Jul 29, 2019
Welcome to the History of Computing Podcast, where we explore the history of information technology. Because by understanding the past, we’re able to be prepared for the innovations of the future! Todays episode is on Digital Equipment Corporation, or DEC.
DEC was based in Maynard Massachusetts and a major player in the computer industry from the 1950s through the 1990s. They made computers, software, and things that hooked into computers. My first real computer was a DEC Alpha. And it would be over a decade before I used 64-bit technology again.
DEC was started in 1957 by Ken Olsen, Stan Olsen, and Harlan Anderson of the MIT Lincoln Laboratory using a $70,000 loan because they could sell smaller machines than the big mainframes to users where output and realtime operation were more important than performance. Technology was changing so fast and there were so few standards for computers that investors avoided them. So they decided to first ship modules, or transistors that could be put on circuit boards and then ship systems. They were given funds and spent the next few years building a module business to fund a computer business.
IBM was always focused on big customers. In the 1960s, this gave little DEC the chance to hit the smaller customers with their PDP-8, the first successful mini-computer, at the time setting customers back around $18,500. The “Straight-8” as it was known was designed by Edson de Castro and was about the size of a refrigerator, weighing in at 250 pounds. This was the first time a company could get a computer for less than $20k and DEC sold over 300,000 of them! The next year came the 8/s. No, that’s not an iPhone model. It only set customers back $10k. Just imagine the sales team shows up at your company talking about the discrete transistors, the transistor-transistor logic, or TTL. And it wouldn’t bankrupt you like that IBM. The sales pitch writes itself. Sign me up!
What really sold these though, was the value engineering. They were simpler. Sure, programming was a little harder, and more code. Sure, sometimes that caused the code to overflow the memory. But at the cost savings, you could hire another programmer! The rise of the compiler kinda’ made that a negligible issue anyway. The CPU had only four 12-bit registers. But it could run programs using the FORTRAN compiler anruntime, or DECs FOCAL interpreter. Or later you could use PAL-III Assembly, BASIC, or DIBOL.
DEC also did a good job of energizing their user base. The Digital Equipment Corporation User Society was created in 1961 by Edward Fredkin and was subsidized by DEC. Here users could trade source code and documentation, with two DECUS US symposia per year - and there people would actually trade code and later tapes. It would later merge with HP and other groups during the merger era and is alive today as the Connect User Group Community, with over 70,000 members! It is still independent today. The User Society was an important aspect of the rise of DEC and of the development of technology and software for mini computers. The feeling of togetherness through mutual support helped keep the costs of vendor support down while also making people feel like they weren’t alone in the world. It’s also important as part of the history of free software, something we’ll talk about in more depth in a later episode.
The PDP continued to gain in popularity until 1977, when the VAX came along. The VAX brought with it the virtual address extension for which it derives its name. This was really the advent of on-demand paged virtual memory, although that had been initially adopted by Prime Computer without the same level of commercial success. This was a true 32-bit CISC, or Complex Instruction Set Computer. It ran Digital’s VAX/VMS which would later be called OpenVMS; although some would run BSD on it, which maintained VAX support until 2016. This thing set standards in 1970s computing. You know Millions of instructions per second (MIPS) - the VAX was the benchmark. The performance was on par with the IBM System/360.
The team at DEC was iterating through chips at a fast rate. Over the next 20 years, they got so good that Soviet engineers bought them just to try and reverse engineer the chips. In fact it got to the point that “when you care enough to steal the very best” was etched into microprocessor die. DEC sold another 400,000 of the VAX. They must have felt on top of the world when they took the #2 computer company spot! DEC was the first computer company with a website, launching dec.com in 85.
The DEC Western Research Library started to build a RISC chip called Titan in 1982, meant to run Unix. Alan Kotok and Dave Orbits started designing a 64-bit chip to run VMS (maybe to run Spacewar faster). Two other chips, HR-32 and CASCADE were being designed in 1984. And Prism began in 1985. With all of these independent development efforts, turf wars stifled the ability to execute. By 1988, DEC canceled the projects. By then Sun had SPARC, and were nipping at the heels.
Something else was happening. DEC made mini-computers. Those were smaller than mainframes. But microcomputers showed up in the 1980s with he first IBM PC shipping in 1981. But by the early 90s they too were 32-bit. DEC was under the gun to bring the world into 64-bit. The DEC Alpha started at about the same time (if not in the same meeting as the termination of the Prism project. It would not be released in 1992 and while it was a great advancement in computing, it came into a red ocean where there were vendors competing to set the standard of the computers used at every level of the industry. The old chips could have been used to build microcomputers and at a time when IBM was coming into the business market for desktop computers and starting to own it, DEC stayed true to the microcomputer business.
Meanwhile Sun was growing, open architectures were becoming standard (if not standardized), and IBM was still a formidable beast in the larger markets. The hubris. Yes, DEC had some of the best tech in the market. But they’d gotten away from value engineering the solutions customers wanted.
Sales slumped through the 1990s. Linus Torvalds wrote Linux on a DEC Alpha in the mid-late 90s. Alpha chips would work with Windows and other operating systems but were very expensive. X86 chips from Intel were quickly starting to own the market (creating the term Wintel). Suddenly DEC wasn’t an industry leader. When you’ve been through those demoralizing times at a company, it’s hard to get out of a rut. Talent leaves. Great minds in computing like Radia Perlman. She invented Spanning Tree Protocol. Did I mention that DEC played a key role in making ethernet viable. They also invented clustering. More brain drain - Jim Grey (he probably invented half the database terms you use), Leslie Lamport (who wrote LaTex), Alan Eustace (who would go on to become the Senior VP of Engineering and then Senior VP of Knowledge at Google), Ike Nassi (chief scientist at SAP), Jim Keller (who designed Apple’s A4/A5), and many, many others.
Fingers point in every direction. Leadership comes and goes. By 2002 it was clear that a change was needed. DEC was acquired by Compaq in the largest merger in the computer industry at the time, in part to get the overseas markets that DEC was well entrenched in. Compaq started to cave from too many mergers that couldn’t be wrangled into an actual vision. So they later merged with HP in 2002, continuing to make PDP, VAX, and Alpha servers. The compiler division was sold to Intel, and DEC goes down as a footnote in history.
Innovative ideas are critical to a company surviving after the buying tornadoes. Strong leaders must reign in territorialism, turf wars and infighting in favor of actually shipping products. And those should be products customers want. Maybe even products you value engineered to meet them where they’re at as DEC did in their early days.