Part 1: The eight-bit era You'll find below an informal roundup of the slip-ups and missteps that stick in the mind of The Reg FOSS desk, from the dawn of the microcomputer industry onwards. We are certain that we've missed plenty - let us know your favorites.

We often hear Commodore fans saying that the company had the worst management in the history of the industry… and it did make its fair share of bad decisions. Even so, there are plenty of rivals for that dubious honor.

Before we recount some other catastrophically bad moves, we thought we'd recap a few of Commodore's awful decisions, for the benefit of non-Commodore types. (Like us: The Reg FOSS desk wasn't a Commodore owner in their prime - this vulture owned Sinclair, then Amstrad, then Acorn kit.)

Those crazy Commies

Commodore sold tens of millions of its multiple different computers, and descendants, offshoots, and successors, not to mention newer versions of its last OS are still available today.

A great place to start is the wonderful Secret Weapons of Commodore site. It lists an amazing variety of devices that Commodore never got around to launching. Few companies have had so many different cancelled products, let alone the many that did make it to market.

Of the models that did reach the market, Commodore's early eight-bit machines followed a natural enough evolution. The original Commodore PET was very early for an all-in-one microcomputer. It was expensive enough already, without adding luxuries that would have been prohibitively expensive in the 1970s, like color and sound. Then came a brutally cost-cut home machine, the very limited - 5kB RAM, 22 columns of text - but even so very successful VIC-20.

After these, Commodore came up with the Commodore 64, arguably the most successful eight-bit of all time. For some mystifying reason, it even offered a portable model, the SX-64. (We suspect this demonstrates that Commodore was day-dreaming about the higher-margin business market.)

The C64 was an enhanced, similar but mostly incompatible, improved, and expanded VIC-20, with a very competitive 64 kB of RAM, superb graphics and sound for 1982. Its only real drawbacks were its expensive and horribly slow disk drives, but then most early C64 owners didn't get those - because their parents didn't buy them. The other limitation was its terrible BASIC: essentially the same one as in the PET, because that meant Commodore didn't have to pay any extra and it could fit into a measly 20kB of ROM. Again, games-playing children didn't care if their BASIC had no graphics or sound support.

It's what came next that was when things started to go strange. Rather than capitalizing on its multimillion-selling range, Commodore launched a new range that was both incompatible and, in important ways, inferior: the 264 series, notably the C16 and Plus/4. They were cheaper, but they couldn't run C64 games.

It followed that with the C128, a C64 with more memory and an even better BASIC - and a built-in Z80 second processor, so the C128 could run CP/M, that famous monochrome text-only OS for 1970s business computers. Why a Z80? Because the improved model was incompatible with Commodore's C64 CP/M cartridge. A more sensible move might have been a new-model Z80 cartridge, saving most owners the cost of a whole CPU they'd never use.

The C128 was arguably sexy, sure, but we feel its two models should never have happened. What most owners could have been delivered in incremental improvements to the C64 was bigger, faster disk drives, better graphics, 80-column text, and so on. Instead, the company designed, built, branded, and marketed these huge wastes of effort. They had a better BASIC, but that wasn't hard. Even so, BASIC was irrelevant to most owners of the greatest gaming eight-bit. Even less relevant was CP/M, which bloated the price for no useful gain. Just what a pre-teen playing video games in their bedroom wanted: dBase II and WordStar!

What most C64 owners really wanted was a better games machine: faster, more colors, better sound, more and quicker storage. Eventually, and probably much too late, Commodore came up with the answer: the Commodore 65. Over three times faster, with twice the memory, stereo sound, a faster and more capacious 3.5 inch floppy drive, 4096 colors, and BASIC 10.

That, of course, is the successor it cancelled: the C65 never launched.

Sinkers or swim

But it wasn't just in America that the home-computer pioneers didn't know what to do. In the early days of the microcomputer industry, most companies' management had no idea where things would go, and many things that are obvious in hindsight really were not apparent at the time. Pretty much every manufacturer flailed around.

Sinclair had vast successes in the Z80 market (although they were just too cheap for the American market). It sold millions of the ZX81 and ZX Spectrum, but less visibly until after the Iron Curtain fell, the Spectrum also inspired dozens of clones.

When Timex launched the US version of the ZX Spectrum, the Timex-Sinclair 2068, it fixed the biggest problem affecting Spectrum games: the dreaded attribute clash. Although the Speccy had 256×192 graphics, in a clever hack to save expensive RAM, the resolution for colors was limited to the text resolution of 32×24 - so when graphics moved across the screen, colors flashed and changed and didn't line up. Timex boosted the color resolution to lines of pixels rather than characters: 32×192. Much better color, but only needing about 6kB more of that precious 48kB of memory.

The TS2068 could also run CP/M and had a "high" resolution (512×192) monochrome mode, too. Timex's big blunder, though, was that it implemented these changes without regard for keeping compatibility with the original model Spectrum. The result was a more capable machine, but one that couldn't run the majority of the UK ZX Spectrum's thousands of games.

Even more bizarrely, though, when Sinclair Research launched the ZX Spectrum 128 in 1986, it adapted a Spanish design - adding the same, better sound chip as the US model (but, naturally, mapped to different, incompatible memory addresses). It also had its own, equally incompatible, banked memory scheme, which unlike the US model couldn't map RAM in place of the BASIC ROM. This made the machine unable to run CP/M - which unlike its Commodore rivals, the Spectrum had the appropriate CPU to handle. That wasn't fixed until Amstrad launched its ZX Spectrum +3 in 1987.

But much more relevant to the Spectrum's budget-gamer market, Sinclair - and its Spanish partner Investronica - failed to add the US model's better graphics. This vulture still owns his original 1987 128K Spectrum and, 38 years later, is still somewhat bitter about being deprived of 64-volume mode.

From tiny Acorns…

It wasn't just Sinclair, of course. Its arch-rival Acorn was just as guilty. Its classic BBC Micro introduced millions of school pupils to computers, especially thanks to the BBC's Computer Literacy Project. But with just 32kB of RAM, it was chronically short of memory. Mode 2 graphics used 20kB, meaning that the entry-level 16kB BBC Model A couldn't use that mode at all.

For both Model A and Model B owners, the most common answer was to use Mode 7, with teletext-style graphics produced by the Mullar SAA5050 chip. With most of the picture generation handled in hardware, this gave crisp 40-column text in a now iconic font, and blocky but colorful graphics - in one-eighth of the memory of any of the other modes.

When the company launched the cut-price Acorn Electron computer, guess which chip it left out…

The Electron bombed horribly and nearly bankrupted the company, before making arguably the greatest comeback in the history of the industry - but we'll get to that.

Coming next

Many of the same companies exhibited equally poor decision-making when it came to the 16-bit era, which we will move on to in Part 2. ®

https://www.theregister.com//2024/08/28/where_computing_went_wrong_feature_part_1/