It is a fairly well-known, but perhaps not broadly appreciated, fact that Apple’s Macintosh could have been a very different computer. Sometimes known as the father of the Mac, sometimes as its eccentric uncle, the project was originally under the direction of computer scientist Jef Raskin. He managed to avoid Steve Jobs’s ire for a while by not telling Jobs about the project, but after the Lisa failed and with Woz recovering from a plane crash, Steve needed something to do and checked in on what the former director of the documentation group was up to.
Outwardly, the Macintosh would have looked a lot like the one that was released: a little friendly box with a screen and a floppy drive on the front. It is what is on the inside that counts, and that would have been radically different. From a hardware perspective the Mac would have looked a lot more like an Apple ][ than a Lisa, reflecting Raskin’s belief that the computer should be accessible to the masses (and by extension, affordable). But the software would have been entirely different.
To see Raskin’s vision in practice, you have to find a Canon Cat which is very difficult because Canon’s typewriter division did not do a very good job of selling any. But to understand it, you can pick up his book, “The Humane Interface”.
Let us start with the simple things: no Caps Lock key on the Cat (or Macintosh-that-was not, or the Swyft to give it Raskin’s post-Apple name) so no CCIDENTAL SHOUTING WHEN YOU MISS THE A KEY. Instead, there is a dedicated “undo/redo” button; redoing the last undone action is always an exact negation. In front of the space bar are two LEAP keys, that you hold while typing to search forward or backward through your document and go (leap) to the matching text.
The absence of a Caps Lock key reflects a deeper, more pervasive design choice: Jef’s computer would avoid modes as far as possible. If you are entering numbers, then you have access to all of the numeric functions like maths operations. If you are entering text, then you would have access to all of the text functions. You can explicitly and temporarily switch modes, as with the LEAP keys, but at all times any particular keypress or combination of keypresses always does the same thing. No trying to predict, as a vi user must, whether typing the letters o, d, d will insert the word odd on the current line or, the word dd on the next line.
This lack of modality runs deep, because modes make things easier for the programmer and harder for the user, and Raskin wanted to make things as easy for the user as possible. This extends to strict design restrictions on application software; after all, applications are modes that the whole computer enters. This is most obvious on a smartphone or tablet, which is not a single UI but N different UIs dependent on how many apps you have. Applications in a Humane Interface are all based on a small number of different high-level, task-centred organisations and are invoked from the data they work with, not by double-clicking on an icon.
Raskin hated double-clicking intensely, due to “dysclicksia” (the inability to double-click accurately due to the precise movement and timing constraints imposed on the user by the computer during a double-click gesture). Indeed he originally observed that the whole idea of a mouse was difficult to pick up metaphorically, noting that on seeing one for the first time many users would literally try to pick it up like Scotty in Star Trek IV. Once the world had been subjected to the mouse he relented, but thought that the single-button design of the Macintosh mouse was confusing. He proposed a three-button design, with buttons for select, act, and a temporary ‘grab’ mode for moving things. Incidentally Jobs also gave up on the one-button mouse: NeXT computers had two buttons, with the second button activating the current application’s menu like an Amiga. Double-clicking puts the computer into a temporary second mode (you have to guess whether you will single-click twice or double-click), and is to be avoided.
Speaking of which, “on” and “off” are two different modes. You do not know whether your keypresses are going to cause text entry, or nothing. So they are to be avoided too. Not entirely, for electrical reasons, but “off” and “dormant” should be indistinguishable. The Cat/Mac/Swyft uses the Smalltalk, or Oberon, model of having all of the user’s active work in memory rather than in files that they must manage on storage. Turning the machine off writes all of the memory to disk, and turning it on loads all of the memory from disk, so the computer is in exactly the same state as it was before it was switched off. On and off are also indistinguishable in that you are allowed to connect devices whatever state the machine is in. This is normal with USB now but was not at the time, when metal-shielded connectors could arc when brought close together and crash or even damage the computer. The connectors would go one better than USB though, being designed so that either end of a cable could be used at either end of a connection, and with a “hermaphroditic” rather than plug/socket arrangement so that any cable was guaranteed to be the right one (and they could be extended by plugging two or more together).
Cover photo by the author.