Could computers make better international development decisions than humans?
Now at first, those present laughed off this question. It borders on fantasy to think a computer could take in the many social and cultural histories, divine the subtleties of donors and the parliaments behind them, and introduce innovations that have long-term impact on notoriously unpredictable humans. Or that’s what we thought until someone brought up the impact of computer algorithms on stock market activity.
Sign up to join us in future debates
Stock Market Algorithms
Back in the day, when you placed an order to buy or sell a stock, the trade went through floor traders at the New York Stock Exchange. Then, computerization of trades began in the early 1970s, and by 2009, high frequency trading, where computers initiate orders automatically based on algorithms, accounted for as much as 70% of all US equity trading volume. Man had been replaced by machine.
Even more, machine was better than man. The New York Times reports that high frequency trading routinely beats human-driven retain investors by $5 per trade, and added up to over $45,000 per month in profits, per trader.
The Role of Data
All those high frequency trading algorithms are made possible by the reams of data available on stock prices going back decades. Add in data that could influence stock markets, like bond, currency and other financial markets, and weather, agricultural, and industrial data, and there is a sea of numbers to analyze and thousands of stocks to use that data on.
With the exponential increase in data sets generated by new and cheaper sensors and the open data movement releasing that data into the wild where others can analyze it, new industries can start to utilize data to create their own versions of high frequency trading. We see the first step in this direction for the contractor community in DC with the launch of GovTribe.
Data Driven Development
Now let’s bring this home to international development. What do you think is the best indicator of height in Africa and India, where height is a proxy for early childhood health (good nutrition, low disease, etc)? Would you have ever thought that it was the level of open defecation in your community? Thanks to data, we know:
Child height is even more strongly associated with the average number of people per square kilometer in a country who practice open defecation. The density of open defecation per square kilometer… can account for 64% of international variation in child height.
And if we let computers find and act on this data set, our solution to stunting would not be to focus on food security to alleviate malnutrition, but on sanitation and sewer systems to reduce open defecation. And yet, what does the majority of childhood health funding support?
Data Beats Humans on Humans
Now some will still say that big data does fine at the policy level, but it could never understand the uniqueness of humans at the individual level. Many will hold up the gold standard of medical care, a primary care doctor working one-on-one with patients, as better than any cold-hearted computer. And they would be wrong.
Indiana University researchers found that they could reduce medical costs by 50% and increase patient outcomes by 40% if a computer chose the treatment via predictive modeling techniques vs. a doctor acting alone.
Your Days are Numbered
Think your job is safe because you do in-person capacity building? Talk to university professions in the age of MOOC. Believe nursing is safe because only humans can help humans heal? This is the future. And do you rely on taxi drivers to help you navigate a new city? Better speak code.
I, for one, welcome our new computer overlords. That’s why I now work at Development Gateway, where Open Data is the future.
Sign up to join us in future debates