Listen to Linus Torvalds when he says that the PC is so easy compared to the nightmare of all those incompatible arm SOCs.
I dream of a smartphone PC.
Entirely accidental. What really started this was the reverse engineering of the BIOS. This allowed any company to enter the market with a PC-compatible computer. IBM tried, but clean-room proved to be an effective defense.
> Without it, we would have many weak walled gardens
Kind of. There were Apple II clones with varyingly clever schemes around copying the original ROMs, and it provided slots so many manufacturers could provide hardware for the machines. Microsoft was one of them, with a Z-80 card that allowed the II to run CP/M software. Atari had a serial bus that allowed external expansion - less popular than the II, but still allowed other manufacturers to expand it. And let's not forget the ubiquitous S-100 bus, with hundreds of different companies making hardware for. In 1983 you could assemble a computer around your choice of processor, graphics (or not), IO, mass-storage, and so on.
> incompatible with each other
Kind of. Prior to MS-DOS, CP/M was the de-facto 8-bit standard for business computers. It had a multi-user version called MP/M, and versions for processors other than the 8080/85 and Z-80. In the higher end, Unix-compatible small computers started to appear - the TRS-80 model II could get a 68000 board, for instance.
It looks like the vast majority of clones used a (near bit-by-bit) copy of the original ROMs. I've only heard of one company using licensed ROMs and one company using reverse engineered ROMs in the American market.
> And let's not forget the ubiquitous S-100 bus, with hundreds of different companies making hardware for. In 1983 you could assemble a computer around your choice of processor, graphics (or not), IO, mass-storage, and so on.
These machines, in conjunction with CP/M, were the closest thing to what the IBM compatible would become, in terms of it being an open platform. Unfortunately, they were also notorious for being incompatible with each other. As you noted, they didn't have to use even remotely similar processors. Heck, you couldn't even exchange disks between many variants (the format was incompatible).
> Entirely accidental. What really started this was the reverse engineering of the BIOS.
Not really. IBM didn't intend the clones, but the effort to create clones was intentional. Not only did you have the outright rip-offs and famously cleanroom reverse engineered ROMs, but you had many companies making machines with various levels of compatibility while trying be legal. Once the PS/2 arrived, IBM's attempt to regain control of the platform, there was a concerted effort by their competitors to ensure they (rather than IBM) defined the standard.
I wish this had remained true for its descendants.
Very true, I still have my XT and AT manuals and use them as illustrations of how documentation should be done. Back then, good documentation included both operational and service manuals including maintenance and circuit diagrams (and as you say IBM even included BIOS source code).
IBM documentation up until at least that point had always been suburb across its product range. For example I have the original copy of this IBM typewriter manual even though I've no longer the machine because the manual is so very excellent:
https://archive.org/details/bitsavers_ibmtypewriricTypewrite...
Manufacturers considered their products and manuals proprietary and protected them with patents and copyrights but they were not secret as is so often the case nowadays. Right to Repair wasn't even a concept back then, had anyone suggested such a notion they'd be laughed at, as it was automatically assumed. What counts as documentation nowadays is an utter disgrace.
IBM was not alone, companies such as HP, Tektronix, Racal (and many others) also produced truly excellent documentation—especially so their maintenance and repair manuals. Equipment performance could often be judged on the quality of the manuals alone.
Those paying equivalent prices or higher for a mid spec computer do get much higher quality documentation, such as high end Toughbooks.