Software Engineering Was a Joke Until the Moon Mission Made It

This is the 14th in an exclusive series of 50 articles, one published each day until July 20, exploring the 50th anniversary of the very first moon landing. You can check out 50 Days to the Moon here every day.

As Internet pioneer and venture capitalist Marc Andreessen wrote in 2011, software eats the world.

But that was not the case during the race for the moon in the 1960s.

On the contrary: the software was such a new idea that the spelling hadn’t even been fixed.

Sometimes it was ‘software’, but just as often it was ‘softwear’.

The words “software” and “softwear” were used interchangeably in newspaper articles, in headlines, and in job advertisements for well-established computer companies such as Control Data Corporation. The headline of a big ad in the St. Louis Post-Dispatch of Control Data Corp. on November 4, 1965 read “Computer Softwear Professionals” and listed vacancies in four job categories, including “Softwear Documentation”.

Honeywell, the minneapolis star reported March 7, 1962, had “new software being developed (which) includes an optical scanner, which can read information for the computer, and a data telephone system for communicating with computers”.

“Softwear’s Her Forte,” was the headline of a programmer’s profile in the Davenport, Iowa, newspaper on Quad City TimesMay 24, 1963.

In this interesting time in the world of computer software, it wasn’t just spelling that was messed up. It was not at all clear how to do large, complicated, nested software projects. What kind of discipline was computer programming anyway? Was it like science? Or math? Do you like to write? Or engineering?

There were well-established protocols for managing hardware engineering projects, and had been for half a century or more. The United States built the Hoover Dam in four years (1936), the Pentagon in 16 months (1943), and the Empire State Building rose at the rate of one story per day and was fully completed in 14 months (1931). Not to mention the execution of the Manhattan Project, part of the vast WWII logistics effort.

But there were no standardized practices or management systems for large software projects. In fact, when computer programmers at MIT began writing the software needed to fly to the Moon in 1961, there had been no software development project of the scale and complexity of Apollo. It was the first of an entirely new type of engineering project. MIT worked for more than a year on computers and software for a spacecraft that did not yet exist.

MIT software engineer Margaret Hamilton Morning is often credited with popularizing the term “software engineering”. Hamilton had graduated from college in 1958, joined the MIT Apollo project in 1963, and by 1969, just 11 years out of college, was supervising the command module software. “Software in the early days of (Apollo) was treated like a son-in-law and wasn’t taken as seriously as other engineering disciplines, such as hardware engineering,” Hamilton said, and that was especially true at an established engineering center like MIT. “It was considered an art and like magic, not a science. . . . I started using the term “software engineering” to distinguish it from hardware and other types of engineering. »

The phrase did not immediately catch on. “When I started using this phrase, it was considered quite amusing. It was a joke for a long time. (The phrase ‘software engineering’ started appearing in computer job advertisements in 1966.)

This first major software programming project for Apollo at MIT set two precedents.

First, he went wildly over space and time. Sometime in 1966, coders at MIT had written 40% more software than computers could handle, and NASA feared that MIT and the software would prevent the United States from reaching the Moon before the Moon. deadline of 1969. This kind of schedule performance has, unfortunately, characterized major software projects ever since.

But the MIT software and the computers it ran on were also widely considered a marvel: Computers with less intelligence than your dishwasher today did a masterful job of navigating spacecraft to the Moon and return. In 11 Apollo missions, nine of which flew to the Moon, there were 2,504 hours of spaceflight, more than 100 days. There was not a single software error or hardware issue recorded. The performance of the Apollo computers and software in flight was perfect.

This kind of performance, unfortunately, has not come to characterize software projects on Earth.

Charles Fishman, who wrote for fast business since its inception, has spent the past four years researching and writing One giant step, a book about how it took 400,000 people, 20,000 companies and a federal government to get 27 people to the moon. (You can order it here.)

For each of the next 50 days, we’ll be posting a new Fishman story – one you’ve probably never heard before – about the first effort to get to the Moon that sheds light on both the historic effort and the current effort. . New posts will appear here daily and will be distributed via Quick business’s social networks. (Follow on #50DaysToTheMoon).

Gordon K. Morehouse