Lay pennies in a long row?
heads heads tails heads tails tails tails heads tails…
Or, better, the Apple version, which stores a bit more data in the same length row, and also doesn’t corrode as much. It’s all dimes.
Many years ago some programs were stored on punchtapes – long, skinny rolls of paper tape with holes punched in them like punchcards. Their advantage relative to punchards was that they were easier to store and you never had to worry about getting them out of order. The disadvantage was that any time the slightest thing went wrong with the takeup mechanism you’d have hundreds if not thousands of feet of paper tape spraying all over the room.
Since I don’t see it mentioned in the other thread, this has, in fact, been done. Had to be copied to a tape to actually play it, but it was done. The Thompson Twins Adventure.
Really, the tape storage was not much different than modem transmission. For some computers, the tape was entirely human-controlled. The original TRS80, for example, you had to fiddle with the volume on the cassette recorder output to get the correct volume/range for the computer to read. Often the result was hit-and-miss.
The problem was to find a method to store and read data that was cheap and affordable. Simplehome computers were about $1000, when that was real money. When sales were measured in tens of thousands, peripherals were not cheap. (My commodore PET 2001 cost $1200 with a built in 7-inch screen and built-in tape drive; the dual-disk add-on unit cost $1500; the cable cost $80. Back then that was EXPENSIVE!). Manufacturers tried to make do with what was available. A floopy drive and controller could cost almost as much as the computer, but most households had a portable cassette player already, and they were about $50 or less.
So storage worked with what was available. They worked around the limitations of that. Monitors were basically non-existent, but a TV interface was cheap. However only about 40 characters across per line was readable on most TV’s. Similarly, home cassette recorders were designed for and capable of a limited audio range. The cassette recording speed for computers was an adequate compromise between the bandwidth and speed, plus had to use off-the-shelf chips for processing the input (so no complicated encoding schemes in those days).
The problem with records that I see is simple - playback quality. It would not take long before your audio records were full of pops and clicks from dust, unless you were aggressively clean. A slight accident scratching a record would mean an annoying click on a song, but might ruin the recording of a program. You could maybe program to record blocks with correctible checksums, but that introduced another layer of complexity when bytes of ROM or RAM were at a premium.
A cassette recorder was typically small and portable; most record players in the late 70’s and early 80’s hooked the turntable to the big stereo amp. Either you set up the computer in the living room (or rec room) next to the sound system, or dragged your audio equipment into the study. Then there’s the issue of finding the correct volume setting.
Finally, we’re still talking volume. When the number of a specific computer sold was under 100,000 it was not practical to press records - you’d need to do a big batch run and pay for it. Computers then were not big business. Cassettes could be recorded on demand. Every model of computer used a different program; heck, it was a battle just to get the early MS-DOS computers to be sufficiently compatible. Plus, cassettes recorded in small batches as needed, allowed for bug fixes if the program had to be tweaked.
Apple’s secret weapon was the genius of Wozniak. When other computers were struggling with cassette tapes, he realized a complicated floppy controller board could be made a lot simpler and cheaper if the CPU was programmed to do much of the work in software that others did with hard-wired chips. As a result, Apple began to dominate the pre-PC market and the price of diskette drives fell significantly, from $500 or more to under $100 as time went on.
I suppose data could be distributed on vinyl, but the cassette was more practical. By the time computers were big enough business, the floppy was standard.
I think the big trick in making a cassette or LP data recording is in md2000’s post above. Rather than custom logic, make the CPU do as much of the work as you can. Building a reasonably robust block error correction system in software would not have been all that hard. Some of what was missing was the mathematics of how to do it - it was still buried in an obscure pure mathematics publication, and only made it to public attention when the CD standard used it. Your average hobbyist programmer would not have been aware of Reed Solomon codes, and the idea of such aggressive and effective error correction just not on the radar.
That said, the Kansas City standard was arguably well below what could have been achieved with only a few dollars of components. But it was good enough when many computers measured their total memory capacity in kilobytes. Knowing what we do now, I would be reasonably confident that we could build a cassette interface for less than $50 in parts for a Z80 or 6502 and 1975 technology, that would reliably get 10’s of kilobytes per second. But back then, that is a different matter.
Curiously this bit isn’t true. The cost to create a master and press a set of records was remarkably cheap. So much so that most pressing plants would do vanity pressings with runs of a hundred odd. I remember in 1979 my uni’s comedy theatre troop (essentially a clone of the Cambridge Uni Footlights) did such a pressing, and hawked it around after performances. When you were only selling a few tens of copies, cassettes were more cost effective, but LPs took over reasonably swiftly once you got into the hundreds.