Yeah right... The Feds are there to get a piece of the action. "Oh sure we'll make sure this system is much more closely controlled by applying our top-secret patch. Nothing to see here. Move along.":)
Having the theoretical ability is one thing, but to actually make use of it is worse...HOW MUCH OF A FREAKIN' BONEHEAD DO YOU HAVE TO BE TO CONFRONT THEM WITH EVIDENCE OBTAINED BY SUCH QUESTIONABLE MEANS?
Exactly.
It's easy to pose a shaky, but ultimately successful, argument for installing remote-activated cameras in the laptops. Let's see: anti-theft, child welfare investigations (since abuse reports often come through the school), think of the children(!), etc. But the dynamic duo of principal and sysadmin can't foresee their own (or their colleagues) patent stupidity. This is the why seemingly great ideas, like the full-body scanners in airports, are actually awful: because they are great until the *inevitable* critical mass of stupid is reached. Nobody wants to hear that though, "Your idea is good in theory, but in reality some bonehead (possibly even you!) will abuse it and the cost will be greater than whatever benefits we gained along the way." So we live and don't learn.
Of course, I have to be a jerk about it, but your (probably joking) recommendation to shoot school admins who fail to learn the Bill of Rights is exactly one such idea. We can all get behind that idea and say "YEAH! SCREW THOSE JERKS! MAKE SURE THEY LEARN!!" And then we all forget to ask "wait, who is doing the grading?" Granted we aren't likely to implement that idea... but there are plenty of morons who would try!
FAIL.
I've been a 'professional radio operator', and you are confusing skill with procedure. We worked with numerous volunteer HAM operators with years of experience who basically showed us how it's done in the field. Yes, there are tons of agency-specific rules and procedures that you have to practice 40+ hours a week to have down. But when technology (inevitably) went sour, the HAM operators were the ones able to keep working.
I have seen Dr Strangelove, and Fail-Safe (the less comical version, arguably based on the same book). I think both of these films reinforce my point. The entire story revolves around the hours BEFORE detonation. To quote Dr S: "The whole point of a Doomsday Machine is lost, if you keep it a secret!" The corollary being that the entire point of a Doomsday machine is its existence, not its use. Consider how the outcome of Dr Strangelove would have been different if one or both sides unwittingly launched duds. The ideal Doomsday machine is 100% credible and 0% effective... but that's pretty well impossible.
Of course, maybe that was your point, in which case: Whoosh! to me.
The catch-22 of post-WWII nuclear warfare is that there is no such thing as launch without retaliation. If we find a rogue nation with a lone nuke or two, we attack with conventional weapons, because the risk incurred by escalation is too great. If a threat is substantial enough to warrant a nuclear attack (as the Soviet Union may have been), they are completely capable of retaliating while our birds are still in the air, what with early detection and all. That's where MAD (mutually assured destruction) comes in. LAUNCHING a nuclear weapon is what causes MAD... by the time of detonation, everyone's fate was sealed several minutes ago.
Maybe I'm being naive, but detonation never seemed all that central to the value of nuclear weapons. Let's face it, if we're ever in the situation where we decide Armageddon is the best option available, whether or not OUR weapons detonate is a triviality. Nuclear weapons are most effective when they AREN'T being used and everyone wants to keep it that way. So unless there's some a priori outward indication that our weapons definitely won't work, thus inviting an attack... nobody (including our enemies) really wants to find out the messy way. Then again, maybe I'm assuming too much rationality for the men with the launch keys...
To me, this translates into "we've made a big unspecialized neural network and we're watching the weights update as we try to classify corporate logos with it".
I think the hope is that this system will show some unique emergent properties that could not be observed in smaller models. If all they wanted to do was recognize logos, they could have done that simulation on a laptop. I haven't read the actual paper, but I'm sure the researchers used some architecture beyond "giant net" or the generalization results would have sucked (rule of thumb: the bigger the net, the weaker the generalization; the smaller the net, the greater the errors)
I should also point out that they are only simulating the cerebral cortex, which is the 'wrinkly' outer portion of the brain. There is a great deal more to the brain than the cerebral cortex, but we generally associate it with what makes us human. Humans have a uniquely large cerebrum compared to our mid-brains. The rest of the brain becomes increasingly important the farther you venture from Homo sapiens in taxonomy. It's becoming increasingly apparent that even the highest order human behaviors (like language) depend on sub-cortical organs, like the putamen. Therefore, while TFA is a great step for neural simulation... it's nothing like a robot cat.
This project is basically a massive neural network simulation with a number of nodes and connections comparable to the estimated totals in a cat's brain. In short, there is nothing cat-like about this system apart from its raw processing power.
Not to reduce the value of this feat, by any means! There are tons and tons of neural network simulations that can produce roughly human-like results in very, very narrow domains, but as the quote below explains, these simulations are decades (or more) from connecting the behavior of tiny subsystems (a few hundred neurons) with the overall phenomenon of 'mind' (conscious and unconscious cognition). The expectation is that a network of this size will show some new emergent properties that will give us clues about the intermediate "higher than cells, lower than interviewing a human" order of processing.
Jim Olds, a neuroscientist and director of the Krasnow Institute for Advanced Study at George Mason University, called the new research a "tremendous step." Olds, who was not involved in IBM's work, said neuroscientists have been amassing data about how the brain works much like "stamp collectors," without a way to tie it together.
"We've made tremendous advances in collecting data, but we don't have a collective theory yet for how this complex organ called the brain produces things like Shakespeare's sonnets and Mozart's symphonies," he said. "The holy grail for neuroscientists is to map activity from single nerve cells, which they know about, into how billions of nerve cells act in concert."
Or to elaborate a bit, I wonder if we're not neglecting a bigger problem in the other direction. It seems like we're constantly discovering greater degrees of mutualism between humans and the micro-organisms swarming all over (and through) our bodies. A common example is our digestive dependence on bacteria in the intestines, and the recently discovered role of the appendix in maintaining the intestinal culture [1,2].
While I'm not aware of any short term (longest stay in space 400-500 days) effects, what major biological functions might change over many years if the bacterial cultures on our skin (for instance) are weakened or eliminated in hyper-sterile or otherwise non-earth-like colony environments? I recall some speculation recently that bacterial by-products might play a role in altering our emotional states day to day. Imagine the unforeseen psychological effects when a currently unidentified bacteria suddenly vanishes from our bodily ecosystems due to habitat change...
I'm definitely not against manned space exploration or even colonization, but before we start bathing our astronauts in hand sanitizer, somebody needs to consider our physiological dependence on the bacterial ecosystem, not just our war against it.
r value alone does not tell one whether the relationship is statistically significant
You're right, and I should have made that point explicit. The Pearson r does not make any indication of significance. I was assuming that the reviewer was aware of the p-value and based his statement on that, but failed to report it. My purpose was to point out that converse, a low r value does not exclude significance either. There's another thread running to the same point.
You're definitely right about that. Since the review doesn't give a p-value, I took it on good faith that it was covered in the book... but then again, a reviewer who doesn't note the p-value probably didn't know what significant meant anyway... hrm...
I think you're confusing correlation (R) with coefficient of determination (R-squared). Only 10-30% (cautiously limiting to 1 sig fig) of the variation is accounted for, but this is still huge. See my other post.
suggest that about 10-30% (r-squared) or more of the variation
Correction: Cut the "or more". It was left over from a previous version of that comment... if you can believe that I actually read and revise my comments before posting! (although apparently not well enough)
TFS never claimed it was a strong correlation. It's a highly SIGNIFICANT correlation (meaning that the probability that the result occurred by chance and not systematically is very low, less than 5%).
Now, whether or not.33 is a STRONG correlation is another matter. By most definitions, it is not, although.52 would be a moderate correlation. However, the correlation does suggest that about 10-30% (r-squared) or more of the variation in subjects' decisions was accounted for by their social security numbers (accounted for != caused by, but we can make inferences based on the experimental design). Over a lifetime, 10% variation due to random irrelevant factors (like SS number) is serious, and 30% is HUGE. In that sense, it is a meaningful result, even if the correlation is not a "strong" one in terms of proportion.
If a truck is tail-gating you, there's also a good chance it's your fault. In areas with any substantial terrain, trucks have a lot more difficulty modulating their speed than your sedan/pickup. I've seen plenty of people think that having a small car made it OK to slip in a couple yards ahead of a rig's front fender. Those people shouldn't be surprised when the trucker is angrily riding their tail all the way down the hill. Good defensive driving means having a reasonable notion of what's going on in the cars around you, not just your own right-to-the-road (not suggesting that this is parent poster's attitude).
I began wondering about this same problem as several questions were distinctly medical (and sounded pretty urgent!). The good news is that QuestionBox is trying to recruit medical professionals to assist with these calls (can't find the link now, but it was on the QuestionBox website).
I'm sure there are plenty of $topic_of_interest geeks out there who would love to volunteer a bit of their time in this sort of capacity, and being telephone-based it's a highly distributable service. Of course, nothing beats a good research librarian in general background knowledge and ability to sift out the garbage.
Slashdot Mirror
You say that now, but you'll be glad Mothra's around when King Ghidorah arives.
Yeah right... The Feds are there to get a piece of the action. "Oh sure we'll make sure this system is much more closely controlled by applying our top-secret patch. Nothing to see here. Move along." :)
Exactly.
It's easy to pose a shaky, but ultimately successful, argument for installing remote-activated cameras in the laptops. Let's see: anti-theft, child welfare investigations (since abuse reports often come through the school), think of the children(!), etc. But the dynamic duo of principal and sysadmin can't foresee their own (or their colleagues) patent stupidity. This is the why seemingly great ideas, like the full-body scanners in airports, are actually awful: because they are great until the *inevitable* critical mass of stupid is reached. Nobody wants to hear that though, "Your idea is good in theory, but in reality some bonehead (possibly even you!) will abuse it and the cost will be greater than whatever benefits we gained along the way." So we live and don't learn.
Of course, I have to be a jerk about it, but your (probably joking) recommendation to shoot school admins who fail to learn the Bill of Rights is exactly one such idea. We can all get behind that idea and say "YEAH! SCREW THOSE JERKS! MAKE SURE THEY LEARN!!" And then we all forget to ask "wait, who is doing the grading?" Granted we aren't likely to implement that idea... but there are plenty of morons who would try!
FAIL. I've been a 'professional radio operator', and you are confusing skill with procedure. We worked with numerous volunteer HAM operators with years of experience who basically showed us how it's done in the field. Yes, there are tons of agency-specific rules and procedures that you have to practice 40+ hours a week to have down. But when technology (inevitably) went sour, the HAM operators were the ones able to keep working.
Of course, maybe that was your point, in which case: Whoosh! to me.
The catch-22 of post-WWII nuclear warfare is that there is no such thing as launch without retaliation. If we find a rogue nation with a lone nuke or two, we attack with conventional weapons, because the risk incurred by escalation is too great. If a threat is substantial enough to warrant a nuclear attack (as the Soviet Union may have been), they are completely capable of retaliating while our birds are still in the air, what with early detection and all. That's where MAD (mutually assured destruction) comes in. LAUNCHING a nuclear weapon is what causes MAD... by the time of detonation, everyone's fate was sealed several minutes ago.
Maybe I'm being naive, but detonation never seemed all that central to the value of nuclear weapons. Let's face it, if we're ever in the situation where we decide Armageddon is the best option available, whether or not OUR weapons detonate is a triviality. Nuclear weapons are most effective when they AREN'T being used and everyone wants to keep it that way. So unless there's some a priori outward indication that our weapons definitely won't work, thus inviting an attack... nobody (including our enemies) really wants to find out the messy way. Then again, maybe I'm assuming too much rationality for the men with the launch keys...
I think the hope is that this system will show some unique emergent properties that could not be observed in smaller models. If all they wanted to do was recognize logos, they could have done that simulation on a laptop. I haven't read the actual paper, but I'm sure the researchers used some architecture beyond "giant net" or the generalization results would have sucked (rule of thumb: the bigger the net, the weaker the generalization; the smaller the net, the greater the errors)
I should also point out that they are only simulating the cerebral cortex, which is the 'wrinkly' outer portion of the brain. There is a great deal more to the brain than the cerebral cortex, but we generally associate it with what makes us human. Humans have a uniquely large cerebrum compared to our mid-brains. The rest of the brain becomes increasingly important the farther you venture from Homo sapiens in taxonomy. It's becoming increasingly apparent that even the highest order human behaviors (like language) depend on sub-cortical organs, like the putamen. Therefore, while TFA is a great step for neural simulation... it's nothing like a robot cat.
This project is basically a massive neural network simulation with a number of nodes and connections comparable to the estimated totals in a cat's brain. In short, there is nothing cat-like about this system apart from its raw processing power.
Not to reduce the value of this feat, by any means! There are tons and tons of neural network simulations that can produce roughly human-like results in very, very narrow domains, but as the quote below explains, these simulations are decades (or more) from connecting the behavior of tiny subsystems (a few hundred neurons) with the overall phenomenon of 'mind' (conscious and unconscious cognition). The expectation is that a network of this size will show some new emergent properties that will give us clues about the intermediate "higher than cells, lower than interviewing a human" order of processing.
Easy. She's upstairs.
Yes, but we already have Fermilab in the States.
yes.
Or to elaborate a bit, I wonder if we're not neglecting a bigger problem in the other direction. It seems like we're constantly discovering greater degrees of mutualism between humans and the micro-organisms swarming all over (and through) our bodies. A common example is our digestive dependence on bacteria in the intestines, and the recently discovered role of the appendix in maintaining the intestinal culture [1,2].
While I'm not aware of any short term (longest stay in space 400-500 days) effects, what major biological functions might change over many years if the bacterial cultures on our skin (for instance) are weakened or eliminated in hyper-sterile or otherwise non-earth-like colony environments? I recall some speculation recently that bacterial by-products might play a role in altering our emotional states day to day. Imagine the unforeseen psychological effects when a currently unidentified bacteria suddenly vanishes from our bodily ecosystems due to habitat change...
I'm definitely not against manned space exploration or even colonization, but before we start bathing our astronauts in hand sanitizer, somebody needs to consider our physiological dependence on the bacterial ecosystem, not just our war against it.
[1]http://sciences.surgery.duke.edu/wysiwyg/downloads/BillSection1SecondInsert.pdf
[2]http://www.msnbc.msn.com/id/21153898/
AC, you've inspired me to a new sig, less inflammatory than it's predecessor. May the mod points flow.
You're right, and I should have made that point explicit. The Pearson r does not make any indication of significance. I was assuming that the reviewer was aware of the p-value and based his statement on that, but failed to report it. My purpose was to point out that converse, a low r value does not exclude significance either. There's another thread running to the same point.
You're definitely right about that. Since the review doesn't give a p-value, I took it on good faith that it was covered in the book... but then again, a reviewer who doesn't note the p-value probably didn't know what significant meant anyway... hrm...
[1] Previous post explaining the difference
I think you're confusing correlation (R) with coefficient of determination (R-squared). Only 10-30% (cautiously limiting to 1 sig fig) of the variation is accounted for, but this is still huge. See my other post.
Correction: Cut the "or more". It was left over from a previous version of that comment... if you can believe that I actually read and revise my comments before posting! (although apparently not well enough)
TFS never claimed it was a strong correlation. It's a highly SIGNIFICANT correlation (meaning that the probability that the result occurred by chance and not systematically is very low, less than 5%).
Now, whether or not .33 is a STRONG correlation is another matter. By most definitions, it is not, although .52 would be a moderate correlation. However, the correlation does suggest that about 10-30% (r-squared) or more of the variation in subjects' decisions was accounted for by their social security numbers (accounted for != caused by, but we can make inferences based on the experimental design). Over a lifetime, 10% variation due to random irrelevant factors (like SS number) is serious, and 30% is HUGE. In that sense, it is a meaningful result, even if the correlation is not a "strong" one in terms of proportion.
You mean fMRI. EEG has nothing to do with the salmon study.
That said, TFA cites fMRI also, so your comparison to the salmon study is plausible, if not terribly informative.
I think you forgot this.... right??
If a truck is tail-gating you, there's also a good chance it's your fault. In areas with any substantial terrain, trucks have a lot more difficulty modulating their speed than your sedan/pickup. I've seen plenty of people think that having a small car made it OK to slip in a couple yards ahead of a rig's front fender. Those people shouldn't be surprised when the trucker is angrily riding their tail all the way down the hill. Good defensive driving means having a reasonable notion of what's going on in the cars around you, not just your own right-to-the-road (not suggesting that this is parent poster's attitude).
Do not shoot off screen to reload!!!!!
I began wondering about this same problem as several questions were distinctly medical (and sounded pretty urgent!). The good news is that QuestionBox is trying to recruit medical professionals to assist with these calls (can't find the link now, but it was on the QuestionBox website).
I'm sure there are plenty of $topic_of_interest geeks out there who would love to volunteer a bit of their time in this sort of capacity, and being telephone-based it's a highly distributable service. Of course, nothing beats a good research librarian in general background knowledge and ability to sift out the garbage.
FWIW, QuestionBox appears to have internships, of what nature I don't know:
http://www.questionbox.org/you.html