So many strawmen, so little time...
"The Pi's strength is its cheapness and the simplicity of its hardware, but at heart, software-wise, it's a PC... <rant about ARM vs x86>"
This is an odd complaint. At heart, the Pi is a mobile-phone chipset married to a low-end ARM chip, and it will run whatever OS is provided. It only takes a few seconds of looking at the official website (https://www.raspberrypi.org/downloads/) to see that there's a number of "officially approved" OS builds available for it, ranging from various flavours of Linux to Windows 10 /and/ RISC OS. And it doesn't take more than a few seconds to find ports of FreeBSD, Android and even more obscure OSs such as Haiku.
It's also worth noting that the Pi isn't bundled with an OS by default, which means that people are actually choosing to run Linux on it - as indeed, are many other "non-PC" devices, especially in the IoT landscape. After all, it's free and there's lots of existing dev tools and support libraries.
"There were some missed opportunities in creating the Raspberry Pi. The Foundation could have brought harmony and a de facto standard for firmware on ARM hardware"
The Pi was never intended to be a high-volume device. Instead, it was intended to be a relatively low-volume educational device, and it wasn't clear until after it had launched how popular it would become. Setting industry standards were never part of the foundation's remit.
Also, the Pi had only sold 5 million units as of February this year. Even if we assume that volumes have since managed to doubled to 10 million, that's a drop in the bucket compared to the "billions" of other ARM-based devices which the article itself notes have been sold in the same timeframe. So the Pi is hardly in a commanding market position!
Finally: as the article itself comments, the Pi deliberately sidestepped the firmware issue. What it doesn't mention is that this was for several pragmatic reasons - the impact on manufacturing costs being the main one. Because, once again, it was intended to be a low-cost, low-volume educational device.
"Failing that, the Foundation could have bundled RISC OS with it"
It's available on the website for free, and there was a fair amount of excitement/publicity when the Pi first launched about the fact that RISC OS was available. Which suggests that, as fun as tinkering with obscure OSs can be, people actually wanted to use an OS which has lots of existing tools and libraries available...
"Pi project founder Eben Upton fondly recalls his first machines, a BBC Micro and an Amiga 600. A kid could fathom those; I did, with my ZX Spectrum."
Ah, the humble Speccy - the grey +2 was my introduction to the wonderful world of computing. And in truth, I think it's a lot easier to learn how to use a computer these days. The 8-bit machines did offer a BASIC prompt on startup, but there was generally little or no support structure for people other than the official manual, whatever the local library had in stock and the odd magazine type-in (which quickly died off as the commerical world moved towards the use of machine code). These days, you can use the internet to search for documentation/prior examples, or post queries to somewhere like stackoverflow.
I'd also argue that it became significantly more difficult to learn how to code when the 16-bit era landed. You no longer had BASIC bundled with the machine and commercial C/Pascal compilers were relatively rare, underperformant and usually badly documented. So you had to either learn assembly or pick up a third-party program such as AMOS.
Then too, if your code crashed or went into an infinite loop back in the 8/16-bit days, you generally crashed the entire computer and lost all your hard work in the process. And let's not go into the time-cost of backing up to tape or floppy disk - especially the latter, since most home coders used repurposed magazine cover disks with distinctly variable levels of quality control...
"Twenty-first century Unix is too big, too weird, too full of arcane 1960s strangeness."
"Conventional wisdom is that this complexity is an unavoidable consequence of modern software's age and maturity, and that it's worth it. You just ignore the stuff you don't need.".
The ZX Spectrum was basically a 16k ROM bolted to 16/48k of ram, a 4mhz z80 CPU, and a custom ULA which did some magic to reduce component counts (and led to the infamous color-clash issues).
To take the current "high-end" Pi, the Pi 2 features 512mb of ram, a multi-core processor, an OpenGL capable GPU, an audio chip, a DMA controller (and an MMU), a mass media controller, a serial controller and a few other things for good measure. All essentially built into the one chip. The complexity of modern software goes hand in hand with the fact that the hardware is so much more capable. And since you can't chisel bits of silicon off the CPU, you pretty much have to ignore the stuff you don't need...
"Which brings me to the other cheap little educational computer you've never heard of: the OberonStation ... No, it won't do the 1,001 things Linux will, but it's no toy ... But what it shows is that complete, working, usable operating systems – not single-trick ponies – can be built, essentially single-handed, and can be understandable by ordinary mortals"
Hmm. An effectively proprietary OS, no USB ports, no soundcard, no network capabilities, PS/2 keyboard/mouse ports and VGA-only output. That sounds like a toy to me!
From a quick glance at the manual, Oberon was a vanity/sabbatical project built by two people in the eighties. I.e. it's pretty much ideosyncratic by definition and was designed back before the concept of networked computers/IoT became mainstream. Also, the manual states that the system "can be understood by a reasonably determined person", which is definitely a step beyond being understandable by an "ordinary mortal"! So I really can't see any justification for using it these days. Especially since any OS-level skills/knowledge you pick up can't be reused on other devices.
So no, Oberon shouldn't grab the Pi's crown. If there's even a crown to grab. Which there probably isn't, since there's so many competitors out there, starting with the millions of Arduino devices out there. The fact that the Pi Zero is so low cost may well cause it to grab some more "makers" market share from Arduino and others of the same ilk, but there's still plenty of choice out there!