Oh, but that was Muslims, you see. Everyone knows they're ignorant and backward by nature. Completely different from our situation. No comparison at all.
As a Christian and an Engineer who loves both God and science (yes, we exist - shock, horror), guys like this sadden and infuriate me. It may surprise some of the readership of/. but there are a lot of us who can actually both be rational and have faith, and who would deliberately vote against someone like this being on a science committee. I am starting to feel a little in the minority though: it's extremists in every direction.
Oh, stop whining. We know perfectly well that there are religious people who are capable of rational thought. We also know--and this is the point to which you seem to be wilfully blinding yourself--that there are a whole lot of people who deliberately allow their religious beliefs to override whatever capability they have for rationality, and that their more rational co-believers are doing nothing to stop them. Before you start slinging mud at nonbelievers ("extremists in every direction," as though there were any equivalence here) you might want to recall a certain line in your own holy book about motes, beams, and eyes.
The idea that we live in a world, or a nation, which can in any meaningful way be described as "post-religion" is one of the funniest things I've ever heard.
Constitution allows government certain functions and everything else is forbidden. Whatever is not expressly allowed is not allowed and running a committee on space etc., it's not expressly allowed.
Article 1, Section 5: "Each House may determine the Rules of its Proceedings..." The formation of committees is entirely within the scope of this power.
Wow, it took 7 whole minutes for a fallacious slippery-slope comment to appear. It used to take a lot less; you must be slipping.
And 3 minutes after that, we get a post displaying the common fallacy of assuming slippery slope arguments are automatically fallacious. So congratulations, you're not slipping at all.
The 60-70% of Americans who are not right-wing nutballs have been making the critical mistake, for quite some time now, of believing that the 30-40% who are right-wing nutballs don't mean what they say, because nobody's really that crazy, right? I mean, maybe a few people, but not tens of millions of them, right? Right?
Except that yes, they are. And while they may be a minority in absolute terms, there are enough of them to constitute a majority of the Republican Party--which means that roughly half of the American political establishment is under the absolute control of these loons. Non-crazy Republicans, of whom there are still quite a few, have wilfully blinded themselves to this situation, and continue to vote to give the nutballs power. The only way to stop this is for the rest of America, the slim majority (hopefully) which is not in the thrall of either ideology or party loyalty, to recognize what's going on, unify against the nutballs and all who associate with them, and send them back to the fringes where they belong.
According to FedEx CIO Rob Carter, that need to analyze events in real time has resulted in an effort to âoeradicallyâ decompose monolithic vertical applications into sets of core granular services, which the company will then mash into any number of analytics applications. The ultimate goal: a matrix of IT services that functions with the speed and flexibility of a brain, freeing FedEx from a system dependent on files strewn across any number of databases kept on disk storage systems too slow to support advanced, real-time analytic applications.
Dear God, I think this man just achieved the Buzzword Singularity. If we can harness this power...
what you just said is an assumption, which is why you collect statistics: to find out if an assumption is valid or not.
It's an observation based on experience. On a case-by-case basis, yes, you gather statistics to test a particular hypothesis. On a more general level, you know that the significance of correlation is a pretty good guide to which hypotheses are worth testing and which should just be discarded--and that if you see a whole bunch of hypotheses that aren't significant presented together, then probably none of them are worth a whole lot of further effort.
The point I think you're missing is that simple phenomena are amenable to simple models, while more complicated phenomena require more complicated models. The propagation of light is, in and of itself, a simple thing, and the luminiferous ether model and the photon model which replaced it are pretty simple too--which is why it was possible to choose one over the other based on simple experiments.
The interaction of gene regulation within a living organism, on the other hand, is tremendously complicated, and any simple model will simply be too simple to make useful predictions. ("All models are wrong, but some are useful.") There are complicated processes that require complicated models in physics too, you know--ask an astrophysicist about galaxy formation sometime. And it's possible that some experiment (or more likely, observation, in the case of astrophysics) will someday prove that these things don't happen at all: that genes do not, in fact, express RNA that directly or indirectly (via translation into protein) regulate the expression of other genes, or that the things we think are galaxies are actually just random collections of stars that appear to be grouped together thanks to tricks of perspective, like constellations, and therefore are not required to "form" at all. Of course, neither one seems particularly likely at this point; we're about as sure that genes regulate other genes, and that galaxies are real aggregations of stars, as we are that light propagates from point to point.
The model-building is therefore in the details, and there a whole lot of those details, every single one of which is subject to revision and occasionally complete disposal. This is where the simple models, the type you seem to consider the only admissible ones, come in. If biology has an analogue to the modern model of the propagation of light, it's probably the translation of mRNA into protein in the ribosome. And believe me, there were a whole lot of competing models for this in the early days, each rigorously tested and repeatedly modified; the bad ones were discarded, and we're left now with the current model which seems to best explain all the evidence--but could still be thrown out tomorrow if someone devises an experiment like Michelson-Morley which shows that it's just completely wrong. In my work, I pretty much take it at face value, just as an astrophysicist studying galaxy formation takes the photon model at face value, because these aren't the problems we're trying to solve.
You could say that you are building a model in the physical science sense if you could look at your network and say: there must be a gene/gene group that acts here, here and here, because this is how the kinetics of the system are when stressed, and the network I built does not have the required dynamics.
Kind of, yeah. Specifically, my model attempts to account for certain kinds of perturbations in the network (genetic knockouts for key transcription factors and cofactors in the Drosophila embryo Hedgehog pathway in one instance, and artificially elevated and depressed levels of growth factors in the transition from fibroblasts to myofibroblasts in porcine and human hearts in the other, if you care) by choosing the topology and parameters which best fit the data. If the data change, so does the fitted model. This is pretty standard practice, and I don't see the problem with it, or why you think it's different from what scientists in other fields do.
Now maybe this is what you are doing, and in which case I apologise, but in my limited experience, such papers are few and far between, and are certainly not bio_med_ papers
"Biomed" is unfortunately a vague term that seems to mean pretty much whatever the person using it wants it to mean; if you're going to argue that models relating to human biology (a.k.a. "medicine") tend to be less well-formed than those relating to other kinds of biology, I tend to agree with you, but
The REAL peers are the folks doing work in the profession day in and day out. As a rule most peer reviews are conducted by people with a decidedly academic focus - the experts in the field are working day jobs that don't afford them time to participate in silly self congratulatory exercises.
And in most scientific fields, those folks are overwhelmingly to be found at academic institutions, and most of those who aren't in academia are in government. Corporate R&D is almost all "D" these days. There used to be a lot more research and publication, and peer review, by people outside academia--in light of your username, you might want to consider the history of Bell Labs, and how sad that history's been in recent years.
Notice how very few of the correlations that turn up are statistically significant? That should tell you something: a statistically significant correlation (and if you're looking at a bunch of possible correlations at once, your bar for significance should be pretty high) usually does mean there's some kind of causal relationship somewhere, whether it's A causing B, B causing A, or some unmeasured C causing both.
TFA does a pretty good job of explaining why. Here's something I'd like to add: no, correlation does not imply causation, in the strict mathematical meaning of "imply"; in mathematical parlance, "A implies B" means that if A is true, B will always be true as well, and of course "X is positively correlated with Y" does not mean "an increase in X causes an increase in Y." But there's another meaning of "imply," and, like the common confusion about the meaning of the word "theory" in creationist arguments, it causes a lot of problems.
In common usage, "imply" carries a lot of ambiguity with it. In fact, it's almost never used to connote mathematical certainty. If you ask me, "Did John say he stole my money?" and I reply, "He implied that he did," that is a very different response from "Yes, that's what he said." In this usage, "A implies B" means that A is something which increases our estimate of the probability of B; if A is true, we're more likely to believe that B is true as well than if we had no information about A at all.
And in this sense, yes indeed, correlation does imply causation, and if you don't understand this then you should probably stop pretending that you understand the English language. Furthermore, it makes perfect mathematical sense. If you have data on both A and B, then if you can show a positive correlation, the hypothesis of a causal relationship will be much, much stronger than if you can't. And if you show a negative correlation, then forget about it. In other words, while "correlation implies causastion" isn't true in mathematical terms, the converse statement, "causation implies correlation," is true. Correlation is necessary though not sufficient for establishing causation.
Perhaps most importantly, every author of every peer-reviewed paper published in a respectable journal knows this. Next time you read some pop-sci reporting on any study in any field, and are tempted on that basis to dismiss it with "Correlation isn't causation, don't those dumb scientists know that?"... stop. Think. Read the abstract. And if you want to discuss the results in any detail, read the paper, and understand the methodology. If it's paywalled, find a way to get access (I guarantee you that you can). And if you're unwilling to do this, then you should probably just keep quiet, because you do not know what you need to know to form an informed opinion.
BTW, the link in the summary goes to the second page of a two-page article. Here is a link to the single-page version.
Thanks for providing a fine example of the short-term thinking that's endemic to the private sector. This is exactly why governments can, and do, accomplish useful things that the private sector can't. Or do you really believe that everything important can be done in less than five years?
Ah, the p-value less than or equal to alpha=0.05, the Holy Grail of Statistics 101 students and the bane of working statisticians.
It's not a bad standard for most uses, because "a one in twenty chance that you're wrong" corresponds pretty well to most people's idea of acceptable risk most of the time. But there's nothing magical about it, and in applications relating to human safety, it's quite common to use a more conservative standard, e.g. alpha=0.01 for safety and alpha=0.05 for efficacy in clinical trials. IOW, we're willing to accept a higher risk that the drug in question is essentially a placebo than that the drug will actually kill patients. A similar standard applies, seems to me, when it comes to turning autonomous and potentially lethal machines loose without human control.
I don't actually have a big problem with the waiting period, although I prefer the 6-month policy of the other agencies to the 12-month policy of NIH. It's not true that year-old publications are "hardly of use to scientists"; e.g., looking at the reference list on a paper I submitted just last month, I find that out of 38 papers referred to, only 6 were published in 2011 or 2012, and half of those were published in OA journals. YMMV, of course, but I don't consider this an undue burden. I admit that I may be biased by being in academia, and thus having access to the newest stuff if I need it.
Ideally I'd like to see all journals run like JMLR, or like the BE Press journals were before they were sold to de Gruyter, but we're not there yet, and I'm not at all sanguine that this approach will move us in that direction. We'll see.
... it strikes me as unnecessarily playing ball with the journal publishers. I'd rather see OA simply enforced by the funding agencies. In my field, bioinformatics, the bulk of the funding in the US and UK comes from four sources: NIH (US government), MRC (UK government), HHMI (US private foundation), and Wellcome Trust (UK private foundation). All of them have open access policies for publications prepared with their money, and they don't much care how you do it: you can post the article in a public repository regardless of how it was published, publish in an OA journal--for which the funding agency will generally pay the publication fee--or publish in a traditional journal and make sure the publisher makes a copy freely available. You can bet the publishers grumble about that last one, but they've mostly gone along with the requirement, because the alternative is saying "we won't publish papers describing research funded by ___," and if they did that they'd cease to exist.
If someone hasn't written a "generate a thought-free anti-Facebook screed from randomly strung-together cliches" page, they should. It would save so much time!
Nope, sorry, that doesn't make a conveniently dehumanizing enough sound bite to repeat ad nauseum whenever the poster needs a quick jolt of smug self-importance for not using Facebook/Google/etc. You'll have to do better than that.
This. "You are not the customer, you are the product" may have been a really insightful observation once upon a time. Now it's become just another sound bite.
Slashdot Mirror
Just because it is probabilistic does not mean it is not a model
There seem to be a large number of /.ers who will never, ever believe this. I'm not sure why.
Oh, but that was Muslims, you see. Everyone knows they're ignorant and backward by nature. Completely different from our situation. No comparison at all.
As a Christian and an Engineer who loves both God and science (yes, we exist - shock, horror), guys like this sadden and infuriate me. It may surprise some of the readership of /. but there are a lot of us who can actually both be rational and have faith, and who would deliberately vote against someone like this being on a science committee. I am starting to feel a little in the minority though: it's extremists in every direction.
Oh, stop whining. We know perfectly well that there are religious people who are capable of rational thought. We also know--and this is the point to which you seem to be wilfully blinding yourself--that there are a whole lot of people who deliberately allow their religious beliefs to override whatever capability they have for rationality, and that their more rational co-believers are doing nothing to stop them. Before you start slinging mud at nonbelievers ("extremists in every direction," as though there were any equivalence here) you might want to recall a certain line in your own holy book about motes, beams, and eyes.
Post-religion, we have ...
The idea that we live in a world, or a nation, which can in any meaningful way be described as "post-religion" is one of the funniest things I've ever heard.
Constitution allows government certain functions and everything else is forbidden. Whatever is not expressly allowed is not allowed and running a committee on space etc., it's not expressly allowed.
Article 1, Section 5: "Each House may determine the Rules of its Proceedings ..." The formation of committees is entirely within the scope of this power.
Wow, it took 7 whole minutes for a fallacious slippery-slope comment to appear. It used to take a lot less; you must be slipping.
And 3 minutes after that, we get a post displaying the common fallacy of assuming slippery slope arguments are automatically fallacious. So congratulations, you're not slipping at all.
The 60-70% of Americans who are not right-wing nutballs have been making the critical mistake, for quite some time now, of believing that the 30-40% who are right-wing nutballs don't mean what they say, because nobody's really that crazy, right? I mean, maybe a few people, but not tens of millions of them, right? Right?
Except that yes, they are. And while they may be a minority in absolute terms, there are enough of them to constitute a majority of the Republican Party--which means that roughly half of the American political establishment is under the absolute control of these loons. Non-crazy Republicans, of whom there are still quite a few, have wilfully blinded themselves to this situation, and continue to vote to give the nutballs power. The only way to stop this is for the rest of America, the slim majority (hopefully) which is not in the thrall of either ideology or party loyalty, to recognize what's going on, unify against the nutballs and all who associate with them, and send them back to the fringes where they belong.
According to FedEx CIO Rob Carter, that need to analyze events in real time has resulted in an effort to âoeradicallyâ decompose monolithic vertical applications into sets of core granular services, which the company will then mash into any number of analytics applications. The ultimate goal: a matrix of IT services that functions with the speed and flexibility of a brain, freeing FedEx from a system dependent on files strewn across any number of databases kept on disk storage systems too slow to support advanced, real-time analytic applications.
Dear God, I think this man just achieved the Buzzword Singularity. If we can harness this power ...
"I'll probably get modded down for this, but ..." (comment goes to +5 immediately)
Research Triangle Park, at a guess, since it's in North Carolina.
what you just said is an assumption, which is why you collect statistics: to find out if an assumption is valid or not.
It's an observation based on experience. On a case-by-case basis, yes, you gather statistics to test a particular hypothesis. On a more general level, you know that the significance of correlation is a pretty good guide to which hypotheses are worth testing and which should just be discarded--and that if you see a whole bunch of hypotheses that aren't significant presented together, then probably none of them are worth a whole lot of further effort.
The point I think you're missing is that simple phenomena are amenable to simple models, while more complicated phenomena require more complicated models. The propagation of light is, in and of itself, a simple thing, and the luminiferous ether model and the photon model which replaced it are pretty simple too--which is why it was possible to choose one over the other based on simple experiments.
The interaction of gene regulation within a living organism, on the other hand, is tremendously complicated, and any simple model will simply be too simple to make useful predictions. ("All models are wrong, but some are useful.") There are complicated processes that require complicated models in physics too, you know--ask an astrophysicist about galaxy formation sometime. And it's possible that some experiment (or more likely, observation, in the case of astrophysics) will someday prove that these things don't happen at all: that genes do not, in fact, express RNA that directly or indirectly (via translation into protein) regulate the expression of other genes, or that the things we think are galaxies are actually just random collections of stars that appear to be grouped together thanks to tricks of perspective, like constellations, and therefore are not required to "form" at all. Of course, neither one seems particularly likely at this point; we're about as sure that genes regulate other genes, and that galaxies are real aggregations of stars, as we are that light propagates from point to point.
The model-building is therefore in the details, and there a whole lot of those details, every single one of which is subject to revision and occasionally complete disposal. This is where the simple models, the type you seem to consider the only admissible ones, come in. If biology has an analogue to the modern model of the propagation of light, it's probably the translation of mRNA into protein in the ribosome. And believe me, there were a whole lot of competing models for this in the early days, each rigorously tested and repeatedly modified; the bad ones were discarded, and we're left now with the current model which seems to best explain all the evidence--but could still be thrown out tomorrow if someone devises an experiment like Michelson-Morley which shows that it's just completely wrong. In my work, I pretty much take it at face value, just as an astrophysicist studying galaxy formation takes the photon model at face value, because these aren't the problems we're trying to solve.
You could say that you are building a model in the physical science sense if you could look at your network and say: there must be a gene/gene group that acts here, here and here, because this is how the kinetics of the system are when stressed, and the network I built does not have the required dynamics.
Kind of, yeah. Specifically, my model attempts to account for certain kinds of perturbations in the network (genetic knockouts for key transcription factors and cofactors in the Drosophila embryo Hedgehog pathway in one instance, and artificially elevated and depressed levels of growth factors in the transition from fibroblasts to myofibroblasts in porcine and human hearts in the other, if you care) by choosing the topology and parameters which best fit the data. If the data change, so does the fitted model. This is pretty standard practice, and I don't see the problem with it, or why you think it's different from what scientists in other fields do.
Now maybe this is what you are doing, and in which case I apologise, but in my limited experience, such papers are few and far between, and are certainly not bio_med_ papers
"Biomed" is unfortunately a vague term that seems to mean pretty much whatever the person using it wants it to mean; if you're going to argue that models relating to human biology (a.k.a. "medicine") tend to be less well-formed than those relating to other kinds of biology, I tend to agree with you, but
In biomed, there are not models to invalidate.
That statement is one of the stupidest things I've ever read on Slashdot ... which is pretty impressive.
Okay, I'm going back to building models of developmental gene regulation now. You're welcome.
The REAL peers are the folks doing work in the profession day in and day out. As a rule most peer reviews are conducted by people with a decidedly academic focus - the experts in the field are working day jobs that don't afford them time to participate in silly self congratulatory exercises.
And in most scientific fields, those folks are overwhelmingly to be found at academic institutions, and most of those who aren't in academia are in government. Corporate R&D is almost all "D" these days. There used to be a lot more research and publication, and peer review, by people outside academia--in light of your username, you might want to consider the history of Bell Labs, and how sad that history's been in recent years.
There's a difference between the actual probability and our estimate of that probability; statistical significance refers to the latter.
Notice how very few of the correlations that turn up are statistically significant? That should tell you something: a statistically significant correlation (and if you're looking at a bunch of possible correlations at once, your bar for significance should be pretty high) usually does mean there's some kind of causal relationship somewhere, whether it's A causing B, B causing A, or some unmeasured C causing both.
TFA does a pretty good job of explaining why. Here's something I'd like to add: no, correlation does not imply causation, in the strict mathematical meaning of "imply"; in mathematical parlance, "A implies B" means that if A is true, B will always be true as well, and of course "X is positively correlated with Y" does not mean "an increase in X causes an increase in Y." But there's another meaning of "imply," and, like the common confusion about the meaning of the word "theory" in creationist arguments, it causes a lot of problems.
In common usage, "imply" carries a lot of ambiguity with it. In fact, it's almost never used to connote mathematical certainty. If you ask me, "Did John say he stole my money?" and I reply, "He implied that he did," that is a very different response from "Yes, that's what he said." In this usage, "A implies B" means that A is something which increases our estimate of the probability of B; if A is true, we're more likely to believe that B is true as well than if we had no information about A at all.
And in this sense, yes indeed, correlation does imply causation, and if you don't understand this then you should probably stop pretending that you understand the English language. Furthermore, it makes perfect mathematical sense. If you have data on both A and B, then if you can show a positive correlation, the hypothesis of a causal relationship will be much, much stronger than if you can't. And if you show a negative correlation, then forget about it. In other words, while "correlation implies causastion" isn't true in mathematical terms, the converse statement, "causation implies correlation," is true. Correlation is necessary though not sufficient for establishing causation.
Perhaps most importantly, every author of every peer-reviewed paper published in a respectable journal knows this. Next time you read some pop-sci reporting on any study in any field, and are tempted on that basis to dismiss it with "Correlation isn't causation, don't those dumb scientists know that?" ... stop. Think. Read the abstract. And if you want to discuss the results in any detail, read the paper, and understand the methodology. If it's paywalled, find a way to get access (I guarantee you that you can). And if you're unwilling to do this, then you should probably just keep quiet, because you do not know what you need to know to form an informed opinion.
BTW, the link in the summary goes to the second page of a two-page article. Here is a link to the single-page version.
Thanks for providing a fine example of the short-term thinking that's endemic to the private sector. This is exactly why governments can, and do, accomplish useful things that the private sector can't. Or do you really believe that everything important can be done in less than five years?
Ah, the p-value less than or equal to alpha=0.05, the Holy Grail of Statistics 101 students and the bane of working statisticians.
It's not a bad standard for most uses, because "a one in twenty chance that you're wrong" corresponds pretty well to most people's idea of acceptable risk most of the time. But there's nothing magical about it, and in applications relating to human safety, it's quite common to use a more conservative standard, e.g. alpha=0.01 for safety and alpha=0.05 for efficacy in clinical trials. IOW, we're willing to accept a higher risk that the drug in question is essentially a placebo than that the drug will actually kill patients. A similar standard applies, seems to me, when it comes to turning autonomous and potentially lethal machines loose without human control.
I don't actually have a big problem with the waiting period, although I prefer the 6-month policy of the other agencies to the 12-month policy of NIH. It's not true that year-old publications are "hardly of use to scientists"; e.g., looking at the reference list on a paper I submitted just last month, I find that out of 38 papers referred to, only 6 were published in 2011 or 2012, and half of those were published in OA journals. YMMV, of course, but I don't consider this an undue burden. I admit that I may be biased by being in academia, and thus having access to the newest stuff if I need it.
Ideally I'd like to see all journals run like JMLR, or like the BE Press journals were before they were sold to de Gruyter, but we're not there yet, and I'm not at all sanguine that this approach will move us in that direction. We'll see.
... it strikes me as unnecessarily playing ball with the journal publishers. I'd rather see OA simply enforced by the funding agencies. In my field, bioinformatics, the bulk of the funding in the US and UK comes from four sources: NIH (US government), MRC (UK government), HHMI (US private foundation), and Wellcome Trust (UK private foundation). All of them have open access policies for publications prepared with their money, and they don't much care how you do it: you can post the article in a public repository regardless of how it was published, publish in an OA journal--for which the funding agency will generally pay the publication fee--or publish in a traditional journal and make sure the publisher makes a copy freely available. You can bet the publishers grumble about that last one, but they've mostly gone along with the requirement, because the alternative is saying "we won't publish papers describing research funded by ___," and if they did that they'd cease to exist.
Then why do half of all scientists consider themselves Christian?
At least if you're talking about in the US, I think you're overstating the number quite a bit. But even if it's true, considerably more than half of the US population as a whole identifies as Christian, so there's still a pretty strong negative correlation between being a scientist and being a Christian, or religious in any form.
Hawking does cosmology with one eyeball, and you are stuck on one-handed gaming? Try ping pong, or darts.
Tell you what, how about we chop off one of your arms--you even get to choose which one--and we'll see if you maintain that attitude.
If someone hasn't written a "generate a thought-free anti-Facebook screed from randomly strung-together cliches" page, they should. It would save so much time!
Nope, sorry, that doesn't make a conveniently dehumanizing enough sound bite to repeat ad nauseum whenever the poster needs a quick jolt of smug self-importance for not using Facebook/Google/etc. You'll have to do better than that.
This. "You are not the customer, you are the product" may have been a really insightful observation once upon a time. Now it's become just another sound bite.