My Favorite Mistake: Paul Allen

Paul Allen at the opening day ceremony of the Flying Heritage Collection at Paine Field in Everett, Washington, on June 6, 2008. Michael O'Leary / The Herald-AP

In the summer of 1972, when I was home in Seattle from college and Bill Gates was heading into his senior year at Lakeside School, we were casting for our maiden business venture. Bill had contracted with a company that measured traffic patterns by counting the car wheels that ran over pressure-sensitive rubber tubes. Every 15 minutes, a machine would punch a sequence of holes onto a customized paper tape, with each pattern representing a number of cars. The tapes had to be manually read and then repunched onto batch-loaded computer cards.

I wondered instead about using a minicomputer. Intel had a new eight-bit microprocessor, the 8008, that I thought could process traffic-flow data analysis. We tracked down an electrical-engineering student, and soon he had a workable sketch for Traf-O-Data, the name Bill proposed. Armed with our data charts on hourly traffic flow, any county would know just where to install stoplights or focus road repairs. Bill and I scraped together $360 and picked up an 8008 chip at a local electronics store.

Two years later, we had a working prototype machine (built on a $1,500 budget). There was just one catch. Despite efforts to sell our wares as far afield as South America, we had virtually no customers. Traf-O-Data was a good idea with a flawed business model. It hadn't occurred to us to do any market research, and we had no idea how hard it would be to get capital commitments from municipalities. Between 1974 and 1980, Traf-O-Data totaled net losses of $3,494. We closed shop shortly thereafter.

Since then, I have made my share of business mistakes, but Traf-O-Data remains my favorite mistake because it confirmed to me that every failure contains the seeds of your next success. It bolstered my conviction that micro-processors would soon run the same programs as larger computers, but at a much lower cost. It also sparked my idea to simulate the 8008 microchip environment on a mainframe, which led to Altair BASIC—the first high-level language designed to run on a microprocessor. This was the essential step toward a personal computer that anyone could use, and the keystone for the creation of Microsoft.