Computer scientists periodically experiment with systems that do away with the software barriers on which today’s computers are based, but these systems are rarely successful in the marketplace. Both the Lisp Machine and the Canon Cat encouraged developers to create programs that ran in the same workspace, rather than dividing the computer up into different applications. The Apple Newton stored information not in files but in “soups”-little object-oriented databases that could be accessed by many different programs, even at the same time. The commercial failure of these systems does not vindicate today’s way of computing but rather is testimony to just how dangerous the dominant-paradigm trap actually is.Consider files and directories. The hierarchical directory system used by Windows, MacOS, and Unix made sense to computer pioneers who grew up using paper and filing cabinets. But why limit today’s computers with 40-year-old metaphors? Computers have fantastic search capabilities. Some documents logically belong in multiple places; why not eliminate the folders and store all of the computer’s information in one massive data warehouse? That’s the way computers in the movies seem to work.
It’s not such a far-fetched notion. It wouldn’t take much to enable today’s computers to store every version of every document they have ever been used to modify: most people perform fewer than a million keystrokes and mouse clicks each day; a paltry four gigabytes could hold a decade’s worth of typing and revisions if we stored those keystrokes directly, rather than using the inefficient Microsoft Word document format. Alas, the convenient abstractions of directories and files make it difficult for designers to create something different.With a little thought, though, we could do far better. Hollywood has dreamed it; now Silicon Valley needs to make it real.