I'm not sure how you are reading emotions like skepticism into my tone, but I assume it's just the way it's worded...
I suppose I should apologize for sounding a little exasperated. I genuinely enjoy sharing whatever I can about the era you're interested in. (... And of course, everything I say should be taken with plenty of grains of salt, because my view of what was happening is certainly going to be incomplete and distorted by the point of view I was taking it in from, which was as a nerdy *kid* who devoured a lot about these new computer things but certainly wasn't working in the industry or whatever.) I think maybe I'm just getting a little frustrated that you keep coming around to inserting "CP/M" into the discourse as if it was essentially synonymous with "pre-IBM PC era computer", because, well, it just
wasn't. Not in the US, and not *anywhere*, really, at least to the sort of degree it would have had to have been for your thesis to make sense.
To be clear, I'm not saying that there *wasn't* a "CP/M era", there was. In fact, if we dig deep enough I think we can make the case there were *multiple* CP/M eras. (I'll get back to this.) The mistake you keep perpetuating in this conversation is the idea that CP/M ever really represented a monolithic unified platform that can meaningfully be compared to what 80x86 ecosystem kicked off (on a wide scale) by the introduction of the IBM 5150 properly gelled into sometime around the dawn of 1984, IE, the "PC Compatible". You're working *so hard* to compare these two things as if they were apples and oranges, when really what you're comparing is apples (and not even all apples, we're pretty specific here; let's just go with Red Delicious, because it's really popular despite being kind of mediocre) and a table full of fruit salads. Not *all* fruit salads, mind you. There are lots of other fruit salads out there, made with whipped cream, sugar syrups, honey, orange juice, olive oil, Jello, you name it, but the only ones that count in your equation is the ones made with mayonnaise. Because the fruit salads are all 8-bit computers. CP/M is the mayo.
One Red Delicious apple is, outside of minor factors like size, ripeness, and whether it's already been thrown at someone at a comedy club, is interchangeable with any other Red Delicious. It's a known quantity. If you go buy a Red Delicious you know what you're getting, and any apple-related accessories like slices and peelers are going to work with it. This is the PC Compatible's deal.
On the other hand, the only thing you know when you buy a CP/M computer is you'd better like mayo, because you'll be tasting it underneath whatever Mandarin oranges and avocado slices were dumped into the bowl to make it. And the spoons you use to eat it aren't going to be compatible with your neighbor's fruit salad unless they bought it from the same store as yours. I mean, sure, I guess it's probably going to take less conversion to swap spoons than it would be for you to eat something from your other neighbor's house, IE, the guy who bought the whipped cream Ambrosia salad, but it's still almost as annoying swapping between the various manufacturers' mayo salads as the non-mayo ones...
Okay, that analogy sort of fell apart at incompatible spoons, but maybe you get the drift. In the murky origins of the First Age of Personal Computers there was no unified single standard, no One Ring to Rule them All, there was simply chaos. A "PC" was a collection of rickety circuit boards built by your local madman with his own three hands, and software consisted of novelty renditions of mediocre Beatles songs laboriously keyed into the front panel switches because that's about all you could do with it. Obviously that state of affairs didn't last and by mid-1975 If You Were a Rich Man (like the Beatles) you could have a "thing" had a keyboard and some kind of display and enough memory to run Altair 4K BASIC, and within a few months of that if you were truly in the Cadillac set you might be in the market for one of those floppy disk drives that until recently you probably needed a contract with IBM to have heard of. At this point it's around 1976 or so outside, and wouldn't you know it, it turns out that there was a guy working for Intel a couple years ago with access to a development system that cost about half the price of your house that had written an operating system for the 8080 that could, with a little work, be adapted to your Altair (if that's what you had happened to buy) and run on your disk system that now only cost about as much as your car. Hoorah! And thus the legend of CP/M being the Founding Father of PC operating systems was born because *obviously* it was immediately the one true choice for every disk-based computer made since. (Until IBM came along and ruined it, obviously.)
Again, here's the problem with view of history: Remember that part about how a CP/M compatible disk system in 1976 cost more than your car? Hard numbers for how many Personal Computers were sold prior to 1977 are pretty hard to come by, but a reasonable figure seems to be around 40,000 units, total, by the end of 1976. Optimistically around 80% of those were 8080 based, and thus at least *potentially* a target for CP/M, but it's a very certain bet that a lot fewer than 80% of those had disk drives before 1977. Meanwhile, in 1977 a funny thing happened: the "Byte Trinity" of the TRS-80, Commodore PET, and Apple II appeared, and by the end of 1977 those three systems combined sold about as many units as all S100 computers up to that point had because, well, they were immensely cheaper and far easier to use "out of the box" than just about any previous system, at least systems without disk drives. (which, again, still roughly tripled the cost of a basic computer to add them.) Now, the existence of these systems certainly didn't *stop* DRI from licensing CP/M to the companies making expensive disk based computers, since these had started finding a profitable niche selling to businesses that could afford them, and sales of these systems kept ramping up, but at a substantially slower rate than these new user-friendly toy/home computers.
Over the next couple years, of course, 1978 through 1979, these cheap computers (the sales of which were continuing to ramp up through the roof) gained the option of disk drives themselves but... funny thing here, none of these cheap computers could run CP/M, either because of using the wrong CPU or having the wrong memory map (the TRS-80's memory map with ROM on the bottom was, strictly speaking, the "right" way to make a cheap-as-possible Z80 based computer with a ROM based OS). And, wouldn't you know it, it didn't matter, all of these machines went on to be successful anyway. They in fact set a a pretty clear precedent: the things that mattered to be successful in the home computer market were price and features, not the specifics about what CPU you used or what OS it ran. The very *concept* of being "compatible" with other systems just didn't really enter into the equation; you couldn't even complain much about these systems lacking compatibility compared to the more expensive business computers, because they weren't very compatible with each other either, for all the already cited reasons. Incompatibility was the
norm in the personal computer market. The fact that CP/M computers were a shade less mutually incompatible than non-CP/M computer really wasn't much of a selling point.
(I mean, sure, it wasn't *nothing*; CP/M spawned some of the first libraries of "public domain" software, a lot of which ended up being uploaded to the early public online information systems like Compuserve, but paying $6 an hour to access Compuserve with your $4,000 computer wasn't exactly mainstream, so the impact of this was pretty limited.)
With this background maybe now you can process why IBM, when vetting CPUs for the 5150, restricted themselves to 16 bit CPUs like the 8088, TMS9900, and Motorola 68000: they looked at the current state of the Personal Computer market and had the wisdom to see that
compatibility didn't matter. Full stop. What mattered was price, performance, features, and value, and like it or not, the 5150, with its incompatible CPU and all, matched or beat most CP/M business machines in nearly every category, in addition to coming along with a solution to a problem that was starting to grate on 8-bit business computers, IE, the awkwardness of expanding past 64K of RAM. (Its price was too high to compete with most *home* computers initially, but the expansion of the semiconductor and computer industry eventually solved that problem.) It was in large part an accidental stroke of genius that IBM happened to blunder into commissioning an OS from Microsoft that made it extra easy for existing CP/M business software to be ported to the IBM PC, but I honestly suspect it would have been just as successful if it hadn't included that token bit of CP/M API compatibility. (Like I noted earlier, completely different APIs, or CPUs, were rarely that much of a barrier for porting the tiny programs of the 8-bit era.)
In short, CP/M never was as big a deal as you think it was. It was a significant part of computer history, yes, and it served a valuable role in developing the concept of microcomputers for business automation (vs. either the “old ways” or much more expensive minicomputers), but in the First CP/M Age it never really was a ”personal”, IE, “home” computer OS, at least in the US, with a following even remotely comparable to that of other 8-bit platforms.
