Sunday, January 06, 2019

How not to debunk IQ

"It is via jjvb not via positiva."
Don't add Latin just to look smart.
 "Designed for learning disabilities, it ends up selecting exam-takers, paper shufflers, obedient IYIs (intellectuals yet idiots), ill adapted for “real life”. "
And yet it predicts for almost all positive results in society from income to marriage stablity.
 "The concept is poorly thought out mathematically (a severe flaw in correlation under fat tails,"
Which again can't be too bad because it is the best predictor of almost all good results.

"fails to properly deal with dimensionality,"
It doesn't need to. Again it's a pretty good predictor of a number of things, that demonstrates it's useful and therefore not pseudoscience. Sure a series of measure for various types of mental functioning would be better. However so far you certainly haven't been able to provide such a series of measures.

"treats the mind as an instrument not a complex system"
OK firstly you know an instrument can be a complex system right? And in any case making a simplifying measurement of a complex system doesn't mean you're not treating it as a complex system. It just means you're concentrating on a subset of it's properties so as to more easily analyze it. You know, science?

"and seemed to be promoted by racists/eugenists,"
It doesn't matter who promotes it.

"people bent on showing that some populations have inferior mental abilities based on IQ test=intelligence;"
Nobody argues that IQ test=intelligence and in fact many people pushing the idea that disparate outcomes are explained by cognitive differences go BEYOND the IQ test.  They refer to 'g', the hypotheticized general level of intelligence and point out that questions that are 'g-loaded' are even more racially differentiated than general IQ questions.

" those have been upset with me for suddenly robbing them of a “scientific” tool, as evidenced by the bitter reactions to the initial post on twitter/smeara cmpaigns by such mountebanks as Charles Murray."
Given that Charles Murray is the most smeared academic I've heard of it's rich for you to complain that he responds with smears.  When we're talking about twitter the evidence is easily accessed, so screenshots or it didn't happen.

" psychometrics peddlers looking for suckers (military, large corporations) buying the “this is the best measure in psychology” argument when it is not even technically a measure"
I don't think you know what a measure is.  Nor do I think you understand why large organizations would want a proxy for something, which is not the same as a measure.

"it explains at best between 13% and 50% of the performance in some tasks( those tasks that are similar to the test itself),"
Earning an income isn't similar to the test itself.  Maintaining a stable marriage isn't similar to the test itself.  Health outcomes aren't similar to the test itself.  Yet all of these are predicted to some extent by IQ.

Far more important is the fact that 13% let alone 50% of the performance in a task is a HUGE difference!  Imagine if you have to pick someone for a task where success can be objectively measured.  Suppose that results from the first attempt form bell curve with a mean of 100 and a standard deviation of 20.  If you had a test that predicted 13% of the outcome that would mean that it had the task would have a mean of 102.6 standard deviation of 17.4 merely from hiring people who score at or above +1 standard deviation.   So over 2% extra productivity merely from applying one test and being somewhat selective on the basis of it.

"minus the data massaging and statistical cherrypicking by psychologists;"
Pics or it didn't happen.

"it doesn’t satisfy the monotonicity and transitivity required to have a measure"
Neither is required to be useful and again, SHOW YOUR WORK!

"(at best it is a concave measure)."
I have no idea what a concave measure is.  I looked up what you meant by "concave" and you named it stupidly (it should be "concave up").  How a measure can be concave I have no idea.

"No measure that fails 60-95% of the time should be part of “science” "
Actually it should the sign should just be reversed.  But please tell me one test where the predictions of IQ fail 60% of the time?  Show me something where if I make a prediction based on how IQ is presently understood I would not just fail as often as chance but 10-45% MORE OFTEN.  I think you're just lying now.

"The graph that summarizes the first flaw (assuming thin tailed situations),showing that  “correlation” is meaningless in the absence of symmetry. We construct (an intelligence test (horizontal), that is 100% correlated with negative performance (when IQ is, say, below 100) and 0% with upside, positive performance. We progressively add noise (with a 0 mean) and see correlation (on top) drop but shift to both sides. Performance is on the vertical axis."
Except for that to work you have to ASSUME that the actual correlation is 0% above a certain point.  This is a testable hypothesis.  It's predictions do not pan out.  If there was no effect of greater IQ past a certain point then extremely high IQ individuals would not have different results than those of moderate (but above threshold) IQ.  This is not the case.

" The problem gets worse with the “g” intelligence based on principal components."
No it doesn't.  I know it doesn't because you claimed it did and didn't support it.

"There is no significant correlation (or any robust statistical association) between IQ and wealth."
So you've found one measurement that didn't corelate to IQ, so what?  Considering that it almost certainly correlates to income and income corelates to IQ at about .23 that means that IQ correlates to income BETTER when wealth is taken into account.

"Most “achievements” linked to IQ are measured in circular stuff s.a. bureaucratic or academic success,"
If that were true it would be irrelevant since "most" is not the same as "all".  Otherwise we could just construct a million tests and say "Most achievements linked to IQ are measured in circular stuff." even if literally thousands of legitimate measures were linked to IQ.

"things for test takers and salary earners.;
What is wrong with earning a salary?  Why is that somehow an indication you aren't intelligent or aren't using intelligence to succeed.

"Wealth may not mean success but it is the only “hard” number,"
No it's not.  Income, conviction rates, accident rates, these are all hard numbers.  You are simply lying how.

" not some discrete score of achievements."


" You can buy food with a $30, not with other “successes” s.a. rank, social prominence,  or having had a selfie with the Queen."
Do you honestly think that nobody ever got free food because they had social prominence?  You don't human very well do you?

"Some argue that IQ measures intellectual capacity real worldr esults come from, in addition, “wisdom” or patience, or “conscientiousness”, or decision-making or something of the sort. No. It does not even measure intellectual capacity/mental powers."
Yes it does.  It's not prefect but if you want a task done that is cognitively complex IQ is a good predictor of whether someone can do it.  If you pick the highest IQ person out of a pair to do such a task you will be right significantly more than 50% of the time.


"If you want to detect how someone fares at a task, say loan sharking, tennis playing, or random matrix theory, make him/her do that task;"
If you want the task actually done on the other hand, particularly if it's a task that causes bad effects if done badly use all the useful measures to determine if they're going to be good first.  That includes IQ.

"we don’t need theoretical exams for a real world function by probability-challenged psychologists.  Traders get it right away: hypothetical P/L from “simulated” paperr strategies doesn’t count Performance=actual. What goes in people’s head or reaction to a screen image doesn’t exist (except via negativa)."
So the billions of hours spent in  simulators was a waste of time?  The idea that the only way to determine if someone might be good at something is to  have them actually do it is suicidally stupid.  I wonder if this guy will be good at running a nuclear power plan?  Well no way to tell let's just see.

Even in situations showing presence of a correlation Income-IQ, you seeM ONSTROUS noise. Even at low IQ! From Zagorsky (2007)."
So what?  Yes with .23 corelation there is a lot of noise.  That doesn't mean that ignoring IQ is a good idea.  It just means that it's not perfect.  You seem to think that a measure not being perfect means it's useless.  Or rather you pretend to think that.

"Why are we talking about it? This truncates the big upside, so we not even seeing the effect of fat tails. Fat Tails If IQ
is Gaussian by construction and if real world performance were, net, fat tailed (it is), then either the covariance between IQ and performanced oesn’t exist or it is uninformational."
I know you pretend to think you've shown that but you really haven't.

"It will show a finite number in sample but doesn’t exist statistically. Another problem: when they say "black people are x standard deviations away”. Different populations have different variances, even different skewness and these comparisons require richer models."
They really don't.  If blacks are say 5 IQ points lower a different variance might change exactly how many blacks you'd expect to be CEOs of major corporations, but it doesn't change that you wouldn't expect them to have substantially less than proportionality.  At least not unless you're saying that they have s.d. at least a point higher than whites.

"These are severe, severe mathematical flaws (a billion papers in psychometrics wouldn’t count if you have such a flaw).

See the formal treatment in my next book. Mensa members: typically high “IQ” losers
in Birkenstocks."
And here he was criticizing others for smears.  Of course mensa members are self-selected and not necessarily typical of high-IQ individuals.  How you quantify "loser" is not specified and I suspect not specifiable.

"But the “intelligence” in IQ is determined by academic psychologists like the
"paper trading” we mentioned above, via statistical negative performance (as it
was initially designed to detect learning special needs)"
You know they've changed the tests since then right?

"but then any measure would work there."
Yeah we measured whether they liked banana pancakes and it detected special learning needs.

"A measure that works in left tail not right tail IQ decorrelates as it goes higher) is problematic."
But that doesn't mean it's not useful.  You keep acting like this is a killer argument, it's not.  It would be if you could demonstrate that say, IQ 120 people were as likely to IQ 150 people to achieve any results.

"We have gotten similar results since the famous Terman longitudinal study,"
And yet you cite not a single such results, either relating the the Terman study or not.  Frankly from what I've read the Terman's study did show that the "gifted" children succeeded far above what chance would suggest, even correcting for the students being well educated and white.

" even with massaged data for later studies."
Show don't tell.

"The statistical spin, as a marketing argument, is that a person with an IQ of
70 cannot prove theorems, which is obvious for a measure of unintelligence but they fail to reveal how many IQs of 150 are doing menial jobs)."
Because that's irrelevant.  The relevant number is what proportion are doing extremely complex mental tasks successfully compared to the proportion at other IQ levels.  A proxy for this is holding highly paid and prestigous jobs, being published in scientific journals etc.

"It is a false comparison to claim that IQ “measures the hardware” rather than
the software."
Did anyone say it did?

" It can measures some arbitrarily selected mental abilities (in a testing environment) believed to be useful."
And shown to be by the consistent corelation of IQ to success at all levels, even if it decreases at higher levels.  Again, useful doesn't equal perfect.

"However, if you take a Popperian-Hayekian view on intelligence, you would realize that to measure it you would need to know the mental skills needed in a future
ecology,"
No because pattern recognition is valuable regardless of environment.  All successful prediction is based on pattern recognition.  In any case you're assuming that IQ doesn't corelate at all to general intelligence, that is the general ability to complete cognitive tasks.  It does.

"which requires predictability of said future ecology. It also requires the skills to make it to the future (hence the need for mental biases for survival)."
But IQ is corelated to getting to the future.

" 1) When someone asks you a question in the real world, you focus first on “why is he/she asking me that?”, which shifts you to the environment (see Fat Tony vs Dr John in The Black Swan) and detracts you from the problem at hand. Only suckers don’t have that instinct."
No most people in most contexts focus first on helping the person who appears to need help.  Sure in some contexts I might ask myself why they need the help, particularly if they are people who have shown themselves to act against my interests or who are behaving strangely by asking.  Some questions might invite that response.  For instance "Where would you hide a Supreme Court judge's body if you only had a few hours to shift it?" would tempt me to consider the askers motives.  "Where's the bus stop?" probably wouldn't.

" 2) Real life never never offers crisp questions with crisp answers"
Should you go back to the ex lover who raped one of your friends once and you multiple times?  That gets a fairly crisp no.

" most questions don’t have answers;"
How would you know that?  Where is your evidence?  And why is that relevant?  After all most questions are never asked because nobody cares.  What was the amount of katsup spilled on Chicago Cubs fans sitting in prime numbered seats during home games in the 1997 season?  Nobody knows and nobody cares.  There are an infinite amount of other questions I could ask that are equally irrelevant.

"perhaps the worst problem with IQ is that it seem to selects for people who don’t like to say “there is no answer, don’t waste
time, find something else”. "
Where is your evidence for this?  You say "seems" as though it's just something you came up with sans any specific evidence.  Like saying "Asians can't drive." because you remember seeing a lot of bad Asian drivers without considering any actual scientific data on Asian driving.

"3) It takes a certain type of person to waste intelligent concentration on classroom/academic problems."
Firstly note the assumption that these problems are a waste of time.  He shows no evidence of this and in fact there are literally thousands of examples of problems academics were interested in that were not a waste of time.  Things like imaginary numbers for instance.  Without understanding those we wouldn't have Euler's equation which is vital to understanding electrical waves, which this computer runs on.

"These are lifeless bureaucrats who can muster sterile motivation."
So they are "lifeless" because they can concentrate on something you can't see the point of, but which other people want them to do?  Wow, you're a really horrible person you know that?  I mean condemning people for being able to do a job someone else wants them to do.

"Some people can only focus on problems that are real, not fictional textbook
ones (see the note below where I explain that I can only concentrate with real
 not fictional problems)."
Then you're kinda stupid.  I mean if you can't think something through merely because you can't relate it to a physical entity your imagination is severely limited.

"4) IQ doesn’t detect convexity (by an argument similarto bias-variance you need to make a lot of small inconsequential mistake in order to avoid a large consequential one. See Antifragile and how any measure of “intelligence” w/o convexity is sterile edge.org/conversation/n)."
It doesn't need to.  Convexity is a quality of the world, not the minds inside the world.

To explain convexity as you uses the term is a situation where the costs from acting are likely to be small and the benefits big so making multiple mistaken acts for each beneficial one.

"To do well you must survive, survival requires some mental biases directing to some errors."
How is that relevant to IQ?

"5) Fooled by Randomness: seeing shallow patterns in not a virtue leads to naive interventionism."
And why would high IQ people see a higher ratio of shallow patterns to deep patterns than lower IQ people?  Sure they would see more patterns regardless of whether they were shallow or deep.  But you yourself point out the value of large numbers of attempt at something even if most are failures.  Hell you wrote a book about it.  So higher IQ people seeing more patterns is a good thing, even if most of them are shallow.

"Some psychologist wrote back to me: “IQ selects for pattern recognition, essential for functioning in modern society”. No. Not seeing patterns except when they are significant is a virtue in real life."
But nobody can see only the significant patterns, not even the smartest scientist trained for years.  What happens in fact is that we see a pattern and try to figure out if it's significant.  Now it's true that highly intelligent people will spend time trying to figure out if more chance patterns are in fact valid signifiers of a relationship.  That's OK if they also figure out more valid signifers of a relationship.  False positives are not as bad as false negatives.

" 6) To do well in life you need depth and ability to select your own problems
 and to think independently."
You're not going to define "do well in life" or "depth" are you?  While thinking independently can be valuable if you can't think WELL it's not helpful.  New ideas from an idiot are generally not worth the time to hear.  Present company excepted.

“ Upper bound: discount the massaging and correlation effects. Picked up from the highly
 unrigorous Intelligence: All That Matters by S. Ritchie.”
Note that again he does nothing to back up his claims that everyone else is disingenuously unrigorous etc.  Note that this is from the man who claimed that "most questions" don't have "crisp" answers without a single reference from any field.

 “Functionary Quotient: If you renamed IQ , from “Intelligent Quotient” to FQ
 “Functionary Quotient” or SQ “Salaryperson Quotient”, then some of the
 stuff will be true. It measures best the ability to be a good slave.”
No it measures the ability to complete complex tasks that have no direct benefit to you in an environment where completing that task pleases others.  In other words it demonstrates that you can work for other people, which in modern society is kinda important.  That's not being a slave.  It's being a worker and if you can't see the difference you're an idiot.  It does not indicate that these people are more subserviant, just that they are prepared to do what others want in certain specific circumstances.

 “The argument that “some races are better at running” hence [some inference about
 the brain] is stale:”
Arguments are valid or not, there is no such thing as "stale".

 “mental capacity is much more dimensional and not defined in the
 same way running 100 m dash is.”
How many dimensions something has determines how easy it is to measure, not whether it could have evolved to different extents in different races.  Nor is how something is defined relevant.  Evolution works on things that have real effects however we define things and whether we define them at all.  If we had never tried to define intelligence it would still have a differential effect on our survival and alleles for it would still reproduce at different rates than other alleles.

 ;The Flynn effect should warn us not just that IQ is somewhat environment
 dependent,”
Nobody denies this.

“ but that it is at least partly circular.”
????

 “If you looked at Northern Europe from Ancient Babylon/Ancient Med/Egypt, you would have
 written the inhabitants off... Then look at what happened after 1600.”
So what?  The fact that a group did not succeed compared to some other group in a different environment thousands of years ago doesn't mean anything about it's IQ or it's effect now.

 “The same people hold that IQ is heritable,”
Everyone in the field who isn't a complete joke holds that it's heritable.  The evidence is overwhelming.

“ that it determines success, that Asians have higher IQs than
 Caucasians, degrade Africans, then don’t realize that China for about a
 Century had one order of magnitude lower GDP than the West.”
So what?  Nobody said that high IQ in a population was capable of overcoming the massive problems that culture and government can create.  Find one person who believes in racial IQ differences that believes that whites in the most horrible political system imaginable would be able to thrive.

“ Reactions to this piece in the Alt-Right Media:”
Name one of the people you describe who identifies as alt-right or who qualifies under Richard Spencer's definition.

“all they got is a psychologist who still hasn’t gotten to the basics of noise/signal.”
Again stop making claims without evidence.


 “Mathematical Considerations
 CURSE OF DIMENSIONALITY A flaw in the attempts to identify “intelligence”
 genes. You can get monogenic traits, not polygenic (note: additive
 monogenic used in animal breeding is NOT polygenic).”
Out of the four ways you could be wrong in the above quote you got one right, and it was the one that favored your opposition.  It's possible to detect genes that affect a  polygenic trait.   That's what I think you meant to say since of course we can detect polygenic traits being tall is a polygenic trait and I can tell if you're tall.  There doesn't seem to be such a thing as "additive monogenic" traits.   Traits with many genes that add up to affect them are called "polygenic".  What I think you meant by polygenic is that a gene will affect a trait differently depending on the presence of absence of other genes.  These can be detected too, easily.

These "synergistic" genes each have an average effect on a trait equal to the probability that the combination of other genes they synergize with are present and the effect on the trait if they are.  So if having genes A, B and C have an effect on IQ of +1 and the probability of genes B and C being present given that gene A is is 0.01 the average effect of gene A is +0.01.  This if there is a population of 100 million people with Gene A the group average IQ will have an s.d. or 15/10,000 or about one sixth the size of the gene's effect.  So Gene A will be six sigmas out.  That's detectable.  Of course it depends on the numbers of people with the gene combination and the size of the effect, but that's true of non-synergistic effects too.  All that matters is a)  how probable is the genetic situation in the population, b) how big an effect does it have on IQ and c) how big is the population you study?   

I know you don't like to think about theorectical situations but it really helps to do so before commenting on them.

No comments: