On this day in 2008 Bill Gates spent his last formal day working with Microsoft.
But, as a salute, let’s consider how it all started forty years earlier – back in 1968:
Excerpt from The PC Pioneers
Bill Gates had an early introduction to programming from the age of thirteen. The Lakeside school he and Paul Allen attended had the foresight to raise funds to offer students access to computers by buying time from General Electric for use of its PDP-10. But Gates and his schoolmates used up all the time available far too quickly.
A new arrangement was made by the school with the Computer Center Corporation (CCC) but this too was abused when the students managed to crash the computer several times and eventually broke into the computer’s systems to alter their logged usage time so they could achieve more access than they were allotted.
They were discovered and banned from using the computer. This prompted four Lakeside students, including Gates and Allen, to form the Lakeside Programmers’ Group and seek out early applications on which they could apply their programming skills.
The very first application they found led them back to CCC. The group was asked to assist with tightening security, in return for unlimited computer access – a ‘no-brainer’ for Gates and his crew. CCC later had problems and went out of the business by 1970.
Other opportunities came along when the group was approached by Information Sciences Inc to write a payroll program. For this they were promised royalties on sales of any packages they produced. Shades of things to come!
The group also did work for TRW, a defence contractor, eliminating the bugs in its computer system. All these tasks expanded their access to computers, developed their knowledge and confirmed their desire to get into the business of computing.
Traf-O-Data – working with emulation
An interesting development came about when they learned of an application used by traffic engineers and local governments – simple rubber tube devices using pneumatics to count passing traffic.
The system counted the number of axles of passing vehicles and recorded it onto a 16-bit paper tape device. The standard at the time used on Teletype machines was 8-bit and this traffic device output could not be read directly into currently available systems. Instead various third parties were hired for a fee to manually work through the tapes and provide the data.
Gates and Allen believed they could automate the process and persuaded their schoolmates to transcribe the tape onto computer punched cards. Via Allen’s father they then gained access to the University of Washington’s computer to process the flow analysis. In this way they managed to undercut the existing suppliers of the service.
But they sought a more elegant approach and decided to develop a microprocessor-based device to read the tapes directly. They had no hardware or engineering skills so approached a friend of a friend, Paul Gilbert, an electrical engineering student at the University of Washington.
Gilbert based the approach upon the Intel 8008 MPU and took much of a year to develop the device. His brother Miles Gilbert designed the logo for the resultant business, Traf-O-Data. This was founded with Gates owning 43% of the equity, Allen with 36% and Gilbert with 21%.
I can find no article on the reasoning for these very odd percentages. If Allen was the one developing the emulation and Gilbert the one doing the hardware, why then did Gates get more, and why 43%? If it was that he was providing the funding then why did he not seek the ratio 51-49 in relation to the others? Why 21% for the hardware designer and not 20% or 25%?
The really interesting point for us is that while the Traf-O-Data computer did not yet exist, how could they develop the software other than to wait on the finished product? This was what prompted Paul Allen to try out computer emulation on a PDP-11, which would later prove to be the approach that they adopted with confidence when they developed BASIC for the Altair 8800.
Computer emulation had been developed at IBM by an engineer called Larry Moss. He wanted to run software from earlier mainframes on the IBM System/ 360 and developed a package for it to perform as if it was the earlier computer. He decided that emulation sounded better than simulation, the word implying that, while it ran the earlier software, it did so with the improved performance of the 360.
On the PDP-11 Gates and Allen, writing in assembly language, created and tested the Traf-O-Data in parallel with the hardware development. Unfortunately at the demonstration of the first unit the reader failed and soon after this the State of Washington offered to run the analysis for its cities at no charge, thereby effectively eliminating the traffic-logging requirement.
Yet the experience was to prove invaluable. When they encountered the Altair computer they were able to prepare a BASIC language for it without every using one – and Microsft was born!