The USB That Wasn’t

Today’s GIF is of Stewart Cheifet holding an ACCESS.bus dongle. Honestly I’m shocked that all of our GIFs aren’t “Stewart Cheifet holds object.”

ACCESS.bus: The forgotten attempt to give us a universal bus for all our serial ports

You can always tell how obscure something is when your search results come up almost completely empty. And when I search for ACCESS.bus, even surrounded by scare quotes, Google brings up a bunch of pictures of actual buses targeted for accessibility.

Put simply, if you’re looking for ACCESS.bus, you’re in for a deep dive unless you immediately go to Google Books.

How obscure are we talking here? Unlike nearly every other technology thing on the internet, there are no modern videos about the protocol, just a reference in a 31-year-old episode of Computer Chronicles. And doing a search on Hacker News, a traditional hotbed of all things technical, brought just ONE result referencing it in that site’s 18-year history. (If this post ends up there, as my posts sometimes do, there will now be two.)

That spare result led to a Twitter thread by Foone Turing, the digital researcher and technology collector whose tweets became one of the best parts of that platform in the late 2010s. Foone’s tweets are locked, just like mine, so if you’re not already following them, I unfortunately cannot share their thread. But as a follower of hers, I still have access, so I’ll share the gist in a bit. (They’re on Mastodon these days, still worth a follow.)

Anyway, now that we’ve established the obscurity of ACCESS.bus, let’s explain what it was trying to do. In the mid-1980s, Apple had firmed up its own serial system, called Apple Desktop Bus, which was developed by no less an Apple employee than Steve Wozniak. It was a huge advantage for the company at the time—while not as fast as USB or Firewire, it was highly versatile thanks to its ability to daisy-chain multiple devices to a single port. (It could support 16 in total, if you were wondering.)

It wasn’t the only system of its kind—the Atari SIO interface, which emerged in 1979 with the Atari 400 and 800, worked similarly, and that bus’ developer, Joe Decuir, later played a key role in developing USB. But before USB was a glimmer in anyone’s eye, Apple’s ADB connector was proving a source of inspiration for the next generation of serial ports.

It likely also inspired port developers in another way. See, ADB largely only worked with Apple hardware, along with some of Apple’s descendants, like NeXT. And meanwhile, on the IBM side of things, you were either stuck with the complicated non-plug-and-play state of affairs of D-Bus-style serial ports, or you were looking longingly at the ADB-like ports IBM added to its PS/2 line. They looked similar, but weren’t quite as useful.

If you were a manufacturer, this state of affairs kind of sucked. It meant that if you wanted to sell a mouse, that mouse would not just work on every desired platform. The device had to be built to work for that specific platform, with its own dedicated driver software. That increased development time, complicated manufacturing, and created consumer confusion. (Picking up a peripheral for your PC, only to realize you accidentally grabbed the Mac version? A real pain.)

But it was a problem that the industry had solved previously, in the early 1980s, with I²C. As buses go, the underlying approach was dead simple, essentially creating a line of communication between components and the primary machine, as explained in the initial patent filing, submitted in 1981:

It is an object of the invention to enable a single two-wire line to be used for the interconnection of an unlimited number of stations. Reliable synchronization is obtained and simple signalling of the beginning and the end of a communication operation is possible. The stations are interconnected by a clock bus and a data bus. Each of the buses operates to form a wired logic function between the stations. Thus, each of the buses includes means (for example a pull-up resistor) which urges the bus toward a second voltage state in the absence of a forcing input from one of the stations.

Simply put, I²C’s reason for being is to provide the most minimalist synchronous communication mechanism possible—a line for tracking timing, a line for tracking data, and (when necessary) a power source. And it’s a standard many of the era computers already supported. Many computers today support I²C components. Naturally, it was a great choice to build a serial bus on.

It just needed the right advocate, and I²C’s was Digital Equipment Corporation (DEC). The former mainframe giant did not survive the ’90s unscathed, but it had a few success stories during the period, including AltaVista and the DEC Alpha.

It was not a company whose work in PCs really shaped the computing industry—rather, the computing industry worked around them—but it wasn’t for lack of trying.

And ACCESS.bus, which effectively expanded the I²C protocol to external peripherals, was one of those attempts. A 1991 article from the Digital Technical Journal explained the mission of the ACCESS.bus project:

ACCESS.bus provides a simple, uniform way to link a desktop computer to a number of low-speed I/O devices such as a keyboard, a mouse, a tablet, or a three-dimensional (3-D) tracker. Designed from the beginning as an open desktop bus, ACCESS.bus facilitates cooperative solutions using equipment from different vendors.

Rather than building some complex new protocol to do this, DEC’s engineers piggybacked on Philips’ existing (and widely supported) I²C standard. (Philips and DEC ended up working together on the final result.)

Leaning into I²C had some advantages, notably the fact that it was easy to implement across devices and platforms. And for a while, it looked like the future of computing would flow through ACCESS.bus’ daisy chain.

Some struggled to hide their excitement. For example, Robin Stacy, an editor and technology columnist at The Macon Telegraph in Macon, GA, wrote multiple columns about emerging serial bus standards.

(Side note: Robin appears to have had one of the best technology columns around, given that it covered emerging standards that most PC magazines were barely covering at the time. I was fascinated enough that I looked him up in hopes of saying hello. Unfortunately, he died in 2023, but what a cool dude.)

Stacy seemed particularly interested in what DEC was doing, which meant he was early to the ACCESS.bus trend. In a 1994 column, he noted that the technology benefited from the simplicity and low cost of the I²C standard.

“The new standard could also be implemented pretty cheaply. To add the small jacks and simple controller chip to peripherals shouldn’t add more than a dollar to the cost,” he wrote. “ACCESS.bus cables require only four wires, so they shouldn’t be expensive, and the controller at the PC end should be as cheap as today’s serial cards.”

So, why didn’t it take off? Well, technically, it did—just not the way DEC probably expected.

ACCESS.bus still sort of lives, but not anywhere you might expect

As you might have noticed, we do not live in a world where we are daisy-chaining computers through an ACCESS.bus interface, and both DEC and Philips Semiconductor each got acquired out of existence.

But wait. Has ACCESS.bus disappeared off the face of the planet, another failed standard on top of all the others? Technically, yes: USB and FireWire each upstaged the standard from a speed and manufacturer buy-in standpoint, and USB gradually upstaged FireWire.

The downside of I²C’s simplicity is that it is a bad option for delivering lots of data—and during the heyday of ACCESS.bus, it topped out at a mere 100 kilobits per second, roughly the speed of ADB and significantly slower than even the initial version of USB. Had ACCESS.bus taken off, it’s likely the flash drive might not have succeeded in the way it did.

But there was a specific feature of ACCESS.bus that appealed to one specific type of computer hardware manufacturer: monitor-makers.

See, before the mid-1990s, most monitors functioned as dumb tubes that didn’t talk back to the computer. There was no on-screen display, nor any way for the computer to control the monitor’s brightness and contrast. But it turned out that I²C, delivered via ACCESS.bus, was an excellent medium for this.

And NEC, a major producer of monitors at the time, had gone all-in on ACCESS.bus. The company’s MultiSync monitor series, dating to 1994/1995, promoted the idea of connecting components through the monitor, rather than the computer. (If, of course, your computer already had an ACCESS.bus adapter.)

They weren’t alone, but there wasn’t that much else out there. However, at least one device gracefully proved the use case: the Forte VFX1 Headgear, an early virtual reality system dating to the mid-1990s.

Mostly forgotten today, except by hardcore Descent fans, the device was an important bridge between the VR we had then and what we’re seeing now from Meta and Apple.

A promotional video from the time highlights how competing VR systems would plug into your average beige box—with a seemingly endless number of cables. However, in the case of the VFX1, all the end user had to do was plug the VR device’s VGA cable into their desktop, then plug their controller into the VR device. It was a use case that obviously made sense, even if VR itself didn’t at the time.

(Of course, this created problems later on, as ACCESS.bus wasn’t directly supported in modern computers, requiring a lot of hacking to resolve, but in 1995, I’m sure it was awesome.)

Soon enough, USB and FireWire were making the far-slower ACCESS.bus irrelevant, and it became clear that maybe slapping on a phone jack-like connector on monitors wasn’t a use case manufacturers or consumers cared about. (The growing interest in laptops would have made it irrelevant anyway.) But the ability to communicate through the same line we were shipping video through, utilizing an existing standard in a new way? That made sense!

That was the thinking that led the Video Electronics Standards Association (VESA) to co-opt the work for ACCESS.bus into the Display Data Channel (DDC). That standard allowed monitors to talk back to the PC and the computer to directly communicate with the display.

And it’s that standard that Foone Turing pointed out on Twitter in that locked-down 2019 thread: We have long left ACCESS.bus, our would-be contender for the USB throne, into the dustbin of history, but DDC is still at the root of how both HDMI and DisplayPort communicate with our computers.

“So that’s the interesting thing about the DDC2 interface still supported over HDMI/DVI today: it was, in part, originally designed to carry ACCESS.bus connections, a failed universal peripheral bus,” she wrote.

If you have a monitor plugged into your computer, you are technically using ACCESS.bus right now. Good luck finding any other peripherals that support it, though.