Nervous System: The Tale of the Tape

While the modern computer user might be tempted to giggle at the seemingly anachronistic sight of tape reels used for computer data, the technology never went away.

In movies and TV shows of the mid-twentieth century, the iconic sight of spinning wheels of magnetic tape became the defining image of high-tech computers. To some degree, this was the result of filmmakers fixating on the only obvious moving part of an otherwise highly powerful but visually undramatic object. At the same time, filmmakers were in a position to point cameras at banks of reel-to-reel tape machines because computers of that era really did use them.

Prior to the concept of a “stored-program computer,” a computer, like any other tool, was engineered to a specific purpose. Its capabilities were literally hardwired into it. To change the programming of the thing meant disconnecting its myriad wires and reconnecting them to suit the new purpose.

By contrast, a stored-program computer had basic, generic instructions preinstalled. The user would load in a program and associated data, and the device would perform accordingly. A stored-program computer had unlimited functionality. This is not to say that such a device had unlimited resources—but rather, that its limited resources could be deployed on an unlimited number of applications. To change it over to a new application did not require manual rewiring—merely the loading of a new program and data.

This new relationship between hardware and software was mirrored in the relationship between live memory and persistent memory inside a stored-program computer. Live memory, used to hold programs and data in the process of being manipulated, prioritized speed over capacity. Persistent memory, used to store the programs and data to be loaded into active memory, prioritized capacity over speed.

When building the first stored-program computers, engineers J. Presper Eckert and John W. Mauchly opted to use punch cards for persistent data storage. It was a sensible, commercial, uncontroversial choice.

The use of perforated punch cards to store binary information dated back at least to the origins of computer science, when Charles Babbage and Ada Lovelace used them to store prototypes of programs, roughly a hundred years before any device capable of using those programs came along. Lasting international standards for paper punch cards had been established in the late 1880s, and many of the potential customers for the new generation of digital computers emerging in the mid-twentieth century already had punch-card readers.

When it came to developing the second generation of stored-program computers, however, Eckert and Mauchly put a different thumb on the scale. Neither the BINAC (BINary Automatic Computer) in 1949 nor the UNIVAC (UNIVersal Automatic Computer), which debuted in 1951, had a punch-card reader. Instead, Eckert and Mauchly had opted for a drive that used half-inch magnetic tape.

Magnetic tape was first developed in 1928 as a recording and storage medium for radio and the recording industry. By the 1950s, that industry had developed enough competition that viable competitors to supply high-quality, dependable tape were competing on price.

In 1951, a single reel of digital tape from IBM could store just over 1 megabyte of information. As paltry as this might sound to modern ears, this was equivalent to 10,000 punch cards. This level of compact, low-cost data storage was a game changer.

That is, as long as “cost” was calculated purely in terms of bytes per dollar. Measured in terms of time, magnetic tape could not be said to be efficient. Tape storage required the system to scan forward and backward through reels to locate specific units of data. The exciting images of spinning reels in movies and TV resulted from this physical fact—reading data into memory meant rolling forward and backward over and over again in sequential fashion.

To accommodate this fundamental inefficiency, data scientists developed new organizational tools to structure electronic data in a format that minimized the need to scan back and forth through a reel. Some of the organizational concepts behind relational database design, for instance, emerged from the desire to structure data in a form suited to storage on magnetic tape.

The decision to omit punch-card readers hurt the UNIVAC’s potential sales. Many customers, already comfortable with and committed to card readers, objected. Those that did install UNIVAC computers had to convert existing paper card data onto tape in order to use it.

At the same time, the choice of tape as a medium of preference meant that the new Computer Age would enjoy the benefits of increasing data density. The density of information one could pack onto a punch card was a known, immutable limit, imposed by a set of standards adopted for the 1890 Census. Manufacturers of magnetic tape, however, would find new ways of packing information more tightly together. Current tape systems boast as much as 15 terabytes on a single cartridge. While there is ultimately a limit to how densely magnetic information can be packed together before it loses fidelity, industry analysts estimate that tape manufacturers have decades of improvement to go before approaching that limit.

While the modern computer user might be tempted to giggle at the seemingly anachronistic sight of tape reels used for computer data, the technology never went away. Tape storage remains the preferred medium for archiving large data volumes, especially in the entertainment industry.

The views and opinions expressed in this article are those of the author and do not necessarily reflect the opinions, position, or policy of Berkeley Research Group, LLC or its other employees and affiliates.