You're not wrong, but there is a bit more subtlety here.
Classic MacOS had no multitasking at all until Multifinder arrived, several years after the Mac was released. I was going to say it appeared in System 6 but I checked and it was actually System 5 in 1987.
Similarly, the "Jackintosh", the multi-million-selling Atari ST with DR GEM on TOS (launched 40 years and 2 months ago, in June 1985) also was single-tasking.
That was normal for the mid-1980s. DOS was single-tasking. The ST shipped before Windows 1.0.
But RISC OS 2 shipped in 1988, the same year as the Mac System 6, and it had cooperative multitasking built-in as standard from Day 1.
(RISC OS 1 was "Arthur" in 1987 and its multitasking was basically just multiple Clock apps.)
Secondly, yes, the Amiga had full multitasking even at launch (the month after the ST) but it also had no memory protection at all. It was not very stable at all. Amiga users got very used to "Guru meditations."
Co-operative multitasking was pretty common in the 1980s because it's much more efficient. Microcomputers then barely had enough RAM and CPU to display a GUI and they didn't have enough for process isolation.
Windows 2, 3.0, and 3.1 could pre-empt DOS apps, but used co-op multitasking for all GUI apps.
Windows NT 3.1 (1993) was the first version that could pre-empt GUI apps. Win32 apps had pre-emptive multitasking, but 16-bit apps (that is, Windows 3 apps being run on NT) were co-op multitasked, unless you were rich, had a £5000+ PC with lots of RAM (like 32MB+) in which case there was a ticky-box to enable running a particular Win16 app in its own memory space. All the rest shared an instance. But this feature ate RAM.
I installed and supported Win NT 3.1 in production and used the option to run Excel in its own memory space, so that an errant copy of Word or MS Mail couldn't crash Excel. It was in a stockbroker and Excel was our single most important Windows app.
But we only had 1 guy with an NT PC at first and that box cost about £8000.
So, yeah, co-op multitasking wasn't as reliable, wasn't as stable, but it was much more memory-efficient and used less CPU, it worked fine on limited 16-bit CPUs, and so it was the norm for almost all computers.
Win NT was too expensive and needed a very expensive PC. OS/2 flopped. It wasn't until Windows 95, a full decade later, that most computers got pre-emptive multitasking.
Though like AmigaOS, Windows 95/98/ME still lacked memory protection. It wasn't until Windows 2000 brought NT to the mainstream that memory protection became common, and probably Windows XP until it became ubiquitous after 98/ME stopped being shipped on lower spec home PCs.
Interesting to wonder what an alternative timeline would have looked like where Acorn used QNX instead of trying to build ARX (the Mach microkernel based system abandoned for Arthur.)
ARM had memory management from the outset, but memory was too expensive to consider anything like unix for the base systems (the alternative RISC iX BSD derived OS required 4MB.) And the context switching overhead would probably have made the desktop experience worse even if they could afford the memory.
Classic MacOS had no multitasking at all until Multifinder arrived, several years after the Mac was released. I was going to say it appeared in System 6 but I checked and it was actually System 5 in 1987.
https://en.wikipedia.org/wiki/MultiFinder
Similarly, the "Jackintosh", the multi-million-selling Atari ST with DR GEM on TOS (launched 40 years and 2 months ago, in June 1985) also was single-tasking.
That was normal for the mid-1980s. DOS was single-tasking. The ST shipped before Windows 1.0.
But RISC OS 2 shipped in 1988, the same year as the Mac System 6, and it had cooperative multitasking built-in as standard from Day 1.
(RISC OS 1 was "Arthur" in 1987 and its multitasking was basically just multiple Clock apps.)
Secondly, yes, the Amiga had full multitasking even at launch (the month after the ST) but it also had no memory protection at all. It was not very stable at all. Amiga users got very used to "Guru meditations."
Co-operative multitasking was pretty common in the 1980s because it's much more efficient. Microcomputers then barely had enough RAM and CPU to display a GUI and they didn't have enough for process isolation.
Windows 2, 3.0, and 3.1 could pre-empt DOS apps, but used co-op multitasking for all GUI apps.
Windows NT 3.1 (1993) was the first version that could pre-empt GUI apps. Win32 apps had pre-emptive multitasking, but 16-bit apps (that is, Windows 3 apps being run on NT) were co-op multitasked, unless you were rich, had a £5000+ PC with lots of RAM (like 32MB+) in which case there was a ticky-box to enable running a particular Win16 app in its own memory space. All the rest shared an instance. But this feature ate RAM.
I installed and supported Win NT 3.1 in production and used the option to run Excel in its own memory space, so that an errant copy of Word or MS Mail couldn't crash Excel. It was in a stockbroker and Excel was our single most important Windows app.
But we only had 1 guy with an NT PC at first and that box cost about £8000.
So, yeah, co-op multitasking wasn't as reliable, wasn't as stable, but it was much more memory-efficient and used less CPU, it worked fine on limited 16-bit CPUs, and so it was the norm for almost all computers.
Win NT was too expensive and needed a very expensive PC. OS/2 flopped. It wasn't until Windows 95, a full decade later, that most computers got pre-emptive multitasking.