But, perhaps surprisingly, electronic discovery software has not thrown paralegals and lawyers into unemployment lines. In fact, employment for paralegals and lawyers has grown robustly. While electronic discovery software has become a billion-dollar business since the late 1990s, jobs for paralegals and legal-support workers actually grew faster than the labor force as a whole, adding over 50,000 jobs since 2000, according to data from the U.S. Census Bureau. The number of lawyers increased by a quarter of a million.

Something similar happened when ATMs automated the tasks of bank tellers and when barcode scanners automated the work of cashiers: Rather than contributing to unemployment, the number of workers in these occupations grew.

These are not special exceptions. On average, since 1980, occupations with above-average computer use have grown substantially faster (0.9 percent per year), as shown in this chart:

The Average Annual Percentage Growth in Jobs Between 1980 and 2013

U.S. Census Bureau

How can this be? It might seem a sure thing that automating a task would reduce employment in an occupation. But that logic ignores some basic economics: Automation reduces the cost of a product or service, and lower prices tend to attract more customers. Software made it cheaper and faster to trawl through legal documents, so law firms searched more documents and judges allowed more and more-expansive discovery requests. Likewise, ATMs made it cheaper to operate bank branches, so banks dramatically increased their number of offices. So when demand increases enough in response to lower prices, employment goes up with automation, not down. And this is what has been happening with computer automation overall during the last three decades. It’s also what happened during the Industrial Revolution when automation in textiles, steel-making, and a whole range of other industries led to a major increase in manufacturing jobs.

But not all of the news about computer automation is good. Some of that growth in computer-using occupations has come at the expense of other occupations. As depicted in the chart below, desktop publishing systems have meant fewer jobs for typographers, as graphic designers took over their work. Computerized phone lines meant fewer jobs for telephone operators, but more jobs for receptionists. These aren’t examples of computers “stealing” jobs; they are cases where computers helped some workers take work from others. Workers with computers frequently substitute for workers in non-computerized jobs.

Number of Graphic-Design Jobs vs. Typographer Jobs, in Hundreds of Thousands

James Bessen; U.S. Census

So, while computers create jobs in some occupations, they also reduce employment in others. The total effect on unemployment depends on which tendency is stronger. Some of my research shows that the net effect, across the economy, is a wash: Computers create about as many jobs as they eliminate. In other words, automation is not causing persistent unemployment.