They did study a relatively cosy community, which was predominantly white. They removed as many families and evidence of distantly-related friends as they could using the Manichaikul et al. 2010 kinship metric, which has been cited 34 times (a fairly good number of hits for a two-year interval.)
The upshot of this study, in my opinion, is that the things that make people compatible friends happen to be reflected (to a degree) in certain markers in the genome. It doesn't necessarily indicate anything about race or relatedness, and you can probably predict that two people of radically different ethnicities may like each other as friends because of similarity simply due to the subset of markers that are important. It's a bit misleading of the authors to imply that this is somehow a pan-genome phenomenon when obviously there are plenty of things in the genome that could change without any detectable impact.
Almost all of the participants in the dataset were Americans of European descent. They controlled for the possibility of distant relatedness as aggressively as they could, which is a well-understood requirement of many GWAS experiments.
To eliminate the possibility that the results are influenced by people tending to make friends with distant relatives, we use only the 907 friend pairs where kinship <=
0 (recall that
kinship can be less than zero whe
n unrelated individuals tend to have negatively correlated
genotypes). This procedure ensures that pairs of friends in the GWAS
are not actually
biologically related
at all.
Can either of them defeat Panopticlick? I don't see anything on Epic's site about hiding font lists. (And on that point, Epic is a bad name choice since it's vaguely synonymous with the death of objectivity in news reporting.)
You'd think this would be a great time to use that "read more" snip feature that's typical of reviews and interviews... or maybe this is a cue to start using Larry Drebes's signature everywhere?
At any rate, it's a little sad to see this OpenID provider going because it means less diversity in the single sign-on landscape, which is the whole point. At least OpenID itself will still be around!
...Fun fact: they're also homophones in a Japanese accent, so that may've been a double joke. A lot of Chinese and Japanese speakers come up with inventive ways to deal with certain non-nasal consonants at the ends of syllables, which can get fantastically distracting when they speak quickly. "Labo" (for "lab") is one I hear a lot; another is "hoteru" (the actual Japanese word for "hotel".)
Knowledge doesn't exist until you can prove it, and two mutually exclusive statements cannot both be true. I will respect your right to hold to your beliefs and to put an end to this conversation, but the decision to do so is no acceptance or validation of your position, which remain illogical. Your justification for why psychology is not a science depends on science having the same flaws that you accuse psychology of having.
It appears to me that you don't have a formal logic education, and that you probably haven't been exposed to a balanced survey of philosophical thought. You're a wonderfully articulate person, but the claims you've been making remain unjustified; at several points you say things that lead to the rejection of the very notion of an axiom, which means that the very act of existing makes you hypocritical. I would strongly encourage you to reach out and consider other perspectives.
That all being said, I wish you well, and I understand if you want to stop discussing this.
We can know the ways in which something behaves without understanding what it is. The unanswerable questions about the nature of the universe do not prevent us from recognizing the patterns of behaviour or the interactions of things that occur within it. This is the third time you've tried to argue from an incorrect generalization.
I'll get back to the rest of your post once we sort this disaster out:
The fact remains that human behavior [...] is not predictable.
The brain is not capable of producing any phenomena that violate the laws of physics, and to the best of our knowledge does not exploit any quantum phenomena that are beyond a computer's ability to simulate or represent with sufficient precision. Where is this profound unpredictability coming from?
So what, exactly, do you think an IQ score is, if not a measurement? Bear in mind that the uncertainty in a quiz is no different than the uncertainty in a physical measuring instrument; they are governed by the same family of statistical distributions. Just because we don't know exactly how IQ relates to pattern recognition ability doesn't make it irrelevant; science is laden with metrics that are meaningful and objective but not direct representations of the underlying phenomenon (for example, we only have a rough idea about how absorbed radiation dosage over time equates to health risks.)
You may want to read up on Behaviorism—widely disparaged as inhumane and simplistic, but completely empirical, reproducible, and appropriately well-funded by some governments. Along with forbidden experiments, it quickly becomes apparent that only ethical necessity forces psychology to lean on philosophical insight for understanding certain areas; wherever possible, experimental methodology is employed and venerated with ever-increasing rigour. The same can be said of any of the social sciences.
Many parts of psychology are rigorous and measurable, and these sub-problems within the vague umbrella of "intelligence" are about as solid as you can get, since they're all analytical skills. Even if the tests are indirect, they're still objectively meaningful.
I don't know why you've brought that up; at no point have I been trying to assert that psychology is founded upon evidence-based hypothesis testing. Philosophy has probed the human mind for thousands of years, and while most of its predictions have been wrong, it is rare if ever that something which is correct has been passed by except due to external interference. There are plenty of things that experimentalism is insufficient to study, most obviously mathematics. This does not make them invalid. Science is only one part of how we understand the world; to assume everything in life must be scientifically-founded is to reject the whole of human civilization and the thinkers who invented natural philosophy. Psychology must make do because it is illegal and unethical to perform proper experiments—and even your second link concedes: "Some research is far more scientifically rigorous. And the field often yields interesting and important insights."
More to the point, all of the domains I brought up, with the exception of emotional intelligence, are problems studied in cognitive science, the field at the juncture between AI, psychology, and neuroscience that seeks to build a bottom-up model of how the brain works. They are all either backed with rigorous, repeatable experiments, or are theoretically airtight, or both.
I have to disagree with you about suggesting those other things are subjective, though. They have objective definitions within psychology—definitions that comprehensive enough to include everything that you or I might consider part of intelligence. They're anything but vague.
To flesh out the examples I gave: emotional intelligence has a very specific meaning and can be tested for, pattern recognition is extremely well-understood and can easily be accomplished by a computer (and tested with the standard IQ test!), decision-making efficiency can be clearly measured in a variety of contexts and is simply the speed at which decisions are made under varying degrees of situational complexity, intuition is skill in recognizing and reasoning about the nature of a concept or phenomenon (and is part of the exam in any UI design course), and spatial reasoning is already a problem heavily described and being studied by AI researchers.
The issue with intelligence is that it's too broad and vague a concept to be defined, not that it's absolute but unknown (as with angstroms) or inherently subjective (as with how a colour is interpreted and represented in an individual's field of vision after retinal processing, i.e. long after the frequency of incoming photons stops being relevant.) They're not quite apt metaphors, hence the confusion, but I understand your point and apologise for the cumbersomeness of the conversation thus far.
I'd actually argue that we should throw out the idea of intelligence altogether and just discuss the elements of it that we actually intend to indicate—pattern recognition, efficiency in decision-making, intuition, spatial reasoning ability, emotional self-awareness, and so on.
I'm not sure that's an entirely apt comparison; the angstrom has a well-defined scientific value (even if an individual is ignorant of it), whereas we may never have a satisfactory way to quantify or qualify the true perception of colour. If we did, then it would no longer be a truly subjective phenomenon.
Fret not, I'm exhaustively familiar with the problem of subjectivity. The trick is to recognize that subjective phenomena fall into two categories: (a) those that we can describe and (b) those that don't matter. If two fully trichromat (three colour-sighted; normal vision) individuals perceive red differently and yet draw the same conclusions about it (e.g. it goes well with what I think is orange) then the distinction doesn't matter; our visual cortices are simply taking different routes to the same destination. There may be a trillion little differences in how two brains process information, but like two bug-free programs written to a very tight specification, the differences aren't important.
The thing about the IQ test is that there's no definition of intelligence to test against. The first tests consisted heavily of common knowledge dependent questions structured as reasoning problems, but these have pitfalls—e.g. if you don't know math, then you can't do arithmetic-dependent problems (an obvious problem for children's IQ tests), and if you expect someone to reason about an everyday situation, then you have to hope they haven't thought about the question before and that their cultural background is compatible. This last one is a real sticking point, and what led to the idea of IQ testing being derided as racist for a long period of time, but the differences can be even more subtle: a colleague in psychology once recounted an incident where a young girl was told to find "what was missing" in a cartoon depicting a suburban setting—the correct answer was "shadows," but she felt it was much more important that the house didn't have a pool, because all of the houses in her neighbourhood did!
As a result, the modern IQ test has retracted into dealing with one type of challenge: "pick which shape comes next in the sequence." This eliminates the ambiguity of what's being tested—it's culturally neutral, reliable calibratable against other versions of the IQ test that do the same thing, and easily repeated—but it doesn't address the core question, which is: "what is intelligence?" The idea of testing pattern recognition as a metric of intelligence goes back to the assumption that it's a core part of the learning process, but if so it must be a very primitive, Pavlovian one—certainly, reading from a textbook or listening to a lecture requires no ability to detect underlying repetition.
On top of that, as I said earlier, this sort of IQ test is something you can practice for, at least if you haven't done it for a while. Pattern recognition may be in part a function of innate ability, but it's definitely also a skill that can be improved, otherwise a number of games would be totally unplayable. As a result a null hypothesis (no change) for the IQ test should predict a slight increase, as the participants review and come to terms with the first test, and you should distrust any study that claims a 1-2 point improvement, as these are almost certainly spurious. However, a negative change, or a significant positive change could still potentially be statistically meaningful results. That still leaves room for misunderstood causation (e.g. cannabis users are all stressed out and don't have time to learn, and so their skills atrophy) but it's definitely legitimate correlation.
What I explained does not make the typical IQ test subjective, it makes it misunderstood: it is simply a test of pattern recognition skill, rather than some sort of generalized, inherent intelligence (which probably doesn't even exist, as people can always be taught to learn better.) If individuals display a consistent drop in performance during IQ testing, trusting that the tests have been constructed in the same manner with comparable problems, then that really does say something coherent and meaningful, even if it's been wrongly described or isn't clearly understood—for example, that the subjects have been less motivated to hone their learning ability.
The study is actually very aggressive and very thorough about including a diverse collection of individuals from different backgrounds (both users and non-users), and can safely be assumed to be properly controlled for most describable variables, as the dataset was collected for a multitude of applications.
If the IQ test were completely uninformative or unrepeatable, it would've been thrown out long before its incompleteness was recognized.
I don't know all that much about photosystems, so I'll have to trust your first two statements at face value—but what you're describing has nothing to do with statistics at all; that's just a misunderstanding of the physics at hand. Statistical calculations are only valid if the hypothesis is valid—in this case, that high absorption is proportional to chlorophyll activation.
You don't have a problem with statistics, you have a problem with ignorance of the facts—that's perfectly normal and healthy, and is necessary in all sciences. There is no causal link between people who are poorly informed and people who back up their statements with statistics; it's just the case that overly simple hypotheses, for which it is easy to derive the relevant statistics, are also easy to arrive at.
In bioinformatics we use statistics at just about every waking moment, and they do matter quite a lot when considering the false positive rate for tests. A microarray containing a million wells that has an error rate of 10^-5 will generally have about a thousand bogus values in it—cases where genes either activated completely or not at all simply because of hardware or procedural defects. Similarly, the likelihood of a random valid open reading frame (start and stop codon spaced 3 * k nucleotides apart, for some value k) below about 100 nt has a higher chance of being spontaneous than being an actual gene. These are things that can easily be demonstrated to be true physically, and yet are perfectly predictable through statistical procedures.
So go easy on the math. Yes there are serious problems in the social and medical sciences with flawed and shill studies, and yes there are plenty of figures thrown around in politics that are derived through questionable methods, but what matters is really that the people generating the figures are fools and scoundrels, not the fact that they framed the results as statistical measures or used some mathematical framework to produce them. The most shameful uses of data collection and extrapolation generally aren't even statistically or empirically sound, as antivirus companies constantly remind us. Rely on your gut instinct that they're slimeballs, not that they tried to dress up their garbage to sound scientific.
The bit about young mathematicians being ground-breakers isn't really proof; that age has gone up steadily as the size of the field (and the amount of learning required to truly reach the top) has increased. A medal devoted to recognizing the work of young mathematicians, the Fields Medal, occasionally gets boycotted because of this ageism.
Age seems to be linked to decreased intelligence because of societal roles more than anything innate: most people stop learning and move into positions in life where they're expected to either pass on knowledge or maintain and manage others. In academia, the tendency for this has been diminishing (no doubt because of ever-lengthening career ladders) and as a result it's been possible to defy that norm. A wonderful example: Bertrand Russell was still active in political activism well into his late nineties.
Slashdot Mirror
They did study a relatively cosy community, which was predominantly white. They removed as many families and evidence of distantly-related friends as they could using the Manichaikul et al. 2010 kinship metric, which has been cited 34 times (a fairly good number of hits for a two-year interval.)
The upshot of this study, in my opinion, is that the things that make people compatible friends happen to be reflected (to a degree) in certain markers in the genome. It doesn't necessarily indicate anything about race or relatedness, and you can probably predict that two people of radically different ethnicities may like each other as friends because of similarity simply due to the subset of markers that are important. It's a bit misleading of the authors to imply that this is somehow a pan-genome phenomenon when obviously there are plenty of things in the genome that could change without any detectable impact.
Almost all of the participants in the dataset were Americans of European descent. They controlled for the possibility of distant relatedness as aggressively as they could, which is a well-understood requirement of many GWAS experiments.
Those are social degrees of separation, not ancestral.
To eliminate the possibility that the results are influenced by people tending to make friends with distant relatives, we use only the 907 friend pairs where kinship <= 0 (recall that kinship can be less than zero whe n unrelated individuals tend to have negatively correlated genotypes). This procedure ensures that pairs of friends in the GWAS are not actually biologically related at all.
The data is from Framingham, Massachusetts. Where did you see any mention of the Midwest?
Can either of them defeat Panopticlick? I don't see anything on Epic's site about hiding font lists. (And on that point, Epic is a bad name choice since it's vaguely synonymous with the death of objectivity in news reporting.)
You'd think this would be a great time to use that "read more" snip feature that's typical of reviews and interviews... or maybe this is a cue to start using Larry Drebes's signature everywhere?
At any rate, it's a little sad to see this OpenID provider going because it means less diversity in the single sign-on landscape, which is the whole point. At least OpenID itself will still be around!
Sincerely,
Larry
—
Larry Drebes, CEO, Janrain, Inc.
Since when has the truth not been harmful in diplomacy, politics, or espionage?
Sadly, no mention of the christening in any of today's stories! Possibly the name was only for rc1 though.
I'm not familiar with that quirk; do you mean the caught/cot distinction? Can you give an example?
...Fun fact: they're also homophones in a Japanese accent, so that may've been a double joke. A lot of Chinese and Japanese speakers come up with inventive ways to deal with certain non-nasal consonants at the ends of syllables, which can get fantastically distracting when they speak quickly. "Labo" (for "lab") is one I hear a lot; another is "hoteru" (the actual Japanese word for "hotel".)
Knowledge doesn't exist until you can prove it, and two mutually exclusive statements cannot both be true. I will respect your right to hold to your beliefs and to put an end to this conversation, but the decision to do so is no acceptance or validation of your position, which remain illogical. Your justification for why psychology is not a science depends on science having the same flaws that you accuse psychology of having.
It appears to me that you don't have a formal logic education, and that you probably haven't been exposed to a balanced survey of philosophical thought. You're a wonderfully articulate person, but the claims you've been making remain unjustified; at several points you say things that lead to the rejection of the very notion of an axiom, which means that the very act of existing makes you hypocritical. I would strongly encourage you to reach out and consider other perspectives.
That all being said, I wish you well, and I understand if you want to stop discussing this.
We can know the ways in which something behaves without understanding what it is. The unanswerable questions about the nature of the universe do not prevent us from recognizing the patterns of behaviour or the interactions of things that occur within it. This is the third time you've tried to argue from an incorrect generalization.
I'll get back to the rest of your post once we sort this disaster out:
The fact remains that human behavior [...] is not predictable.
The brain is not capable of producing any phenomena that violate the laws of physics, and to the best of our knowledge does not exploit any quantum phenomena that are beyond a computer's ability to simulate or represent with sufficient precision. Where is this profound unpredictability coming from?
So what, exactly, do you think an IQ score is, if not a measurement? Bear in mind that the uncertainty in a quiz is no different than the uncertainty in a physical measuring instrument; they are governed by the same family of statistical distributions. Just because we don't know exactly how IQ relates to pattern recognition ability doesn't make it irrelevant; science is laden with metrics that are meaningful and objective but not direct representations of the underlying phenomenon (for example, we only have a rough idea about how absorbed radiation dosage over time equates to health risks.)
You may want to read up on Behaviorism—widely disparaged as inhumane and simplistic, but completely empirical, reproducible, and appropriately well-funded by some governments. Along with forbidden experiments, it quickly becomes apparent that only ethical necessity forces psychology to lean on philosophical insight for understanding certain areas; wherever possible, experimental methodology is employed and venerated with ever-increasing rigour. The same can be said of any of the social sciences.
Many parts of psychology are rigorous and measurable, and these sub-problems within the vague umbrella of "intelligence" are about as solid as you can get, since they're all analytical skills. Even if the tests are indirect, they're still objectively meaningful.
I don't know why you've brought that up; at no point have I been trying to assert that psychology is founded upon evidence-based hypothesis testing. Philosophy has probed the human mind for thousands of years, and while most of its predictions have been wrong, it is rare if ever that something which is correct has been passed by except due to external interference. There are plenty of things that experimentalism is insufficient to study, most obviously mathematics. This does not make them invalid. Science is only one part of how we understand the world; to assume everything in life must be scientifically-founded is to reject the whole of human civilization and the thinkers who invented natural philosophy. Psychology must make do because it is illegal and unethical to perform proper experiments—and even your second link concedes: "Some research is far more scientifically rigorous. And the field often yields interesting and important insights."
More to the point, all of the domains I brought up, with the exception of emotional intelligence, are problems studied in cognitive science, the field at the juncture between AI, psychology, and neuroscience that seeks to build a bottom-up model of how the brain works. They are all either backed with rigorous, repeatable experiments, or are theoretically airtight, or both.
It's a pretty common name—I was also murdered four years ago in Scotland.
I have to disagree with you about suggesting those other things are subjective, though. They have objective definitions within psychology—definitions that comprehensive enough to include everything that you or I might consider part of intelligence. They're anything but vague.
To flesh out the examples I gave: emotional intelligence has a very specific meaning and can be tested for, pattern recognition is extremely well-understood and can easily be accomplished by a computer (and tested with the standard IQ test!), decision-making efficiency can be clearly measured in a variety of contexts and is simply the speed at which decisions are made under varying degrees of situational complexity, intuition is skill in recognizing and reasoning about the nature of a concept or phenomenon (and is part of the exam in any UI design course), and spatial reasoning is already a problem heavily described and being studied by AI researchers.
The issue with intelligence is that it's too broad and vague a concept to be defined, not that it's absolute but unknown (as with angstroms) or inherently subjective (as with how a colour is interpreted and represented in an individual's field of vision after retinal processing, i.e. long after the frequency of incoming photons stops being relevant.) They're not quite apt metaphors, hence the confusion, but I understand your point and apologise for the cumbersomeness of the conversation thus far.
I'd actually argue that we should throw out the idea of intelligence altogether and just discuss the elements of it that we actually intend to indicate—pattern recognition, efficiency in decision-making, intuition, spatial reasoning ability, emotional self-awareness, and so on.
I'm not sure that's an entirely apt comparison; the angstrom has a well-defined scientific value (even if an individual is ignorant of it), whereas we may never have a satisfactory way to quantify or qualify the true perception of colour. If we did, then it would no longer be a truly subjective phenomenon.
Fret not, I'm exhaustively familiar with the problem of subjectivity. The trick is to recognize that subjective phenomena fall into two categories: (a) those that we can describe and (b) those that don't matter. If two fully trichromat (three colour-sighted; normal vision) individuals perceive red differently and yet draw the same conclusions about it (e.g. it goes well with what I think is orange) then the distinction doesn't matter; our visual cortices are simply taking different routes to the same destination. There may be a trillion little differences in how two brains process information, but like two bug-free programs written to a very tight specification, the differences aren't important.
The thing about the IQ test is that there's no definition of intelligence to test against. The first tests consisted heavily of common knowledge dependent questions structured as reasoning problems, but these have pitfalls—e.g. if you don't know math, then you can't do arithmetic-dependent problems (an obvious problem for children's IQ tests), and if you expect someone to reason about an everyday situation, then you have to hope they haven't thought about the question before and that their cultural background is compatible. This last one is a real sticking point, and what led to the idea of IQ testing being derided as racist for a long period of time, but the differences can be even more subtle: a colleague in psychology once recounted an incident where a young girl was told to find "what was missing" in a cartoon depicting a suburban setting—the correct answer was "shadows," but she felt it was much more important that the house didn't have a pool, because all of the houses in her neighbourhood did!
As a result, the modern IQ test has retracted into dealing with one type of challenge: "pick which shape comes next in the sequence." This eliminates the ambiguity of what's being tested—it's culturally neutral, reliable calibratable against other versions of the IQ test that do the same thing, and easily repeated—but it doesn't address the core question, which is: "what is intelligence?" The idea of testing pattern recognition as a metric of intelligence goes back to the assumption that it's a core part of the learning process, but if so it must be a very primitive, Pavlovian one—certainly, reading from a textbook or listening to a lecture requires no ability to detect underlying repetition.
On top of that, as I said earlier, this sort of IQ test is something you can practice for, at least if you haven't done it for a while. Pattern recognition may be in part a function of innate ability, but it's definitely also a skill that can be improved, otherwise a number of games would be totally unplayable. As a result a null hypothesis (no change) for the IQ test should predict a slight increase, as the participants review and come to terms with the first test, and you should distrust any study that claims a 1-2 point improvement, as these are almost certainly spurious. However, a negative change, or a significant positive change could still potentially be statistically meaningful results. That still leaves room for misunderstood causation (e.g. cannabis users are all stressed out and don't have time to learn, and so their skills atrophy) but it's definitely legitimate correlation.
What I explained does not make the typical IQ test subjective, it makes it misunderstood: it is simply a test of pattern recognition skill, rather than some sort of generalized, inherent intelligence (which probably doesn't even exist, as people can always be taught to learn better.) If individuals display a consistent drop in performance during IQ testing, trusting that the tests have been constructed in the same manner with comparable problems, then that really does say something coherent and meaningful, even if it's been wrongly described or isn't clearly understood—for example, that the subjects have been less motivated to hone their learning ability.
The study is actually very aggressive and very thorough about including a diverse collection of individuals from different backgrounds (both users and non-users), and can safely be assumed to be properly controlled for most describable variables, as the dataset was collected for a multitude of applications.
If the IQ test were completely uninformative or unrepeatable, it would've been thrown out long before its incompleteness was recognized.
Already used up that excuse on asteroids, alas. Maybe if China says they're gonna do it?
I don't know all that much about photosystems, so I'll have to trust your first two statements at face value—but what you're describing has nothing to do with statistics at all; that's just a misunderstanding of the physics at hand. Statistical calculations are only valid if the hypothesis is valid—in this case, that high absorption is proportional to chlorophyll activation.
You don't have a problem with statistics, you have a problem with ignorance of the facts—that's perfectly normal and healthy, and is necessary in all sciences. There is no causal link between people who are poorly informed and people who back up their statements with statistics; it's just the case that overly simple hypotheses, for which it is easy to derive the relevant statistics, are also easy to arrive at.
In bioinformatics we use statistics at just about every waking moment, and they do matter quite a lot when considering the false positive rate for tests. A microarray containing a million wells that has an error rate of 10^-5 will generally have about a thousand bogus values in it—cases where genes either activated completely or not at all simply because of hardware or procedural defects. Similarly, the likelihood of a random valid open reading frame (start and stop codon spaced 3 * k nucleotides apart, for some value k) below about 100 nt has a higher chance of being spontaneous than being an actual gene. These are things that can easily be demonstrated to be true physically, and yet are perfectly predictable through statistical procedures.
So go easy on the math. Yes there are serious problems in the social and medical sciences with flawed and shill studies, and yes there are plenty of figures thrown around in politics that are derived through questionable methods, but what matters is really that the people generating the figures are fools and scoundrels, not the fact that they framed the results as statistical measures or used some mathematical framework to produce them. The most shameful uses of data collection and extrapolation generally aren't even statistically or empirically sound, as antivirus companies constantly remind us. Rely on your gut instinct that they're slimeballs, not that they tried to dress up their garbage to sound scientific.
The bit about young mathematicians being ground-breakers isn't really proof; that age has gone up steadily as the size of the field (and the amount of learning required to truly reach the top) has increased. A medal devoted to recognizing the work of young mathematicians, the Fields Medal, occasionally gets boycotted because of this ageism.
Age seems to be linked to decreased intelligence because of societal roles more than anything innate: most people stop learning and move into positions in life where they're expected to either pass on knowledge or maintain and manage others. In academia, the tendency for this has been diminishing (no doubt because of ever-lengthening career ladders) and as a result it's been possible to defy that norm. A wonderful example: Bertrand Russell was still active in political activism well into his late nineties.
This is a very fascinating statement and I wish to understand how you arrived at it.