Saturday, January 30, 2016

The value of outsiders - a bit of big-data

There are many examples from the past where  recruiting someone from an entirely different specialization into an organization pushed the envelope of knowledge. It is quite remarkable to see how organizations rapidly turn dull when narrow policies are in place. Too many establishments languish because of inbreeding and the lack of a broad vision. Outsider perspectives are passively discouraged or even actively blocked out.

I have written in the past on Reginald Moreau and his contribution to ornithology thanks to being imported from accounting by C. B. Williams. Here are a couple of other stories from closer home.

A recent buzzword is "big-data". Like other fashionable words in business, this too will pass. But I like to extoll some of the great ideas related to data analysis that took root in India and became less important in later times. One of these was the Baconian ideal of evidence-based reasoning. It is interesting how India became a place where the idea of "facts" (an older word for data) and their collection grew in importance rather early. One of the early forces was W. H. Sykes, the man who brought the "Dukhun" (Deccan) into worldview, was perhaps the earliest data fancier. Sykes was a "statistical reporter" and founding member of the Statistical Society of London and when he returned to England he became an MP and a director of the East India Company where he championed methods of data collection for decision making. Sykes was amazingly interdisciplinary in his early data-collection work. Zoologists know him for the birds and fish that he collected but he was also a great collector of other kinds of data and a bit of an analyst. He calculated the efficiency of the E.I.C. and its army - comparing costs with those of the French. He looked at life expectancy, insurance and a number of ideas that were still unknown.

Sir Gilbert "Boomerang" Walker
Another place where data gathering became quite important was in the study of weather. The Indian Meteorological Department as we have seen earlier was established by an order signed by A. O. Hume (there are claims that Hume "established" the IMD but this is probably incorrect and it seems like more credit might be due to someone Hume did not like - Sir Richard Strachey) and was headed by a geologist Henry Francis Blanford (whose brother, another geologist, would plunge into zoology and be the founding editor of the Fauna of British India series).  One of the things that Blanford unearthed was a link between the amount of snow falling in the Himalayas and the monsoon in the subsequent season. Large scale patterns like the depression in barometric pressure across the country had already been detected before this. It soon became clear that someone with skills beyond traditional meteorology was needed and that was when Gilbert T. Walker known for his mathematical gifts left a teaching position at Cambridge to join the Indian Meteorological Department. He had already become a Fellow of the Royal Society thanks to his contributions to the mathematics of electromagnetism. Early in his career he worked on so much mathematics that he fell ill and needed to take breaks in Switzerland, a period in which he became an expert ice skater! Walker had other interests including boomerangs and had earned himself the nickname of Boomerang Walker. Walker like E.H. Hankin, the microbiologist, took an interest in the soaring of birds in India. When in India, he was frequently seen throwing boomerangs on the grounds of Annandale in Shimla. Here he worked on the physics of boomerangs and the mathematics behind the paths taken and examined parameters such as the twist and angle of the surfaces. Walker also began to examine weather data and developed methods to deal with time-series. Autoregression models today use what are called the Yule-Walker equations. The other contributor was Udny Yule who worked on sunspot patterns. (Interestingly it turns out that Walker's contributions were overlooked for years and one reviewer who had dismissed him was Herman Wold who helped developed the multivariate technique of PLS!) Using these methods Walker found large scale weather patterns across the southern hemisphere - what is now referred to as the ENSO (El Nino Southern Oscillation).

Thomas Nelson Annandale, a director of the Zoological Survey of India, was broad-minded enough to get someone with a statistical and mathematical perspective to examine data collected by physical anthropologists (the main collector was a Herbert Risley who has also been called a "scientific racist"). Annandale chose to hand out the dataset to P.C. Mahalanobis and that led to a major leap in multivariate analysis. The eponymous measure of distance (i.e. the opposite of similarity), Mahalanobis distance, is today one of the most widely used ways to examine the similarity of an object to a predefined group of objects and examine the likelihood of its membership to the group on the basis of a set of variables measured on a continuous scale.

Know an organization that needs to changes its valuation of specialists from other fields?