The GNU project was trying to create a free version of Unix - the GNU system - and was going about it in a systematic fashion, one tool at the time. The kernel was left until last, and Linux simply happened to come at the right moment, when most of the system was already up and running but the kernel wasn't.
where's my "-1 Wrong" modifier? saying GNU was trying to create a free Unix is simply laughable. you know what it stands for, right? HURD was never intended to be unix, it was intended to be something else entirely. the GNU toolset was developed on Unix because that was the best development environment available, and happened to also be the most similar to GNU's vision for HURD. the fact that the GNU tools look vaguely unix-like (at best; --some-really-long-option, for example, is decidedly un-unixy) is almost more of a coincidence. they were explicitly not trying to create a free Unix.
As it happens, the GNU project does have a working kernel of their own, HURD. HURD never really took off, mainly because Linux got the snowball effect going - it got some users, some of whom began co-developing it, making it better, which in turn gained it more users and more developers and so on. Linux has almost all the developers, so HURD has almost none.
HURD never took off due to the lack of a clearly articulated vision, lack of leadership, and the fact that Linux filled most of the true goals of the contributors: "gimme something free i can use that mostly works". sure, Linux getting all the flash in the media and schools biases people towards working on that over other kernels, but HURD can't put all their failings on Linux's shoulders. the various BSD's - hell, even Plan 9 - manage to progress further and more quickly than HURD does, despite HURD being the pet project of GNU's nominal head.
But thinking that Linux is the true success story and the GNU project just a less important side path is absurd. It's the GNU project that made Linux possible, not the other way around.
really? can you document that? to me, it sounds like speculation. i disagree, and think it's the other way around: without a useful core, the tools would've remained just a curiosity. by contrast, by 1994 (at the latest; possibly as early as 1991), Linux could've just used the BSD user-land stuff (and, IMHO, we'd all have been better off). Linux didn't make GNU possible, that's true - but it made it relevant.
You think that Linux - a single operating system kernel - is going to have more lasting influence than the whole free software movement, of which the Linux kernel is just a part of ?
that's not what the parent said. he said the future belongs to people like Linus more than people like RMS. that's true. even within the GNU world there's these differences in outlook and the results become apparent.
Especially when what allowed Linux to grow in the first place was the development model made possible by the GPL ?
stop drinking GNU's cool-aid. what allowed Linux to grow initially was the open space for a free unix-like system on the most common cheap hardware. initially, the "development model" was Linus hacking on things. then it was a bunch of people hacking on things, and Linus putting the bits together. the development model was not new, and was not invented by GNU.
i agree with your concern in general, but it's worth pointing out that one can, in fact, file a provisional patent which lasts up to one year. i'm not sure what the expense involved is, and you have to provide more than just a description (which is a good thing, in my opinion), but the bar is much lower than a full patent app.
This is definitely at the expense of the inventor, and would also make invalidation of obvious patents much harder, since prior art would no longer apply.
this law sucks, but not for this reason. no, it wouldn't. prior art is only applicable today if it's disclosed and available: otherwise, the PTO can't reliably tell whether the claim of prior art is on the level or not ("i had a one-click system in 1985, i just didn't show it to anyone! honest!"). how disclosed or available is open to interpretation, but the only real impact moving from first-to-invet to first-to-file would have would be the inclusion of a new set of prior art: that between the invention and the filing.
the bible (including Genesis) changes over time, just phenomenally slowly. look at the scholarship around the J, P, E, and other authors in the torah. or the book selection process during the protestant reformation, or the fact that the roman and orthodox churches use different sets of books in "the" bible. and that's to say nothing of translation and editing issues, which make the biggest difference in the modern world (for christians, anyway; jews are affected much less by current translation issues, muslims practically not at all).
no, it doesn't. it summarizes the fatal flaw in using religion to replace science. religion's a very useful tool for understanding some aspects of the world, including human nature and psychology, even if you don't believe it's a useful or true description of the metaphysical. it's just not a replacement for science, nor a good means for understanding, say, physical phenomenon. don't confuse arguments against using religion to teach science with arguments against religion in general.
yes, this works, at least partly. speech recognition is very tricky stuff, and i doubt even this would solve all the problems, but it would help. but there's a more fundamental problem: in theory, he's demo'ing the real product. most people's computers don't have any provision for plugging in two mics to do that sort of noise cancelation; that sort of processing is non-trivial - i wouldn't want it chewing up CPU cycles generally, and most computers won't have special hardware for it; and most people don't have two mics. to overcome these problems, MS would have to make the demo differ significantly from the actual product.
it's unfair that you've been mod'd flamebait. i even have mod point, but i commented elsewhere on this story before seeing your comment. sorry about that. no, you're not flamebait; it would even be insightful - if it were true. microsoft, at any point, did very little real research on speech recognition systems. the innovation in that field has always come from other places. by your reasoning, Microsoft should've been the front-runner on speech systems until, say, 1996. they weren't. until around 1996 or 1997, the stuff coming out of Bell Labs probably was. around 1997, they stagnated and IBM made some very interesting advances. in terms of actual products, the two of them were the leaders until around 2001, when i lost track of such things. various universities also have produced very promising research and prototypes. but MS has never been out front on that.
that's the point. this isn't really about the technology in these demos. as someone who's worked building speech systems, i'll tell you that type of demo was a big risk. sure, ambient noise seemed low to us, but who knows what it was like on the stage? did he have a monitor speaker up there? and echo - at least enough to throw off the mic used for recognition - seems entirely possible. had they run the demo in exactly that sort of environment? with a simulated audience? had the presenter trained the system himself? speech recognition's hard, especially dictation-type systems. still, i have to agree with the anchor at the end: "live television's rough; welcome to our world." if you aren't sure you can pull it off, it's generally foolish to take that kind of risk. we in the tech world generally agree that MS hasn't gotten where they are because of superior technology; i've often heard them referred to (and referred to them myself) as a marketing company rather than a technology company. so why do they get this stuff wrong so often? where are the high-profile botches from, say, Apple?
i agree on a few counts: the goal focus for projects probably comes (at least in significant part) from the corporate world, and projects which focus on community rather than goals tend to stick around longer. but your conclusion from this strikes me as totally backwards. the result of a strictly community-focused group is one which sticks around, but never accomplishes anything. the OpenDarwin directors considered the state of things a failure because they didn't accomplish their goals. i'm not even sure how else you'd define "failure". it became, instead, a hosting site - which there are already plenty of, which do a perfectly fine job, run by people for whom that is their goal.
building a community is a perfectly reasonable goal, of course. but this idea that goals are somehow "corporate" and therefore antithetical to community spirit or some nonsense like that is one of the most detrimental (and pervasive) ideas in the open source world.
That vision disappeared pretty much the day George Washington left office.
no it didn't. at the earliest, it disappeared at the conclusion of the Civil War; more likely, during reconstruction. a strong federal government which essentially dominated the states was the most significant outcome of that war.
this is a common misunderstanding, even by Bell Labs employees and management. Bell Labs never just meant the research folks. originally, way back when, all development was done by an organization called Bell Labs, then handed over to the business units, basically to market and sell. later, the development shops were pushed off into the business units. the employees were no longer under the head of Bell Labs on the org chart, but were still Bell Labs employees - all AT&T (later Lucent) technical employees are. i believe that was around divestiture in 1984. at that point, "core" Bell Labs (i don't believe that was ever an official term, but it was certainly common enough that everyone knew what you meant) consisted of Research and AT (i think it stood for Advanced Technology), which was essentially an in-house contracting shop. with the major restructuring of Bell Labs a little under a year ago by the then-new head (the largest restructuring since 1984, at least), AT is gone, Research is still there, and now they finally have groups with the explicit charter of taking research - real research - and bringing it into the product shops, something which had been missing since the development staff were moved under the business units.
and, in my opinion, that's what primarily caused Lucent's collapse: a decade and a half worth of disconnect between possibly the most brilliant research organization on the planet and Lucent's product shops. it started in 1984, and the inertia was just so incredible that it took a decade and a half to catch up with them.
the "rent your phone forever" model made a lot more sense when they were manufactured to last forever, cost proportionally more, AT&T controlled who could connect devices to their network, repairs/replacements were free forever with good turn-around time, and so on. the decline of that started probably in the early '70s, and certainly by divestiture in 1984 it no longer made sense for folks to lease their phones. from at least that point, your grandmother had the option to buy a $35 Radio Shack phone, or whatever else she preferred, to do the job.
believe it or not, some folks are still renting their phones. AT&T (and Lucent, who i think actually manages the phone leasing business now, with a name under license from AT&T) certainly understands that this isn't a wise thing to do, but that doesn't mean they're going to say "no thanks" to people who'd rather give them lots of money.
besides, your general point is misguided: they came up with plenty which was practical for their customers. things like phone rentals - relatively low cost, steady income stream - help subsidize other things like the development of direct-dial long distance. and that was pretty cool for their customers.
Re:How does it compare with the SavaJe OS
on
Driving Plan 9
·
· Score: 1
huh. outside of conversations i've had with one other guy here at work, i've not heard anybody mention SavaJe in years. brief history:
around 1996, some of the folks at Bell Labs who worked on Plan 9 started a related project called Inferno. it took the same ideas contained in Plan 9 (and much of the current implementation) and wrapped them up in a virtual machine (Dis), new application programming language (Limbo), different graphics programming system (tk), and some other changes. Plan 9 and Inferno fed each other healthily for quite some time. Inferno was handed over to a business unit in Lucent to try to make a product out of. despite significant technical progress, the BU failed at pretty much all their business goals, and subsequently collapsed. Inferno was used internally in a few projects, and Vita Nuova got the rights to do ongoing development, worldwide distribution, and so on. a few folks from the Inferno BU went off and founded SavaJe, taking the then-currnet copy of the source with them to use as a base. SavaJe has evolved entirely independently since then, and isn't really related now.
i don't really like Java, but for what it's worth, i think if you're going to build a java-bassed device what those guys are doing is exactly the right way to do it: build it in low level, push for full J2EE (not J2ME).
ah, where i can't reply. lovely. well, since my reading of that is that you're not particularly interested in the debate, just name calling and posturing, i'll keep this (relatively) brief.
you simply assert, without justification or follow-up, that a Mac is not as "programmer-friendly" as a Linux box; i honestly can't make sense of this statement. at worst, they're on equal footing (factoring out things which are purely personal taste, like desktop environment), with the Mac having nearly all the Linux tools, libraries, and APIs available to it, plus a set more which developers may either use or safely ignore. you talk about "missing the future" or something (you're missing a verb there; maybe "know" around the parens?), and then go on to talk about Vista. again, your argument is totally unclear, but i think you're implying that the only alternative to believing that Linux is "cutting edge" is to believe Windows is. which, of course, betrays your total lack of depth in computer science. yeah, Windows is a botch, through and through. but what kind of excuse is that? comp sci has moved on, at least in the research and academic communities. the Unix ideas are currently most clearly embodied in Plan 9 or Inferno today, and taken much farther, to very interesting effect; vastly different system architectures like Oberon were very promising, if not downright wonderful. yet your response to a criticism that Linux might not be "cutting edge" focuses on Windows. which is exactly the problem.
yeah, MS Windows stalled computing science progress in industry for probably about a decade and a half. and now it's Linux's turn, re-inventing things that were old when Windows 95 was just a gleam in Gates' eye.
and IBM, and Sun. from a business perspective, IRIX and GL were simply ways for SGI to sell boxes, the same way OS X is "just" a way for Apple to sell boxes. grandparent needs to realize that stand-along OS or system software companies are the exception (Microsoft being the only really successful one, and that's largely due to a collection of other market forces; more commonly, they end up like SCO). application software is somewhat different; there's a more viable market there. but SGI's not known for any of that; are they really likely to go head-to-head with folks like Adobe and Apple and win, starting nearly from scratch? no, hardware remains the best bet for SGI to recover, if there's any way at all to pull it off.
this has been tried several times by several different folks. the most famous examples, i think, are AT&T's Hobbit processor and the various efforts at a Java processor or JVM-on-silicone. the Java case is interesting, since the design of the virtual machine is particularly mismatched from that of most modern processors (note the comment comparing the JVM and Dis, the Limbo/Inferno VM, at the end of the wikipedia article; it's maybe a little misleading now since the JVM's had millions of dollars worth of effort poured into optimization, but it's still mostly true). in every case, the outcome has been the same: the market simply doesn't support the effort. every once in a while, someone with the resources to do chip design and fab stumbles upon this idea and has a go of it, but it always works out the same: the incentive to switch - increased performance for applications written in a particular language - simply isn't worth the costs - mostly, toolchain development/maintenance and porting costs, along with the usually-tremendous performance hit running applications in any other language.
first, for most people security just isn't worth very much. they want to be able to check it off on a list, but that's about it. MS Office says it's secure? ? done. compatibility - not just the ability to read, but the ability to look 100% visually the same - is a much, much bigger deal for most corporate folks outside of engineering. second, you're assuming a rational consumer. that is an invalid assumption that leads to the undoing of loads of business models. "consumers" should under no circumstances be understood to be, as a class, rational beings. they are primarily emotional beings. you don't have to watch much television or read many mainstream magazines to realize that the advertising industry generally understands this. in the role of "consumer", people are driven by flash, by fear, by mob thinking.
when a given tradeoff becomes too painful, rational people reconsider their options; consumers go shopping.
telnet was just intended to be the simplest example. sure, service providers probably don't want me telneting into my phone, but they very much do want me doing interesting things with it - things involving moving data around in both directions. the same question goes for pushing, say, streaming video, IM invitations, you can't do that to an email address. and providers most certainly want that (provided they get money for it; data plans are a first cut at that, and they're looking for other ways).
unfortunately, in our company (and in many others, from folks i've talked to), that "killer feature" is the ability to create something which looks 100% "correct" on what the majority of people you communicate with use. we don't get MS Office for our engineers, and instead issue them OpenOffice (really NeoOffice, since we're a largely Mac shop); they don't prepare Office docs for consumption outside the company, only rarely for inside the company (outside System Engineering, which i'm embarrassed to say produces their requirements in Word), and it's not important that what they see or produce be rendered 100% visually correct, as long as it's close enough to efficiently and correctly communicate the content. OO (and NeoOffice) fit that bill just fine. but for the more customer-facing types - sales, marketing, business development - the "look" is important. it needs to be slick. so all those folks, as well as the top management across the board, get MS Office (about 2/3 on Mac, the rest on Windows).
this problem is certainly not specific to OO; MS Office is frequently not compatible with MS Office. as i said, we're mainly a Mac shop and the only thing that occasionally jumps up and bites us about that is version incompatibilities between MS Office files. we've seen MS Office (on Mac) create files that various versions of MS Office (on Windows) simply can't open, and vice versa; that's rare, but rendering errors are common. this is even true between various versions of MS Office on Windows, but the frequency of the cases seem to increase geometrically the farther you get from using the same version.
one of the business development guys i work with finally got fed up with the situation last week and said "we should just be PDFing these things anyway." i wanted to hug him.
how does one telnet to 12223334444@serviceprovider.com? that's not any sort of endpoint address. as the intelligence of mobile devices continues to grow, the mindset you seem to be trapped in for addressing them will become less and less tenable (it's already a royal pain; buy me lunch some time and i'll tell you all about the problems we've had building a mixed-communication-mode system trying to talk to mobiles). NAT works remarkably well for a significant number of cases. but it's no magic bullet. SRV records in DNS actually could make the arbitrary portmapping you seem to be advocating a lot more powerful, but converting all the web infrastructure in the world to no longer assume port 80, but instead look up a SRV record - and then repeating that for ssh, telnet, ftp, &c - is a herculean job, as well. lacking that, NAT is virtually useless for providing services. and "providing services" is a much broader thing than just "being a server" - VoIP, P2P, game traffic, and so on all qualify. it's an ongoing engineering sink to continue to work around NAT. NAT is great if you're happy with the producer-consumer model prevalent in old-school media, but lots of people aren't.
i'm not entirely clear on what your point is. i emphatically agree with everything your quoted study claims. from context, it seems you were presenting it as a counter to something you misunderstood me to be saying. my only claim is that when home-schooling does fail - and, like any system, it inevitably does at least sometimes - the most common cause is the parent overlooking the need for proper socialization. i make no claim about the frequency of such failures. were i to speculate, based on personal experience with youth i've had the opportunity to work with i'd say they're probably not very common, somewhat less common than the failures of the western public school system.
sibling comments about only including online persons in the count for the final number are likely valid. another possibility is the fact that one email sent != one email received. mailing lists are the most common case of this. of course, that still depends on how the 50M's being counted.
Uhm. Since when is school the only place for kids to socialize?
it's not the only place, certainly, but it's an important one. at the latest, since the introduction of modern western public education, that's a very important part of the purpose (likely for earlier educational systems, but i don't really know much about educational systems between the ancient greeks and our modern western variety). socialization can certainly be done elsewhere, but it requires specific attention on the part of the parent. when home schooling fails, it's generally because the parent overlooks that.
School is supposed to be for learning, anyhow. Let the kids socialize on their own time. Maybe if they actually taught kids things in school instead of "socializing" them, things would work better in the first place.
you have an overly-narrow definition of learning. yes, school is for learning, but in addition to the material written on the curriculum, there are many, many more lessons to be learned - arguably far more important ones. these range from simple meta-accademic concerns - for example, you do homework not just to memorize the specific facts/formulas/&c, but to learn how to learn more difficult things - to more abstract life skills, like how to work with people with different interests and backgrounds, how to understand differing viewpoints, and so on. perhaps if people spent more time in school focusing on those, instead of an obsession with test scores, things would work better in the first place.;-)
note also that the balance between "life skill" type learning from socialization and academic learning changes over the course of a student's life. the charter school in question is K-8, when most folks agree the balance is in favor of the former.
Slashdot Mirror
i agree with your concern in general, but it's worth pointing out that one can, in fact, file a provisional patent which lasts up to one year. i'm not sure what the expense involved is, and you have to provide more than just a description (which is a good thing, in my opinion), but the bar is much lower than a full patent app.
the bible (including Genesis) changes over time, just phenomenally slowly. look at the scholarship around the J, P, E, and other authors in the torah. or the book selection process during the protestant reformation, or the fact that the roman and orthodox churches use different sets of books in "the" bible. and that's to say nothing of translation and editing issues, which make the biggest difference in the modern world (for christians, anyway; jews are affected much less by current translation issues, muslims practically not at all).
no, it doesn't. it summarizes the fatal flaw in using religion to replace science. religion's a very useful tool for understanding some aspects of the world, including human nature and psychology, even if you don't believe it's a useful or true description of the metaphysical. it's just not a replacement for science, nor a good means for understanding, say, physical phenomenon. don't confuse arguments against using religion to teach science with arguments against religion in general.
heh. that was great. the "massive branch prediction logic" bit was my favorite.
yes, this works, at least partly. speech recognition is very tricky stuff, and i doubt even this would solve all the problems, but it would help. but there's a more fundamental problem: in theory, he's demo'ing the real product. most people's computers don't have any provision for plugging in two mics to do that sort of noise cancelation; that sort of processing is non-trivial - i wouldn't want it chewing up CPU cycles generally, and most computers won't have special hardware for it; and most people don't have two mics. to overcome these problems, MS would have to make the demo differ significantly from the actual product.
it's unfair that you've been mod'd flamebait. i even have mod point, but i commented elsewhere on this story before seeing your comment. sorry about that.
no, you're not flamebait; it would even be insightful - if it were true. microsoft, at any point, did very little real research on speech recognition systems. the innovation in that field has always come from other places. by your reasoning, Microsoft should've been the front-runner on speech systems until, say, 1996. they weren't. until around 1996 or 1997, the stuff coming out of Bell Labs probably was. around 1997, they stagnated and IBM made some very interesting advances. in terms of actual products, the two of them were the leaders until around 2001, when i lost track of such things. various universities also have produced very promising research and prototypes. but MS has never been out front on that.
that's the point. this isn't really about the technology in these demos. as someone who's worked building speech systems, i'll tell you that type of demo was a big risk. sure, ambient noise seemed low to us, but who knows what it was like on the stage? did he have a monitor speaker up there? and echo - at least enough to throw off the mic used for recognition - seems entirely possible. had they run the demo in exactly that sort of environment? with a simulated audience? had the presenter trained the system himself? speech recognition's hard, especially dictation-type systems.
still, i have to agree with the anchor at the end: "live television's rough; welcome to our world." if you aren't sure you can pull it off, it's generally foolish to take that kind of risk. we in the tech world generally agree that MS hasn't gotten where they are because of superior technology; i've often heard them referred to (and referred to them myself) as a marketing company rather than a technology company. so why do they get this stuff wrong so often? where are the high-profile botches from, say, Apple?
In the future, please look at link referents before clicking on them if you care about what's going to be returned.
i agree on a few counts: the goal focus for projects probably comes (at least in significant part) from the corporate world, and projects which focus on community rather than goals tend to stick around longer. but your conclusion from this strikes me as totally backwards. the result of a strictly community-focused group is one which sticks around, but never accomplishes anything. the OpenDarwin directors considered the state of things a failure because they didn't accomplish their goals. i'm not even sure how else you'd define "failure". it became, instead, a hosting site - which there are already plenty of, which do a perfectly fine job, run by people for whom that is their goal.
building a community is a perfectly reasonable goal, of course. but this idea that goals are somehow "corporate" and therefore antithetical to community spirit or some nonsense like that is one of the most detrimental (and pervasive) ideas in the open source world.
this is a common misunderstanding, even by Bell Labs employees and management. Bell Labs never just meant the research folks. originally, way back when, all development was done by an organization called Bell Labs, then handed over to the business units, basically to market and sell. later, the development shops were pushed off into the business units. the employees were no longer under the head of Bell Labs on the org chart, but were still Bell Labs employees - all AT&T (later Lucent) technical employees are. i believe that was around divestiture in 1984. at that point, "core" Bell Labs (i don't believe that was ever an official term, but it was certainly common enough that everyone knew what you meant) consisted of Research and AT (i think it stood for Advanced Technology), which was essentially an in-house contracting shop. with the major restructuring of Bell Labs a little under a year ago by the then-new head (the largest restructuring since 1984, at least), AT is gone, Research is still there, and now they finally have groups with the explicit charter of taking research - real research - and bringing it into the product shops, something which had been missing since the development staff were moved under the business units.
and, in my opinion, that's what primarily caused Lucent's collapse: a decade and a half worth of disconnect between possibly the most brilliant research organization on the planet and Lucent's product shops. it started in 1984, and the inertia was just so incredible that it took a decade and a half to catch up with them.
the "rent your phone forever" model made a lot more sense when they were manufactured to last forever, cost proportionally more, AT&T controlled who could connect devices to their network, repairs/replacements were free forever with good turn-around time, and so on. the decline of that started probably in the early '70s, and certainly by divestiture in 1984 it no longer made sense for folks to lease their phones. from at least that point, your grandmother had the option to buy a $35 Radio Shack phone, or whatever else she preferred, to do the job.
believe it or not, some folks are still renting their phones. AT&T (and Lucent, who i think actually manages the phone leasing business now, with a name under license from AT&T) certainly understands that this isn't a wise thing to do, but that doesn't mean they're going to say "no thanks" to people who'd rather give them lots of money.
besides, your general point is misguided: they came up with plenty which was practical for their customers. things like phone rentals - relatively low cost, steady income stream - help subsidize other things like the development of direct-dial long distance. and that was pretty cool for their customers.
huh. outside of conversations i've had with one other guy here at work, i've not heard anybody mention SavaJe in years. brief history:
around 1996, some of the folks at Bell Labs who worked on Plan 9 started a related project called Inferno. it took the same ideas contained in Plan 9 (and much of the current implementation) and wrapped them up in a virtual machine (Dis), new application programming language (Limbo), different graphics programming system (tk), and some other changes. Plan 9 and Inferno fed each other healthily for quite some time. Inferno was handed over to a business unit in Lucent to try to make a product out of. despite significant technical progress, the BU failed at pretty much all their business goals, and subsequently collapsed. Inferno was used internally in a few projects, and Vita Nuova got the rights to do ongoing development, worldwide distribution, and so on. a few folks from the Inferno BU went off and founded SavaJe, taking the then-currnet copy of the source with them to use as a base. SavaJe has evolved entirely independently since then, and isn't really related now.
i don't really like Java, but for what it's worth, i think if you're going to build a java-bassed device what those guys are doing is exactly the right way to do it: build it in low level, push for full J2EE (not J2ME).
ah, where i can't reply. lovely. well, since my reading of that is that you're not particularly interested in the debate, just name calling and posturing, i'll keep this (relatively) brief.
you simply assert, without justification or follow-up, that a Mac is not as "programmer-friendly" as a Linux box; i honestly can't make sense of this statement. at worst, they're on equal footing (factoring out things which are purely personal taste, like desktop environment), with the Mac having nearly all the Linux tools, libraries, and APIs available to it, plus a set more which developers may either use or safely ignore. you talk about "missing the future" or something (you're missing a verb there; maybe "know" around the parens?), and then go on to talk about Vista. again, your argument is totally unclear, but i think you're implying that the only alternative to believing that Linux is "cutting edge" is to believe Windows is. which, of course, betrays your total lack of depth in computer science. yeah, Windows is a botch, through and through. but what kind of excuse is that? comp sci has moved on, at least in the research and academic communities. the Unix ideas are currently most clearly embodied in Plan 9 or Inferno today, and taken much farther, to very interesting effect; vastly different system architectures like Oberon were very promising, if not downright wonderful. yet your response to a criticism that Linux might not be "cutting edge" focuses on Windows. which is exactly the problem.
yeah, MS Windows stalled computing science progress in industry for probably about a decade and a half. and now it's Linux's turn, re-inventing things that were old when Windows 95 was just a gleam in Gates' eye.
and IBM, and Sun. from a business perspective, IRIX and GL were simply ways for SGI to sell boxes, the same way OS X is "just" a way for Apple to sell boxes. grandparent needs to realize that stand-along OS or system software companies are the exception (Microsoft being the only really successful one, and that's largely due to a collection of other market forces; more commonly, they end up like SCO).
application software is somewhat different; there's a more viable market there. but SGI's not known for any of that; are they really likely to go head-to-head with folks like Adobe and Apple and win, starting nearly from scratch?
no, hardware remains the best bet for SGI to recover, if there's any way at all to pull it off.
this has been tried several times by several different folks. the most famous examples, i think, are AT&T's Hobbit processor and the various efforts at a Java processor or JVM-on-silicone. the Java case is interesting, since the design of the virtual machine is particularly mismatched from that of most modern processors (note the comment comparing the JVM and Dis, the Limbo/Inferno VM, at the end of the wikipedia article; it's maybe a little misleading now since the JVM's had millions of dollars worth of effort poured into optimization, but it's still mostly true).
in every case, the outcome has been the same: the market simply doesn't support the effort. every once in a while, someone with the resources to do chip design and fab stumbles upon this idea and has a go of it, but it always works out the same: the incentive to switch - increased performance for applications written in a particular language - simply isn't worth the costs - mostly, toolchain development/maintenance and porting costs, along with the usually-tremendous performance hit running applications in any other language.
first, for most people security just isn't worth very much. they want to be able to check it off on a list, but that's about it. MS Office says it's secure? ? done. compatibility - not just the ability to read, but the ability to look 100% visually the same - is a much, much bigger deal for most corporate folks outside of engineering.
second, you're assuming a rational consumer. that is an invalid assumption that leads to the undoing of loads of business models. "consumers" should under no circumstances be understood to be, as a class, rational beings. they are primarily emotional beings. you don't have to watch much television or read many mainstream magazines to realize that the advertising industry generally understands this. in the role of "consumer", people are driven by flash, by fear, by mob thinking.
when a given tradeoff becomes too painful, rational people reconsider their options; consumers go shopping.
telnet was just intended to be the simplest example. sure, service providers probably don't want me telneting into my phone, but they very much do want me doing interesting things with it - things involving moving data around in both directions. the same question goes for pushing, say, streaming video, IM invitations, you can't do that to an email address. and providers most certainly want that (provided they get money for it; data plans are a first cut at that, and they're looking for other ways).
unfortunately, in our company (and in many others, from folks i've talked to), that "killer feature" is the ability to create something which looks 100% "correct" on what the majority of people you communicate with use. we don't get MS Office for our engineers, and instead issue them OpenOffice (really NeoOffice, since we're a largely Mac shop); they don't prepare Office docs for consumption outside the company, only rarely for inside the company (outside System Engineering, which i'm embarrassed to say produces their requirements in Word), and it's not important that what they see or produce be rendered 100% visually correct, as long as it's close enough to efficiently and correctly communicate the content. OO (and NeoOffice) fit that bill just fine. but for the more customer-facing types - sales, marketing, business development - the "look" is important. it needs to be slick. so all those folks, as well as the top management across the board, get MS Office (about 2/3 on Mac, the rest on Windows).
this problem is certainly not specific to OO; MS Office is frequently not compatible with MS Office. as i said, we're mainly a Mac shop and the only thing that occasionally jumps up and bites us about that is version incompatibilities between MS Office files. we've seen MS Office (on Mac) create files that various versions of MS Office (on Windows) simply can't open, and vice versa; that's rare, but rendering errors are common. this is even true between various versions of MS Office on Windows, but the frequency of the cases seem to increase geometrically the farther you get from using the same version.
one of the business development guys i work with finally got fed up with the situation last week and said "we should just be PDFing these things anyway." i wanted to hug him.
how does one telnet to 12223334444@serviceprovider.com? that's not any sort of endpoint address. as the intelligence of mobile devices continues to grow, the mindset you seem to be trapped in for addressing them will become less and less tenable (it's already a royal pain; buy me lunch some time and i'll tell you all about the problems we've had building a mixed-communication-mode system trying to talk to mobiles).
NAT works remarkably well for a significant number of cases. but it's no magic bullet. SRV records in DNS actually could make the arbitrary portmapping you seem to be advocating a lot more powerful, but converting all the web infrastructure in the world to no longer assume port 80, but instead look up a SRV record - and then repeating that for ssh, telnet, ftp, &c - is a herculean job, as well. lacking that, NAT is virtually useless for providing services. and "providing services" is a much broader thing than just "being a server" - VoIP, P2P, game traffic, and so on all qualify. it's an ongoing engineering sink to continue to work around NAT.
NAT is great if you're happy with the producer-consumer model prevalent in old-school media, but lots of people aren't.
i'm not entirely clear on what your point is. i emphatically agree with everything your quoted study claims. from context, it seems you were presenting it as a counter to something you misunderstood me to be saying. my only claim is that when home-schooling does fail - and, like any system, it inevitably does at least sometimes - the most common cause is the parent overlooking the need for proper socialization. i make no claim about the frequency of such failures. were i to speculate, based on personal experience with youth i've had the opportunity to work with i'd say they're probably not very common, somewhat less common than the failures of the western public school system.
sibling comments about only including online persons in the count for the final number are likely valid. another possibility is the fact that one email sent != one email received. mailing lists are the most common case of this. of course, that still depends on how the 50M's being counted.
note also that the balance between "life skill" type learning from socialization and academic learning changes over the course of a student's life. the charter school in question is K-8, when most folks agree the balance is in favor of the former.