Understanding Polygenic Scores (PGS)

In the years to come it will be very important to understand the concept of your polygenic score for traits such as height, weight, intelligence, personality and many others. In chapter 12 of his book, "Blueprint: How DNA Makes Us Who We Are" Robert Plomin gives a gentle introduction to the PGS concept which I excerpt here.

---

"Because polygenic scores are the basis for the DNA revolution in psychology, it is essential to understand what they are. A polygenic score is like any composite score that psychologists routinely use to create scales from items, such as those on a personality questionnaire. The goal of a polygenic score is to provide a single genetic index to predict a trait, whether schizophrenia, well-being or intelligence.

To get a concrete understanding of a polygenic score, consider a personality trait like shyness. A questionnaire to assess shyness includes multiple items in order to tap into different facets of shyness. For example, a typical shyness questionnaire will have items about how anxious you are in social situations and how much you avoid these situations for example, going to a party, meeting strangers and speaking up at a meeting. You might be asked to respond using a three-point scale (0 = not at all, 1 = sometimes, 2 = a lot).

A shyness score is created by adding these items, taking care to ‘reverse’ items as needed so that a high score means a high degree of shyness. If our shyness measure had ten items scored 0, 1 and 2, total scores could vary from 0 to 20. Simply adding the items like this treats each item as if it is equally useful, but all items are not equally useful. For this reason, items are often added after they are weighted by some criterion of their usefulness at capturing the construct of shyness.

This is exactly how polygenic scores are created, except that, instead of items on a questionnaire, we add up SNP genotypes. Like the three-point rating scale for shyness, SNP genotypes are scored as 0, 1 or 2, indicating the number of ‘increasing’ alleles, as in the example of the FTO SNP [a polymorphism implicated in weight gain].

In the same way that we can add up alleles for one SNP to create a genotypic score, we can also add up alleles for many SNPs to create a polygenic score, just as we add questionnaire items to create a shyness score.

The results from genome-wide association studies are used to select SNPs and to assign weights to each SNP. For example, in the GWA analysis of weight, the FTO SNP accounts for much more variance than other SNPs, so it should count for much more in a polygenic score for weight.

The following table shows how one individual’s polygenic score is created from ten SNPs. For the first SNP, this individual’s genotype is AT. For this SNP, the T allele happens to be the increasing allele that is positively associated with the trait. So, the individual’s genotypic score for this SNP is 1 because the genotype has only one increasing T allele.

Across the ten SNPs, the individual has a total of nine increasing alleles for the trait out of a possible score of 20. So, this individual would have a polygenic score just below the population average score of 10 for this trait.

This score merely adds the number of increasing alleles, which works reasonably well as a polygenic score.

However, we can increase its precision by weighting the genotypic score for each SNP by how much the SNP correlates with the trait. The correlation between each SNP and the trait is taken from the GWA analysis. If one SNP correlates five times more with the trait than another SNP such as SNP 1 versus SNP 10 it should count for five times as much in the polygenic score.

The weighted genotypic scores in the last column of the table are the product of the genotypic score for each SNP and the correlation with the trait. The sum of these weighted genotypic scores for the ten SNPs is 0.023.

This number isn’t as interpretable as the unweighted genotypic score of 9, which is just the sum of the ‘increasing’ alleles. However, both the unweighted polygenic score of 9 and the weighted score of 0.023 can be expressed simply as a percentile in the population. For this individual, both types would indicate a polygenic score just below average.

How many SNPs should go into a polygenic score? Initially, polygenic scores were created using only the genome-wide significant ‘hits’ from a GWA study. For weight, ninety-seven independent SNPs reached genome-wide significance. Creating a polygenic score from these top ninety-seven SNPs explains 1.2. percent of the variance in weight in independent samples. This is only slightly better than the prediction from the FTO SNP by itself, which explains 0.7 per cent of the variance.

Using only genome-wide significant hits is like demanding that each item in our shyness scale predicts significantly on its own. We don’t do this for other psychological scores because it is unrealistic to expect each item to stand on its own. The goal is to have a composite scale that is as useful as possible.

A better idea is to do what we do when we create other psychological scores: keep adding items as long as they add to the reliability and validity of the composite in independent samples. For polygenic scores, the key criterion is prediction. The new approach to polygenic scores is to keep adding SNPs as long as they add to the predictive power of the polygenic score in independent samples.

This is the strategy that has paid off in the last two years in producing powerful polygenic scores for psychological traits. Some false positives will be included in the polygenic score but that is acceptable as long as the signal increases relative to the noise, in the sense that the polygenic score predicts more variance.  ...

To interpret polygenic scores, it is important to keep in mind that they are always distributed like a bell-shaped curve, that is, a normal distribution. This bell-shaped curve is dictated by the fundamental law of probability, the central limit theorem, which is the basis for all statistics.

The normal distribution is found when many random events contribute to a phenomenon, like flipping a coin and counting the number of times the coin comes up heads. If you flip a coin ten times, you could get no heads or ten heads in a row, but most of the time the total number of heads will be between four and seven. If you do this many times, you will get a perfectly normal bell-shaped distribution, peaking at five, which will be the average number of heads. Flipping coins and counting heads is exactly analogous to counting the numbers of ‘increasing’ alleles from SNPs to construct polygenic scores for many individuals.

I will describe all my polygenic scores in terms of percentiles in the normal distribution. That is, to what extent is my polygenic score above or below the average polygenic score in the comparison sample, the 50th percentile?

It turns out that my polygenic score for height is at the 90th percentile. So, based on my DNA alone, knowing nothing else about me, you could predict that I am tall. And, in fact, I am 6 feet 5 inches. Of course, you can easily see that I am tall if you saw me, but with DNA you could tell that I am tall without even looking at me.

Most importantly, you could have predicted when I was born that I would be tall. Unlike any other predictors, polygenic scores are just as predictive from birth as from any other age because inherited DNA sequence does not change during life. In contrast, height at birth scarcely predicts adult height.

The predictive power of polygenic scores is greater than any other predictors, even the height of the individuals’ parents. Another advantage of polygenic scores over family resemblance is that parental height provides only a family-wide prediction that is the same for any child born to those parents.

In contrast, polygenic scores provide a prediction specific to each individual. In other words, my polygenic scores at birth would have predicted that I would be taller than expected on the basis of the average height of my parents.

Before looking at my other polygenic scores, one other general point needs to be highlighted about predicting individuals. My actual height is at the 99th percentile but my polygenic score is at the 90th percentile. Are polygenic scores sufficiently accurate for prediction?

For example, in TEDS [Twins Early Development Study - 1994 onwards], the polygenic score for height predicts 15 percent of the variance in actual height in these young adults. But 15 per cent is a long way from 100 per cent.

In fact, polygenic scores can never predict 100 per cent of the variance of any trait, because the ceiling for prediction is heritability. For height, heritability is 80 percent, but for psychological traits heritability is 50 percent, which means that polygenic score prediction is always going to be way south of perfect.

The big question is the extent to which polygenic scores will be able to predict all the heritable variance of traits. This gap is called missing heritability, and is described in the Notes section at the end of this book. "

--- [end of text extract] ---

Plomin's scatter plot looks rather messy but it hides the extent of clustering when you consider each decile separately.

As mentioned in the legend above, the vertical lines indicate the 95% confidence intervals. They clearly illustrate the linear trend line. Also, I suspect the limitations on sample size (20,000 pairs of twins in TEDS) for genetic studies.

For the top and bottom deciles, the following chart shows the extent of overlap.

Yet at the extremes the differences are very large. This is a general truth as regards traits whose values are are normally distributed.

Plomin finishes this section by emphasising yet again that due to the current lack of power (too few SNPs identified) and the ceiling of heritability, the PGS prediction is just that - a prediction with an error distribution around it. It is not deterministic.

I would comment that the error bars may be long right now, but as sample sizes get larger and non-additive effects are factored in, they can be made considerably smaller.

This is a small excerpt from an excellent book, by the way, which I reviewed here.

The dream of 'designer babies'

Some things in the media are so mind-numbingly stupid that I feel frozen into dumbness: 'Where would I start?'

Where is the politician or pundit who simply states - in exasperation:

"Why wouldn't we want designer babies? Is random better? And when you choose your partner with care, aren't you trying - in  part - to optimise your children? Listen people, we already do designer babies! And that's good and indeed evolutionarily obvious."

OK, I never heard that. Ever.

But we will soon be able to make more informed choices about the genomes of our offspring. Three obstacles:

1. We need to understand the phenotypical effects of alleles (existing or even new)
2. We need certainty about the effects of genetic engineering - no mistakes
3. We need to make it easy and convenient to access and then implant the modified cell.

So, once fixed, my question is: what do we want to do?

---

Issues

Firstly, there seems to be a biological recognition of kin-similarity. Babies swapped unknowingly at birth who grow up with unrelated 'parents' do seem to feel that something is wrong. I don't think we understand causally-genomically what is going on here and changing too many alleles, increasing the genomic distance between parents and child, may start to impact the relationship.

Secondly, you can't optimise everything. It may be true - as Steve Hsu has insisted - that high IQ tends to correlate with general health and superior performance in most areas. And if it's a question of minimising genetic load you can see why.

But beyond a certain point, high IQ seems to require that more of the brain is devoted to the processing of abstractions leaving less for other things. This is probably part of the underlying etiology for different personality types.

There seem to be few highly-intellectual axe-wielding warriors.

Thirdly, there is a feeling that although some parents might want their children to be great athletes or painters or musicians or novelists or priests or therapists or .. parents, there is something special about IQ.

Without intelligence we are sunk, none of the rest is going to work as our civilization will collapse. And that has been true up to now: all that smart fraction stuff - that you need a lot of folks with IQ 105+ just to run a complex society. And that the real innovation comes from those world-class people with IQs in excess of 160.

But .. I have not the slightest doubt that by the time we are able to routinely engineer the genomes of our offspring, we will have AI systems which are conceptually-competent way beyond the smartest humans that we can envisage.

We know what really smart people do: they internalise and extend a vast set of abstractions and manipulate them in interesting and complex ways to engage with problems. The lower foothills of this space are already colonised by deep-learning systems: their future seems pretty scalable.

It's not at all clear that the destiny or destination of the human race is or should be unbounded smartness, once we correct the errors of mutational load.

---

I've just finished reading "The Strangest Man: The hidden Life of Paul Dirac, Quantum Genius" by Graham Farmelo.

Amazon link

Dirac was plainly one of the smartest people who have ever lived. He is also generally assessed as autistic (Asperger's syndrome). In Farmelo's account, although not chronically unhappy, he hardly seems to have led a fulfilled life. His total, obsessive focus on developing theory seems to have led to a final disillusionment with the state of physics when he died (1984) - and with his own life's work.

I suspect that the first epoch of empowered genomic engineering will not be a mad rush for ever higher IQ, with the target of the ubiquitous production of von Neumann equivalents. Instead, I suspect we will edit out the obvious errors and reinforce the talents already latent in the specific underlying genome.

Let a thousand flowers bloom.

---

I suspect the second epoch of genomic engineering will be entirely different.

Selling capitalism to the feudal nobility

Steve Hsu is pessimistic about the terminal age of decadence in which we live:

"The empires Glubb studied had a lifespan of about ten human generations, or two hundred and fifty years, despite changing factors such as technology. Glubb describes a pattern of growth and decline, with six stages: the Ages of Pioneers, Conquest, Commerce, Affluence, Intellect and Decadence. He pointedly avoided writing about India or China, focusing rather on middle and western Eurasia, stating that his knowledge was inadequate to the task.

Note that six stages in 10 generations means that significant change can occur over one or two generations -- a nation can pass from one age to the next, as I believe we have in America during my lifetime.

... There does not appear to be any doubt that money is the agent which causes the decline of this strong, brave and self-confident people. The decline in courage, enterprise and a sense of duty is, however, gradual. The first direction in which wealth injures the nation is a moral one. Money replaces honour and adventure as the objective of the best young men. Moreover, men do not normally seek to make money for their country or their community, but for themselves.

Gradually, and almost imperceptibly, the Age of Affluence silences the voice of duty. The object of the young and the ambitious is no longer fame, honour or service, but cash. Education undergoes the same gradual transformation. No longer do schools aim at producing brave patriots ready to serve their country. [ Or to discover great things for all mankind! ] Parents and students alike seek the educational qualifications which will command the highest salaries. ...

Duty, Honor, Country:

The unbelievers will say they are but words, but a slogan, but a flamboyant phrase. Every pedant, every demagogue, every cynic, every hypocrite, every troublemaker, and I am sorry to say, some others of an entirely different character, will try to downgrade them even to the extent of mockery and ridicule.

The 21st century American reality (the Age of Decadence):

"Yeah, I calculated the NPV, and, you know, it's just not worth it for me. I really believe in your project, though. And, I share your passion. Good luck."

The description is of the decline of asabiyyah, as complacency and selfish individualism possess the elites. Yet there is still something superficial about this account.

---

It seems plausible that Duty, Honor, Country are the paramount virtues of a vigorous and rising polity. They are communitarian traits as in Jonathan Haidt's MFT, almost the opposite of the value set of liberalism. Undoubtedly they base themselves on evolutionarily-ancient components of the human psyche, selected for group cohesion.

Pre-capitalist formations, such as the empires of antiquity and those of feudalism, codified and celebrated duty, honour and country/empire. The elites knew they had to hang together or they'd hang separately, given their explicit social position as oppressors. When social solidarity failed, rebellions soon followed. Peter Turchin has written books about this.

In capitalism it's somewhat different. Economic exploitation is hidden behind the veil of equal formal rights for all. Most people sign up to the elite idea that capitalism is not a class society. The elites do not generally rely upon the threat of explicit oppression, but on the atomisation of labour, free to flow where fluid capital requires it (in terms of geography, roles and skills).*

Liberal individualism is the soul of modern capitalism but it doesn't really stir the heart .. and it doesn't glue society together.

---

I'm reading Charles Stross's The Bloodline Feud: The Family Trade and The Hidden Family (Merchant Princes Omnibus Book 1) where the heroine (a feisty tech journalist who is also feudal royalty) attempts to kickstart capitalism in a parallel feudal world. Stross is the pre-eminent writer of economics science-fiction and his book has received plaudits from leading economists such as Paul Krugman.

Amazon link

From a feudal point of view, capitalism as an overarching system looks incredibly weird. The world is run by merchants, who have no interests apart from enlarging their capital again and again?

'What then is life for?' they would ask**.

---

* This is at the root of the 'somewheres' vs. 'anywheres' distinction we saw with Brexit. The Remainers cannot conceive how anyone could be opposed to their vision of a uniform transnational community of right-thinkers embracing an enlightened globalised capitalism; Leavers conversely can't understand why our hard-built and largely pleasant British island community should be subordinated to more powerful European nation-states with their own somewhat inimical interests.


** A quote I once heard: "No-one ever gave their life for IBM."

James Damore gets fired

The rainbow colours of diversity

The ex-Google employee was terminated for observing (in an internal memo) that genders differ in their aptitude for, and interest in, computer science.

• Razib Khan predicted: "whoever wrote that memo will surely be fired".
• Scott Alexander rebutted the Google official line indirectly (wrong about aptitude).
• Greg Cochran simply mused about 'The Index'.

Steve Hsu has this to say, while Scott Aaronson has written an elegant and oblique post.

---

Blank-slate ideology asserts that all humans - of whatever gender and/or race - are cognitively biologically-identical, and that therefore all detrimental differences of outcome are solely due to oppression. When elites defend plainly-wrong ideas with the utmost ferocity, I am reminded of Marx's point, "The ideas of the ruling class are in every epoch the ruling ideas."

Why would the ruling elite want to believe it? Because it would be so dangerously divisive to look at any human subgroup and affirm that group's lower average cognitive performance or competence. Even identifying plainly outperforming groups - such as the Ashkenazim - risks opening the door to the concept of differences. So extreme is this impetus to faux-egalitarianism that many people deny even physical performance differences between genders or races. Social cohesion is the game.

I keep reminding myself (I already wrote about this) that globalisation is the economic driver for blank-slate ideology. Its young, educated beneficiaries are eager converts (arguably against their own longer-term interests) to causes so apparently egalitarian, culturally left-wing .. and economically convenient for globalised capital.

At least we haven't reintroduced the auto-de-fé yet or I wouldn't be writing this.

---

The Google CEO's sophistry was particularly disingenuous: see "If Google did heavy lifting ...".

Smart pets? An *obvious* spinoff ...

From Steve Hsu today at Information Processing:

"Work on cognitive enhancement will probably be done first in monkeys, proving Planet of the Apes prophetic within the next decade or so :-)"

Many (most?) of the alleles implicated in superior cognitive functioning will doubtless have other physiological effects. This 'multiple and diverse impact' of genes is called Pleiotropy.

For this reason - when we have a list of the relevant IQ alleles - we would be ill-advised to shift into genome-editing mode and just flip all those genes to their 'enhancement' variants.

Naturally, there'll be loads of research first on mice, monkeys (expensive) and .. cats.

Link

Contra Wittgenstein, I may yet share my life with a truly biologica Aineko before I die.

---

(Nothing worse than being incessantly nagged by an entitled creature which believes itself master or mistress of the universe and which additionally possesses claws).

"No .. human has touched the edge of the truth of Go"

A minority interest of mine is the game of Go. I hope to study it .. at some point.

“After humanity spent thousands of years improving our tactics, computers tell us that humans are completely wrong,” Mr. Ke, 19, wrote on Chinese social media platform Weibo after his defeat. “I would go as far as to say not a single human has touched the edge of the truth of Go.”

From this unfailingly interesting article (via Marginal Revolution).

---

A lot of people have picked up on this, or related issues:

- Scott Alexander writes about a recent stellar AI conference
- Steve Hsu tells our masters about the future of AI and genetic engineering.

How smart are software developers?

It's quite hard getting figures but based on this raw data, it appears that the average IQ of students applying for PhD programmes at good universities in computer science is around 128.

That sounds right to me. Most software development companies don't take people without good first degrees or a higher degree in a STEM subject - and that's just the pool we're talking about here.

"Logic: Form and Function": J. A. Robinson

John Alan Robinson is a semi-forgotten hero of artificial intelligence (he died four months ago). He discovered the resolution rule (and rediscovered Herbrand's concept of unification) which made automated theorem-proving a computational reality. And he wrote a book which was a classic on how to implement a resolution theorem prover.

Amazon Link

His 1979 book, Logic: Form and Function, is out of print and now only available second-hand. It's sunk into obscurity, as you can see from Amazon's terrible image above.

After some efforts, I have managed to order a copy and will do my best to write a review (Google reports no accessible reviews on the Internet). Here is some background on Robinson (Wikipedia).

"Robinson was born in Halifax, Yorkshire, England in 1930 and left for the United States in 1952 with a classics degree from Cambridge University. He studied philosophy at the University of Oregon before moving to Princeton University where he received his PhD in philosophy in 1956. He then worked at Du Pont as an operations research analyst, where he learned programming and taught himself mathematics.

He moved to Rice University in 1961, spending his summers as a visiting researcher at the Argonne National Laboratory's Applied Mathematics Division. He moved to Syracuse University as Distinguished Professor of Logic and Computer Science in 1967 and became professor emeritus in 1993.

It was at Argonne that Robinson became interested in automated theorem proving and developed unification and the resolution principle. Resolution and unification have since been incorporated in many automated theorem-proving systems and are the basis for the inference mechanisms used in logic programming and the programming language Prolog.

---

Automated theorem proving (ATP) is the engine of classical AI in the way that neural nets are the engine of contemporary AI.

Deep Learning, which makes everything into a classification problem, has superseded brittle, classical AI not least because high-level features emerge from the weight-training process under some optimisation criteria, rather than being hand-crafted by the programmer.

Check this brilliant article from the NYT (via Steve Hsu).

But if your ambitions are somewhat modest, ATP takes you a long way .. and you can do it at home, kids.

---

Here is the table of contents from "Logic: Form and Function".

And finally, here is the text from Chapter 13 where Robinson documents his (very old-fashioned) Lisp code for the hyper-resolution theorem prover (PDF).

My apologies for the poor quality, but it is readable. To make sense out of how it works, I suspect you need to buy the book and to read the preceding chapters.

Today's links

Around the Internet today for your amusement and education.

1. 10 Essentials of Quantum Mechanics

As the title suggests, Backreaction lists ten key points. It's a bit subjective as to which is the greatest misconception corrected here, but I'd vote for number 4.

"4. There is no spooky action at a distance

"Nowhere in quantum mechanics is information ever transmitted non-locally, so that it jumps over a stretch of space without having to go through all places in between. Entanglement is itself non-local, but it doesn’t do any action – it is a correlation that is not connected to non-local transfer of information or any other observable.

"It was a great confusion in the early days of quantum mechanics, but we know today that the theory can be made perfectly compatible with Einstein’s theory of Special Relativity in which information cannot be transferred faster than the speed of light."

2. A spat about increasing human IQ

Why did Scott Alexander (Slate Star Codex), Garrett Jones (Hive Mind), and Razib Khan (GNXP) even bother to alert Steve Hsu to this pernicious post, and who is PZ Myers anyway?

Whatever - Professor Hsu gives a reasoned and educational reply.

"Myers is both confused and insulting in his blog post, but I'll refrain from ad hominem attacks, and just focus on the science.

Myers seems to think that humans with much better cognitive abilities than our own can't exist. Sort of like a farmer in 1957 claiming that chickens that are bigger and faster maturing than his own could not exist [...] . I urge Myers to read some books on population genetics before returning to this discussion.

"The argument for why there are probably genomes not very different from our own, but which lead to much better cognitive ability, is very simple, and I went through it in a post called Explain it to me like I'm five years old [...] ."

Worth following the link even if you're older than five.

3. A spat about the KKK on campus (Indiana University)

Via Breitbart (h/t Steve Sailer).

 'Students be careful, there's someone walking around in kkk gear with a whip.'

"Residential hall advisor Ethan Gill quickly wrote an email to his students, warning them of the “threat” on campus: “There has been a person reported walking around campus in a KKK outfit holding a whip. Because the person is protected under first amendment rights, IUPD cannot remove this person from campus unless an act of violence is committed. Please PLEASE PLEASE be careful out there tonight, always be with someone and if you have no dire reason to be out of the building, I would recommend staying indoors if you’re alone.”

"Later in the evening, Gill was forced to retract his warning on his Facebook page, where he clarified that the purported Klansman was actually just an innocent priest dressed in liturgical garments. The “whip” turned out to be the clergyman’s robe-like belt that was tied around his waist."

Are there no safe spaces anywhere any more? Will no-one think of the children?

Offspring IQ vs parental midpoint IQ

We had Alex and Adrian with us over the Easter and fell to talking .. as you do .. about the correlation between parental IQ and the intelligence of their offspring. I was trying to recall this post, from Steve Hsu, from which I reproduce the key material below.

There are three key ideas in Steve's post:

• You start from the parental midpoint IQ (the average of father and mother)
• To get the mean, or expected value of the offspring IQ multiply by h²  (Breeder's Equation)
• The distribution of offspring IQ has a tightened standard deviation, 12 rather than 15 IQ points.

---

"Assuming parental midpoint of n SD above the population average, the kids' IQ will be normally distributed about a mean which is around +.6n with residual SD of about 12 points. (The .6 could actually be anywhere in the range (.5, .7), but the SD doesn't vary much from choice of empirical inputs.)

"So, e.g., for n = 4 (parental midpoint of 160 -- very smart parents!), the mean for the kids would be 136 with only a few percent chance of any kid to surpass 160 (requires +2 SD fluctuation). For n = 3 (parental midpoint of 145) the mean for the kids would be 127 and the probability of exceeding 145 less than 10 percent.

"No wonder so many physicist's kids end up as doctors and lawyers. Regression indeed! ;-)
 ...
"Assuming bivariate normality (and it appears that IQ has been successfully scaled to produce this), the offspring density function is normal with mean n*h² and variance 1-(1/2)(1+ρ)h², where ρ is the correlation between mates attributable to assortative mating and h² is the narrow-sense heritability. *

"I put h² between .5 and .7. Bouchard and McGue found a median correlation between husband and wife of .33 in their review many years back, but not all of that may be attributable to assortative mating. So anything in (.20, .25) may be a reasonable guesstimate for ρ.
...

"Note: Some people are confused that the value of h² = narrow sense (additive) heritability is not higher than (.5 - .7). You may have seen *broad sense* heritability H² estimated at values as large as .8 or .9 (e.g., from twin studies). But H² includes genetic sources of variation such as dominance and epistasis (interactions between genes, which violate additivity). Because children are not clones of their parents (they only get half of their genes from each parent, and in a random fashion), the correlation between midparent IQ and offspring IQ is not as large as the correlation between the IQs of identical twins."

* Putting in the numbers: σ = 15 √(1 - 0.5 * 1.225 * 0.6) = 12.
___

The other discussion we had was how to estimate the IQ of my parents, now both dead.

Neither ever took an IQ test as far as we know. We have my mother's DNA with 23andMe so eventually with full genome sequencing we may expect to read this off (when the research ...).

We don't have my father's DNA although we might forensically get it one day via his belongings.

Still, all is not lost. Life is an IQ test and we have the biographies. We can also do some reverse correlations from the children, (my brother, sister and myself). IQ test data is not available here either, but we can still make biographical estimates.

AI really is applied neurobiology

When I was researching AI back in the 1980s, we'd all heard of Geoffrey Hinton. He was the key pioneer of artificial neural nets, a field which at the time wasn't making much progress. He never turned up at the main AI conferences where we discussed logics, theorem-proving and symbolic planning programs. A different paradigm entirely.

I never set eyes on him.

Professor Hinton has had the last laugh. His work, and that of close colleagues, underpins the recent AI successes of Google (AlphaGo!), Baidu, Facebook and Microsoft. Whether it's speech recognition, automatic translation or face recognition, deep-learning neural nets are powering it along behind the scenes.

Here's Professor Hinton's recent address to the Royal Society (h/t Steve Hsu). I don't normally find time to watch other people's recommended videos, but I made an exception for this one. Hinton rather reminds me of Richard Dawkins in appearance and style. He's lucid, understated and staggeringly smart. Here he engagingly tells the story of the fall and rise again of the neural net approach to artificial intelligence.

At almost the end of the talk, he puts up this slide for almost two seconds .. and then hides it.

The "secret slide"

I'm sure he just felt it wasn't quite aligned to his audience which didn't seem packed with AI specialists.

There was always a tendency within AI which made a distinction between our language for describing agents as knowing, believing, wanting entities - using intentional, symbolic terms, and the presumed internal agent architecture which caused behaviour - and which need not involve pushing symbols around at all.

We've long been aware of the awesome computation underlying animal/human unconscious situational competences. We've long failed to replicate such abilities using 'Good Old-Fashioned AI'. Perhaps it's time to conclude, with Prof. Hinton, that the architecture which realises such capabilities really has to be that of the deep-learning neural net.

Prof. Hinton is at pains to point out that the most sophisticated current systems fall well short of human brain structure both in terms of quantity of neurons and complexity of interconnection and communication.

We should keep reminding ourselves that brains are doing important stuff at the granularity of small groups of molecules: they are the essence of nanotech.*
___

* In fact, we should probably be using our best AI neural nets to figure out what our own neurons are actually doing. The massive connectivity found in brains may implement features and structures so complex as to be beyond unaided human comprehension.

[In an interesting analogy, you may recall the 'killer app' for quantum computers is said to be the simulation of quantum systems themselves, intractable with conventional computers.]

The aliens amongst us

Ludwig Wittgenstein famously wrote:

"If a lion could speak, we would not be able to understand him". This is on the grounds that language only acquires meaning through a community of speakers using it as part of their 'form of life' (way of life). Hence beings with a radically different way of life would not be able to make sense of the others' utterances."

I am unconvinced. The lion inhabits the same spatio-temporal world as we do, lives in a similar planetary environment, has the same mammalian drives. We share all that stuff. No, we'd understand the talking lion only too well.

The real aliens are the ones we're building.

Back in the dawn era of artificial intelligence, researchers worked on chess endgames. They built a database storing every possible play in the final 20-30 moves, marking those which led to a win. Their AI program simply looked in the database to find the optimal response to any human move.

It was a curious experience playing against such a program. The computer made moves that seemed incomprehensible, which defied understanding, but which by some strange alchemy led inexorably to its victory. It was impossible to understand what the program was thinking of.

If that program could have talked, all it would have said was:

"My last move was marked optimal in my database of quite a few possible moves."

It would have had to say that sentence every time, and its human opponent's understanding would never have improved. Donald Michie coined the term 'Human Window' - and these programs were outside it.

These days we do better. Our AI programs no longer look up their actions in mammoth, static databases - they learn features, they chunk the phenomena.

Chunking it may be, but not as we know it.

Here's "fhe" on Hacker News:

"When I was learning to play Go as a teenager in China, I followed a fairly standard, classical learning path. First I learned the rules, then progressively I learn the more abstract theories and tactics. Many of these theories, as I see them now, draw analogies from the physical world, and are used as tools to hide the underlying complexity (chunking), and enable the players to think at a higher level.

"For example, we're taught of considering connected stones as one unit, and give this one unit attributes like dead, alive, strong, weak, projecting influence in the surrounding areas. In other words, much like a standalone army unit.

"These abstractions all made a lot of sense, and feels natural, and certainly helps game play -- no player can consider the dozens (sometimes over 100) stones all as individuals and come up with a coherent game play. Chunking is such a natural and useful way of thinking.

"But watching AlphaGo, I am not sure that's how it thinks of the game. Maybe it simply doesn't do chunking at all, or maybe it does chunking its own way, not influenced by the physical world as we humans invariably do. AlphaGo's moves are sometimes strange, and couldn't be explained by the way humans chunk the game.

"It's both exciting and eerie. It's like another intelligent species opening up a new way of looking at the world (at least for this very specific domain). and much to our surprise, it's a new way that's more powerful than ours."

Steve Hsu sardonically quotes  DeepMind CEO Demis Hassabis:

"Over the summer DeepMind will look at the internal representations used in the valuation engine to see how they correspond to expert human intuitions about Go."

"This is like peeking into the mind of an alien creature that evolved fighting for territory in a 2D world with discrete spacetime :-)"

If we design minds which induce features from spaces which do not share our human geography and agency, their mental concepts will massively fail to intersect with our own. We will have no referents for their words; it will be like talking general relativity to a four year old.

If such an AI could speak, we would not be able to understand it.*

---

* You might be inclined to say, '... without considerable effort.'

  I might agree with you .. up to a point.

Some really good-bad books

Fay Weldon writes one of my favourite classifications:

"In the nineties, moved by the new GOO (‘GOOd read’) classification in Camden public libraries, I devised my own classification scheme as follows.

Good-good books – the best of contemporary literary novels, plus classics which were best-sellers in their day and have withstood the passage of time: all engaging with the intellect, if ‘difficult’.

Bad-good books – pretentious and dreadfully boring, yet taken seriously by the occasional reviewer (usually a friend of the author) and funded by the Arts Council.

Good-bad books – intensely readable, unpretentious and seldom reviewed.

Bad-bad books – worthy only to be hurled into the corner or dropped in the bath."

Sometimes you just need to switch off your intellect and enjoy a piece of page-turning escapism. If you are a science-fiction fan, you could do considerably worse than turn to B. V. Larson, for example his Undying Mercenaries series of exceedingly good-bad books.

The first volume ...

Larson also writes insightfully about the craft of successful, self-published writing ("Advice Concerning the Self-Publishing Game ").

---

Steve Hsu has an interesting post on the interpretation of quantum mechanics (PDF here). Basically he does a good job of explaining where the many worlds interpretation comes from, and how decoherence plays a critical role in 'separating branches'. He then sheds a really clear light on the issue of 'where do the Born Rule probabilities come from?' concluding this is still an unresolved problem.

As a bonus, the comments feature the infamous Luboš Motl who is coaxed to clearly explain his own views on quantum reality. As a hard-line 'Copenhagenist', Motl appears to believe that reality is both ill-defined and lacks objective (classical-ish) reality in the absence of sentient observers - a highly counter-intuitive view for sure! Here is how he explains things:

"The conceptually right to describe a world without sentient beings is that an unspecified and unknown initial wave function evolves unitarily according to Schrödinger's equation and never collapses because it's only measurements that may collapse and there are none in your theory. The complete "diffusion" of the wave function (into the linear superposition of dead and alive cats and all objects, small and big, in the most general superpositions of all conceivable states) may be said to be a problem - but another problem is that the initial state is totally unknown, too.

"It makes no sense to say that the initial wave function is a particular thing because one may only say that the wave function is a particular thing [if] something is [a] measurement - if a sentient being becomes aware of the result of some measurement. This is not happening in a universe without sentient beings. So there's no specific science to discuss in a universe without sentient beings at all. The laws may still be the same as they are in our world but they won't be applied in any particular situation because there are no particular situations or particular special wave functions in a world where no one ever measures anything.

"Einstein asked whether there is any Moon over there if no one looks. In practice, classical physics is a good enough approximation, so one may assume that the Moon is pretty much there even before observers look etc. But conceptually, if you care about similar objects for which the quantum effects are strong, the right answer is that the Moon just isn't at any particular location and has no other particular properties if no one looks. The wave function isn't a real object of any type. Its amplitudes can't be measured in a single repetition of the situation. It is only a template storing information allowing to predict probabilities of things that actually can be measured - the observables."

This is all accessible to anyone who has taken and understood QM at an undergraduate level.

Tom Wolfe (1971) - on intimidation and shake-downs

Steve Hsu points to this Tom Wolfe piece from 1971: ' Mau-Mauing the Flak Catchers'*.

"... Wolfe's book is set at the Office of Economic Opportunity in San Francisco which was in charge of administering many of the anti-poverty programs of the time. Wolfe presents the office as corrupt, continually gamed by hustlers diverting cash into their own pockets. The essay centers on the irony of these failed programs fortifying not the diets but the resentment and contempt of the Black, Chicano, Filipino, Chinese, Indian, and Samoan communities of San Francisco.

"Wolfe describes hapless bureaucrats (the eponymous Flak Catchers) whose function was reduced to taking abuse, or "mau-mauing" (in reference to the intimidation tactics employed in Kenya's anti-colonial Mau Mau Uprising) from intimidating young Blacks and Samoans, who are seen as reveling in the newfound vulnerability of "the Man".

The flak-catchers smile pathetically, allowing their tormentors to indulge themselves in abuse; the process is seen as a farcical but useful expedient, condescending toward the resentment of these communities. He described one mau-mauer who would show up at the offices and hand over ice-picks, switch-blades and straight-razors that he said were taken from gangs, in exchange for payments from the program. As a result, much of the money of these programs was not reaching its intended recipients, rendering the programs largely ineffective."

Tom Wolfe is, of course, a great writer (although this piece is too long). He seems devoid of political correctness, and his writing is energetic, and great at painting a scene. What I wondered was how on earth he managed to get all this detail (names, places, events, back stories - all fly-on-the-wall stuff).

Did he sit in the 'Office of Economic Opportunity' interview room cataloguing the horrors? Did he wander the ghettos watching pimps and dealers up close and personal?

Tom Wolfe is an effete, middle-class white man who, one imagines, wouldn't last two minutes in a tough 'hood.

Tom Wolfe: author and 'ghetto expert'

I'm often reminded how economics, despite its many flaws, is still the best of that sorry bunch of subjects, the social sciences. In the anti-poverty programmes one sees the essence of public choice theory applied to special interests.

---

* The Tom Wolfe piece is here - shame about the intrusive ad at the bottom of the page. And the poor formatting. It reads much better on a tablet.

"Twenty Two!"

You will be surprised to discover that occasionally Clare and myself have arguments: politics usually. I think it's fair to say that emotions are sometimes engaged and we get a little red-faced. But now, if I come close to winning, Clare defiantly spits back: "Twenty Two!".

Back-up to this report from OKCupid. Christian Rudder, the 39-year-old president and co-founder of the online dating site, has written a book, “Dataclysm: Who We Are When We Think No One’s Looking”. The book describes trends he's found analysing OKCupid's enormous database.

An example: here's how straight women rate the men on OKCupid based on their age. Rudder reports: “Women who are, say, 28 find guys who are also 28 about the most attractive, and so forth. Up until about 40, when that’s getting too old.”

The dotted line is the 'equal age' line. You can see that young women find men a few years older more attractive. Around 29-30 it all evens out. As women get into their forties, they want a younger specimen, a moderately-toy boy.

Now, you will naturally be interested in the similar chart for men. As men age, what is the age of the women they find most attractive? And this data is, of course, also available.

The feminist rests her case.

---

Via Steve Hsu.

A Modest Proposal

Success in life strongly correlates with intelligence. Mass university education for both sexes has led to mass assortative mating, the smart hitching up with the smart. They transition into cognitively demanding jobs such as higher management, finance, accountancy, the law, architecture, IT, science and engineering where pay and prospects are good. Smart couples have smart offspring (the heritability of intelligence is high). In consequence, we are in danger of breeding a stable, stratified society with an intellectually-gifted upper class and a rather dim, under-employed lower class.

Some people don't care about this. Others do care, but don't see how we can avoid it. But Toby Young has a rather bizarre idea as to how to fix it. You should read the whole of his rather long article (here, via Steve Hsu who is name-checked) but the essential proposition is this:

"I’m thinking in particular of the work being done by Stephen Hsu, Vice-President for Research and Professor of Theoretical Physics at Michigan State University. He is a founder of BGI’s Cognitive Genomics Lab. BGI, China’s top bio-tech institute, is working to discover the genetic basis for IQ. Hsu and his collaborators are studying the genomes of thousands of highly intelligent people in pursuit of some of the perhaps 10,000 genetic variants affecting IQ. Hsu believes that within ten years machine learning applied to large genomic datasets will make it possible for parents to screen embryos in vitro and select the most intelligent one to implant. Geoffrey Miller, an evolutionary psychologist at New York University, describes how the process would work:

Any given couple could potentially have several eggs fertilized in the lab with the dad’s sperm and the mom’s eggs. Then you can test multiple embryos and analyze which one’s going to be the smartest. That kid would belong to that couple as if they had it naturally, but it would be the smartest a couple would be able to produce if they had 100 kids. It’s not genetic engineering or adding new genes, it’s the genes that couples already have.

It’s worth repeating this last point, because it deals with one of the main reservations people will have about this procedure: these couples wouldn’t be creating a super-human in a laboratory, but choosing the smartest child from the range of all the possible children they could have. Nevertheless, this could have a decisive impact. “This might mean the difference between a child who struggles in school, and one who is able to complete a good university degree,” says Hsu.

My proposal is this: once this technology becomes available, why not offer it free of charge to parents on low incomes with below-average IQs? Provided there is sufficient take-up, it could help to address the problem of flat-lining inter-generational social mobility and serve as a counterweight to the tendency for the meritocratic elite to become a hereditary elite. It might make all the difference when it comes to the long-term sustainability of advanced meritocratic societies.

At first glance, this sounds like something Jonathan Swift might suggest and, of course, there are lots of ethical issues connected with “designer babies”. But is it so different from screening embryos in vitro so parents with hereditary diseases can avoid having a child with the same condition? (This is known as a pre-implantation genetic diagnosis.) I don’t mean that a low IQ is comparable to a genetic disorder like Huntington’s, but if you allow parents to choose which embryo to take to term, whatever the reason, you’ve already crossed the Rubicon. And screening out embryos with certain undesirable genes is legal in plenty of countries, including Britain.

In an article for Nautilus, Stephen Hsu argues that making this new technology widely available will be essential to prevent it being exploited by the privileged few, thereby exacerbating inequality:

Almost certainly, some countries will allow genetic engineering, thereby opening the door for global elites who can afford to travel for access to reproductive technology. As with most technologies, the rich and powerful will be the first beneficiaries. Eventually, though, I believe many countries will not only legalize human genetic engineering, but even make it a (voluntary) part of their national healthcare systems. The alternative would be inequality of a kind never before experienced in human history.

Hsu isn’t being paranoid. Some rich people, like the movie star Jodie Foster, have already used artificial insemination to try and maximise their children’s IQs utilising the sperm of Nobel Prize-winners. If high-achieving couples in London, Paris and New York are prepared to make their children listen to Mozart in the hope of boosting their intelligence, even though there’s no evidence it has any effect, they wouldn’t hesitate to make use of a technology that actually worked.

Hsu’s solution is to make it freely available to everyone, but that would only help to prevent it making existing inequalities even worse. After all, if people from all classes used it in exactly the same proportions, all you’d succeed in doing would be to increase the average IQ of each class, thereby preserving the gap between them. Wouldn’t it be better to limit its use to disadvantaged parents with low IQs? That way, it could be used as a tool to reduce inequality.

This technology might actually be more effective than anything else we’ve tried when it comes to tackling the issue of entrenched poverty, with the same old problems—teenage pregnancy, criminality, drug abuse, ill health—being passed down from one generation to the next like so many poisonous heirlooms. In due course, why not conduct a trial in a city like Detroit and see if it works? It has become a cliché to point out that the disadvantages of being brought up in a low-income family are apparent when a child is as young as eighteen months, so it shouldn’t take long to see if increasing the IQs of children from deprived backgrounds makes an impact. It would be inexpensive, too, so wouldn’t involve a massive hike in taxation. “The cost of these procedures would be less than tuition at many private kindergartens,” says Hsu.

In a sense, what I’m suggesting is a form of redistribution, except the commodity being redistributed is above-average intelligence rather than wealth. This is a way of significantly reducing end-state inequality that should be acceptable to conservatives (at least, non-religious conservatives) because it doesn’t involve the use of coercive state power. Participation would be entirely voluntary. Let’s call this policy “g-galitarianism”. (For those unfamiliar with the jargon, “g” is commonly used by psychologists and geneticists to stand for “general factor of cognitive ability” and is often used as a synonym for “IQ”. It was first given this designation by Charles Spearman, a British army officer, at the turn of the last century.)

A lot of the resistance to this idea will come from a visceral dislike of anything that smacks of eugenics, for understandable historical reasons. But the main objection to eugenics, at least in the form it usually takes, is that it involves discriminating against disadvantaged groups, whether minorities or people with disabilities. What I’m proposing is a form of eugenics that would discriminate in favour of the disadvantaged. I’m not suggesting we improve the genetic stock of an entire race, just the least well off. This is a kind of eugenics that should appeal to liberals - progressive eugenics."

Young concludes:

"Liberals who deny that there’s any genetic basis to intelligence shouldn’t have a problem with this trial since, according to their logic, the in vitro procedure I’m proposing won’t have any effect on IQ. That doesn’t mean they won’t object, of course. Herrnstein points out this inconsistency in the Appendix to IQ in the Meritocracy: “Thus, the very same people one day abhor the idea of tampering with people’s genes may, the next day, vigorously deny the conclusion that human society involves genetic factors.”

I suspect that Young has his tongue firmly in cheek here. A technique which is guaranteed to give you the smartest kids you can feasibly produce? The middle and upper classes will be all over it!

'Super-Intelligent Humans Are Coming' - be a little concerned

Last year Steve Hsu wrote an article entitled: "Super-Intelligent Humans Are Coming". Here's an excerpt:

"Genetic engineering will one day create the smartest humans who have ever lived.

"Lev Landau, a Nobelist and one of the fathers of a great school of Soviet physics, had a logarithmic scale for ranking theorists, from 1 to 5. A physicist in the first class had ten times the impact of someone in the second class, and so on. He modestly ranked himself as 2.5 until late in life, when he became a 2. In the first class were Heisenberg, Bohr, and Dirac among a few others. Einstein was a 0.5!

"My friends in the humanities, or other areas of science like biology, are astonished and disturbed that physicists and mathematicians (substitute the polymathic von Neumann for Einstein) might think in this essentially hierarchical way. Apparently, differences in ability are not manifested so clearly in those fields. But I find Landau’s scheme appropriate: There are many physicists whose contributions I cannot imagine having made.

"I have even come to believe that Landau’s scale could, in principle, be extended well below Einstein’s 0.5. The genetic study of cognitive ability suggests that there exist today variations in human DNA which, if combined in an ideal fashion, could lead to individuals with intelligence that is qualitatively higher than has ever existed on Earth: Crudely speaking, IQs of order 1,000, if the scale were to continue to have meaning.

"... Does g predict genius? Consider the Study of Mathematically Precocious Youth, a longitudinal study of gifted children identified by testing (using the SAT, which is highly correlated with g) before age 13. All participants were in the top percentile of ability, but the top quintile of that group was at the one in 10,000 level or higher. When surveyed in middle age, it was found that even within this group of gifted individuals, the probability of achievement increased drastically with early test scores. For example, the top quintile group was six times as likely to have been awarded a patent than the lowest quintile. Probability of a STEM doctorate was 18 times larger, and probability of STEM tenure at a top-50 research university was almost eight times larger. It is reasonable to conclude that g represents a meaningful single-number measure of intelligence, allowing for crude but useful apples-to-apples comparisons.

"... Once predictive models are available, they can be used in reproductive applications, ranging from embryo selection (choosing which IVF zygote to implant) to active genetic editing (for example, using CRISPR techniques). In the former case, parents choosing between 10 or so zygotes could improve the IQ of their child by 15 or more IQ points. This might mean the difference between a child who struggles in school, and one who is able to complete a good college degree. Zygote genotyping from single cell extraction is already technically well developed, so the last remaining capability required for embryo selection is complex phenotype prediction. The cost of these procedures would be less than tuition at many private kindergartens, and of course the consequences will extend over a lifetime and beyond.

"The corresponding ethical issues are complex and deserve serious attention in what may be a relatively short interval before these capabilities become a reality. Each society will decide for itself where to draw the line on human genetic engineering, but we can expect a diversity of perspectives. Almost certainly, some countries will allow genetic engineering, thereby opening the door for global elites who can afford to travel for access to reproductive technology. As with most technologies, the rich and powerful will be the first beneficiaries. Eventually, though, I believe many countries will not only legalize human genetic engineering, but even make it a (voluntary) part of their national healthcare systems.

"The alternative would be inequality of a kind never before experienced in human history."

Steve Hsu focuses on the additive nature of most genes:

"In plant and animal genetics it is well established that the majority of phenotype variance (in complex traits) which is under genetic control is additive. (Linear models work well in species ranging from corn to cows; cattle breeding is now done using SNP genotypes and linear models to estimate phenotypes.) There are also direct estimates of the additive / non-additive components of variance for human height and IQ, from twin and sibling studies. Again, the conclusion is the majority of variance is due to additive effects.

"There is a deep evolutionary reason behind additivity: nonlinear mechanisms are fragile and often "break" due to DNA recombination in sexual reproduction. Effects which are only controlled by a single locus are more robustly passed on to offspring. Fisher's fundamental theorem of natural selection says that the rate of change of fitness is controlled by additive variance in sexually reproducing species under relatively weak selection.

"Many people confuse the following statements:

"The brain is complex and nonlinear and many genes interact in its construction and operation."

"Differences in brain performance between two individuals of the same species must be due to nonlinear effects of genes."

"The first statement is true, but the second does not appear to be true across a range of species and quantitative traits."

The brain is a computational system of somewhat bounded size. There are two ways its performance can be improved: firstly by making neurons and neuron connectivity more efficient, for example, increased speed of operation, better insulation. This doesn't necessarily increase the size requirements of the brain.

The second is by increasing the number of neurons dedicated to certain tasks, for example abstract reasoning. Given a bounded skull size, this necessarily decreases the number of neurons available for other tasks. I don't think it's an accident that theoreticians are often clumsy, or that great athletes are often not renowned as deep conceptual thinkers.

Because of these trade-offs, the quest for greater intelligence is likely to raise all boats while issues in the first category are addressed (we'd all benefit from removing mistakes and inefficiencies in our neuronal blueprints) but will accentuate stereotypical nerdism and 'absent-minded professor' syndrome as we tweak alleles in the second category (not so good).

Do we know which alleles code for which category of enhanced intellectual performance? No.

"What's my IQ?" (You ask yourself)

Lots of people haven't taken IQ tests, or maybe they did but their company declined to share the results. But calm yourselves: there is a way. Recall that one of the major utilities of IQ test results is that they correlate strongly with competence in occupations requiring intelligence. So naturally, you can reverse-engineer your IQ from your occupation or education, or at least establish a personal floor level.

Let's start with university. In the UK, along with many industrialised societies, something like 40% of the cohort go to university. This sets the entrance level at the 60th percentile. Looking up the value of 0.1 (corresponds to 0.6 = 0.5 + 0.1) in a normal distribution table tells us that this occurs at 0.25 standard deviations to the right of the mean. Since IQ is normed at mean = 100, standard deviation = 15, the minimum IQ to get into university these days is about 104.

Obviously if this is your IQ you're going to be doing media studies or drama, not economics, mathematics or physics. It's a floor.

Now we turn to another useful chart, this time from Steve Hsu via Lubos Motl. I reproduce part of Lubos' rather satirical post below. Steve Hsu's original data was GRE (Graduate Record Examinations) scores, which he then translated into IQ scores. The GRE are taken by most American university graduates seeking to do post-graduate study to PhD level. Anyway, here's Lubos:

"Steve Hsu has found a very interesting table with the average GRE scores computed for various concentrations. He has also defined a linear map translating the average V-Q-A scores into a more familiar IQ scale. This convention looks natural to me and I will follow his scale although it is not guaranteed that it is equally calibrated as other IQ measurements.

"Disclaimer: these cold numbers expressing typical IQ for different occupations must be interpreted very carefully. They don't necessarily imply anything. The outcome depends on the character of the question, discrimination, etc. Despite different numbers, all of us are equal. Blah blah blah. And so on.

"The results are: [A PhD student's typical IQ for this subject]

130.0 Physics
129.0 Mathematics
128.5 Computer Science
128.0 Economics
127.5 Chemical engineering
127.0 Material science
126.0 Electrical engineering
125.5 Mechanical engineering
125.0 Philosophy
124.0 Chemistry
123.0 Earth sciences
122.0 Industrial engineering
122.0 Civil engineering
121.5 Biology
120.1 English/literature
120.0 Religion/theology
119.8 Political science
119.7 History
118.0 Art history
117.7 Anthropology/archeology
116.5 Architecture
116.0 Business
115.0 Sociology
114.0 Psychology
114.0 Medicine
112.0 Communication
109.0 Education
106.0 Public administration"

Since these are graduate entry levels, they're creaming off the top end of undergraduates. So it's kind of an 'undergraduate good 2:1' floor we're seeing here. Oh, and scary & surprising about US doctors, don't you think? In the UK, the average IQ of medical students is reckoned to be 125 (page 61).

At the very top-end, Steve Hsu has observed:

"I doubt that the average among eminent scientists (averaging over all fields) is 160; probably a bit lower like 145."

Finally I draw your attention to "Smart Fraction" Theory, which examines the notion that to run a modern, complex, industrialised society requires a mass of people with (verbal) IQs greater than around 106. Our best societies today have a smart fraction just under 50% (see graph at the bottom of the linked article).

Three good pieces

1. Best review of Iain M. Banks's Culture sequence

Matt Hilliard, a smart and insightful reviewer at Strange Horizons wrote this piece on his blog: the best assessment of what on earth Banks was trying to do with his Culture novels. His take on Peter Watts' Echopraxia is equally superior.

2. Steve Hsu's popular and accessible article on genetic engineering for super-intelligence

Steve Hsu in action

"Super-Intelligent Humans Are Coming - genetic engineering will one day create the smartest humans who have ever lived".

'Stephen Hsu is Vice-President for Research and Professor of Theoretical Physics at Michigan State University. He is also a scientific advisor to BGI (formerly, Beijing Genomics Institute) and a founder of its Cognitive Genomics Lab.'

His blog is here.

3. What makes the quadcopter design so great for small drones?

Remember all those camera drones with four rotors arranged in a square? Amazon's PR stunt for drone delivery? Why do we still go for helicopters in the large then? Is the future the quadcopter?

Here's why not.

A helicopter drone

Thanks to Jess Riedel's blog for the last two links.
---
Oh, and here's a laugh-out-loud bonus from The War Nerd.

Take me to Istanbul ..

Smart genes from stupid people?

Intelligence, as measured by IQ, is largely - but not completely - under genetic control. For example, this from Wikipedia:

"There are some family effects on the IQ of children, accounting for up to a quarter of the variance. However, adoption studies show that by adulthood adoptive siblings aren't more similar in IQ than strangers, while adult full siblings show an IQ correlation of 0.6.

"Conventional twin studies reinforce this pattern: monozygotic (identical) twins raised separately are highly similar in IQ (0.86), more so than dizygotic (fraternal) twins raised together (0.6) and much more than adoptive siblings (~0.0)."

This means that when you meet someone smart they're likely to have the benefit of a good set of genes (for IQ) and some lucky environmental input. Apart from basic stuff like good nutrition, lack of serious diseases and a non-traumatic upbringing, no-one is too clear what a 'lucky environment' actually is. However, only the genes-part makes it through to the next generation so without the luck the kids are likely (but not certain) to be a little less bright than their luckier parent.

Equally, someone a little bit dim is likely to have been dealt a poor hand in the genes-for-high-IQ department, but also to have had ill-luck environmentally speaking. Their kids will not inherit the bad luck, on average, and will therefore tend to be brighter.

This phenomenon is called regression towards the mean.

We can put some numbers behind all this, as Steve Hsu explains. Recall that IQ is conventionally standardised (for European Caucasians) as mean = 100 and standard deviation (SD) = 15 IQ points.

"Assuming parental midpoint of n SD above the population average, the kids' IQ will be normally distributed about a mean which is around +0.6n with residual SD of about 12 points. (The .6 could actually be anywhere in the range (.5, .7), but the SD doesn't vary much from choice of empirical inputs.)

"So, e.g., for n = 4 (parental midpoint of 4 x 15 = 160 -- very smart parents!), the mean for the kids would be 100 + 0.6 x 4 x 15 = 136 with only a few percent chance of any kid to surpass 160 (requires +2 SD fluctuation). For n = 3 (parental midpoint of 145) the mean for the kids would be 127 and the probability of exceeding 145 less than 10 percent.

"No wonder so many physicist's kids end up as doctors and lawyers. Regression indeed! ;-)"

This gives us a chance to revisit Cinderella.

Prince Charming is, frankly, not the sharpest tool in the box. More a lover than a fighter, his IQ is around 105 - below average for someone of his royal socio-economic status.

Cinders, naturally, lives amongst the dregs of society. She is part of half the population with below average IQ. In fact, when she was tested, her IQ was only 90. However, she is known as the pretty-but-dim one of the family so there is hope that she just got unlucky: perhaps those evil step-sisters deprived her of essential nutrients as a child.

So, prince and poor-girl get together. What can we say about the likely IQ of their kids?

Their parental midpoint IQ is (105 + 90)/2 = 97.5 which is 2.5/15 = 1/6 of a standard deviation below population average: this is the 'n' in the formula above. The mean IQ for their kids is therefore predicted to be 100 - 0.6 x (1/6) x 15 = 98.5 with a standard deviation of 12 IQ points. There is a 45% chance that a child of this fairytale marriage will have an IQ of 100+.

Reasonable odds of not being a dreg .. even a royal dreg!

---

Table of IQ requirements for various jobs (not including royalty).