My paean to the IBM 1403, with which I've spent many loving days and nights:
The clunky printer attached to the IBM 709 "mainframe" computer was a slow, lumbering monster. But the practice, in the day, was to use the smaller (only $250,000) IBM 1401 computer to load decks of program/data punched cards onto tape, the tape "mounted on the IBM 709" "mainframe" for execution, then the program's output to be written to tape (our 709 had 8--later 12--729 tape drives), and carried back to the IBM 1401 for printing of results. A "job ticket" specified which card decks went to tape, and which tapes would then be sent to the IBM 1403 attached to the 1401. And, that was the marvel: It could print several hundred pages in just a few minutes, often as graphs composed of asterisks, dashes, and other symbols, representing the points on the axes and the data points computed. Crude graphs, to be sure, but very effective to show non-technical executives. All in marvelous black (or blue) on white paper
The 1403 was the star of the show. Nobody much cared about the support task of copying boxes of punched cards to tape. They loved watching the lights on the huge "front desk" of the 709, the source of most TV footage of "a computer at work," in the day. But, they loved the speed, efficiency, quality, and distinctive (but relative quiet of the closed-box printer cabinetry) sound of that 1403. It meant we had results to see! Those of us who moved beyond FORTRAN (the preferred language on the big 709) found the 1401 computer a delight to program, with a memory structure of variable-length words with a "word mark" bit to distinguish the end of a string of characters...an architecture I'd love to see revived.
But, the 1403 was the workhorse of the business, and its' star performer. When results of huge warfare simulation models, or Linear Programming model forecasts of macroeconomic possibilities, often with foot-high stacks of large, wide pages emerged from the back of the 1403...faster than one could read them...everyone looked for the "macro trends" of big areas of ink (or barren spans of white), they gave insight into the likely success or failure of the most recent changes in the models...and, occasionally presaged teentsy bugs that had created hugely errant results.
Given the technology of the day (the laser was yet to be invented) all these technologies in the emergent era of modern computers were marvels, and the IBM 1403 was the most effective tool of them all. Without that ability to produce massive reams of output for later analysis by mathematicians and programmers, and executives, and analysts, we'd've never made the subsequent leaps that have led to the cellphones we have today.
ANECDOTE: True Side-Story about the masses of blinking lights on the 709. We were hard by the Pentagon, and contractors used our "service bureau" at C-E-I-R for doing warfare modelling. Most programmers cleverly used control over some of the 709's console lights to indicate progress, or other information. At $800/hour (in the 1960's) it was important to know of the results were likely to be good or bad, so we could quickly terminate the latter to save money. One fellow was building naval warfare simulation models, considering different weapons and tactics to maximize the achievement of battle outcomes, and he used one bank of lights to indicate which kinds of targets were being destroyed in the simulated battle. One day, I'm watching the lights flickering at a decent rate, when the programmer in charge of building and testing the model was watching those lights blinking, and suddenly, leapt out of his chair, reaching for the "kill" button, exclaiming, at the top of his lungs, "The Damned Thing's Attacking CARGO Ships!!!", as he pressed the button to reset the computer! No 1403 output from THAT job. :-)