Most people think that science and math and engineering are found everywhere, like soccer, but actually, they are regional practices, more like hurling or tossing the caber. In the map, countries are resized according to the number of scientific papers they produce. Population size plays a role, but average productivity matters more. Note that Singapore, with a population of 5 million, looks bigger than Indonesia, with 240 million people.
You might think that scientific productivity depends on science funding, rather than regional differences in talent. But you would be wrong. It depends to some extent on money, but it depends more on the distribution of talent. You can get a feel for the influence of money by comparing China or India and their diasporas in the West. You see far higher scientific productivity in those diaspora populations than in their homelands. Partly this is due to selective immigration, but much of it has to do with with economic (and cultural) differences).
Like the Chinese and Indians, some ethnic groups show mediocre results in their benighted homelands and better results in Western countries. On the other hand, other groups do poorly everywhere. People whose ancestors are from sub-Saharan Africa produce very few scientific papers – on average – no matter where they live. This is also the case (to a lesser extent) for mestizos from central and South America, for Filipinos and Malays, and most of the Indian castes. You don’t see much out of the Middle East, either - although Armenians are an exception, and there may be others. Interestingly, Armenians were generally thought (by themselves and others) to be smarter than the average bear back in Byzantine times.
Generally the pattern is about what you would expect from the world distribution of IQ (note the correlation with latitude), coupled with the notion that science is generated by people out in the tail of the intelligence distribution. About 2% of a population with an average IQ of 100 scores above IQ above 130, about 0.1% above 145. For a population with an average of 85, only 0.1% will score above 130 – 20 times fewer. I am, for the moment, disregarding fat tails and some actual population differences in the standard deviation. The same thing is happening with the Ashkenazi Jews: a modest shift in mean (about 0.8 standard deviation above the European average) causes a big change in the fraction that exceeds a high threshold.
The officially approved theory in the US, and in most other Western countries, is that all these differences are environmentally caused and can be environmentally remedied. Of course no one has actually managed to do this, anywhere. In Malaysia, there is a similar-sized IQ gap (about 15 points) between the Chinese and the Malay majority, but their attitude is different: no one seems to think that anything can be done about it. They may be wrong….
Efforts to change this pattern can be fun to watch. I know of a mathematician who admitted to a feeling of guilt when his department repeatedly fulfilled its Hispanic affirmative action goals by hiring Ashkenazi Jews from Argentina.
You might suspect that the pattern would change if we raised the bar, looked at more extreme and important examples of scientific creativity, rather than counting every paper. It does. If you look at Nobel prizes in the sciences, or Fields medals, the differences become starker. Whole nations disappear. Single individuals outdo whole civilizations. The only noticeable deviation from the simple IQ-distribution pattern is somewhat lower-than-expected scientific productivity, at the highest levels, in northeast Asians. We will see if this persists as China modernizes.
Some of these patterns are clearer when you look at mathematics, since it doesn’t require expensive facilities, and has a strong tradition of amateurism. At the limit, you can find people who never had any formal training in mathematics but somehow produced work of the highest quality, like Srinivasa Ramanujan or George Green. As far as I can tell, these amazing amateurs almost always originate from the same populations that produce large numbers of more conventional scientists.
Lastly, if we look at the demographics, it seems that every population that produces serious scientific players has sub-replacement fertility.