Indian kids are getting dumber at maths!

OK, that headline is a bit of a clickbait! They probably aren’t getting dumber, if anything it could be the opposite if the Flynn effect is really true. However, I specifically had the declining performance of Indian kids at the world’s foremost mathematics competition in mind, better known as the International Mathematics Olympiad or IMO, when I wrote that headline.

A bit of personal also-ran history is involved here as I did compete to join the Indian team at the turn of the millennium, but the competition was so fierce that I could not manage to get into the national-level top six that represent each country at the IMO. And that was just as well, as all the guys were clearly brighter and I did not deserve to be in that peer group. Nonetheless that teenage experience of competitive problem-solving (and failing to make the cut) informs my desire to keep a close watch on the Indian team’s performance at the IMO. I also occasionally try to solve IMO problems on boring London tube commutes, i.e. when I manage to get a damned seat, and share them with colleagues at work. Those interested can try them here.

The 2017 IMO recently concluded in late July, and the Indian team showed its worst performance this year since 1990 – the year it first started competing in this annual mathematical jousting event. Since this is brownpundits I tried to put the declining performance of Indians into context by comparing it, over the years, with our brown South Asian neighbours, Bangladesh and Pakistan. Throwing in Iran and United Kingdom as controls to add a bit of perspective. All the country-wise data can be accessed here.

Annual IMO Rank per country

The graphs are telling! India was up there with Iran and UK as its peers all through the 90s decade to the early 2000s. Indians were slightly worse-off than Iran, but by the turn of the millennium we were doing better. My own school-leaving cohort (and a couple of years around that) soundly beat both the Iranians and Brits. Yet 2005 marks a regime shift for the worse in average Indian performance at IMO and the data seem statistically significant.

I am at loss to explain this clearly worsening trend of performance by India’s brightest millennials. Did Indian parents really start begetting a dumber brood from the 90s onwards? I hope not! Good feeder schools and rigorous mathematical training play a big part in preparing high-school kids for such competitions and it is possible that some silly policy change (that I am unaware of) by the Indian government may have been a causal factor.

But there’s some hope in the same data for our Eastern cousins. Bangladeshi kids (and their mathematics training programme) seems to have shown phenomenal improvement(!) over the same period and now easily better India. India’s coincident decline does not help matters either. Bangladesh started competing at the IMO in the same year as Pakistan with similar laggardly results, but the subsequent improving trend in performance is clear as day. As for Pakistan, well, let’s just say that their national priorities leave a lot to be desired…

Why Democracy?

The idea to write this blog post on Democracy arose out of the need to describe what it is in context of Brexit. For more on the Brexit referendum itself see this. In this post I am trying to distill my own understanding of Democracy and have included the results of a numerical experiment I ran to quantify some ideas around the concept.

Democracy is essentially an algorithm to correct political error. In that respect Democracy belongs to a special class of algorithms, with Darwinian evolution, scientific peer review or machine learning being other notable members of the same class. The kinship between these disparate and very fundamental processes is not coincidental. It is explained by Popperian epistemology, which makes the existence and mitigation of error central to the idea of any knowledge generation.

Any discussion of the process of knowledge creation may seem like a digression at this point. However, please persevere for the next three paragraphs as setting this context is important for the central thesis on Democracy. According to Popper, knowledge itself can be understood as explanations, i.e. guesses or conjectures with two major criteria for goodness: falsifiability and parsimony. Any knowledge creator (sentient or otherwise) must therefore create knowledge in exactly this manner: creatively produce guesses or conjectures (including even, what look like, wild ones) and criticise them to remove those that are erroneous. Two immediate corollaries of this theory arise: a) existence of error is a permanent feature of any form of knowledge. Claims of knowledge that are perfect (e.g. a manual revealed by so-called prophets) are therefore, for want of a better word, baloney. And b) boundless knowledge-generation must require the ability or enabling culture to air seemingly wild guesses and criticise even ostensibly unimpeachable maxims. Continue reading “Why Democracy?”

The Indo-Aryan question nearing resolution

India Today published my review of the current state of the genetics and genomics of the Indian subcontinent, and what it can tell us about the ethnogenesis of South Asians generally. In the piece I tried to be very circumspect and stick to what we know with a high, if not perfect, degree of certainty. Here I will add some comments where I reduce the threshold of certainty somewhat. That is, I’m going to include here my beliefs where I think I’m right, but in some details wouldn’t be surprised if I was wrong.

First, the title is Aryan wars: Controversy over new study claiming they came from the west 4,000 years ago. Writers don’t get to choose titles, and this is not one I would have chosen. But I am not in a position to care or know what draws clicks. Let’s note that this “controversy” is restricted mostly to India. Outside of India it’s not controversial, but a matter of the science, because people don’t have any political or social investment in the topic. It reminds me of debates about genetics and intelligence in the West, where emotions get overwrought and lies fly wildly with abandon.*

Second, there is a reference in the figures to an “Out of India” (OIT)  model. That is, the Aryans migrated out of India, and implicitly the Indo-European languages derive from South Asia. I don’t think this theory has any support at all. That is, I think it is rather clear that proto-Indo-European probably emerged neither in Europe proper, nor in South Asia, but in the Inner Eurasian spaces between. But for an Indian audience ignoring OIT would seem a peculiar lacunae, so there was a reference added to the figure on that account (I pushed back against this, but do not make ultimate decisions on figures).

But I do think it was plausible up until 2009’s Reconstructing Indian History to suggest that most modern South Asian ancestry dates to the Pleistocene. In this framework the Indo-Europeanization of the subcontinent was primarily a cultural one, where small groups of Central Asians imposed their language on the native population. What the genome-wide work has shown is that South Asians are the product of a large-scale mixing process between a population very distant from West Eurasians (“Ancestral South Indians”, ASI) and a population which was indistinguishable from other West Eurasians (“Ancestral North Indians”, ANI).

Since ANI is indistinguishable from West Eurasians I hold it is clearly a West Eurasian population in provenance. Those who reject this position from a scientific perspective believe that there could have been some sort continuous zone of “ANI-like” habitation from northwestern South Asia up into northern Inner Eurasia (and perhaps toward West Asia as well) dating from the late Pleistocene. I do not that believe this is plausible, and I will tell you that prominent researchers who I have brought up this idea to are somewhat incredulous.**

Third, there are major unresolved issues genetically in relation to the dates and the total number of mixing populations. I am quite confident saying around half of the total South Asian genomic ancestry today derives from populations who were living outside of South Asia on the Holocene-Pleistocene boundary 11,700 years ago. Much of that ancestry probably flourished between the Caucasus and Zagros mountains. The remainder somewhere in the vast swath of territory between the Baltic and Siberia (perhaps further south, toward the Pamirs?).

But I am not confident of the relative balances of contribution to the ANI. It does seem that the northern component, which is derived in part from the southern component, is much more prominent in upper castes and northwestern populations. In contrast the southern component is found throughout the subcontinent.

In Genomic insights into the origin of farming in the Near East there is analysis of South Asia in the supplements. The author concludes that ANI can not be modeled as a single population (Zack Ajmal and I were saying this in 2010). The top hits for the sources of ANI tend to be the genomic sample from the Zagros, in western Iran (before subsequent admixture with Levantine farmers), and a population similar to the Yamna culture of the steppe. The issue seems to be that later steppe populations which harbor a fair amount of “Early European Farmer” ancestry (e.g., LBK in Central Europe) due likely to back migration aren’t good model fits.

Below are two plots, one showing a scatter of South Asian groups with their Iran_N (a sample from ~10,000 years ago) vs. Yamna (from ~5,000 years ago), and another with the ratios.


DO NOT TAKE THE PROPORTIONS LITERALLY.  My intuition is that these models are overestimating the proportion of steppe ancestry, but my confidence in my intuition is low.

There are two groups enriched for Iran_N ancestry:

  1. Lower caste groups, especially from South India.
  2. Populations in southern Pakistan.

The reasons differ. If you have done genetic analysis of the Pakistani populations it seems quite obvious that unlike other groups in South Asia Pakistani groups facing the Arabian sea across from Oman have genuine Near Eastern ancestry. This affinity declines as you go north in Pakistan rather rapidly. Notice though one South Indian group: Jews from Cochin. This population clearly has recent Near Eastern ancestry.

The Kharia are an Austro-Asiatic Munda group. For whatever reason Austro-Asiatic groups seem to consistently have very little steppe ancestry. The Mala are Dalits from South India. The further up you go on the modal Iran_N-Yamna cline you see the populations are either upper caste, or, they are from the far northwest of the subcontinent.

The conclusion I derive from this is that first there was an early migration of West Eurasian populations consisting of Iranian farmers. This group mixed with the ASI element. The Indo-Aryans, which probably correlates with the Yamna-like component, arrived later as an overlay (and nearly half of their ancestry was derived from Iranian farmers). Then many South Asian populations have modifications on this base model of compound ANI + ASI; Munda and Bengali have later East Asian ancestry, while populations on the Arabian sea have Near Eastern ancestry.

Fourth, the story in India Today leans heavily on Y chromosome of R1a1a lineage. It is true we are Lords of the Steppe and destined to drive our enemies before us. But, it is not the primary story. And yet Y chromosomal phylogenies are easy for the public to understand. But they only make sense in light of the above framework. R1a1a is found in South Indian tribal populations. It seems likely that Indo-Aryan paternal lineages were highly invasive across the subcontinent, just as they were in Europe. In many cases they likely extended far beyond domains where Indo-European acculturation occurred.

I’m probably wrong on some of the details. But I suspect the final story will not be so different from this.

Finally, I will mention the cultural element here. There is a fair amount of the discussion of the form “so you are saying the ancestors of Indians are Europeans?” or “does this mean Hinduism is not Indian?”

The piece was about genetics and demography, not my opinions about culture. So I will say this:

  1. The “West” as an entity is no older that Classical Greece. 500 BC. My own personal position, strongly held, is that the West should indicate cultures and societies which descend from the European societies which adhered to the Western Church around ~1000 AD (some nations, like Lithuania, became absorbed into this cultural complex hundreds of years later). So Russia is not the West. And Merovingian Francia is not the West.
  2. Indian civilization of what we term the Hindu variety coalesced in the period between between 500 BC and 500 AD, from before the Mauryas, up to the Guptas. Obviously the period before 1000 BC was important in setting the ground-work, but I do not believe it was Indian as we’d understand it in anything but the geographical sense, nor was it Hindu in any way we’d recognize it today (similarly, Shang dynasty China was not China as we’d understand, which came into being after 500 BC).

These positions mean that I think nationalist passions are in the “not even wrong” category. Indian Hindu civilization is indigenous by definition, since it was synthesized in situ on the edge of historical perception and attestation (for the record, I think Adi Shankara was critical in the completion of a crystalized self-conception of Hindu religio-philosophical thought, but its origins predate him). Similarly, Indian civilization was not seeded by white Europeans because white Europeans were only coming into being in Europe when the Indus Valley civilization was collapsing.

That is all (for now).

Addendum: The first tranche of ancient DNA should be out in a few months. Also, there is another paper on Indian genetics in the work from the usual suspects. There won’t be anything totally surprising (or so I’ve been told).

* By lies, I mean the contention that intelligence is an “invalid” instrument in relation to predictiveness, or, if it is valid, it is not genetically heritable. People routinely lie about these facts in discussion or spread lies because there are socially preferred positions which they conform to. Similarly, many questions about Indian history seem to hinge on widely promoted lies.

** This model needs to also confront the massive mixing of the last 4,000 years. If it is true then it is ASI which is mostly likely intrusive, because it is not creditable that these two populations were in nearby proximity for tens of thousands of years without exchanging genes.

Indian genetics, the never-ending argument

I am at this point somewhat fatigued by Indian population genetics. The real results are going to be ancient DNA, and I’m waiting on that. But people keep asking me about an article in Swarajya, Genetics Might Be Settling The Aryan Migration Debate, But Not How Left-Liberals Believe.

First, the article attacks me as being racist. This is not true. The reality is that the people who attack me on the Left would probably attack magazines like Swarajya as highly “problematic” and “Islamophobic.” They would label Hindu nationalism as a Nazi derivative ideology. People should be careful the sort of allies they make, if you dance with snakes they will bite you in the end. Much of the media lies about me, and the Left constantly attacks me. I’m OK with that because I do believe that the day will come with all the ledgers will be balanced. The Far Left is an enemy of civilization of all stripes. I welcome being labeled an enemy of barbarians. My small readership, which is of diverse ideologies and professions, is aware of who I am and what I am, and that is sufficient. Either truth or power will be the ultimate arbiter of justice.

With that out of the way, there this one thing about the piece that I think is important to highlight:

To my surprise, it turned out that that Joseph had contacted Chaubey and sought his opinion for his article. Chaubey further told me he was shocked by the drift of the article that appeared eventually, and was extremely disappointed at the spin Joseph had placed on his work, and that his opinions seemed to have been selectively omitted by Joseph – a fact he let Joseph know immediately after the article was published, but to no avail.

Indeed, this itself would suggest there are very eminent geneticists who do not regard it as settled that the R1a may have entered the subcontinent from outside. Chaubey himself is one such, and is not very pleased that Joseph has not accurately presented the divergent views of scholars on the question, choosing, instead to present it as done and dusted.

I do wish Tony Joseph had quoted Gyaneshwer Chaubey’s response, and I’d like to know his opinions. Science benefits from skepticism. Unfortunately though the equivocation of science is not optimal for journalism, so oftentimes things are presented in a more stark and clear manner than perhaps is warranted. I’ve been in this position myself, when journalists are just looking for a quote that aligns with their own views. It’s frustrating.

There are many aspects of the Swarajya piece I could point out as somewhat weak. For example:

The genetic data at present resolution shows that the R1a branch present in India is a cousin clade of branches present in Europe, Central Asia, Middle East and the Caucasus; it had a common ancestry with these regions which is more than 6000 years old, but to argue that the Indian R1a branch has resulted from a migration from Central Asia, it should be derived from the Central Asian branch, which is not the case, as Chaubey pointed out.

The Srubna culture, the Scythians, and the people of the Altai today, all bear the “Indian” branch of R1a. First, these substantially post-date 6000 years ago. I think that that is likely due to the fact that South Asian R1a1a-Z93 and that of the Sbruna descend from a common ancestor. But in any case, the nature of the phylogeny of Z93 indicates rapid expansion and very little phylogenetic distance between the branches. Something happened 4-5,000 years ago. One could imagine simultaneous expansions in India and Central Asia/Eastern Europe. Or, one could imagine an expansion from a common ancestor around that time. The latter seems more parsimonious.

Additionally, while South Asians share ancestry with people in West Asia and Eastern Europe, these groups do not have distinctive South Asian (Ancestral South Indian) ancestry. This should weight out probabilities as to the direction of migration.

Second, I read some of the papers linked to in the article, such as Shared and Unique Components of Human Population Structure and Genome-Wide Signals of Positive Selection in South Asia and Y-chromosomal sequences of diverse Indian populations and the ancestry of the Andamanese. The first paper has good data, but I’ve always been confused by the interpretations. For example:

A few studies on mtDNA and Y-chromosome variation have interpreted their results in favor of the hypothesis,70–72 whereas others have found no genetic evidence to support it.3,6,73,74 However, any nonmarginal migration from Central Asia to South Asia should have also introduced readily apparent signals of East Asian ancestry into India (see Figure 2B). Because this ancestry component is absent from the region, we have to conclude that if such a dispersal event nevertheless took place, it occurred before the East Asian ancestry component reached Central Asia. The demographic history of Central Asia is, however, complex, and although it has been shown that demic diffusion coupled with influx of Turkic speakers during historical times has shaped the genetic makeup of Uzbeks75 (see also the double share of k7 yellow component in Uzbeks as compared to Turkmens and Tajiks in Figure 2B), it is not clear what was the extent of East Asian ancestry in Central Asian populations prior to these events.

Actually the historical and ancient DNA evidence both point to the fact that East Asian ancestry arrived in the last two thousand years. The spread of the first Gokturk Empire, and then the documented shift in the centuries around 1000 A.D. from Iranian to Turkic in what was Turan, signals the shift toward an East Asian genetic influx. Alexander the Great and other Greeks ventured into Central Asia. The people were described as Iranian looking (when Europeans encountered Turkic people like Khazars they did note their distinctive physical appearance).

We have ancient DNA from the Altai, and those individuals initially seemed overwhelmingly West Eurasian. Now that we have Scythian ancient DNA we see that they mixed with East Asians only on the far east of their range.

The second paper is very confused (or confusing):

The time divergence between Indian and European Y-chromosomes, based on the closest neighbour analysis, shows two different distinctive divergence times for J2 and R1a, suggesting that the European ancestry in India is much older (>10 kya) than what would be expected from a recent migration of Indo-European populations into India (~4 to 5 kya). Also the proportions suggest the effect might be less strong than generally assumed for the Indo-European migration. Interestingly, the ANI ancestry was recently suggested to be a mix of ancestries from early farmers of western Iran and people of the Bronze Age Eurasian steppe (Lazaridis et al. 2016). Our results agree with this suggestion. In addition, we also show that the divergence time of this ancestry is different, suggesting a different time to enter India.

Lazaridis et al. accept a mass migration from the steppe. In fact, the migration is to such a magnitude that I’m even skeptical. Also, there couldn’t have been a European migration to South Asia during the Pleistocene because Europeans as we understand them genetically did not exist then!!!

I assume that many of the dates of coalescence are sensitive to parameter conditions. Additionally, they admit limitations to their sampling.

Ultimately the final story will be more complex than we can imagine. R1a is too widespread to be explained by a simple Indo-Aryan migration in my opinion. But we can’t get to these genuine conundrums if we keep having to rebut ideologically motivated salvos.

Related: Ancient herders from the Pontic-Caspian steppe crashed into India: no ifs or buts. I wish David would be a touch more equivocal. But I have to admit, if the model fits, at some point you have to quit.

Indian media is finally reporting on the Aryan migration into South Asia

For various ideological reasons in India there has been a strong resistance to the idea that Aryans came from outside of South Asia. When David Reich’s Reconstructing Indian Population History was published 2009 the Indian media had a weird response. For example, Aryan-Dravidian divide a myth: Study.

Though Reich’s paper was equivocal, it was clear to me that it was likely going to be the launching point for a resurrection of the Aryan migration theory. Now Tony Joseph in The Hindu has published a pretty good survey of the literature, How genetics is settling the Aryan migration debate. Nothing new for readers of this weblog, but he some good quotes:

The avalanche of new data has been so overwhelming that many scientists who were either sceptical or neutral about significant Bronze Age migrations into India have changed their opinions. Dr. Underhill himself is one of them. In a 2010 paper, for example, he had written that there was evidence “against substantial patrilineal gene flow from East Europe to Asia, including to India” in the last five or six millennia. Today, Dr. Underhill says there is no comparison between the kind of data available in 2010 and now. “Then, it was like looking into a darkened room from the outside through a keyhole with a little torch in hand; you could see some corners but not all, and not the whole picture. With whole genome sequencing, we can now see nearly the entire room, in clearer light.”

In relation to online debates I have had Indian interlocutors tell me flat out that they believe in the papers published between 2005 and 2010. It is nice to get the scientists who actually published this work now admit that new results overturn the older theories.

Note: I am going to refer to this as a migration, because “invasion” seems to connote too much specificity as to how it happened. But I have a difficult time imagining that it was a peaceful process.

The last days of pre-ancient DNA Indian population genomics

If anyone wants to know about the population genetics of South Asia, I recommend three papers (all are open access):

Genetic Evidence for Recent Population Mixture in India

A genetic chronology for the Indian Subcontinent points to heavily sex-biased dispersals

The promise of disease gene discovery in South Asia

In the near future ancient DNA will do for South Asia what has been done for Europe, and to a lesser extent the Near East. It will pull back our veil of ignorance. But until then we have genomic inference from larger data sets with a greater number of markers. What can we say now?

– The 2009 work that modern South Asians are broadly a compound of two streams of the out of Africa populations is correct. One is much like other West Eurasians. Another is distantly related to other East Eurasians, with possible affinities to Paleolithic Southeast Asian hunter-gatherers.

– The West Eurasian ancestry of South Asians, the “Ancestral North Indians” (ANI), does likely seem to be a mixture at minimum between two groups. One element is related to the eastern farmers who first adopted agriculture on the slopes of the Zagros ~10,000 year ago. Another stream is closely related to the Yamna people who flourished on the Eurasian steppe north of the Black Sea ~5,000 years ago.

– The Munda peoples seem to have a distinct Southeast Asian component that ties them with other Austro-Asiatic peoples. Their migration was almost certainly tied to the Neolithic migration of rice farmers. They are likely not the primal aboriginals of South Asia.

– The R1a1a-Z93 Y chromosomal lineage found across much of South Asia, and especially the higher castes and the north, increased in frequency within the last 4,000 years. It is almost certainly exogenous to South Asia; ancient DNA from the steppe finds the Z93 in Iranic peoples, but no Indian ancestry in these groups.

As I said, ancient DNA will clarify lots of things. I expect that to happen in the next few years.