Thursday, March 31, 2016

Nuke 'em!

From Wikipedia:
"Gregory M. Cochran (born 1953) is a physicist and adjunct professor of anthropology at the University of Utah, known for hypotheses in evolutionary medicine and genetic anthropology. He argues that cultural innovation resulted in new and constantly shifting selection pressures for genetic change, thereby accelerating human evolution. He is co-author of the book 'The 10,000 Year Explosion'."
In a typically curmudgeonly post, Cochran imagines a robust response to jihadists, over at West Hunter.
"Now and then I contemplate the possible outcomes if the United States got really, really angry, say at jihadists, if they went too far and struck a nerve. Crazed fury. Jihadists seem to think that enraging the western powers is strategically sensible, but they might be wrong. I’m not talking about little things like the invasion of Iraq or Afghanistan: no, I mean really angry. You should picture Uncle Sam turning green and bursting out of his Uncle Sam suit ... .

"There’s the old reliable, nuclear weapons. There wouldn’t be a lot left of the Arab world, especially when you consider little tactical enhancers like blowing up the Aswan Dam, or nuking a nuclear reactor. You could simply drop enough medium-life-time radioactive dust to make a region uninhabitable for months, or years, or decades, which could make the Haj pretty difficult.

"Even semi-conventional war could become lot more intense: we haven’t done fire raids lately, but we still can (B52s can carry a huge payload). It’s hard to make laser weapons work for most purposes (atmospheric transmission) – but it’s easy to make ones that blind. We’re working on smart bullets: right now you have to fire thousands per hit, but that’s going to change.

"Nerve gas? effective.

"Germ warfare? Amazing things are now possible: we could probably tailor agents to hit particular ethnic groups (there is always leakage: I’m not saying that we wouldn’t get our hair mussed.) . They could kill – swiftly, or agonizingly slowly, They might trigger Creutzfeldt–Jakob disease: not just for cannibals, anymore. You might see agents that cause insanity, or sterility, or damn-foolishness (hmmmmm).

"Once CRISPR goes to war, you would see agents that cause germ-line genetic changes – nasty changes with built-in genetic drive that spread to the whole population."
I suppose we should add this to the debate about what to do about the new Caliphate. Peter Turchin has a series of articles on the same issue, applying his mathematical-history theory of 'cliodynamics'.

Perhaps understandably, Turchin is less keen on “ethnocide”.

As Turchin points out in the relevant article, world - and European - history is chock-full of examples of the kind of stuff Prof. Cochran playfully advocates. But the West doesn't seem to do that anymore, mostly because elite opinion has been captured by empathic compassion.  This almost-Buddhist ideology (bien-pensant liberalism to its enemies) is a natural fit to the rather weak government model of capitalist democracies, where it's important to keep all interest groups onside to ameliorate civil strife and get re-elected.

Government policy trundles along the vector sum of pressure group lobbying; the masses grudgingly acquiesce in the frequent stupidities which follow.

Until they don't.

Wednesday, March 30, 2016

Diary: foot problem + Game of Thrones + AlphaGo + genetics

I vaguely recall jamming my foot against the side of the bed while hoovering last week. By Easter Sunday the entire right foot was red, the skin taut and swollen over the toe joint and I was limping around in socks. The last few days have seen a slow improvement but I think it may be the weekend before I can claim to be back to normal. Some observations.
  1. I haven't taken any pain medication, believing that pain is a signal which I do well to heed and analgesics probably mess up self-healing. I recognise I may be in a minority on this.

  2. The body's response to damage seems to affect many aspects of its functioning, not just the topical region. I feel a bit tired and - strangely - a bit more relaxed, like I'm let off worrying about stuff. Interesting.

  3. Doctor Google was reassuring. The alternative suggestion of gout was excluded - no causal pathway from over-indulgence in Easter eggs.

A Game of Thrones (book vol 1) is highly addictive, once you've printed off the family trees of the various noble houses. It reminds me how natural the ties of family, personal loyalty and honour are, and how alien the cool, depersonalised, transactional styles of modern urban capitalism. No wonder American politicians and business people engaged with negotiators from traditional societies talk past each other in mutual incomprehension.

What would the noble protagonists of GoT make of "The Martian", which we saw on DVD Saturday evening?
The hero is some kind of insolent, wise-cracking, artisan-monkey who refuses to die quietly on Mars as he should. His liege-lord commander shows weakness by beating herself up over leaving him (she did her duty: so problem?).

The world actually cares about this minion, and the powers-that-be indulge their idiotic sentimentality. Could never happen.

Thank the Gods it's only far-fetched speculative fiction.

I thought this was a good article about AlphaGo - assessing its significance now the dust has settled a bit.


This from Professor Greely, Director at the Center for Law and the Biosciences, Stanford University, as set out in his book, 'The End of Sex and the Future of Human Reproduction'.
“In 20 to 40 years, when a couple wants a baby, he’ll provide sperm and she’ll provide a punch of skin,” Prof Greenly told The Times.

"He said the female skin sample will be used to create stem cells, which can in turn be used to create eggs.

"These eggs can then be fertilised with the sperm cells, resulting in a selection of embryos.

"Prof Greenly predicts these embryos will be studied for any signs of malady.

“The prospective parents will be told, ‘These five have really serious diseases, you don’t want them’.

"Of the other 95, they will be given the pluses and minuses,” he said.

"He said that after weighing up the prospective advantages and disadvantages of the healthier embryos, the parents will choose one to be implanted into the woman, which will become their child.

“Parents will get the embryos grouped by categories,” Prof Greenly said.

“One category will be very severe, untreatable, nasty diseases. This will affect one to two per cent of embryos.

“Another category will be other diseases.

“The third is cosmetics: hair, eyes, shape, whether the hair goes white early. We don’t know much about this yet, but we will.

“A fourth category is behavioural. I think here information will be limited. We won’t be able to say, ‘This child is in the top one per cent of intelligence’. We probably will be able to say, ‘This child has a 60 per cent chance of being in the top half’.”
He's being judiciously careful here. In 20 years time we'll be able to read off from the genome both IQ and personality type more accurately than current psychometric testing can.

Tuesday, March 29, 2016

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 h (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.

Thursday, March 24, 2016

Boaty McBoatface is racist

Oops! My inner SJW just there. That and a weakness for clickbait.

When I first encountered this story, like everyone else I was spontaneously LOL - almost ROFL. And then, being the pathetic obsessive that I am, I fell to wondering just why it was so funny.

Saga Briggs describes humour as an “incongruity that is recognised and resolved in some way.” Other writers emphasise a situation vaguely established as threatening, but then reframed as benign. We're all familiar with black or gallows humour.

"Boat Boatface" would not really have been that amusing as the proposed name of a polar research vessel. Adding the suffix -y - Boaty - lends an air of indulgent affection, as if to a child. The ship has been contextualised as cuddly.

"Boatface" is an insult, the direct opposite of the cuddlyness of "Boaty" - but the real genius (for those immersed in English culture) is the "Mc" prefix, which makes the persona of the boat Scottish .. and therefore glum and dour.

The cognitive dissonance is complete: a soppy, cuddly prefix for the boat's first name, immediately followed by an insult suggesting the diametric opposite.

The overall effect is to ridicule and wrong foot the efforts of the vessel's pompous owner, presumably looking for "Endeavour" or "Polar Explorer" or something equally bland dignified.

We feel vaguely threatened by being invited to jump through hoops by authority. "Boaty McBoatface" subverts that pretty comprehensively: at the expense of a shocking racist stereotype of the Scotsman, who as we all know ... .

No to racist boat names, I say.

Wednesday, March 23, 2016

My automated PA (© Google Next)

Because I review science-fiction books on Amazon, new authors email me asking for a review of their latest work. It's a bit embarrassing knowing how to reply, but luckily I'm a beta-user of Google's next generation personal assistant (Google Next).

Since Google already knows so much about me, it has sensed my disquiet and has learned how to respond on my behalf, like a good PA should. I'm completely hands-off.

Here's an example of a (slightly anonymised) note I received yesterday, and the automated reply.

Hi Nigel,

I saw that you offer a lot of great reviews on Amazon. I'm the author of "V NZ FYRRCYRFF: FVZ 811", a newly released YA Science Fiction novel. I would love to send you a free copy of the ebook in exchange for your honest review.

If you are interested in reading and reviewing my book, I can send you the ebook as a gift through Amazon, as a mobi or as a pdf file. You can also download it for free through the Kindle Unlimited program.

If you're like me, you have a backlog of books you want to read. I'm okay if it takes you months to get to my book and review it.

Thanks for your consideration,


Author Wbuna Gjvff

----  And here is Napa's reply ---

Dear Wbuna,

Thanks for your recent email. You asked if Nigel Seel would review your science fiction book in exchange for a free copy. Please note that Nigel Seel does not check any books in exchange for free copies because he believes that this puts him under an implicit obligation to be more pleasant than he is normally. If he likes the tone of the book, so he can buy it anyway and review it then. He is unlikely to tell you, but you can see it on Amazon.

Nigel compliments you. On your marketing skills and wishes you every success with your book. It's a crowded market and he knows that a successful author to be very hard work.

Best wishes,

Nigel’s automated personal assistant
Powered by Google Next © 2018.

Tuesday, March 22, 2016

The science of volcanic prediction

I'm refreshing and improving my knowledge of Bayesian Statistics, working through this OU text available at Amazon.

In the first section I came across this:

which I guess makes those unfortunate scientists candidates for a Darwin Award.

A polygenic risk for autism spectrum disorders

The full title of the paper in the news today is "Genetic risk for autism spectrum disorders and neuropsychiatric variation in the general population" authored by a cast of thousands. It's published in Nature Genetics - here's the abstract.
"Almost all genetic risk factors for autism spectrum disorders (ASDs) can be found in the general population, but the effects of this risk are unclear in people not ascertained for neuropsychiatric symptoms. Using several large ASD consortium and population-based resources (total n > 38,000), we find genome-wide genetic links between ASDs and typical variation in social behavior and adaptive functioning.

"This finding is evidenced through both LD score correlation and de novo variant analysis, indicating that multiple types of genetic risk for ASDs influence a continuum of behavioral and developmental traits, the severe tail of which can result in diagnosis with an ASD or other neuropsychiatric disorder. A continuum model should inform the design and interpretation of studies of neuropsychiatric disease biology."
At some level this is not too surprising, there's not a 'gene for autism or Asperger Syndrome' any more than there's a gene for height or intelligence. Instead a number of alleles of small effect push the genome towards or away from autism spectrum traits, with de novo negative mutations adding a sprinkling of gratuitous damage.

There is a dataset of families where one family member has an ASD (autism spectrum disorder) while the the rest are not so classified. This is the Simons Simplex Collection (SSC):
"The Simons Simplex Collection (SSC) is a core project and resource of the Simons Foundation Autism Research Initiative (SFARI). The SSC achieved its primary goal to establish a permanent repository of genetic samples from 2,600 simplex families, each of which has one child affected with an autism spectrum disorder, and unaffected parents and siblings."
There is a test, similar to an IQ test, which measures the level of difficulty experienced by people with an ASD. It's called the Vineland Adaptive Behavior Rating Scale (VABS):
"Raw scores are converted to IQ-type standard scores (mean: 100 sd: 15) for each domain and for the composite adaptive behavior score. Score ranges are as follows:
  • 70-80 borderline adaptive functioning; 
  • 51-55 -70: mildly deficient adaptive functioning ; 
  • 35-50: moderately deficient adaptive behavior; 
  • 20-35: severely deficient adaptive behavior; 
  • less than 20: markedly or profoundly deficient adaptive behavior. 
Scores above 80 are classified in approximately the same ranges (low average, average, above average, superior) as IQ scores."
When you test the SSC non-ASD people (controls) vs. SSC people with ASD (cases) using the VABS, this is what you see:

Figure 2 from the Nature Genetics paper 
You plainly observe the overlap between non-ASD and ASD people on a scale designed to measure degrees of autistic behaviour.

While this histogram does not directly exhibit the underlying genetic analysis, it's exactly what you expect to see from a polygenic syndrome.

Monday, March 21, 2016

Brexit: surviving the car-crash

I'm reading Mervyn King's excellent "The End of Alchemy: Money, Banking and the Future of the Global Economy".

Newspaper reviews have been mixed: some commentators clearly wanted the inside story of the crash of 2008. Instead, the former governor of the Bank of England has provided a thoughtful critique of contemporary macroeconomic thinking, both Keynesian and neoclassical.

He gives short shrift to behavioural economics, choosing instead to focus on what he calls radical uncertainty, the unknown unknowns which regularly jeopardise the assumptions built into macroeconomic models. And subvert the policies of politicians.

Here is what he has to say about the prospects for the Euro, the Eurozone and the EU as a whole at the end of chapter 6.
"But in the euro area the fight for survival has become a battle between politicians and arithmetic. Although the future outcome is unknowable, history is on the side of arithmetic. The tragedy of monetary union in Europe is not that it might collapse but that, given the degree of political commitment among the leaders of Europe, it might continue, bringing economic stagnation to the largest currency bloc in the world and holding back recovery of the wider world economy.

"It is at the heart of the disequilibrium in the world today. The French ambition to curtail the economic power of Germany, and especially its central bank the Bundesbank, by drawing it into a monetary union that would be controlled by French civil servants has failed. The French economy is weaker than that of Germany, and monetary union has increased, not reduced, Germany's political dominance.

"Responsibility for the economic conditions in other member states will be laid at the door of Germany. The idea of a federal union was intended to represent the birth of a new Europe, born out of the common experience of defeat and occupation during the Second World War among all the original members of the European Economic Community.

"Attempts to recreate the Holy Roman Empire have often appealed to a European elite, but have foundered on the resistance of its peoples."
In the previous Brexit post I highlighted suggestions that if the European Union survives in any sense it will be led - dominated - by Germany. All other member states, by definition, will then be subordinate. Any path leading to a stable political pecking order in a future EU will be long and complex, and measured in decades, but that's the destination. If it exists at all.

But it seems unlikely that the EU in anything like its current geometry can survive over a period measured in decades. As Mervyn King points out in other chapters of his book, the Greek debt crisis has not gone away and the inevitable Greek default will be hugely unpopular in Germany. And Germany has lent massively to the rest of 'Club Med' in what might be considered belated war reparations.

Historical institutions tend to last longer than anyone thinks possible, and then to collapse faster.

There is a conceivable strategy for the UK in being close to the Northern European axis in the EU, and facilitating the transition to something more sustainable after the EU car crash. But that would demand a political culture in the UK which is communautaire. Such a culture does not exist. The British transactional paradigm for dealing with its EU partners will be worse than useless as the car crash unfolds.

The coming referendum is not the only occasion when the UK's existential relationship with the EU will be confronted - how could a foundational EU crisis not pose it again? - but proponents for staying in should be indicating what tactical advantages the UK gains given the extraordinarily dire outlook in the longer term.

If the UK (England?) decides to abandon ship, or is ejected at some point by a form of constructive dismissal, then it will operate as I imagine Japan will end up in a couple of decades, navigating between powers (the German-dominated EU, America, China) which are so much more powerful than itself.

One of the dilemmas facing even the most well-informed voter in the forthcoming referendum is that even when the issue is framed in this way, it's by no means obvious what the optimal answer is. We also have to consider that different sectional groups in the UK (employees of financial institutions and multinationals, the self-employed, trades unionists, blue-collar workers, welfare recipients, the Scots) will have variant interests perhaps best served by voting in different ways.

According to the oddschecker website today, the probability of a 'stay' vote is 69% and of a 'leave' vote is 31%.

Sunday, March 20, 2016

Interrogating Salah Abdeslam

So the Belgians caught Salah Abdeslam, the logistician for the Paris attacks, and he will be extradited to Paris for interrogation. I doubt they'll be gentle, but robust interrogation only takes you so far in a Western European democracy.

Whatever happened to this?

I would have thought intelligence agencies around the world would have been falling over themselves to perfect brain scanning for interrogation. Writing last summer, Eli Wolfe found it very difficult to get anyone to state on the record they were working on this:
"But Conolly, who notes that his lab has designed and built all the MPI scanners in North America, is so adamant that his research (including that funded by the National Institutes of Health via the BRAIN Initiative) not be linked in any way to lie detection that he refused to be interviewed about it."
It seems harder than you might imagine:
"For starters, there’s no consensus in scientific circles about what part of the brain controls deception. The prefrontal cortex, which is responsible for regulating higher planning and other executive goals, is one promising candidate. But according to Jack Gallant, a professor of neuroscience at UC Berkeley, the search for this Holy Grail of deception-control in the brain oversimplifies the complexity involved in planning a lie.

“Most things in the brain are distributed over multiple brain areas, and the patterns relating brain activity to any other sort of behavior state tend to be thoroughly complicated,” Gallant said. “I mean, there are different kinds of lies, there are different motives for lying, there are different ways of telling a lie…. Most memories that people have are not accurate—they’re confabulated.”

"Thus it may not come as a surprise that experiments testing fMRI-based lie detection techniques are riddled with confounds. Last year, Anthony Wagner, a professor of neuroscience at Stanford University and a member of the Law and Neuroscience Project, co-authored a meta-analysis on dozens of lab-based studies testing whether or not fMRI could distinguish the lying mental state. The conclusion: Virtually all fMRI-based lie-detection experiments suffer from serious design flaws."
What you'd like is not so much to detect that the subject is lying per se as to read off what they're actually thinking about - directly from the brain scanner. The Economist wrote about this back in 2011 and I embellished it in a piece I did for

Left: what the subject was viewing. Right: computer reconstruction from scanned brain

We must be further along by now, especially as it seems a natural fit to AI deep-learning.

So if you were dispensing funding for military interrogation R&D in this general area, how would you go about it? How would you avoid the liberal scruples (and PR sensitivities) of academic and industry researchers?

I think I might dress it up as prosthetic assistance for injured veterans.

Friday, March 18, 2016

Brexit: the fundamental issue is the UK vis-à-vis Germany

Henry Kissinger once famously observed, “Poor old Germany. Too big for Europe, too small for the world.

The British seem to believe that the EU is perpetually on the brink of collapse due to this or that crisis. In fact the EU subsists in a political 'lowest-energy state' - it continually varies its political geometry to accommodate challenges; it endures, it does not collapse. There is, however, an underlying secular trend towards German dominance.

For well-known historical reasons, Germany has hitherto been reluctant to overtly lead the EU - but things are changing. The EU core zone is evolving into Germany and its satellites, client states if you will. The suggestions of a supersized Holy Roman Empire are not misplaced.

The Holy Roman Empire in the early 16th century

Put to one side the headline issues of euro-idealism, immigration scares, trade-tariff issues, easy currency transactions and free-ish travel arrangements. Political strategy is about interests and power. The strategic issue facing the UK (perhaps I mean England) will be the same in the twenty-first century as it was in the twentieth. What's the best way to deal with Germany?

  1. If the UK stays in the EU, will it end up as a German satellite? 
  2. If it leaves, will it still end up as a German satellite, albeit at arm's length?

It might be argued that other countries were quite prepared to tolerate being 'Germany's running dogs' - in that old, unpleasant communist phrase. Sure, they yap, bicker and whine a bit, but they were keen to join the EU and prepared to embrace the Euro. They knew what they were getting into. Even Greece figured it out in the end.

England, you say, has always made it clear we're not signed up to the EU vision. We don't believe the answer is always 'more Europe'.

From Germany's point of view (and that of its allies) that makes the UK a real pain to deal with. The UK is the Kevin Pietersen of the European Union - a big hitter with no loyalty or esprit-de-corps. There are those in the europroject who would be delighted to see such an awkward member-state just push off.

But in or out, we continue to face the fundamental issues of power relationships and the resolution of conflicts of interest between the UK and a German-led Europe.

My feeling is that there can, in the end, only be one leader of the EU and that's Germany. So the answer to question 1 above is 'yes'.

There are historical precedents for offshore islands maintaining independence from a mainland hegemon (Japan, Taiwan, Singapore) although it's never easy. So my answer to question 2 above is 'not necessarily'.

But don't rush to judgement - perhaps there are worse things in the world than being Germany's competent and loyal senior lieutenant?


Marl Mardell at the BBC website has a perceptive piece on this.

Thursday, March 17, 2016

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.]

Wednesday, March 16, 2016

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.

Tuesday, March 15, 2016

BBC Two and "The City And The City" - China Miéville

Something to look forward to later this year. BBC Two has commissioned "The City And The City" in four hour-long episodes. Inspector Tyador Borlú, of the Extreme Crime Squad, Besźel, will soon be investigating the murder of Mahalia Geary.
"The City & the City takes place in the cities of Besźel and Ul Qoma. The precise location of these cities is not described. However, various references in the book indicate that the cities are perhaps in the coastal Black Sea region of south-central Europe: Besź smugglers traffic in goods from Varna, Bulgaria and Bucharest, Romania; protagonist Tyador Borlú enjoys his coffee in the Turkish style.

"These two cities actually occupy much of the same geographical space, but via the volition of their citizens (and the threat of the secret power known as Breach), they are perceived as two different cities. A denizen of one city must dutifully 'unsee' (that is, consciously erase from their mind or fade into the background) the denizens, buildings, and events taking place in the other city – even if they are an inch away.

"This separation is emphasised by the style of clothing, architecture, gait, and the way denizens of each city generally carry themselves. Residents of the cities are taught from childhood to recognise things belonging to the other city without actually seeing them. Ignoring the separation, even by accident, is called "breaching" – a terrible crime by the citizens of the two cities, even worse than murder."
How will they deal with "unseeing"? Perhaps they'll 'fade out' the other city?

"The City And The City" is my favourite China Miéville novel. It has weaknesses in both plot and characterisation, but I have never seen 'the social construction of reality' better realised.

Update: 7th April 2018. They finally showed it - here's my review.


In other news, Arthur C. Clarke's 'Childhood's End' comes as a three-part series to Sky 1 at 9 pm this Thursday (repeated a few hours later, deep in the night, on Sky 2).

Mr Clarke's rather clunky fiction has largely gone out of fashion - who reads it any more? - but 'Childhood's End' is clever, has genuine plot surprises and is filled with a galactic sense of wonder.

I do hope they do it justice.


As I have my partially-sequenced genome over at 23andMe, I decided to do my bit for genetic research by signing up with DNA.Land.

"DNA.Land is a place where you can learn more about your genome while enabling scientists like us to make new discoveries for the benefit of humanity. The website is not-for-profit and run by the Erlich and Pickrell labs affiliated with Columbia University and the New York Genome Center.

"The purpose of DNA.Land is to enable you to learn more about your DNA and allow you the autonomy to share your data to facilitate important scientific research at the forefront of genome sciences and medicine. Our goal is to help members interpret their data and connect potential participants with research studies.


"After you provide consent, your participation would consist of creating your personal profile and securely uploading your genetic data to DNA.Land. We will use the most cutting edge genetic tools to analyze your data and return your results regarding ancestry, relatives, and different traits.

"As we want to learn about the genetic basis of different traits, we will ask you to fill out surveys relating to your (or your family’s) ancestry and health. You will also have the option to automatically contribute data from your social media profiles for new types of analysis, so we can learn about traits that are dynamic and more difficult to measure, such as social preferences.

"Your profile will also display a badge that summarizes your various contributions to DNA.Land. You can tweet this badge, share it on Facebook, or sew it on your old scout uniform.

"There are no costs associated with taking part in DNA.Land and you will not be compensated for participating."
Where is the value-add over, say, the 23andMe health and ancestry reports? DNA.Land 'impute' missing parts of your genome based on "whole genome sequencing data used to create a dictionary of genomic 'text' (known as haplotypes)."

So how does this help?
"Uploaded genotype files (e.g. from 23andMe) contain between 500,000 to 1 million SNPs. DNA.Land's imputation pipeline imputes (i.e. infers the value of) an additional 38 million SNPs."
I guess this exploits linkage disequilibrium.

After many hours of crunching, DNA.Land will deliver an imputed VCF file (about half a Gigabyte). They attempt to explain how to interpret this ("Understanding VCF values") - but it will plainly take more work on my part to figure it out. The raw detail file - in all its immensity - seems pretty opaque; probably best to run the whole thing through Promethease.

Here's what SNPedia says about DNA.Land with reference to Promethease.

And here's how to run the imputed results through Promethease. Haven't tried it as the file isn't yet available, I think it's just use the "Upload Raw data" button.

Costs $10 for the enhanced ('imputed') report.

So I'm currently waiting for DNA.Land's computers to finish crunching my 23andMe raw data file.


Update: (Tuesday 3.22 pm): Promethease currently uploading 417 MB of dnl13394_inl.imputed.vcf via ADSL. A long wait ahead.


Update: (Tuesday 7.29 pm): OK, so all done.

I have the Promethease report downloaded, all 8 MB of it. Time enough to browse it tomorrow.

I rather wish that Promethease had told me it doesn't accept .bcf files (the binary version of .vcf) before I uploaded half a Gigabyte's worth - how hard could that be, to check a file extension?

It did, however, take .vcf so it all worked in the end.

Monday, March 14, 2016

Cornwall: St Michael's Mount & Marazion

We were in Cornwall a few days last week, near Penzance and opposite St Michael's Mount.

You may be thinking: Cornwall in early March, with arctic temperatures and 60 mph gales - are they mad? Less of the schadenfreude, mes braves! There is no such thing as bad weather, only bad clothing. Also, those weather people always exaggerate. And though the roads are no wider than hat-pins, at least there are few cars coming the other way.

So the promised hurricane on Wednesday was just blustery wind, the promised walls of water and mountains of spray failing to make an appearance. As it was too windy to do the NT gardens we'd pencilled in, we hopped from pub to pub instead - a succession of roaring fires.

Tuesday, in the drizzle and fog, we ventured across the causeway to St Michael's Mount. The castle on top and the roads leading up were closed, leaving merely a trivial perambulation around the picturesque stone cottages at the end of the causeway.

Click on the pictures below to make them larger.

Clare at our hotel room window

St Michael's Mount from the Mount Haven Hotel

The causeway at low tide

According to studies, "the Welsh, followed by the Cornish, remain among the most genetically distinct of all the groups on mainland Britain. They carry more DNA that could date back to the tribes that colonised Britain after the last Ice Age 10,000 years ago."

Our area, Marazion, had a smattering of Cornish language signs and you could make a case that the locals were smaller and visually-distinguishable from the few anglo-saxon tourists; no wonder we're dismissively tagged as 'emmets'.

Ethnic pride's a good thing - so where's Mebyon Kernow then? Bring it on, kothmans.

Sunday, March 13, 2016

AlphaGo and Laplace

On Friday AlphaGo had won its first two matches against Go champion Lee Sedol and I was wondering what was the probability it would win the third (out of five).

1. If we consider AlphaGo and Lee Sedol to be pretty evenly matched, then the chances of two wins to AlphaGo are 25% - hardly unusual - so the chances of a further AI win might still plausibly be 50:50.

But .. two wins in a row?

2. If we adhere strictly to a frequentist dogma, then with two wins in two trials we must assume that AlphaGo would win its next match with probability 1. But that's crazy.

3. Laplace had a rule for this kind of thing: the Rule of Succession. Here's how Wikipedia describes it.
"If we repeat an experiment that we know can result in a success or failure, n times independently, and get s successes, then what is the probability that the next repetition will succeed?"
We count a success as scoring 1 and a failure as scoring zero, then Laplace tells us the probability of a success the next - (n+1)th - trial is:

(s + 1)
(n + 2).
Read the article for the counterintuitive derivation.

So we plug in the numbers: two games (n = 2) and two successes (s = 2) so the probability of AlphaGo winning the third game is 3/4. Sounds about right, I thought.


On Saturday, AlphaGo duly won the third match in a row and the series, but today (Sunday) it lost its fourth match.

The last match is on Tuesday, so what are AlphaGo's chances?  (3 + 1)/(4 + 2) = 2/3.

In case you were thinking of placing a bet.


Update Tuesday: AlphaGo wins the fifth match, and the series 4:1.

Thursday, March 10, 2016

"But yours doesn't work!"

An apocryphal consultant story.

An electronics company and a critical chip; an in-house design that can't be made to work.

A consultant is brought in and shown the specs. The problem is difficult; the consultant can't see a way to solve it.

In the cab back to the airport, an epiphany. The consultant works up his design and returns to present it to the chief engineer.

The internal team examines it. The chief engineer is dismissive: "Your design is slower than mine."

The consultant explodes in exasperation: "But your design doesn't work. If we're allowed to submit designs which don't work, trust me, I can provide you with a design twice as fast as yours!"

Saturday, March 05, 2016

Science-Fiction vs. Speculative Fiction

People argue about the distinction between science-fiction and speculative fiction. For me, science-fiction is situated within a scientific or technological culture; story-settings embedded within a context which makes sense to anyone with the appropriate background. A technical setting may be complex, unusual or even baroque ... but it will possess a logical coherence which informs or drives the narrative.

Speculative fiction, I suggest, is written by humanities people  - who have not had a scientific or technical formation, who have not been immersed within that culture.

When they need unusual ideas for their setting or plot device, they seem to find candidates in the increasingly strange worlds of modern science, technology and engineering. But in the hands of the speculative fiction writer, these borrowed concepts are merely bangles plucked from a mighty Christmas tree. You can tell that they've just taken stuff from popularisations; ideas as mood music: shallow, disposable and frequently misunderstood - not at all integral to the texture of their story.

Claire North, you know who you are!


I suppose I should, for younger readers, make an obligatory reference here to C. P. Snow's famous "Two Cultures".

I'll also mention - again! - the dwarves-elves distinction due to SF writer Stanislaw Lem.

Friday, March 04, 2016

Odds and Probabilities: Clinton vs. Trump

If the odds on something are b:a, then the probability is a/(a+b). Example: if the odds are 2:1, the probability is 1/3.

Here is the website oddschecker's take today on the US Presidential Election in November:
  • For Hillary Clinton, the odds to win are 8:15, a probability of 15/23 = 65%
  • For Donald Trump, the odds are 11:4, a probability of 4/15 = 27%.
Turning now to the British EU Referendum in late June, oddschecker has this to say:
  • The odds for staying in are 4:11, a probability of 11/15 = 73%
  • The odds for Brexit are 5:2, a probability of 2/7 = 29%
Never bet against the bookies. And sure as sure, these odds are not the final word.


Read an interesting paper yesterday (h/t Marginal Revolution): ""Personality Traits and the Dimensions of Political Ideology"" (PDF). Here's a histogram showing the key finding.

This says that you incline to conservative economic political views if you are somewhat extraverted, hard-headed, conscientious, emotionally-stable and of a rather conventional and traditionalist disposition.

On the other hand, you tend to liberal economic political views if you are rather agreeable and 'open to experience'. One thinks of those pleasant, imaginative, emotional luvvies - always so keen to dispense other people's money.

I'm not sure this isn't restating the obvious as regards stereotypical voter types. Still, the sample sizes were large and all science has to start somewhere.

Thursday, March 03, 2016

"The First Fifteen Lives of Harry August" - Claire North

I'm currently reading "The First Fifteen Lives of Harry August" by Claire North to my wife as her after-dinner treat.

[Update (March 15th 2016): after 11 readings I've abandoned this book. The plot is slow and uninvolving; the characters insipid and hard to engage with. At fast reading speed this is tolerable but when reading aloud, it's just too difficult to care.]

Here's what The Guardian had to say about it:
"From first line to beautiful denouement, Claire North's The First Fifteen Lives of Harry August  is a gripping read that is often quietly profound, emotionally affecting and intellectually dizzying.

"Harry August is fated to live his life again and again, a kind of identical reincarnation, born to the same parents, in the same body, but with memories of his previous lives intact. As he lies dying at the end of his 11th life he is visited by a seven-year-old girl with a message from thousands of years in the future that he must deliver to the Cronus Club: "The world is ending."

"The members of the Cronus Club are, like Harry, men and women who live their lives over and over. When, in his next incarnation, Harry duly delivers the message, he is told that there is no hope of effecting change. What follows is Harry's investigations into the eventual apocalypse, and his fateful involvement with his friend Victor, a "fellow traveller" intent on accelerating the world's technological progress for his own ends …

"As might be expected from such a narrative, the novel is an examination of determinism and free will, but also a subtle study of friendship, love and the fluid complexity of existence."
Claire North is the pseudonym of Catherine Webb, a prolific and acclaimed author. The book is well-written though I'd probably give it four stars on Amazon; the text sometimes strays too far into clunky exposition but the characterisation and plot development is not without interest.

This is probably speculative fiction rather than science-fiction: the difference being subtle and maybe even tendentious (literary types don't write SF, even when you think that's exactly what they're doing). Speculative fiction is not written from within a scientific or engineering culture.

I have taken to recording my half-hour readings - yes, I'm compiling a de-facto audiobook, full of clinking coffee cups, unwanted phone interruptions and random audience reaction. The folder is on my Google Drive and can be accessed here .. if you fancy being read to, very badly (format is mp3 and m4a). Eight instalments exist as I write; I'm about 25% through. More to come. (But now stopped - see update above).

Tech note: I found NCH's Switch and WavePad very useful (free download) for audio format conversion and audio-level boosting. Thanks NCH.


Google recently released an uploader program to transfer all your hard drive photos onto the Google Photos cloud. Choose the option where this doesn't come off your free allowance. I have something like 15,000 pictures and videos assembled over the decades, and more than 24 hours later, it's still uploading the last 649.

The results are arranged on a timeline, and transform Google Photos into a super photo album on a tablet: I'm really impressed. I hope that Google gets moving on adding image search tools - I'd like to tag a photo for a person or place (or any number of attributes or properties) and then search my whole, way-too-large-to-manage dataset.

The picture below, of the person read to each evening, emerged without AI assistance.

Unusually cheerful for first thing in the morning - a campsite in the Dordogne 2012

Who knows how much gold is buried in my back-catalogue?

Wednesday, March 02, 2016

"This Census-Taker" - China Miéville

We had some weather this morning:

Hail on the drive; rain on the window

Enough! Let's consider China Miéville's Kafkaesque new novella, "This Census-Taker".

Here's the blurb:
"In a remote house on a hilltop, a lonely boy witnesses a traumatic event. He tries - and fails - to flee. Left alone with his increasingly deranged parent, he dreams of safety, of joining the other children in the town below, of escape.

"When at last a stranger knocks at his door, the boy senses that his days of isolation might be over.

"But by what authority does this man keep the meticulous records he carries? What is the purpose behind his questions? Is he friend? Enemy? Or something else altogether?"
A feature of Miéville's novels is an intense focus on setting and atmosphere. The boy lives with his father and mother in a shack on a hill overlooking a hick country town, 'Bridgetown'. The nearest city is remote, 'on the coast', and the source of unspecified political power remoter still.

The father is a foreigner, a practitioner of some unspecified but esoteric handicraft, who did something terrible in a past war and had to flee; the mother is a local who left Bridgetown for the coast, only to return with her new husband.

Miéville's writing is cool and dislocated; alienation the dominant mode of relationships. Temporal order is subverted and the story-telling lurches from first to second to third person. Through a gathering immersion in the 'traumatic event' and its aftermath, the reader is challenged to figure out what the wider narrative might be. Ostensible resolution comes finally through the Census-Taker but ambiguity persists to the end of the book.

This review could continue with spoilers, perhaps give you a personal opinion of what a plain statement of this narrative would convey, but why undermine the point of reading it? You engage with China Miéville to enter strange, subverted realities and in this novella, to re-experience childhood's loneliness and confusion and a child's naive vulnerability to adult manipulation. And as an adult, to comprehend in your own way more than the main character, the boy, can appreciate.