At his Gates Notes blog, Bill Gates commemorates the 50th anniversary of Microsoft with a story about the company’s first code:
The story of how Microsoft came to be begins with, of all things, a magazine. The January 1975 issue of Popular Electronics featured an Altair 8800 on the cover. The Altair 8800, created by a small electronics company called MITS, was a groundbreaking personal computer kit that promised to bring computing power to hobbyists. When Paul [Allen] and I saw that cover, we knew two things: the PC revolution was imminent, and we wanted to get in on the ground floor.
At the time, personal computers were practically non-existent. Paul and I knew that creating software that let people program the Altair could revolutionize the way people interacted with these machines. So, we reached out to Ed Roberts, the founder of MITS, and told him we had a version of the programming language BASIC for the chip that the Altair 8800 ran on.
There was just one problem: We didn’t. It was time to get to work.
The detail of this story that I hadn’t remembered is that they didn’t have access to the chip. Instead, Paul Allen wrote an Intel 8080 simulator on Harvard’s PDP-10 mainframe so they could develop and test their BASIC interpreter without having actual Altair hardware. It was a technical feat, but Harvard wasn’t amused at the use of academic resources for commercial purposes.
The university didn’t fare so badly in the end. Bill Gates and Steve Ballmer have since donated about $85 million to Harvard, and the Bill & Melinda Gates Foundation has donated another $103 million for medical and educational research.
Along with the first-hand tale and grainy black-and-white photos, I was charmed by how mousing over the text briefly replaces letters with non-alphabetic characters, turning the post into interactive ASCII art that cleverly evokes the character-based displays of the time.
Absolutely, and this was far from the only example.
It’s well known, for example, that Gates/Microsoft promised an operating system to IBM for the PC when they had literally nothing. They ended up buying someone else’s operating system (86-DOS, aka “QDDOS”), changing little more than the name, and delivering that to IBM.
Of course, DOS got rewritten several times since then (there was very little 86-DOS code in MS-DOS 2.0), but it doesn’t change the fact that Microsoft got that contract by selling a product they didn’t actually have and were not yet developing.
Absolutely vaporware, but there’s a big difference between convincing another company that you have something and marketing it to consumers when you don’t.
In the MITS case, Gates and Allen didn’t have anything, but Ed Roberts didn’t pay them anything upfront, and they only got a royalty licensing deal once they showed the successful demo. With IBM, if my quick research is correct, IBM did pay $50,000 upfront but could easily have walked away and gone with CP/M or built their own operating system.
In both cases, Gates may have overpromised, but he also delivered, once by fast turnaround and once by buying 86-DOS.
The 4K interpreter took up about 3200 bytes of 4096.
A 32bit floating point subroutine package with transcendentals took about 1600 bytes.
That means they implemented the rest of the interpreter in 1600 bytes. That is really tight coding!
Mits came out with the Altair 680b that used a 6800 CPU. It ran at half the speed the stock 6800 could run at.
At full speed CPU it would do the benchmark in 3.2 seconds.
The 1 MHz 6800 was faster than the 4 MHz 8080!
The processor internal architectures were different.
Mits was later bought by Pertec. They thought they were getting Basic. Gates was never a Mits employee. As such, and because his father was a lawyer, they were able to sue and declare the license that Mits had was not exclusive. Hence TRS-80 et al.
When Gates and Allen went to Mits they took a punched paper tape with the software on it. It loaded and ran the first time, or so the story goes.