Dude...I said "short-term". Same "short" as in "short-sighted", as in the decisions I was objecting to ignore the same factors of logic you're quite aware of.
Perhaps the W3C's HTML Validator or something similar? Rate the page based on conformance to the HTML specs (say, number of errors divided by length of HTML), in the hopes that this has some correlation to how generally useful the page is (i.e., if they can't be bothered to follow the technical rules, they probably don't have enough of a clue to put out content of genuine use to their users instead of just brochureware or scams or the like)? This wouldn't be perfect, of course, and utility is very much a subjective measure...
To be fair, there aren't many weapons in orbit right now. Ultra-classified test projects, if any, not counting spy satellites (which can't actually hurt anything themselves; they just take good pictures) and unintentional (i.e., unguided) re-entering satellites.
The administration's chief concern is not the welfare of the American people, but the short-term welfare of certain moneyed interests (including themselves). At least, that's the way the evidence points so far. And certain rich aerospace interests profit highly from having space be exclusive, since they're paid however much they can justify it costing plus a guaranteed profit, usually proportional to the cost. Reality and long-term survival of the human race are utterly dismissed in this logic.
Reduce cost to get stuff into space. $10k/lb. to LEO makes everything else too expensive. $100/lb. is apparently achievable even before you get many users (and being reached for by many private efforts today).
Set up some industry in orbit, or on the Moon, to pay for this. Space tourism, mining, automated construction (of solar power satellites, telescopes, and whatever else people will pay you to build up there)...even with what little we know, there are already potentials for business. More opportunities will likely pop up in the course of setting these up, but the venture capitalists want established opportunities - and those do exist.
Set up manned habitation. There are people who will gladly pay to, within limits, be research subjects for the long-term effects of space on the average (non-elite-astronaut) human body, especially if you use the results to build things that minimize these effects (for instance, start out with spin-induced artificial gravity for most of the habitat); if you add more devices later, you can then ask for more people to "test" the improvements. Eventually, expand this habitat to be self-sufficient.
...and there's your humanity in space. Sit back, and let the distributed (if limited in certain ways) intelligence of the masses work its magic.
Well, I can certainly imagine an experiment where two or more people had implants connected to their nervous systems that could send and receive signals over this new bandwidth - even if you could only have "on" and "off", you could still use Morse code or something - but I don't see this being directly suggested by that entry...
Re:If you think popups are bad...
on
Think And Click
·
· Score: 1
Shock of this nature wears off eventually. It could get one, two...maybe a full handful or two of clicks. But, sooner rather than later, most people would come to expect and adapt to the sudden images; I doubt almost any unwilling subject would be trapped for more than a few minutes at the extreme.
Assuming, of course, you can get your artillery in range. They ran into a similar problem with the Predators: they found OBL, but could not get munitions close enough before the Predator had to leave. When they came back, he was gone.
I've seen some Web comics about that - but using spiders, not helicopters. Can't remember the URL just now, though.
Taking it seriously even though you said not to - the main problem is cost. Sure, you and I can build drones that meet these specs for less than $10,000 a pop, maybe even less than $1,000, but the military brass would never trust 'em unless they cost at least $1,000,000 per. They need to be documented (we could write up a page or two easily enough, if we kept the whole system simple) and thoroughly tested (which, again, we could probably do to whatever predetermined, published spec they cared to make available to the public), and...
Hmm. Gee. Defense spending increased to record highs, not-quite-formal declaration of an unlimited war against America's enemies, et cetera and so forth, and you have to ask whether Bush sincerely believes that our best immediate future is a more militarized one?
Well...each synapse firing is a couple floating point operations (add value of firing to target neuron, after multiplying by this synapse's current weight), plus one more FPO to update the synapse's weight upon firing, plus one floating point comparison (fire if total excitation >= set point) which can be mostly ignored for this calculation since O(synapses) >> O(neurons); call it 20 cycles per synapse. So, a 1 GHz CPU (10^9 cycles per second) could service 5*10^7 synapses per second. Given about 10^14 [1] synapses, this calls for 2*10^6 CPUs...and, with bulk discounts on that scale, you might well be able to score 1 GHz processors for $500 each, leading to a total outlay for CPUs only of $1*10^9.
As for the memory requirements...10^14 synapses * 10^10 (=~ 2^30) bits per synapse for 10-digit (slightly limited) floating point integers = 10^24 bits. Proper sequencing of the neurons can probably ensure that memory reads and writes for neurons that have been processed/are about to be processed do not significantly slow down the system (i.e., no more than one read/write happening to any given memory block at a time). A quick price check shows certain models of RAM (made for just-less-than-1 GHz Macs) retailing for about $30 for 10^20 bits, so memory costs should not be a factor relative to CPU costs.
That said, even at this cost scale, you'd probably still want to pay Intel. Their factories actually do cost more than even this, and for practical terms, they can pour some extra effort for free into creating massively parallel systems like this. Besides, you'd want to concentrate on writing the software. This would be a standard large scientific computer; you might be able to get a grant from some university if you could convince them this could model a human brain with any significant degree of accuracy. (To avoid ethical concerns: test it out modelling lower-density brains - start with insects', move up to amphibians, birds, mammals, primates...the ethics should be worked out long before you reach humans, or at least you can claim they should be.)
[1] Given the American vs. British definitions of "million", "billion", et cetera, it is generally much safer to use scientific notation when any degree of precision is desired for figures at or over 10^6.
If you give each neuron a hundred sixty-four bit numbers to model its total "state" (logically, not physically), then, today a billion dollars will buy you a computer that can handle and process the equivalent of anyone's brain.
I agree with the spirit of your post in general, but...all this for only one billion dollars? I was under the impression it cost quite a bit more, today. Numbers, please? (Number of neurons that can be emulated by a certain CPU * price for that CPU model, plus equivalents for RAM and so forth.)
DARPA doesn't have to fund it. Some of the grunts have been building their own 'bots. Though I would like to see an aerial combat contest, if only to have the infamous Predator as the house target^H^H^H^H^H^Hbot.
Great. So, when is anyone not affiliated with NASA - and who is actually working on ways to drive down launch costs, not just throwing lots of money at the problem without much results - going to be able to get their hands on it?
Never, you say? It's too expensive/complicated/restricted-for-national-secu rity
to let us ordinary folks get our hands on it? Why, in the IT industry I come from, son, we have a word for that: "vaporware".
It might improve your response time, if the signal travels faster from brain through reader to cable into computer, than from brain through neural system to hand to mouse to cable to computer.
The upgrade I'm looking forward to is when they can emulate the individual neurons in chips, then replace the neurons (one by one if necessary) and accelerate any discrete subsystems that have been fully replaced. Output to computer should then go a bit faster, not to mention better possibility of revival if my body gets shut down (and I don't mean just sleep).
What does it mean, exactly, to "think about" something? They trained the monkey to produce neural patterns that had some perceived correlation with a certain action, without actually invoking that action. The "mouse" was configured to go off of these neural patterns. What the monkey was actually thinking about, only the monkey actually knows.
Can't see why that was modded off-topic. You'd need nutrient support to keep steak cells alive, probably pretty similar to that which keep stem cells alive. The exact mix of nutrients to best optimize steak growth is not currently known by anyone, to my knowledge.
Turning grass? Heck, let's go one step further: turn solar power, water, and nutrients from the air and soil into meat. Skip the grass step.
True, the bovine methods are adequate, but we can do better. (Legs to acquire the grass and exercise? Production line to deliver the grass at a predictable rate; electrical impulses to stimulate the meat a lot more evenly. Teeth to grind the grass? A mill can do it without getting part of the ground product stuck in the grinding. Reproduction? Nah, we just want the meat to grow: a little genetic engineering, and the muscles can have cancerous growth rates, with bunches of 'em harvested and fixed for human consumption every so often.)
Yes, and that's the threat: victims can spend millions of dollars to defend themselves, or just pay $10K for a license. Granted, the lawsuit's end might involve the ex-patent holder being forced to pay their would-be victim's legal fees, but the victim is still out quite a lot in the mean time.
Field cuisine: take foods X and Y which are supposed to be stored separate but served together, eat half a mouthful of each, and mix in your mouth. I hear it works well for coffee (powdered grounds + hot water)...
What Mt_Honkey said in terms of specifics, but...if NASA thought about cost/return in general, don't you think they'd have invested more than a trifle into reducing the cost of routine space missions?
Slashdot Mirror
One would hope that this is "Stuff that matters", at least to him. ;)
Dude...I said "short-term". Same "short" as in "short-sighted", as in the decisions I was objecting to ignore the same factors of logic you're quite aware of.
Note that this is about the resale of bundled software, so it's not like EULAs are dead
...aww, foo. <sighs> So much for instant-reaction happiness.
Perhaps the W3C's HTML Validator or something similar? Rate the page based on conformance to the HTML specs (say, number of errors divided by length of HTML), in the hopes that this has some correlation to how generally useful the page is (i.e., if they can't be bothered to follow the technical rules, they probably don't have enough of a clue to put out content of genuine use to their users instead of just brochureware or scams or the like)? This wouldn't be perfect, of course, and utility is very much a subjective measure...
To be fair, there aren't many weapons in orbit right now. Ultra-classified test projects, if any, not counting spy satellites (which can't actually hurt anything themselves; they just take good pictures) and unintentional (i.e., unguided) re-entering satellites.
The administration's chief concern is not the welfare of the American people, but the short-term welfare of certain moneyed interests (including themselves). At least, that's the way the evidence points so far. And certain rich aerospace interests profit highly from having space be exclusive, since they're paid however much they can justify it costing plus a guaranteed profit, usually proportional to the cost. Reality and long-term survival of the human race are utterly dismissed in this logic.
Well, I can certainly imagine an experiment where two or more people had implants connected to their nervous systems that could send and receive signals over this new bandwidth - even if you could only have "on" and "off", you could still use Morse code or something - but I don't see this being directly suggested by that entry...
Shock of this nature wears off eventually. It could get one, two...maybe a full handful or two of clicks. But, sooner rather than later, most people would come to expect and adapt to the sudden images; I doubt almost any unwilling subject would be trapped for more than a few minutes at the extreme.
Assuming, of course, you can get your artillery in range. They ran into a similar problem with the Predators: they found OBL, but could not get munitions close enough before the Predator had to leave. When they came back, he was gone.
I've seen some Web comics about that - but using spiders, not helicopters. Can't remember the URL just now, though.
Taking it seriously even though you said not to - the main problem is cost. Sure, you and I can build drones that meet these specs for less than $10,000 a pop, maybe even less than $1,000, but the military brass would never trust 'em unless they cost at least $1,000,000 per. They need to be documented (we could write up a page or two easily enough, if we kept the whole system simple) and thoroughly tested (which, again, we could probably do to whatever predetermined, published spec they cared to make available to the public), and...
Hmm. Gee. Defense spending increased to record highs, not-quite-formal declaration of an unlimited war against America's enemies, et cetera and so forth, and you have to ask whether Bush sincerely believes that our best immediate future is a more militarized one?
Well...each synapse firing is a couple floating point operations (add value of firing to target neuron, after multiplying by this synapse's current weight), plus one more FPO to update the synapse's weight upon firing, plus one floating point comparison (fire if total excitation >= set point) which can be mostly ignored for this calculation since O(synapses) >> O(neurons); call it 20 cycles per synapse. So, a 1 GHz CPU (10^9 cycles per second) could service 5*10^7 synapses per second. Given about 10^14 [1] synapses, this calls for 2*10^6 CPUs...and, with bulk discounts on that scale, you might well be able to score 1 GHz processors for $500 each, leading to a total outlay for CPUs only of $1*10^9.
As for the memory requirements...10^14 synapses * 10^10 (=~ 2^30) bits per synapse for 10-digit (slightly limited) floating point integers = 10^24 bits. Proper sequencing of the neurons can probably ensure that memory reads and writes for neurons that have been processed/are about to be processed do not significantly slow down the system (i.e., no more than one read/write happening to any given memory block at a time). A quick price check shows certain models of RAM (made for just-less-than-1 GHz Macs) retailing for about $30 for 10^20 bits, so memory costs should not be a factor relative to CPU costs.
That said, even at this cost scale, you'd probably still want to pay Intel. Their factories actually do cost more than even this, and for practical terms, they can pour some extra effort for free into creating massively parallel systems like this. Besides, you'd want to concentrate on writing the software. This would be a standard large scientific computer; you might be able to get a grant from some university if you could convince them this could model a human brain with any significant degree of accuracy. (To avoid ethical concerns: test it out modelling lower-density brains - start with insects', move up to amphibians, birds, mammals, primates...the ethics should be worked out long before you reach humans, or at least you can claim they should be.)
[1] Given the American vs. British definitions of "million", "billion", et cetera, it is generally much safer to use scientific notation when any degree of precision is desired for figures at or over 10^6.
Which other facilities shall they use, then? Yes, there are some that are building their own, but they're not ready yet...are they?
If you give each neuron a hundred sixty-four bit numbers to model its total "state" (logically, not physically), then, today a billion dollars will buy you a computer that can handle and process the equivalent of anyone's brain.
I agree with the spirit of your post in general, but...all this for only one billion dollars? I was under the impression it cost quite a bit more, today. Numbers, please? (Number of neurons that can be emulated by a certain CPU * price for that CPU model, plus equivalents for RAM and so forth.)
The concept's been around science fiction for a while. I'd like to see it made real.
"The Imperial Stormtrooper Marksmanship Academy" has been referenced in so many places, for its outstanding lack of quality... ;)
DARPA doesn't have to fund it. Some of the grunts have been building their own 'bots. Though I would like to see an aerial combat contest, if only to have the infamous Predator as the house target^H^H^H^H^H^Hbot.
Great. So, when is anyone not affiliated with NASA - and who is actually working on ways to drive down launch costs, not just throwing lots of money at the problem without much results - going to be able to get their hands on it?
u rity
Never, you say? It's too expensive/complicated/restricted-for-national-sec
to let us ordinary folks get our hands on it? Why, in the IT industry I come from, son, we have a word for that: "vaporware".
It might improve your response time, if the signal travels faster from brain through reader to cable into computer, than from brain through neural system to hand to mouse to cable to computer.
The upgrade I'm looking forward to is when they can emulate the individual neurons in chips, then replace the neurons (one by one if necessary) and accelerate any discrete subsystems that have been fully replaced. Output to computer should then go a bit faster, not to mention better possibility of revival if my body gets shut down (and I don't mean just sleep).
What does it mean, exactly, to "think about" something? They trained the monkey to produce neural patterns that had some perceived correlation with a certain action, without actually invoking that action. The "mouse" was configured to go off of these neural patterns. What the monkey was actually thinking about, only the monkey actually knows.
Can't see why that was modded off-topic. You'd need nutrient support to keep steak cells alive, probably pretty similar to that which keep stem cells alive. The exact mix of nutrients to best optimize steak growth is not currently known by anyone, to my knowledge.
Turning grass? Heck, let's go one step further: turn solar power, water, and nutrients from the air and soil into meat. Skip the grass step.
True, the bovine methods are adequate, but we can do better. (Legs to acquire the grass and exercise? Production line to deliver the grass at a predictable rate; electrical impulses to stimulate the meat a lot more evenly. Teeth to grind the grass? A mill can do it without getting part of the ground product stuck in the grinding. Reproduction? Nah, we just want the meat to grow: a little genetic engineering, and the muscles can have cancerous growth rates, with bunches of 'em harvested and fixed for human consumption every so often.)
Yes, and that's the threat: victims can spend millions of dollars to defend themselves, or just pay $10K for a license. Granted, the lawsuit's end might involve the ex-patent holder being forced to pay their would-be victim's legal fees, but the victim is still out quite a lot in the mean time.
Field cuisine: take foods X and Y which are supposed to be stored separate but served together, eat half a mouthful of each, and mix in your mouth. I hear it works well for coffee (powdered grounds + hot water)...
What Mt_Honkey said in terms of specifics, but...if NASA thought about cost/return in general, don't you think they'd have invested more than a trifle into reducing the cost of routine space missions?