is not "little more than a single molecule", it is a robot built to nanoscale precision. It has a computer on board (or it wouldn't be a robot), sensors, and tools built into it. It can perceive and respond to its environment.
I agree, however, that microscopic nanobots probably aren't going to be all that useful. They can't see the big picture, so they can't really know what's going on around them. They also can't have terribly powerful computers because they're just too small. Power sources will be problematic, as will communication, further limiting their uses.
IMHO, nanobots will show their greatest potential, not as isolated nanites, but in huge differentiated linked masses (closer to mammals than microbes). This will make them better-coordinated and a lot less likely to go berserk on a self-reproducing rampage (think self-expanding, not self-reproducing). They can still have nanoscale tools, they'll just know where they are and how what they're being used for fits into the larger task.
While inbreeding can cause problems, often severe ones, it is not a death sentence. You can create an entire population from a single pair of siblings. Release a mating pair of rabbits on an island with no predators and lots of food, and come back in a few years; if you don't find rabbits, you probably won't find anything green either.
Animals have given birth to implanted fetuses of other species. I can't remember the exact details of an example... I think there was a rare type of cat that another cat of a common species gave birth to...
Of course the mitochondrial DNA was from the host cell. They knew it would be and didn't really care. It's not a big thing. Mitochondria are mitochondria, they change tranportable blood fuel into usable cell fuel (I'm just not up to big words like glucose tonight). A mammoth with modern elephant (or cow, or pig, or sheep) mitochondria is a mammoth as far as I'm concerned.
(now that that's out of my system...)
The Dolly technique is crusty in other ways, but it should work well enough to get some hairy elephants walking around northern Asia. Well, not quite the Dolly technique... this requires something a little more complicated, but IMHO doable in a year or two with enough money (or ten years from now in somebody's back yard).
I'd agree with you on the DNA bit, but they've got a whole mammoth. That's one heck of a DNA sample! They should be able to patch up the cracks with that big a sample.
So you're saying the phone companies are guilty of not subsidizing ISDN so they can charge by the minute for local calls? Of course, though, they can't charge by the minute for local calls, and they will never get approval to do so. Any serious talk of a modem tax is referring to special charges for modem users, either directly or indirectly.
ISDN was probably poorly handled, but it was never that great a technology to begin with. It is complex and therefore costly to install (misconfigurations galore), requires that special lines be laid and special equipment installed, and not really that much better than POTS.
ISDN was never really viable as a standard feature in everyone's home. If the prices were high, it was because the telcos knew they would only make money this way. It had to be profitable on the few customers who would pay a premium for a modest performance increase. They aren't stupid; if they saw potential for installing these lines in most homes, they would have lowered the prices and made a fortune on volume.
Now that better technologies are emerging, and internet connections are quickly approaching a household standard, some phone companies are starting to do exactly what you describe, except with DSL.
Of course the phone companies are greedy. Any for-profit public corporation is required by law to seek maximum profit for its shareholders. That doesn't mean that they are acting maliciously.
I know I'm getting a pretty good deal with my cable access, but I see no reason why it can't continue much longer. Bandwidth and cable modems are going down in price, and the cable was already there anyway. It's a good deal for the huge improvement in performance, but not too cheap to pay for itself.
I think cable is going to keep growing, become profitable, and eventually lower in price. It'll also go in waves of slowing down and speeding up as more people join and the system is expanded (it really scales quite nicely, if not trivially; if usage increases to the point where it is a problem, the funding is there to do some rewiring to solve it).
Here in Canada, we have $50/month cable connections. The price is high enough still that they don't have to charge for installation. Everyone who has it raves about it, and people are signing up like mad. I wouldn't be surprised if it was already turning a healthy profit. I'm sure they'd love to compete with the POTS services, and I bet they'll offer limited bandwidth connections for $20 soon.
The prices on POT lines are based on the way they work without ISPs: occasional short uses, not long-term continual connections. You are on a time-share system, and you are insisting on your right to use it 100% of the time.
The phone system is not built to support half of the population being continually connected across town. With the rate internet use is growing, this could easily happen in a few years.
If you are connecting 24/7/365, you should be paying for a line from you to the other end of the connection, because that is what you are using (aside from all the unnecessary switching hardware in between that you are also taking up). If that means paying ten times a normal second-line fee, so be it (after all, 24/7 is considerably more than ten times the normal usage). It's either the people creating the drain on the system who pay for it, or everybody else. Constant internet connections using telephone lines are a horrible abuse of a system meant for other things.
Direct connections may be more expensive right now, but only for the end user. When you consider all resources consumed, you realize that direct connections can be much cheaper, because they are designed for this purpose. When prices fairly reflect the resources used, the system will change to use the minimum resources. Everyone will start using cable or DSL, or some other dedicated service, the price will drop to or under that of a dedicated POT line, connections will be much faster and more reliable, and the situation will be better for everyone.
That whole thing about "once you've let it go, the move is made" is a silly rule anyway. Suppose you're moving a piece across the board and you twitch involuntarily and lose contact with it. Does it make any sense to force you to leave the piece out in the middle of the board?
Yes, technically, the move was made. That doesn't mean he intentionally cheated, and it doesn't mean he deserved to lose. He only lost touch with it for a split second, and who can say that it was intentional?
I think the move should be official when you hit the clock. Then it would be completely unambiguous.
It's pretty petty to complain over a technicality (after all, in a friendly game you'd let it pass unless you're a complete prick), but what do you expect from people whose lives center around a board game?
I never thought it was processor features that were preventing color screens in palmtops, it's the price and power consumption of color LCDs. It still needs some external logic to support color LCD anyway.
I just don't see any reason to get excited about this.
That was because they were blowing up useless status symbols in defiance of the decadence of a consumer-based culture.
Don't you remember what the man said? "Why do you know what a GUI is? Is it something that's essential to your survival? In ten years you'll be standing here, in leather clothes that'll last you your whole life, watching people typing cryptic commands on text terminals, using computers that don't come in your choice of five fruity colors. You are not special, do not think different, all your bits are part of the same core dump as the rest of us."
When was the last time your XFree86 crashed? When was the last time your computer crashed, and you blamed it on Linux?
I've had Gnome and Enlightenment crash on me plenty of times (hmm... beta software and beta software, what a great choice for the default setup on RedHat), so I switched to wmx and now the only thing that crashes is Netscape (and my own stuff while under development).
Yeah, you need a commercial quality X server about as much as you need a commercial quality OS.
I came up with one of those, too, after four sleepless days of calculus cramming followed by watching Dune. By the end of it I was eating instant coffee crystals.
It is by will alone I set my mind in motion. Through the drug caffeine the thoughts acquire speed, the teeth acquire stains, the stains become a warning. It is by will alone I set my mind in motion.
My parents just bought a nice little computer for $650 (yeah, Cdn) at the local CompuSmart (I'm sure any halfway decent chain has similar offers). That's with win98 and the Corel suite, a modem, ethernet capabilities, and all the other standard trimmings. Only 32 megs of RAM and a 4.3 GB HD and lame hardware 3D, but it's still an awesome machine for anything but the newest games or video creation/editing. They could have a decent monitor and printer for under the $1000 mark (if they didn't already have those from the old computer).
If they'd sell these things without the commercial software, the price could easily be under $600, and you'd have a great little Linux box. Even for the current price I'd like to buy a couple for testing my network code when I can afford it.
They haven't stolen your labor. If you put it under GPL you gave it away and made no requirement that derivatives be returned to you. People are entirely within their rights to modify GPL software for personal or in-house use and never let anyone else have it. So don't talk about being "payed back" by getting the use of their work. At any rate, whatever improvements a commercial software house could produce, they would not produce them if they had to release them under GPL anyway. As I said, it is spiteful, because those who do not want to produce free software will not decide to do so just because there is some good GPL code they want to use, rather they will simply be forced to recreate the code or skip the project entirely.
Proprietary and commercial are one and the same for all intents and purposes. You can't sell non-proprietary software, you can only support it, and others have the exact same freedom to do so.
Proprietary software is not inherently inferior to free software. Most free software took enormously more resources to produce, and most of the good stuff has its roots in proprietary software (through cloning or whatever); just because nobody is summing up the paid time and overhead for all those hackers doesn't mean that those resources weren't used.
When I release free software, I release it into the public domain. I do so because I can't make money from it and I want other people to benefit from it, in whatever way they can. Why should I screw the people who do think they can make money by using part of it? It's a material world, and my fellow programmers have to eat too.
You are forced to put the GPL on your work if your work is upgrades, bugfixes, or other derivatives of GPL'd stuff.
The point is that the originator that puts the work under GPL only did the original work, which he allows to be compiled and run freely, then he grabs everybody else's later work on it, no matter what they wanted for it.
The net effect is not that people with commercial interests in their software release the source, but that they keep their hands off GPL code entirely.
With open source, it is entirely possible for someone with a sufficient commercial interest to recreate any code under GPL (by reading the source and then duplicating the functionality exactly). Forcing them to do so is spiteful, it hurts them without benefiting you.
Anyway, I don't really see anything wrong with someone deciding to GPL their own work, when they make an informed decision to do so. Most people, though, talk about it as if GPL code is "more free" than public domain, which is dead flat wrong. People release their code under the GPL just on reflex, not because they really want the extra restrictions. I just don't think anyone considers public domain any more, when it is the best choice in many cases.
If you want the benefits of something like TGPL, you can release your stuff into the public domain and get even more out of it.
Public domain source can be re-released under any licence (including GPL; from there subsequent mods of the GPL version are GPL, but you could make the same mods to the public domain source if you wanted to and re-release under GPL to put the mod in the public domain) or used in any proprietary projects.
TGPL, OTOH, can only co-exist with GPL, among the open-source licences, because you can't release your source under another licence, even though you could hide it away in a binary.
RMS would hate TGPL because the only reason he created GPL is to prevent FSF code from being used in proprietary software. It does nothing but force all modifications and uses of GPL code to be openly released under GPL. To me, this is spiteful and authoritarian-communist; nobody can "steal" public domain code, they are only free to use it and retain all rights to their own mods and derivatives (whereas the GPL grabs the work of others). Heaven forbid somebody should want to make a buck off his own work!
Actually, I suspect this is rather closely related to how the human brain works in some cases, but perhaps more efficient due to some shortcuts we can take. How could consciousness work without some form of broadcast? Somehow, thoughts are being moved around so different parts all over your brain are hearing the same things.
There is some very definite predefined structure to the brain between the gross and cellular anatomy. It isn't just a raw neural net that is trained by physical pleasure and pain. Somehow, consciousness, intent, recognition of success and failure are built in at the highest level and recognition of the same sound in different pitches and visual recognition independent of image location on the retina are built in at a low level.
I suspect if you could read and analyse all of the connections, it wouldn't look like one big mess of seemingly random connections, but a lot of small neural nets arranged and interconnected in hierarchies, sorting pipes, buses, and the like (which aren't especially neural mechanisms, they're probably better done in the familiar ways we've developed that suit silicon).
I agree about using more neural nets on the problem. I never meant to imply that it would be anything more than a single component in a larger system. At some point in the process, you need to recognize sounds, whatever you do with them later.
What I dislike is the way some people treat neural nets as a magic bullet, as if we only need to make a big enough neural net and it will solve any problem. I think only small neural nets really work well; beyond a few dozen neurons, an external structure is needed to get anything to work.
(IMHO, the most important thing for *-recognition programs to start doing is admitting that they didn't understand and asking for clarification; "best guess" is not a good strategy)
Some people are saying that you can't make a really big neural net efficiently (at least without specialized hardware), but I don't see why you couldn't have hundreds of seperate neural nets each reporting on whether one word was said.
A very tiny, very simple computer could handle the task of managing a few neural nets. You could make it out of a few thousand surface features on a chip, so you could pack thousands of these processors on a chip. For that matter, they probably don't need to be terribly fast, so you could make them like memory chips. Imagine a megabyte chip, but instead of 1024K dumb memory, with 1024 minimal neural processors, each with 512 bytes of RAM.
Broadcasting the incoming data is pretty simple, and I don't think the networking issues of one or two of these processors reporting every few seconds would be too severe.
Training wouldn't be all that hard, either. You need a few man-years of samples, but the training could be done in parallel. It would cost a few million dollars (unless there was a dedicated online effort, which is entirely possible), but not billions. Imagine going down to the mall and asking people if they would read a few hundred words for $20; no problem, just repeat it all over the place so it deals well with accents.
There has never been a task better suited to massive parallel processing.
Oh yeah, I suppose I have to say: hey, we can do it with a Beowulf cluster, |)00|)Z!
Part of my point was that, "supposed to" aside, the MS mouse's shape (along with most other ergonomic mice) is more uncomfortable whatever hand you use it with. Square edges are good, they let you nudge instead of grabbing.
Games are probably responsible for a lot of CTS and other RSI. In a work app, you can always stop and start at will, but when you play a game you can lose if you try to take a break. I'd be inclined to think that this is a much more common cause than people are willing to admit.
I'm right handed but I use my mouse with my left hand. I use a cheap square mouse with 3 buttons.
I don't see why you would need a special left-hand mouse. I only ever barely touch the mouse and I pick it up between the tips of my thumb and last two fingers when I have to, so the shape seems of very little importance compared to the weight of the mouse. It seems much less stressful to me to use my longer middle finger to click the left mouse button, so I never tried to configure it backwards.
If you ask me, lefties have it pretty sweet. Most keyboards have number pads and other junk where the mouse should be, so most right-handers have to either type at an awkward angle or use the mouse at an awkward angle.
I've never liked a mouse that cost over $20 (the store paid me to take my current mouse: they sold it for under the manufacturer's rebate). They have awkward shapes you have to wrap your hand around and are often heavier than the cheaper mice. They rarely have 3 buttons and often the buttons are asymmetrical, misshapen, or oddly textured. For fine work, such as drawing or 3d modelling, a mouse is a bad choice; pen tablets are much better.
Slashdot Mirror
is not "little more than a single molecule", it is a robot built to nanoscale precision. It has a computer on board (or it wouldn't be a robot), sensors, and tools built into it. It can perceive and respond to its environment.
I agree, however, that microscopic nanobots probably aren't going to be all that useful. They can't see the big picture, so they can't really know what's going on around them. They also can't have terribly powerful computers because they're just too small. Power sources will be problematic, as will communication, further limiting their uses.
IMHO, nanobots will show their greatest potential, not as isolated nanites, but in huge differentiated linked masses (closer to mammals than microbes). This will make them better-coordinated and a lot less likely to go berserk on a self-reproducing rampage (think self-expanding, not self-reproducing). They can still have nanoscale tools, they'll just know where they are and how what they're being used for fits into the larger task.
As anyone can obviously see, the old paradigms have failed. Make way for the future, make way for PISC!
That's right, the wave of the future is the Pathetic Instruction Set Computer.
(in all seriousness, if this stuff turns your crank, building a computer from standard TTLs is way cool)
While inbreeding can cause problems, often severe ones, it is not a death sentence. You can create an entire population from a single pair of siblings. Release a mating pair of rabbits on an island with no predators and lots of food, and come back in a few years; if you don't find rabbits, you probably won't find anything green either.
to-MAY-to to-MAH-to
Animals have given birth to implanted fetuses of other species. I can't remember the exact details of an example... I think there was a rare type of cat that another cat of a common species gave birth to...
Argghh! We've been through this.
Of course the mitochondrial DNA was from the host cell. They knew it would be and didn't really care. It's not a big thing. Mitochondria are mitochondria, they change tranportable blood fuel into usable cell fuel (I'm just not up to big words like glucose tonight). A mammoth with modern elephant (or cow, or pig, or sheep) mitochondria is a mammoth as far as I'm concerned.
(now that that's out of my system...)
The Dolly technique is crusty in other ways, but it should work well enough to get some hairy elephants walking around northern Asia. Well, not quite the Dolly technique... this requires something a little more complicated, but IMHO doable in a year or two with enough money (or ten years from now in somebody's back yard).
I'd agree with you on the DNA bit, but they've got a whole mammoth. That's one heck of a DNA sample! They should be able to patch up the cracks with that big a sample.
Whoops, I'm paying $40/month. It's -$10 for being a cable customer here too, but the cable modem rental is included in the price.
So you're saying the phone companies are guilty of not subsidizing ISDN so they can charge by the minute for local calls? Of course, though, they can't charge by the minute for local calls, and they will never get approval to do so. Any serious talk of a modem tax is referring to special charges for modem users, either directly or indirectly.
ISDN was probably poorly handled, but it was never that great a technology to begin with. It is complex and therefore costly to install (misconfigurations galore), requires that special lines be laid and special equipment installed, and not really that much better than POTS.
ISDN was never really viable as a standard feature in everyone's home. If the prices were high, it was because the telcos knew they would only make money this way. It had to be profitable on the few customers who would pay a premium for a modest performance increase. They aren't stupid; if they saw potential for installing these lines in most homes, they would have lowered the prices and made a fortune on volume.
Now that better technologies are emerging, and internet connections are quickly approaching a household standard, some phone companies are starting to do exactly what you describe, except with DSL.
Of course the phone companies are greedy. Any for-profit public corporation is required by law to seek maximum profit for its shareholders. That doesn't mean that they are acting maliciously.
I know I'm getting a pretty good deal with my cable access, but I see no reason why it can't continue much longer. Bandwidth and cable modems are going down in price, and the cable was already there anyway. It's a good deal for the huge improvement in performance, but not too cheap to pay for itself.
I think cable is going to keep growing, become profitable, and eventually lower in price. It'll also go in waves of slowing down and speeding up as more people join and the system is expanded (it really scales quite nicely, if not trivially; if usage increases to the point where it is a problem, the funding is there to do some rewiring to solve it).
Here in Canada, we have $50/month cable connections. The price is high enough still that they don't have to charge for installation. Everyone who has it raves about it, and people are signing up like mad. I wouldn't be surprised if it was already turning a healthy profit. I'm sure they'd love to compete with the POTS services, and I bet they'll offer limited bandwidth connections for $20 soon.
The prices on POT lines are based on the way they work without ISPs: occasional short uses, not long-term continual connections. You are on a time-share system, and you are insisting on your right to use it 100% of the time.
The phone system is not built to support half of the population being continually connected across town. With the rate internet use is growing, this could easily happen in a few years.
If you are connecting 24/7/365, you should be paying for a line from you to the other end of the connection, because that is what you are using (aside from all the unnecessary switching hardware in between that you are also taking up). If that means paying ten times a normal second-line fee, so be it (after all, 24/7 is considerably more than ten times the normal usage). It's either the people creating the drain on the system who pay for it, or everybody else. Constant internet connections using telephone lines are a horrible abuse of a system meant for other things.
Direct connections may be more expensive right now, but only for the end user. When you consider all resources consumed, you realize that direct connections can be much cheaper, because they are designed for this purpose. When prices fairly reflect the resources used, the system will change to use the minimum resources. Everyone will start using cable or DSL, or some other dedicated service, the price will drop to or under that of a dedicated POT line, connections will be much faster and more reliable, and the situation will be better for everyone.
Up with the modem tax!
That whole thing about "once you've let it go, the move is made" is a silly rule anyway. Suppose you're moving a piece across the board and you twitch involuntarily and lose contact with it. Does it make any sense to force you to leave the piece out in the middle of the board?
Yes, technically, the move was made. That doesn't mean he intentionally cheated, and it doesn't mean he deserved to lose. He only lost touch with it for a split second, and who can say that it was intentional?
I think the move should be official when you hit the clock. Then it would be completely unambiguous.
It's pretty petty to complain over a technicality (after all, in a friendly game you'd let it pass unless you're a complete prick), but what do you expect from people whose lives center around a board game?
I never thought it was processor features that were preventing color screens in palmtops, it's the price and power consumption of color LCDs. It still needs some external logic to support color LCD anyway.
I just don't see any reason to get excited about this.
That was because they were blowing up useless status symbols in defiance of the decadence of a consumer-based culture.
Don't you remember what the man said?
"Why do you know what a GUI is? Is it something that's essential to your survival? In ten years you'll be standing here, in leather clothes that'll last you your whole life, watching people typing cryptic commands on text terminals, using computers that don't come in your choice of five fruity colors. You are not special, do not think different, all your bits are part of the same core dump as the rest of us."
...it was something like that anyway.
Just in case you wonder who these people are, their web page is at http://www.xigraphics.com.
When was the last time your XFree86 crashed? When was the last time your computer crashed, and you blamed it on Linux?
I've had Gnome and Enlightenment crash on me plenty of times (hmm... beta software and beta software, what a great choice for the default setup on RedHat), so I switched to wmx and now the only thing that crashes is Netscape (and my own stuff while under development).
Yeah, you need a commercial quality X server about as much as you need a commercial quality OS.
I came up with one of those, too, after four sleepless days of calculus cramming followed by watching Dune. By the end of it I was eating instant coffee crystals.
It is by will alone I set my mind in motion.
Through the drug caffeine
the thoughts acquire speed,
the teeth acquire stains,
the stains become a warning.
It is by will alone I set my mind in motion.
My parents just bought a nice little computer for $650 (yeah, Cdn) at the local CompuSmart (I'm sure any halfway decent chain has similar offers). That's with win98 and the Corel suite, a modem, ethernet capabilities, and all the other standard trimmings. Only 32 megs of RAM and a 4.3 GB HD and lame hardware 3D, but it's still an awesome machine for anything but the newest games or video creation/editing. They could have a decent monitor and printer for under the $1000 mark (if they didn't already have those from the old computer).
If they'd sell these things without the commercial software, the price could easily be under $600, and you'd have a great little Linux box. Even for the current price I'd like to buy a couple for testing my network code when I can afford it.
They haven't stolen your labor. If you put it under GPL you gave it away and made no requirement that derivatives be returned to you. People are entirely within their rights to modify GPL software for personal or in-house use and never let anyone else have it. So don't talk about being "payed back" by getting the use of their work. At any rate, whatever improvements a commercial software house could produce, they would not produce them if they had to release them under GPL anyway. As I said, it is spiteful, because those who do not want to produce free software will not decide to do so just because there is some good GPL code they want to use, rather they will simply be forced to recreate the code or skip the project entirely.
Proprietary and commercial are one and the same for all intents and purposes. You can't sell non-proprietary software, you can only support it, and others have the exact same freedom to do so.
Proprietary software is not inherently inferior to free software. Most free software took enormously more resources to produce, and most of the good stuff has its roots in proprietary software (through cloning or whatever); just because nobody is summing up the paid time and overhead for all those hackers doesn't mean that those resources weren't used.
When I release free software, I release it into the public domain. I do so because I can't make money from it and I want other people to benefit from it, in whatever way they can. Why should I screw the people who do think they can make money by using part of it? It's a material world, and my fellow programmers have to eat too.
You are forced to put the GPL on your work if your work is upgrades, bugfixes, or other derivatives of GPL'd stuff.
The point is that the originator that puts the work under GPL only did the original work, which he allows to be compiled and run freely, then he grabs everybody else's later work on it, no matter what they wanted for it.
The net effect is not that people with commercial interests in their software release the source, but that they keep their hands off GPL code entirely.
With open source, it is entirely possible for someone with a sufficient commercial interest to recreate any code under GPL (by reading the source and then duplicating the functionality exactly). Forcing them to do so is spiteful, it hurts them without benefiting you.
Anyway, I don't really see anything wrong with someone deciding to GPL their own work, when they make an informed decision to do so. Most people, though, talk about it as if GPL code is "more free" than public domain, which is dead flat wrong. People release their code under the GPL just on reflex, not because they really want the extra restrictions. I just don't think anyone considers public domain any more, when it is the best choice in many cases.
If you want the benefits of something like TGPL, you can release your stuff into the public domain and get even more out of it.
Public domain source can be re-released under any licence (including GPL; from there subsequent mods of the GPL version are GPL, but you could make the same mods to the public domain source if you wanted to and re-release under GPL to put the mod in the public domain) or used in any proprietary projects.
TGPL, OTOH, can only co-exist with GPL, among the open-source licences, because you can't release your source under another licence, even though you could hide it away in a binary.
RMS would hate TGPL because the only reason he created GPL is to prevent FSF code from being used in proprietary software. It does nothing but force all modifications and uses of GPL code to be openly released under GPL. To me, this is spiteful and authoritarian-communist; nobody can "steal" public domain code, they are only free to use it and retain all rights to their own mods and derivatives (whereas the GPL grabs the work of others). Heaven forbid somebody should want to make a buck off his own work!
Actually, I suspect this is rather closely related to how the human brain works in some cases, but perhaps more efficient due to some shortcuts we can take. How could consciousness work without some form of broadcast? Somehow, thoughts are being moved around so different parts all over your brain are hearing the same things.
There is some very definite predefined structure to the brain between the gross and cellular anatomy. It isn't just a raw neural net that is trained by physical pleasure and pain. Somehow, consciousness, intent, recognition of success and failure are built in at the highest level and recognition of the same sound in different pitches and visual recognition independent of image location on the retina are built in at a low level.
I suspect if you could read and analyse all of the connections, it wouldn't look like one big mess of seemingly random connections, but a lot of small neural nets arranged and interconnected in hierarchies, sorting pipes, buses, and the like (which aren't especially neural mechanisms, they're probably better done in the familiar ways we've developed that suit silicon).
I agree about using more neural nets on the problem. I never meant to imply that it would be anything more than a single component in a larger system. At some point in the process, you need to recognize sounds, whatever you do with them later.
What I dislike is the way some people treat neural nets as a magic bullet, as if we only need to make a big enough neural net and it will solve any problem. I think only small neural nets really work well; beyond a few dozen neurons, an external structure is needed to get anything to work.
(IMHO, the most important thing for *-recognition programs to start doing is admitting that they didn't understand and asking for clarification; "best guess" is not a good strategy)
Use more neural nets.
Some people are saying that you can't make a really big neural net efficiently (at least without specialized hardware), but I don't see why you couldn't have hundreds of seperate neural nets each reporting on whether one word was said.
A very tiny, very simple computer could handle the task of managing a few neural nets. You could make it out of a few thousand surface features on a chip, so you could pack thousands of these processors on a chip. For that matter, they probably don't need to be terribly fast, so you could make them like memory chips. Imagine a megabyte chip, but instead of 1024K dumb memory, with 1024 minimal neural processors, each with 512 bytes of RAM.
Broadcasting the incoming data is pretty simple, and I don't think the networking issues of one or two of these processors reporting every few seconds would be too severe.
Training wouldn't be all that hard, either. You need a few man-years of samples, but the training could be done in parallel. It would cost a few million dollars (unless there was a dedicated online effort, which is entirely possible), but not billions. Imagine going down to the mall and asking people if they would read a few hundred words for $20; no problem, just repeat it all over the place so it deals well with accents.
There has never been a task better suited to massive parallel processing.
Oh yeah, I suppose I have to say: hey, we can do it with a Beowulf cluster, |)00|)Z!
Part of my point was that, "supposed to" aside, the MS mouse's shape (along with most other ergonomic mice) is more uncomfortable whatever hand you use it with. Square edges are good, they let you nudge instead of grabbing.
Games are probably responsible for a lot of CTS and other RSI. In a work app, you can always stop and start at will, but when you play a game you can lose if you try to take a break. I'd be inclined to think that this is a much more common cause than people are willing to admit.
I'm right handed but I use my mouse with my left hand. I use a cheap square mouse with 3 buttons.
I don't see why you would need a special left-hand mouse. I only ever barely touch the mouse and I pick it up between the tips of my thumb and last two fingers when I have to, so the shape seems of very little importance compared to the weight of the mouse. It seems much less stressful to me to use my longer middle finger to click the left mouse button, so I never tried to configure it backwards.
If you ask me, lefties have it pretty sweet. Most keyboards have number pads and other junk where the mouse should be, so most right-handers have to either type at an awkward angle or use the mouse at an awkward angle.
I've never liked a mouse that cost over $20 (the store paid me to take my current mouse: they sold it for under the manufacturer's rebate). They have awkward shapes you have to wrap your hand around and are often heavier than the cheaper mice. They rarely have 3 buttons and often the buttons are asymmetrical, misshapen, or oddly textured. For fine work, such as drawing or 3d modelling, a mouse is a bad choice; pen tablets are much better.
I guess we're just arguing semantics.