You're overdramatizing. This is a process that will take hundreds if not thousands of years, even with technology helping to accelerate it. It's not like we'll wake up 10 years from now with a unified language and forget how to read today's literature!
By the time we have a unified language, we'll have a whole new set of literature to go along with it. Today's literature will be like ancient greek literature, and yes, it will only be readable by people with special training. It will need to be translated, just like ancient greek is today. What's the big deal? The biggest difference is that only one translation would be needed, and therefore all the translation work could be focused on that.
Furthermore, nobody will be forced to adopt a unified language. It will simply evolve. Words will travel from one language to another. Phrases will creep in from other languages. Languages will become closer, and eventually merge. You can see it happening today; at least the beginnings. It will only continue even faster, as the Internet is here to stay and the growth of the global marketplace shows no signs of slowing.
Academics care about linguistic diversity in an abstract sense, but normal people really don't. People care about it, but in a much more practical sense of everyday communication. People will accept gradual, evolutionary changes to their language, as long as they can express themselves in a way they like. Academics often fight against change, because their theories were all developed to explain the old ways of doing things. They will fight against language unification; luckily I believe they will not be able to prevent it, or even slow it very much. [Note: this is a gross generalization about "academics", please remember that all generalizations are false.]
You ask what's so great about a global language? The removal of all language barriers from everything! Duh!
Maybe you don't personally notice any language barriers right now, but that doesn't mean you couldn't benefit from their removal. Maybe there are some really cool people in China right now doing brilliant work in your field that you just don't know about because it's all in Chinese. Maybe you would benefit from the increased efficiency of a global economy without language barriers. I think it's an indisputable fact that removing language barriers is a great thing.
Who needs to be careful? Hopefully the Internet *will* cause languages to merge. It could be like the Tower of Babel in reverse. Wouldn't it be great if there was a unified global language?
Now I know some people would be quite upset at the horrible "loss" of cultural diversity implied by a single global language. But we can be just as diverse in many other ways that don't cause us to be unable to communicate with each other on a basic level. And IMHO, being able to communicate is much more important than some academic's ideal of "cultural identity".
Because the Internet is not controlled and maintained by the government like the road system is. (Purposely, I might add, and with many benefits. If the government controlled the Internet it would be much different than it is today. It would probably suck.) That is why your analogy is flawed.
The market must act as the force that keeps Microsoft honest. Why the market has not done so is an interesting question. My theory is that since Windows *is* the computer for most people, any problems with Microsoft software are simply blamed on computers in general and seen as unavoidable. Though if you look at interest in Linux, a large chunk has been due to the perception that it is more secure. So there have been some effects. Apple also benefits from this, to a lesser extent I believe.
Only if you type equations into the calculator at high speed, one after the other, constantly. If you use your calculator normally, >90% of the time it's sitting idle or taking typed input. If you're doing really long calculations *and* you're anal about turning it off every time you finish a calculation, you might cut that to 50%. During that time an overclocked calculator will be doing no work but using twice the battery power anyway. Which is why I say it would be useful to overclock only when doing calculations, in the manner of Intel's Pentium M laptops with the adjustable clock speeds.
Graphing complex functions is slow. Calculating integrals is slow. 3D graphs are abysmally slow. Speeding these functions up could be quite useful. Of course, you could just use Virtual TI on your PC if you wanted it to be really fast, or there's always Mathematica. I'm sure overclocking your calculator cuts the battery life in half or worse, which is why they are clocked so low by TI to begin with. Now if he could figure out a way for it to automatically overclock itself only while doing calculations (not waiting for input), then he might be onto something...
The easier fix is to edit the article but not change anything, then link to the version in the page history. The page history can't be edited, so the link is permanent. OTOH, this goes against the Wiki spirit, because this would discourage Slashdot visitors from improving the page as well. (the edit link would be gone, and improvements to the current version wouldn't show up on Slashdot).
You misunderstand again. The beauty of the klik approach is you *don't* need incredibly complex logic. You simply need *one* script for each distro version supported. This script installs the base libraries and doesn't need to be complex at all.
The script goes through the list of base libraries (which is much much smaller than the entire list of possible packages for each distro). For each library it has a location to check and a location to obtain the file if it is not there. It simply downloads the files you are missing and that's it. Very simple. If you want to be fancy it can use the distro's package manager to install the files if possible, but that is not even necessary. In fact, if you don't use a specific distro package manager, this script could probably be portable across several distros. After this simple script finishes you are ready to run any klik package.
There are of course tradeoffs with the klik approach. However I think that it is the most sensible packaging solution to come along yet. It provides a middle ground between "huge self-contained packages with everything included" and "completely modular packages, but you need to download 10 dependencies to install any software". I'll list some of the potential problems I see:
Any library not included with the base must be downloaded in each klik package that uses it.
I believe a sensible base library set can be found such that most applications will be a reasonably-sized download, yet the base library set is still manageable. Yes some libraries will be downloaded twice, but it's not the end of the world.
The base libraries don't change, what if one is upgraded?
If a particular program requires an updated version it can be included in the klik package. I think the longer-term solution, however, is to provide a new version of the base library set every few months. Since it hopefully is kept small, this is still doable.
Because they haven't yet done the work necessary to support installation on non-Debian-based distros. However, the concept is sound and the implementation is straightforward. All that is needed is to compile a list of common base software that should be available on every distro. Then when you install the klik client it can check to make sure you have that software and install it if you don't (using a simple script that runs your distro's package manager or simply downloads the necessary files from klik itself). After that klik would work the same way it does on Debian. It's much simpler and more straightforward than developing an alternate meta-packaging system that does on-the-fly dependency resolution using any of 10 different package management systems; and therefore much more likely to work correctly and much less work to implement and maintain. That's why it's better than Autopackage. Less complexity = better software.
You misunderstand. It does not require a debian-based distribution. The current implementation assumes that some software (mainly KDE) is preinstalled on a user's machine, but it doesn't necessarily need to have been installed from debian packages.
You have not given any reason why autopackage is better than klik. I think klik is better. Firstly, klik has *no* installer. Packages just work the instant you download them. That beats an installer any day of the week. Klik does lots of dependency checking and resolution also, but it works behind the scenes before the package is downloaded, completely invisible to the end user. Furthermore, klik is useful today, with lots of packages that are autogenerated from Debian packages. Autopackage might have a couple of packages now, but nothing near the selection klik has.
Where did you get that information? It isn't in the article, which says instead "running a Linux-based operating system as well as all applications". It isn't on Realm Systems' website, which contains practically no useful information at all (in the grand tradition of corporatespeak "paradigm shift" websites). Furthermore, since when does Embedded XP run on dual PowerPCs? And if all it does is connect using Terminal Services, there's no point in having dual PowerPCs, 64 MB RAM, and a 20 GB hard drive onboard. You can do that with a a regular USB key.
If you want to be believed, post as a real user instead of Anonymous Coward, and cite your sources.
Oh, I forgot to list the disadvantages. The way I see it, hardware compatibility will probably be spotty. Not only does it need to detect and have drivers for all types of PC graphics cards, network/wireless cards, sound cards, modems, and mice; it has to use them all through this weird USB connection, plus it has to dynamically switch to a completely different set without rebooting when it is plugged into a different PC. Even if it works this process can't be very fast, not to mention the fact that it needs to re-setup the Internet connection (if it's not using DHCP you'll need to specify the correct network settings too, ouch) and don't forget that it needs to hibernate Windows too, which takes a while and doesn't always work correctly. I couldn't see the whole process taking less than a minute in a best-case scenario, which makes it a lot more annoying to, say, plug it in for a quick email check.
Either that, or it doesn't really hibernate and take over the host computer like they say, but instead simply runs a proxy program on the host computer's OS to display stuff on the screen and get input from the keyboard/mouse. This would be fast, convenient, and highly compatible (and I suspect for these reasons that this is how it really works); however, it would be completely vulnerable to keyloggers or other malicious software running on the host machine, defeating the whole security aspect of the device.
It's not for you, it's for large companies. Now that I understand what it is, I think it could actually be useful. The article didn't help much, but if you want to know what it really is:
It is a USB hard drive + integrated computer running something like knoppix + fingerprint scanner, in a package not much larger than an iPod. You take it with you anywhere, then plug it into any computer you find. After scanning your fingerprint it automatically takes over that computer and brings up your suspended Linux desktop on the screen, just how you left it. It also sets up a VPN over the Internet to access your company's intranet so you can work just like you were in the office. When you're done you unplug it, which instantly suspends your session until you plug it in again.
The central server is cool because it backs up all of the devices as they are being used, so if somebody loses theirs it's not a problem. You can just copy their backed-up image onto a new device and give it to them. You can also track usage and do security junk like that. I'm not a big fan of that feature, since it will probably lead to stupidly restrictive policies that are automatically enforced with an iron hand by the software. But that's a management problem, not a problem with the technology per se.
I'm not impressed by guesture recognition either, but what about visual feedback? Imagine this: a mouse pointer is on the screen, but it stays stationary as you move the screen around it, allowing you to select things. Or the screen acts as a window onto a large image or web page, which stays stationary as you move the phone around to read different parts. The workability of this scheme would depend greatly on very high-quality motion sensing, though. It remains to be seen just how high-quality Samsung's is. Plus holding your phone up in the air would probably get tiring.
Mozilla should add more web development features in general. The DOM inspector seems like it would be cool, but I've never been able to use it effectively for anything, it's rather cumbersome. Same for the Javascript debugger. They should both be integrated together, get some serious UI work, and then get integrated with Mozilla's HTML editor, which should do HTML validation internally. Mozilla could be an IDE like Visual Studio for HTML/javascript. That would work wonders for web developer mindshare, because even IE-centric developers might be lured to Mozilla by effective debugging of their HTML and Javascript. No browser has ever really provided that before, because they are always in end-user mode, where they try their best to render malformed HTML instead of telling the user where the problems are.
Firefox has an extensive manual, it's in the Help menu. Even "Help for Internet Explorer users" is right on the help menu. It gives advice on how to migrate, what's different in Firefox, and how to use Firefox's new features to their full potential. I don't run Mozilla, maybe it doesn't have that; yet another reason Firefox is better.
Hardly any websites use 600-800 KB files. For the 30 KB images you download for most webpages, this compression would only reduce the load time a tiny bit. For slow dialup connections, it might actually improve load times. Some sites might find it useful. And in a year or two, faster computers will cut the delay in half. Of course the question is moot unless Firefox achieves >90% market share, since Microsoft isn't likely to add IE support for this format anytime soon.
You're assuming the end user isn't going to upload anything. If real multicast becomes available, it will fuel a boom in home serving. Imagine a multicast-enabled bittorrent. You could literally distribute a large file to the entire internet simultaneously from your 384 kbps upload connection, at very little or no cost to you personally. You could have a protocol which acted like HTTP while demand was low, but when requests became frequent it would switch to multicast, enabling you to serve a nearly unlimited number of users from any connection. Slashdotting would become a thing of the past. Gnutella would become waaaay better. The Internet as we know it would change drastically, and there is no doubt in my mind that ISP traffic would increase dramatically despite the increased efficiency.
The idea of switching to multicast when available to encourage ISPs to offer it is extremely interesting. What if BitTorrent added multicast capability? Would backbone ISPs race to cut the 35% of their traffic that is BitTorrent by finally implementing usable multicast?
Multicast needs to be implemented in the network, by ISPs. They haven't. Even if they did, they would charge you through the nose for it, just because they can.
They don't have to worry too much yet. I think the answer to the question posed in the article title is "No." The one station on peercast.org at this time with more than 20 listeners skips like crazy. Furthermore, I suspect that the upstream bandwidth of most listeners is not yet large enough to support decent video content, making peercasting TV infeasable. Certainly you're not going to get HDTV or even normal broadcast quality from this anytime soon.
However, I do have to commend the peercast.org folks for an exceptionally nice user experience for their software. It installs in a snap and works immediately with zero configuration, using my default media players even. That's a big step toward wide adoption. Now if only the the ISPs would stop being so stingy with upload bandwidth, so the concept actually had a chance of working...
Maybe peercasting will be used for something other than traditional content with actors. For example, people who take interesting video of newsworthy events with their camcorders or phones or whatever will be able to broadcast it themselves, hopefully without fear of getting Slashdotted to death.
Interesting. I got to the end of the article before I realized it was written by a woman. I wonder if a man went around and interviewed booth babes in the same way, would they tell him the sorts of things they told this woman? I'll bet it would turn out to be a totally different article.
Cringley talking about VOIP and PBXes got me thinking:
VOIP unlimited calling plan: $19.99/mo
2-phone cell plan with unlimited mobile-to-mobile minutes: $49.99/mo
Connect one cell to your computer and leave it at home; take the other with you. Use PBX software to rig up a bridge between your VOIP and the home cell. Use the home cell as a gateway between your roaming cell and the VOIP number, abusing the free mobile-to-mobile minutes.
Poof! Unlimited cellular anytime minutes to/from anywhere in the US, and low VOIP rates on all your international calls, for $69.98 per month. If you're a heavy user of anytime minutes or international calls, the savings could add up very fast. At least until the phone company figures out what you are doing...
Slashdot Mirror
By the time we have a unified language, we'll have a whole new set of literature to go along with it. Today's literature will be like ancient greek literature, and yes, it will only be readable by people with special training. It will need to be translated, just like ancient greek is today. What's the big deal? The biggest difference is that only one translation would be needed, and therefore all the translation work could be focused on that.
Furthermore, nobody will be forced to adopt a unified language. It will simply evolve. Words will travel from one language to another. Phrases will creep in from other languages. Languages will become closer, and eventually merge. You can see it happening today; at least the beginnings. It will only continue even faster, as the Internet is here to stay and the growth of the global marketplace shows no signs of slowing.
Academics care about linguistic diversity in an abstract sense, but normal people really don't. People care about it, but in a much more practical sense of everyday communication. People will accept gradual, evolutionary changes to their language, as long as they can express themselves in a way they like. Academics often fight against change, because their theories were all developed to explain the old ways of doing things. They will fight against language unification; luckily I believe they will not be able to prevent it, or even slow it very much. [Note: this is a gross generalization about "academics", please remember that all generalizations are false.]
You ask what's so great about a global language? The removal of all language barriers from everything! Duh!
Maybe you don't personally notice any language barriers right now, but that doesn't mean you couldn't benefit from their removal. Maybe there are some really cool people in China right now doing brilliant work in your field that you just don't know about because it's all in Chinese. Maybe you would benefit from the increased efficiency of a global economy without language barriers. I think it's an indisputable fact that removing language barriers is a great thing.
Now I know some people would be quite upset at the horrible "loss" of cultural diversity implied by a single global language. But we can be just as diverse in many other ways that don't cause us to be unable to communicate with each other on a basic level. And IMHO, being able to communicate is much more important than some academic's ideal of "cultural identity".
The market must act as the force that keeps Microsoft honest. Why the market has not done so is an interesting question. My theory is that since Windows *is* the computer for most people, any problems with Microsoft software are simply blamed on computers in general and seen as unavoidable. Though if you look at interest in Linux, a large chunk has been due to the perception that it is more secure. So there have been some effects. Apple also benefits from this, to a lesser extent I believe.
Only if you type equations into the calculator at high speed, one after the other, constantly. If you use your calculator normally, >90% of the time it's sitting idle or taking typed input. If you're doing really long calculations *and* you're anal about turning it off every time you finish a calculation, you might cut that to 50%. During that time an overclocked calculator will be doing no work but using twice the battery power anyway. Which is why I say it would be useful to overclock only when doing calculations, in the manner of Intel's Pentium M laptops with the adjustable clock speeds.
Graphing complex functions is slow. Calculating integrals is slow. 3D graphs are abysmally slow. Speeding these functions up could be quite useful. Of course, you could just use Virtual TI on your PC if you wanted it to be really fast, or there's always Mathematica. I'm sure overclocking your calculator cuts the battery life in half or worse, which is why they are clocked so low by TI to begin with. Now if he could figure out a way for it to automatically overclock itself only while doing calculations (not waiting for input), then he might be onto something...
The easier fix is to edit the article but not change anything, then link to the version in the page history. The page history can't be edited, so the link is permanent. OTOH, this goes against the Wiki spirit, because this would discourage Slashdot visitors from improving the page as well. (the edit link would be gone, and improvements to the current version wouldn't show up on Slashdot).
The script goes through the list of base libraries (which is much much smaller than the entire list of possible packages for each distro). For each library it has a location to check and a location to obtain the file if it is not there. It simply downloads the files you are missing and that's it. Very simple. If you want to be fancy it can use the distro's package manager to install the files if possible, but that is not even necessary. In fact, if you don't use a specific distro package manager, this script could probably be portable across several distros. After this simple script finishes you are ready to run any klik package.
There are of course tradeoffs with the klik approach. However I think that it is the most sensible packaging solution to come along yet. It provides a middle ground between "huge self-contained packages with everything included" and "completely modular packages, but you need to download 10 dependencies to install any software". I'll list some of the potential problems I see:
I believe a sensible base library set can be found such that most applications will be a reasonably-sized download, yet the base library set is still manageable. Yes some libraries will be downloaded twice, but it's not the end of the world.
If a particular program requires an updated version it can be included in the klik package. I think the longer-term solution, however, is to provide a new version of the base library set every few months. Since it hopefully is kept small, this is still doable.
Because they haven't yet done the work necessary to support installation on non-Debian-based distros. However, the concept is sound and the implementation is straightforward. All that is needed is to compile a list of common base software that should be available on every distro. Then when you install the klik client it can check to make sure you have that software and install it if you don't (using a simple script that runs your distro's package manager or simply downloads the necessary files from klik itself). After that klik would work the same way it does on Debian. It's much simpler and more straightforward than developing an alternate meta-packaging system that does on-the-fly dependency resolution using any of 10 different package management systems; and therefore much more likely to work correctly and much less work to implement and maintain. That's why it's better than Autopackage. Less complexity = better software.
You misunderstand. It does not require a debian-based distribution. The current implementation assumes that some software (mainly KDE) is preinstalled on a user's machine, but it doesn't necessarily need to have been installed from debian packages.
You have not given any reason why autopackage is better than klik. I think klik is better. Firstly, klik has *no* installer. Packages just work the instant you download them. That beats an installer any day of the week. Klik does lots of dependency checking and resolution also, but it works behind the scenes before the package is downloaded, completely invisible to the end user. Furthermore, klik is useful today, with lots of packages that are autogenerated from Debian packages. Autopackage might have a couple of packages now, but nothing near the selection klik has.
If you want to be believed, post as a real user instead of Anonymous Coward, and cite your sources.
Either that, or it doesn't really hibernate and take over the host computer like they say, but instead simply runs a proxy program on the host computer's OS to display stuff on the screen and get input from the keyboard/mouse. This would be fast, convenient, and highly compatible (and I suspect for these reasons that this is how it really works); however, it would be completely vulnerable to keyloggers or other malicious software running on the host machine, defeating the whole security aspect of the device.
It is a USB hard drive + integrated computer running something like knoppix + fingerprint scanner, in a package not much larger than an iPod. You take it with you anywhere, then plug it into any computer you find. After scanning your fingerprint it automatically takes over that computer and brings up your suspended Linux desktop on the screen, just how you left it. It also sets up a VPN over the Internet to access your company's intranet so you can work just like you were in the office. When you're done you unplug it, which instantly suspends your session until you plug it in again.
The central server is cool because it backs up all of the devices as they are being used, so if somebody loses theirs it's not a problem. You can just copy their backed-up image onto a new device and give it to them. You can also track usage and do security junk like that. I'm not a big fan of that feature, since it will probably lead to stupidly restrictive policies that are automatically enforced with an iron hand by the software. But that's a management problem, not a problem with the technology per se.
I'm not impressed by guesture recognition either, but what about visual feedback? Imagine this: a mouse pointer is on the screen, but it stays stationary as you move the screen around it, allowing you to select things. Or the screen acts as a window onto a large image or web page, which stays stationary as you move the phone around to read different parts. The workability of this scheme would depend greatly on very high-quality motion sensing, though. It remains to be seen just how high-quality Samsung's is. Plus holding your phone up in the air would probably get tiring.
Mozilla should add more web development features in general. The DOM inspector seems like it would be cool, but I've never been able to use it effectively for anything, it's rather cumbersome. Same for the Javascript debugger. They should both be integrated together, get some serious UI work, and then get integrated with Mozilla's HTML editor, which should do HTML validation internally. Mozilla could be an IDE like Visual Studio for HTML/javascript. That would work wonders for web developer mindshare, because even IE-centric developers might be lured to Mozilla by effective debugging of their HTML and Javascript. No browser has ever really provided that before, because they are always in end-user mode, where they try their best to render malformed HTML instead of telling the user where the problems are.
Firefox has an extensive manual, it's in the Help menu. Even "Help for Internet Explorer users" is right on the help menu. It gives advice on how to migrate, what's different in Firefox, and how to use Firefox's new features to their full potential. I don't run Mozilla, maybe it doesn't have that; yet another reason Firefox is better.
Hardly any websites use 600-800 KB files. For the 30 KB images you download for most webpages, this compression would only reduce the load time a tiny bit. For slow dialup connections, it might actually improve load times. Some sites might find it useful. And in a year or two, faster computers will cut the delay in half. Of course the question is moot unless Firefox achieves >90% market share, since Microsoft isn't likely to add IE support for this format anytime soon.
You're assuming the end user isn't going to upload anything. If real multicast becomes available, it will fuel a boom in home serving. Imagine a multicast-enabled bittorrent. You could literally distribute a large file to the entire internet simultaneously from your 384 kbps upload connection, at very little or no cost to you personally. You could have a protocol which acted like HTTP while demand was low, but when requests became frequent it would switch to multicast, enabling you to serve a nearly unlimited number of users from any connection. Slashdotting would become a thing of the past. Gnutella would become waaaay better. The Internet as we know it would change drastically, and there is no doubt in my mind that ISP traffic would increase dramatically despite the increased efficiency.
Wow, someone needs to tell Bram Cohen about this...
The idea of switching to multicast when available to encourage ISPs to offer it is extremely interesting. What if BitTorrent added multicast capability? Would backbone ISPs race to cut the 35% of their traffic that is BitTorrent by finally implementing usable multicast?
Multicast needs to be implemented in the network, by ISPs. They haven't. Even if they did, they would charge you through the nose for it, just because they can.
However, I do have to commend the peercast.org folks for an exceptionally nice user experience for their software. It installs in a snap and works immediately with zero configuration, using my default media players even. That's a big step toward wide adoption. Now if only the the ISPs would stop being so stingy with upload bandwidth, so the concept actually had a chance of working...
Maybe peercasting will be used for something other than traditional content with actors. For example, people who take interesting video of newsworthy events with their camcorders or phones or whatever will be able to broadcast it themselves, hopefully without fear of getting Slashdotted to death.
Interesting. I got to the end of the article before I realized it was written by a woman. I wonder if a man went around and interviewed booth babes in the same way, would they tell him the sorts of things they told this woman? I'll bet it would turn out to be a totally different article.
VOIP unlimited calling plan: $19.99/mo
2-phone cell plan with unlimited mobile-to-mobile minutes: $49.99/mo
Connect one cell to your computer and leave it at home; take the other with you. Use PBX software to rig up a bridge between your VOIP and the home cell. Use the home cell as a gateway between your roaming cell and the VOIP number, abusing the free mobile-to-mobile minutes.
Poof! Unlimited cellular anytime minutes to/from anywhere in the US, and low VOIP rates on all your international calls, for $69.98 per month. If you're a heavy user of anytime minutes or international calls, the savings could add up very fast. At least until the phone company figures out what you are doing...