An extension of this is how in the 90s, people were in the habit of calling the rectangular box part of a desktop computer (the case and everything inside, the bit that isn’t the monitor or external peripherals) “the CPU”.
1950s: a computer consists of a CPU and peripheral equipment; the CPU is often referred to by its housing, the main frame.
1960s: IBM is forced to allow competitors to make plug-compatible add-ons; as a result, computer manufacturers are typically divided into mainframe manufacturers (who make CPUs) and peripheral equipment manufacturers (who don’t).
[This is the crucial missing link in the etymology that I have not seen before: it explains a lot!]
1970s: the rise of minicomputers and microcomputers creates the need for a word for large computers; since IBM and the seven dwarfs had been called mainframe manufacturers, the large computers they made became called mainframes – but “mainframe” still often referred to the central box of minis and even micros.
1980s: the terminology settles, and “mainframe” becomes specifically “large computer”
Yes, that’s a good summary. (Author here.) Another factor is that terms such as minicomputer didn’t have a solid definition; essentially minicomputer manufacturers such as DEC built minicomputers, while mainframe manufacturers such as IBM built mainframes, even if a high-end DEC computer was more powerful than a low-end IBM computer.
I’ve always heard (read) that “minicomputer” referred to “minimal computer” and came from DEC, where “minimal” meant something small (relatively, considering the size of most other computers) you might dedicate to specific uses, like a specific department or to control tasks.
While that could be an oversimplification, I always sound out “minimal computer” in my head to remember the distinction between computers that cost as much to rent per month as a house costs to buy versus computers that only cost as much to buy as a fancy house.
I think it might be worth adding to the etymology that telephone exchanges had been organised into “frames” for quite some years before electronic or even electromechanical computers existed: the Harvard Mark 1 could easily be viewed as continuing that tradition.
I found myself slightly surprised a few weeks ago when I came across the fact that one of the standard beehive frame sizes (Langstroth) is 19”, which might possibly suggest an origin for the 19” rack used originally for instrumentation but later for modularised computer equipment: different traditions borrowing from each other.
Names like “mini” and “micro” have always been subjective: the Burroughs B700 was labelled a “microcomputer” because (simplified account) they considered that its instruction set corresponded to the microcode on other computers (it actually had “nanocode” in ROM).
It might possibly be best to focus on the “main” in “mainframe”: it was the centralised computer facility indispensable to any self-respecting enterprise, even if it was surrounded by tabulators, datacomms frontends or whatever.
An extension of this is how in the 90s, people were in the habit of calling the rectangular box part of a desktop computer (the case and everything inside, the bit that isn’t the monitor or external peripherals) “the CPU”.
My over-simplified summary:
1950s: a computer consists of a CPU and peripheral equipment; the CPU is often referred to by its housing, the main frame.
1960s: IBM is forced to allow competitors to make plug-compatible add-ons; as a result, computer manufacturers are typically divided into mainframe manufacturers (who make CPUs) and peripheral equipment manufacturers (who don’t).
[This is the crucial missing link in the etymology that I have not seen before: it explains a lot!]
1970s: the rise of minicomputers and microcomputers creates the need for a word for large computers; since IBM and the seven dwarfs had been called mainframe manufacturers, the large computers they made became called mainframes – but “mainframe” still often referred to the central box of minis and even micros.
1980s: the terminology settles, and “mainframe” becomes specifically “large computer”
Yes, that’s a good summary. (Author here.) Another factor is that terms such as minicomputer didn’t have a solid definition; essentially minicomputer manufacturers such as DEC built minicomputers, while mainframe manufacturers such as IBM built mainframes, even if a high-end DEC computer was more powerful than a low-end IBM computer.
I’ve always heard (read) that “minicomputer” referred to “minimal computer” and came from DEC, where “minimal” meant something small (relatively, considering the size of most other computers) you might dedicate to specific uses, like a specific department or to control tasks.
While that could be an oversimplification, I always sound out “minimal computer” in my head to remember the distinction between computers that cost as much to rent per month as a house costs to buy versus computers that only cost as much to buy as a fancy house.
Great article! Thanks :)
I think it might be worth adding to the etymology that telephone exchanges had been organised into “frames” for quite some years before electronic or even electromechanical computers existed: the Harvard Mark 1 could easily be viewed as continuing that tradition.
I found myself slightly surprised a few weeks ago when I came across the fact that one of the standard beehive frame sizes (Langstroth) is 19”, which might possibly suggest an origin for the 19” rack used originally for instrumentation but later for modularised computer equipment: different traditions borrowing from each other.
Names like “mini” and “micro” have always been subjective: the Burroughs B700 was labelled a “microcomputer” because (simplified account) they considered that its instruction set corresponded to the microcode on other computers (it actually had “nanocode” in ROM).
It might possibly be best to focus on the “main” in “mainframe”: it was the centralised computer facility indispensable to any self-respecting enterprise, even if it was surrounded by tabulators, datacomms frontends or whatever.