The Tragedy of The BBC Micro

I remember some period after the BBC Micro – affectionately “the Beeb” – came out that I really wanted one. That lasted a couple of years. My budget couldn't afford it (I was a pre-teen!) But by the time they came out with the Master series it was clear that virtually everyone else had something more interesting at a lower cost, Atari's ST being an obvious example. And Acorn never cut the price.

Meanwhile the Beeb had been competing with the Commodore 64 for 4 years which wasn't for the most part a better computer – though it had twice as much RAM and arguably better game graphics – but was half the price. And don't get me started on the ZX Spectrum...

(I wonder sometimes if Commodore had fixed the serial 488 thing with the 64 before selling it, and maybe added an 80 column mode so it was suitable for any type of application, whether computing in the entire 1980s might have ended up looking very different. Anyway...)

So why was I excited by the BBC Micro in 1981-3?

There were a lot of other features of it I knew I wouldn't ever be able to afford to use, but those came out of the box.

The BBC Micro was, to the British home computer market, roughly what the Apple ][ was to the US market. A fast 6502 based machine with a lot of expansibility, sold at a premium. It was (massively) superior to the Apple ][ series, with the exception of the GS, except in one critical area. It had excellent video, including a 640x256 mode, while it didn't have slots it had chip holders to add floppy disk support (costing around ¼ of the price of the machine if you added them, the floppy costing about half) and networking support, together with a CPU expansion bus, ROM slots, and the aforementioned built-in BASIC and assembler. The OS wasn't bad for an 8 bitter either, and did a lot of things in 1981 that even many contemporary 16 bit OSes didn't do, notably with abstraction.

But...

The only thing it didn't have was enough RAM to do any of this.

With 32k of RAM (Acorn sold a 16k version for a short time but nobody bought it because it was completely useless), there was only one practical video mode for general use, called “Teletext” or “Mode 7”, otherwise a massive chunk of the already meager 32k was used for video. Teletext was 40x25 with limited low resolution graphics characters (the 2x2 pixel grid type) and no user defined characters. Games had to use the other modes, obviously, but were limited in scope as a result between the inefficient 6502 instruction set and having to squeeze into 12-22k of RAM. (Though “Elite” was a BBC Micro thing so not impossible. I'm just saying there's a reason the Spectrum and Commodore 64s had so many games compared to the Beeb.) And for serious software, the lowest memory mode that could support 80 column text used 16k, leaving just 16k for your document, assuming the software itself was provided on a ROM. (Oddly, there was no cartridge slot, but you could semi-permanently install ROMs inside the case on unpopulated ROM chip holders.)

Acorn's original fix was for the memory issue was to recommend people buy a 6502 card for the Beeb's CPU expansion bus, which would have its own memory, and the Beeb itself would then turn into a servant for the CPU card, providing video, a keyboard, disk access (if you had a disk drive), networking, etc. It would run most of the same software, with the Beeb's abstraction of, for example, the graphics system working well here. But second processor cards were expensive. Moreover it wasn't released until late 1983.

Alas Acorn saw the Beeb as a cash cow, and as time went on didn't feel any need to waste money on improvements. The Beeb had a guaranteed market due to the BBC's blessing and the government promoting the BBC Micro to schools. The machine as a whole started to get stale very, very quickly.

By 1982, Sinclair had released the ZX Spectrum, which was inferior in every way (except, in my unpopular view, the Z80, though Beeb advocates would rightly point out the BBC's 6502 was considerably faster), but was also a third of the price, had more memory, and supported a “good enough” spec that it was adequate as a home computer. Commodore released the Commodore 64, which had faults, but cost about 2/3 of the price of the BBC Model B, and had enough memory to be more practical as a real computer. For professional uses, both were poor, neither supporting an 80 column text mode for example, but most people weren't using home computers for professional reasons.

Acorn didn't care. They didn't adjust the specification of the Beeb, and addressed the calls for lower costs instead by releasing a “cheap” crippled BBC architecture micro called the Electron. The Electron ran at between a quarter and half the speed (It had a 4 bit memory bus!) of the Beeb, and had no memory saving “teletext mode” (making the 32k of RAM even more painful)

It was also missing the CPU expansion feature, and a few other little used Beeb features, which wasn't a big deal. On the plus side it was half the price of the Beeb. Alas Acorn messed up getting the machine into stores at a time when the machine might have been, almost, competitive.

Acorn probably could have put 64k of RAM (and given it an 8 bit data bus) fixing the Electron's major issues, but chose not to because it didn't have a similar update ready for the Beeb and presumably didn't want to produce a machine that might have cut into the latter's sales. The question I have is why? Why didn't Acorn add more RAM to the BBC series, and at least make some effort to cost reduce it so they didn't have to increase the cost? The Electron wouldn't have cannibalized sales of the BBC Micro had the BBC Micro kept up. Indeed, done correctly, it might have even improved BBC Micro sales.

Acorn finally addresses the shortcomings

In 1985, four years after the release of the original, and three years and 11 months after it became obvious it didn't have enough memory, Acorn finally upgraded the RAM of the standard BBC to 64k, but bizarrely raised the price by 25%.

The upgrade effectively, once graphics were taken into account, more or less tripled the usable RAM of the machine. At the same time, Atari was selling the 520ST, a 16/32 bit 68000 based computer for around the same price. Amstrad had been selling several Z80 based computers with 64k-128k of RAM, with 128k of RAM, a colour monitor, and a disk drive, costing around the same amount as a bare bones Beeb (not BBC+.) Commodore's heavily over-engineered C128 was the only computer that came close in poor value, and was arguably a better machine than the BBC+. The BBC+ (as it was known) was widely panned given the update was both mediocre and came with a price hike.

In 1986 Acorn released the Master series, which despite being a higher spec and still floating around the 500GBP mark like the BBC+, was in reality mostly a cost-reduced BBC+, albeit with a better keyboard and a larger ROM accommodating some built in software, Commodore +4 style.

In reality, I suspect the Master was a stop gap to prevent schools from buying commodity PCs. By this time Acorn was working on the ARM chip and was a year away from releasing the the Archimedes, something genuinely as exciting as the BBC was in 1981.

At this point, they'd pretty much lost the industry. Businesses had gone with cheap PCs, and may have done so anyway even if Acorn had been more pro-active. The home computer market was dominated by Sinclair/Amstrad and Commodore. Only the education market was still buying Acorns, and that was ending too. My upper school passed the Beebs by and bought nearly-IBM-compatible Nimbus machines from Research Machines instead. Plus a whole job lot of Spectrums because they were cheap. Plus it had a bunch of cast off Memotech MTXes for some reason. But not a single Beeb.

Wrapping up

I feel like this is a bit of a tragedy.

Right now, many readers of this are itching to tell me I'm wrong and that the 32k wasn't a problem and they had a Beeb and loved it. And I believe you loved it and you possibly convinced yourselves that 32k wasn't a problem too. But it was. The reason you're annoyed is because the Beeb was genuinely an exciting machine at the time, and its limitations weren't enough to make it “bad” compared to, say, the ZX Spectrum, with its rubber keyboard, limited BASIC, and poor build quality. But the memory issues crippled it and put it in a weird position of being more expensive than most home computers, and unsuitable for tasks a bare bones business micro, even a CP/M machine, could do.

There's a timeline though where Acorn could have ended up being relevant well into the 1990s. It has Acorn continuing with the ARM, obviously, but in the mean time:

I think there was an arrogance about British tech businesses back then that had people working for them (1) convinced that if they have the best product, people will avoid their competitors, and (2) convinced that if they have the best product in $YEAR, it will still be the best in $YEAR+1, looking constantly for reasons to pretend the newer products aren't better. That mentality possibly explains some of Acorn's errors, but a sizable amount of it seems to be an unwillingness to improve when there was no financial incentive to do.