Comment by Animats | Hacker News Reader

Comment by Animats | Hacker News Reader

Parent Full thread

Nah. The computer was created as a follow-on to the IBM 601.[1]

There's a whole line of development, the IBM 600 series, which gets no respect in computer history, yet led to the first mass produced business computers.

IBM 601 (1939). IBM's first multiplier. All electromechanical. Punched cards in and out.

IBM 602 (1946, World War II stalled development) Electromechanical. Added division. 6 memory locations.

IBM 603 (1946, project resumed from 1937) Vacuum tube version of the 602. At last, electronics. Only about 20 made.

IBM 604 (1948) A bigger, better 603. 5600 manufactured. First electronic computing product produced in volume. Things are getting serious. There are now one-off larger electronic computers, but they're too expensive for commercial use.

IBM CPC. (1949) An IBM 604 or 605, cabled up to several other IBM tabulating machines, allowed programming via punched cards instead of plugboards. About 2500 made.

IBM 650 (1953) Knuth's first computer. Tube computing, drum memory. This was a real general purpose computer, and the first one made in quantity. About 2000 made. Lots of accessories available - printers, card readers, card punches, and, in 1956, disks. Now we're getting somewhere.

IBM 608 (1955) A successor to the 604/CPC line, made with transistors.

That was the end of the IBM 600 line. Next was the IBM 1401 (1959), which was a quite good computer. Over 10,000 made. The mainstay of business computing in the 1960s.

These boring but useful 600 series machines were cost-effective, worked well, and sold well. IBM also did expensive one-offs and low-volume systems, as the 700 series (701 through 7094) that were more bleeding edge. Most early ones were for government-related "supercomputer" customers.

In 1964, IBM announced the IBM System/360, which brought together the "supercomputer" 700 line and the workhorse 600 series. That lives on today as the IBM Z-series.

And that's how computers became a volume product.

[1] https://en.wikipedia.org/wiki/IBM_601

Author here. Sure, there are lots of ways to tell the history of computers, because lots of different lines of development fed into their creation. I hint at this in the prologue.

But I chose to highlight one particular aspect (the origins of the basic components used to build digital computers - relays, tubes, and transistors). This naturally downplays the importance of IBM relative to other possible ways of telling the story. There's a mea culpa somewhere in the footnotes about the neglect of IBM.

On the basic component end, the history of early electronics was a search for gain. Modern components have so much gain you can throw much of it away to make digital switches or negative feedback amplifiers. Early electronics struggled to get enough gain to drive headphones. Lack of good amplifiers limited the range of long distance telephones.

There's a whole history of forgotten attempts to get some gain. Edison's electromotograph. (See if you can find out how that works. Edison discovered variable friction depending on electric current, but the physics wasn't understood until the 1960s or so.) A speaker driver pushing against a carbon microphone (early Bell System long distance). Higher power devices included magnetic amplifiers (the UNIVAC Solid State computer used those) and large Ward-Leonard drives, used for elevator control for many decades.

There was a long struggle on the transmit side of radio to get gain at high power. The Alexanderson alternator, a high frequency AC generator, was the first real success. There's a long history of transmitting tubes, finally maxing out in radar systems for the DEW line that had tubes with their own vacuum pumps.

Another kludge was the super-regenerative receiver, where positive feedback was used to run the same signal through the same stage more repeatedly. Noted for producing loud squealing sounds.

There's also some history on the musical side. Look up the Teleharmonium, a musical instrument the size of a building. One was actually built. Lacking an amp, they had a sizable AC generator for each note. There's also the switchgear used in large pipe organs, with both electrical and pneumatic components. Those were workarounds for not having enough gain.

We're spoiled today. You can hold the semiconductors used for controlling a locomotive's power in your hand. The first half of the 20th century, though, was a struggle to get enough gain to get anything done.

The switch theme is really interesting because I've been thinking about it in the context of quantum--qubits--recently. There's nothing really inherent that makes binary logic rather than ternary logic or analog inherently superior. It just turns out that making on-off switches has historically been relatively easy and cheap. I plan to read the book.

Binary logic is usually composed of analog amplifiers designed to quickly hit some upper limit and saturate. In the off state, little power is dissipated, because current flow is tiny. In the on state, little power is dissipated, because resistance is low. In between, the "linear region", is where heat is generated. The goal is usually to spend as little time in the linear region as possible.

Most qubit designs use high- and low-energy states to encode binary values. In that regime, there is a reason to avoid 3-level systems: the energy gaps are hard to tune, and controlling the coupling between them is even harder.

I particularly liked the way the series covered the way science and engineering progresses, through the almost blundering interaction of geniuses, cranks, fast buck artists, and bureaucrats over time.

Reply on news.ycombinator.com