within reason. However, if Hichiro Suziki is at bat in his home stadium, the ump might give him the benefit of the doubt on the strike zone so that the fans get to see him make a great hit. If a call is close, the ref will err on the side of "what will make the majority of the fans happy?" That is his job, due to the "invisible hand" market effects at play.
Actually, you'll notice that all kinds of things are being pulled off of the PCI bus in recent chipsets. The southbridge on new Intel mobos has separate bus drops for the built-in IDE ports, USB ports, NICs, and the PCI bus so that they don't steal bandwidth from the PCI bus. Once all of the major peripherals have been pulled off of the PCI bus it is fast enough. But if you put all of those on the same PCI bus, you would probably notice the need for a faster standard.
Yeah, but only the full version of the format program will format at the full 4GB setting. The installer's format program refuses to do so. Probably because it is a bad idea in most cases. I mean, a 64KB cluster size is horrible for general use (fine if you are storing mostly large files, but general use means a whole lot of small files).
A few years ago, I had a large gzipped tarball (source for the Modula III compiler) that I wanted to untar. I think I had 2 partitions, each about 3 GB. One was FAT 16, the other was FAT 32. The FAT 16 partition had about 1 GB free, and the FAT 32 partition had about 700 MB free, so I decided to decompress to the partition with more free space -- the FAT 16 partition. I knew that the tarball was only 300 MB decompressed, so I expected things to work ok. They didn't. I ran out of space. So I moved the tarball to the FAT32 partition to make more room, then tried again (untarring to the FAT16 partition). Again, it ran out of space halfway through. But when I tried untarring to the FAT32 partition, it had plenty of room (and untarred more quickly, to boot). That was when I realized that each of the thousands of tiny files in the tarball was taking up a full 64KB on the FAT16 partition, but only 4KB on the FAT32 partition, exploding the disk usage by a factor of 16.
Ok, long story. But the point is that while FAT16 CAN do 4GB, it is generally a bad idea. And that might explain why nobody made a mad rush to upgrade all of the format utilities to support the 4GB partition size. (Of course, it could have just been laziness.)
Slashdot Mirror
I guess I still don't see that happening anytime soon. I think humans will continue for a long time to be pushed into jobs that can't be done effectively or economically by machines.
But you never know... Numbers 1 and 3 would solve a lot of problems permenantly (in the worst sense of the phrase). And number 2 would probably be just as bad -- work is a very important part of our psychology. Any of your 3 choices would probably be the end of humanity.
If option 4 (people always have a place) doesn't exist, we're in trouble in the long term.
Like, duh. Half of the jobs available today will be taken by robots. This is called "industrialization," and it has been happening for hundreds of years. As machines become more capable, they take over in jobs that used to require humans. This has several consequences.
The global workforce is essentially bigger, since a machine was added, but the people it replaced were not removed. This surplus can cause a number of different results: unemployment, fewer working hours, or growth of production.
Fast change leads to unemployment. The situation changes more rapidly than the economy can adjust for. The human psychology seems to prefer working a full day, so while we really could probably be working only 1 hour a day and still maintaining sustenance levels of production, we are still working 8 (or more!) hours for a nice lifestyle. While we work less than most people did 100 years ago, the trend seems to be more for the surplus to move into growth of production.
Since we can only eat so much food, the production has to find more creative outlets. Computer Games, Themed Desktops, screen savers with flying toasters -- these are all technically unnecessary, yet somebody gets paid to make them.
So that is what happened to the jobs that were stolen by the tractor -- they were pushed out of the plow driver market and into the Video Game Industry. It took time, but that is where technology has gotten us.
So I'm not too worried. Yeah, I've got to keep sharp so that some AI doesn't steal my job. But I'm not too worried -- I'll hopefully have found something more fun to do when a computer takes over my current position.
Only if the point is to sell the advertised product. Not all spam need be so direct.
For example: CALL 1-800-SOMEBODY-THE-SPAMMER-HATES AND WE WILL GIVE YOU FREE MONEY!!!!!!!!!!
So people start calling some random business's 1-800 number demanding their free money or complaining about the spam. Phone bill goes through the roof, legit calls get DOS'ed, and the spammer might actually be able to put some small company out of business.
I'm sure the more creative among you could come up with even more fun scenarios than this. But let's not give anybody any ideas, eh?
But you and the rest of the Slashdot crowd don't get to decide what those rules are. That is for judges who know the law to decide. This is what drives me nuts about people on Slashdot who post things like "Microsoft is a convicted criminal, and therefore..." Microsoft is a monopoly, which means that they operate under restrictions that other companies do not have. But those restrictions are decided by judges, not by Slashbots.
Funny you should mention it, but I spent the last few evenings undoing that mistake.
Split your requirements into two computers. One should be a devoted gateway. It should do any routing, NATing (if you do that sort of thing), firewalling, traffic shaping, DNS, DHCP, etc., everything associated with "getting the Internet into my residence". Maybe email fits in as well. If you want, you can even put Squid on it. I got a box that is perfect for this for about $105 (after shipping). Just add a second NIC (everything else is included). Maybe a bit of overkill (400 MHz, 128 MB RAM, 6 GB HDD), but hey, it was cheap. Not as small as the IWill, but since I tend to not mess with my router much, it doesn't really need to be any smaller than it is.
Then have second machine -- perhaps the one referenced in this article -- for everything else. Forward port 80 to it so your web sites appear to come from the main system. Forward port 2222 to 22 so that you can SSH into it from the outside world. Put samba, MySQL, XFree, etc. on it.
The point of this is, put the stuff that you simply set up and forget on one machine. Make it dedicated to its purpose of handling the network needs for your residence. Then for everything else, use one or more other machines. Then, if you crash or screw up or whatever the one machine, you haven't taken down your whole local network. You don't have to worry about accidentally messing up your firewall config, since it is on a box that you never touch. You still get a web (and any other) server that appears to be directly exposed, but you can mess with the web server configuration without risking messing with your whole network. Finally, you can disconnect and take off with your secondary server without messing anything up. Before I split the job over two computers, I wanted to take my computer somewhere else for a few days. But I couldn't -- taking it would take the network down. And to make my routing computer work properly in the new environment, I would have to reconfigure it. And then I would have to configure it again to get it back up and running once I put it back to work as a router. Now, I can just unplug the secondary computer. It is already set up to use DHCP, and it doesn't expect to act as a router, so no reconfiguration. And the network at home stays up -- the web server is gone, but that is all.
Anyway, my suggestion is to give permanent services a permanent, stable home. Give the dynamic services and the stuff you will play with a separate box. Kind of like the kernel-space, userland separation that makes an OS stable...
While clock speed and power usage are correlated, they aren't the same thing. "Given the same performance, power usage is a valid factor in deciding on a processor" is not the same as "given the same performance, clock speed is a valid factor in deciding on a processor." One is true, the other is not.
True. Power consumption is a valid criterion for processor selection. However, that doesn't invalidate my point: low GHz/Flops is not a valid criterion.
Some of the compiler optimizations can make a huge difference, especially on a raw benchmark like this. Using SSE2 or not using SSE2 could make a LOT of difference. Turning on SSE2 flags for GCC doesn't necessarily mean that SSE2 will be used in all of the places that it should be used, while in the real world, the computation intensive stuff will probably make good use of SSE2.
GCC is almost certainly the compiler that PPC apps will use. It is the one that Macintosh uses, distributes, and supports. But it isn't necessarily the one that x86 apps will use -- x86 developers also use MSVC and Intel, among others.
I'm guessing you're trying to imply that the Xeon machine was somehow unstable. That could be true, but we don't have enough information.
Why would the Xeon machine crash? Perhaps:
* The Mac version of Mathematica got recompiled for a 64 bit architecture, so it could handle 64 bit memory space. The Xeon machine didn't have a 64 bit version of Mathematica, and therefore couldn't handle it.
* The Xeon version didn't handle PAE properly and had bugs.
* Any number of other reasons.
It only can offer that power if it can overcome the megahertz disparity.
You can't tell me that a 2 GHz G5 is more powerful than a 3 GHz P4 just because it has higher Flops/GHz. If the performance of the two chips is approximately equal, then they are equal. Case closed. You can't say "It's better because it does it with fewer GHz," since I don't care how many GHz it took. I care how fast it got the job done and how much it cost me.
If IBM is able to bring the clock speed of the G5 up to be on par with the P4 while maintaining its CPI count, then you'll have something to say. Until then, the comparison is pretty meaningless.
That's meaningless. The benchmarks measure performance. If the Apple chips get their performance through RISC, and the Pentium chips don't, who cares? If Apple has the better underlying technology, they should win. If Pentium comes out on top, it means that they have better technology, RISC or no RISC.
They were expecting the article to sit in queue for a day or two before being posted. By some fluke, the article got posted while still relevant. What is SlashDot coming to???
Nope. Windows CE 4.2 just came out, and it is basically what Pocket PC 2003 is based on. Windows CE 5.0 isn't due for a while.
Nope. Windows Consumer Electronics. (That's official.) And it still is.
Cuz of stupid legacy stuff that expects the "current directory" to start with a drive letter.
Use drive manager to tell Windows to mount them at a mount point instead of at a drive letter.
Windows CE supports:
x86
SH3
SH4
Two flavors of Mips
Arm
Anything new added to the Windows CE lineup would have to go through a pretty long process before it got PocketPC support, so you won't see any new processor flavors anytime soon.
Actually, in all seriousness, investigating the sources of a story can lead to "Unfair" metamods. This has happened to me twice now:
1. Somebody posts a comment that on the surface looks reasonable, but is actually completely off base and wrong. You can't tell that the comment is wrong without doing some research or following the links.
2. Moderator actually discovers that the poster is blatantly lying, and mods post as Troll.
3. Metamoderator sees that a perfectly reasonable looking post got marked as Troll, so marks that moderation as Unfair.
That is why the metamoderation system is flawed...
I would agree with the idea that Halo was rushed. The later levels lack the depth and the "WOW" factor that I felt in the first few levels. They definitely get more repetitive (think Library, as well as others).
It was definitely good enough, and it was probably the "correct" decision at the time to ship it instead of wait. At that point, XBox really needed a few good titles to get it off the ground, and Halo was needed. On the other hand, now there are enough titles out there that Halo 2 isn't needed to keep the XBox going. Shipping later will probably cost them sales, as Christmas is a great time for it to come out, but it doesn't affect the strategic positioning of the XBox any -- if anything, it improves the strategic positioning because the XBox market will not have another unfinished game. By not rushing Halo 2, Microsoft is stating "we think we would lose more sales by shipping an unfinished product than we would lose by not shipping in time for Christmas."
Competetive means that if you raise your price, the customer will choose the competition.
Non competetive means that if you raise your price, the customer will choose to not purchase the product at all (more accurately, your competition is literally "nothing").
Non competetive means higher prices, but not infinite prices.
Everybody is different, and I suppose I wander in more conservative circles. But nearly everybody I've ever met who has a tattoo regrets it. Several of them are saving for laser tattoo removal. Remember that while the tattoo may express who you are today, you will be a different person 20 years from now. The tattoo might not express who you will be, but it will still be there expressing itself.
Perhaps there are better ways of expressing yourself. I mean, what does dying your hair or getting a piercing say? Everybody is going to interpret it differently, and many of them will get the wrong meaning. If the idea is to make yourself look a certain way, then I suppose it works. But if the idea is to communicate something, you probably could find a more effective and accurate way to do it.
Remember one thing. While we would all like to think that professional sports are about playing games and showing who has the best team, in reality it is about making money. Think about that for a minute.
If your objective is to make the most money, then the ref's job is to keep the fans happy, not to make fair calls. Obviously, the fans will be disappointed if blatently unfair calls are made, so that helps keep the calls within reason. However, if Hichiro Suziki is at bat in his home stadium, the ump might give him the benefit of the doubt on the strike zone so that the fans get to see him make a great hit. If a call is close, the ref will err on the side of "what will make the majority of the fans happy?" That is his job, due to the "invisible hand" market effects at play.
Whatever happened to *NICE* time between failure?
Actually, you'll notice that all kinds of things are being pulled off of the PCI bus in recent chipsets. The southbridge on new Intel mobos has separate bus drops for the built-in IDE ports, USB ports, NICs, and the PCI bus so that they don't steal bandwidth from the PCI bus. Once all of the major peripherals have been pulled off of the PCI bus it is fast enough. But if you put all of those on the same PCI bus, you would probably notice the need for a faster standard.
Yeah, but only the full version of the format program will format at the full 4GB setting. The installer's format program refuses to do so. Probably because it is a bad idea in most cases. I mean, a 64KB cluster size is horrible for general use (fine if you are storing mostly large files, but general use means a whole lot of small files).
A few years ago, I had a large gzipped tarball (source for the Modula III compiler) that I wanted to untar. I think I had 2 partitions, each about 3 GB. One was FAT 16, the other was FAT 32. The FAT 16 partition had about 1 GB free, and the FAT 32 partition had about 700 MB free, so I decided to decompress to the partition with more free space -- the FAT 16 partition. I knew that the tarball was only 300 MB decompressed, so I expected things to work ok. They didn't. I ran out of space. So I moved the tarball to the FAT32 partition to make more room, then tried again (untarring to the FAT16 partition). Again, it ran out of space halfway through. But when I tried untarring to the FAT32 partition, it had plenty of room (and untarred more quickly, to boot). That was when I realized that each of the thousands of tiny files in the tarball was taking up a full 64KB on the FAT16 partition, but only 4KB on the FAT32 partition, exploding the disk usage by a factor of 16.
Ok, long story. But the point is that while FAT16 CAN do 4GB, it is generally a bad idea. And that might explain why nobody made a mad rush to upgrade all of the format utilities to support the 4GB partition size. (Of course, it could have just been laziness.)