Gross domestic product, perhaps the most commonly used statistic in the world for evaluating economic progress, has some issues.
Increasingly, one of the biggest problems is that GDP generally underestimates the value of free goods and services—checking facts on Wikipedia or sharing photos on Instagram, for instance. GDP is best at measuring the impact of TV and car sales—not of things available for free or that require you to view ads, like broadcast TV or Facebook, explains the Financial Times’s Gillian Tett.
As a new research paper points out, this shortcoming also means GDP may be missing a lot of value created in the form of free programming languages (pdf). The most popular programming languages, like JavaScript and Python, are open source. This means that anyone can use them for free and modify them to develop new programs that they can then offer for free or for sale. JavaScript, for example, is used on about 95% of websites. Python, the most popular tool for data scientists, is used by companies like Google and Facebook to analyze data and develop new products.
What makes GDP buggy
If a company buys proprietary data science software—like SPSS, SAS or STATA—that purchase makes it into GDP. Yet if the company chooses to use a free, open-source language for statistics—like Python or R—that choice never makes into the GDP, even though they are very similar products.
Also, if a company like Google gives a web developer time to work on improving JavaScript or Python, the payments made to the developer make it into the GDP. Yet if a person chooses to contribute to those languages on their own time, as legions of devotees do, that is not directly counted—even though both create value. (This is similar to how unpaid housework, typically done by women, does not make into GDP.)
In their research paper, a set of economists and data scientists led by Carol Robbins of the National Science Foundation, attempt to quantify a share of the value being created by these programming languages. They do this by examining the code used to develop JavaScript, Python, R, and the statistical programming language Julia that is available on the code repository site Github.
Closing the open-source loop
They find that based on the typical pay of computer programmers, the cost to develop the Github code for these languages would be over $3 billion—much of which was unpaid for. This is a very low-end estimate: A great deal of the code is not on Github and these are only four of the many open-source languages. The actual value could be magnitudes greater.
With numbers like these missing from GDP, it’s hard to know exactly how much economic growth is happening. Dour reports of slow growth in worker productivity may be oversold.
It’s good that researchers are beginning to assess the value of “free” goods. It probably won’t completely change our views, but it is likely that the economy is doing a bit better than conventional statistics suggest.