web.me.com/gtownerIn my article "Apple and IBM" in the last issue, I made passing mention of a new data storage technique called "CD ROM" and said that I though it would eventually cover 90% of computers' long-term memory needs. In response to overwhelming requests for clarification (well, actually, my dog barked when I read this part to him), here is an explanation of what I meant.
CD ROM is a technique for laser-writing computer data very densely on a plastic disk that resembles a phonograph record. In fact, CD audio recordings use a nearly identical technique. Present CD ROMs hold 550 million characters ~ roughly equivalent to 100,000 pages of printed text.
The physics of CD ROM does not easily permit erasing and rewriting individual parts of the disk (hence ROM, meaning "Read-Only Memory"). Also there are presently two technological areas that need improvement: the recording function is still expensive, and moving the reading head to specific tracks on the disk is done mechanically, and thus relatively slowly. But it is easy to foresee a day soon when CD ROM record/play devices will be cheap, and will be able to search a disk almost instantaneously by electronically manipulating the laser beam.
The fact that CD ROM disks are not erasable is not a drawback, because the plastic medium is so cheap. In fact, the disk's ability to remember everything that has been recorded on it evokes whole new kinds of computer architecture and opens up new ways of using computers. When I use my present word processor, for example, I "save" my file (usually containing one chapter or article) every ten minutes or so. This means that the computer erases its last-stored copy and writes the updated version over the same space on the disk. With a CD ROM recorder, however, it would write each update as a new file. I would wind up with end-to-end copies of my work in its every stage, from roughest outline to final draft. If I saved every part of a 200-page book 500 times, it would still fit on one disk. Had such a possibility existed earlier, we might today see literary archaeologists spending their lifetimes happily combing through every text incarnation that Joyce created when composing Ulysses.
To take a more useful example, a bank could store all its transaction for a week or a month on one disk. Years' worth of detailed records would scarcely fill a file drawer. The entire archives of the Social Security Administration could live in a corner of a small office.
CD ROM can also affect the way computers work. Anyone who has bought a personal computer has heard of "RAM" — random-access memory, the expensive part of the machine. But most of RAM is used for relatively fixed storage ("nonrelocatable heap objects" in the lingo), which could be satisfied by a quick-access CD ROM. A future desktop computer could operate with less RAM than today, yet deliver most of the power of a gigantic mainframe.
What does all this mean for Apple versus IBM? My thesis was that mainframe technology is about to be swept away by the kind of products that Apple makes. A good test for incipient technological change is to ask: if neither technology were established, but both were competing to fill an established need, which would win out? Would computer makers today develop mammoth disk drives and banks of tape recorders, if they could design around CD ROM instead? I think the answer is obvious.
I predict that the first crack in the dam will occur when a substantial computer-dependent operation (such as a bank or a phone company) is persuaded to empty their dollar-gobbling central computer room and replace it with equipment that fits on a desktop. When that company happily goes on about its business, the revolution will have begun.