be sure, programmable in the usual sense: one cannot sit at a keyboard and type in a program for it to follow. On the other hand, one could certainly change the memory spindles, albeit with some difficulty, and thus reprogram the computer for other games. Imagine a Tinkertoy device that plays go-moku narabe (a game played on an 11-by-11 board in which one player tries to place five black stones in a row while preventing an opponent from creating a row of five white stones). A Tinkertoy computer programmed for go-moku narabe, however, would probably tower into the stratosphere.

     The real lesson the Tinkertoy computer can teach us resides in a rather amazing feature of digital computation: at the very root of a computation lies merely an essential flow of information. The computer hardware itself can take on many forms and designs. One could build perfectly accurate computers not only of Tinkertoys but also of bamboo poles, ropes and pulleys [see "Computer Recreations," SCIENTIFIC AMERICAN, April, 1988], plastic tubes and water­even, strange to think, electrical components. The lastnamed are preferred, of course, because of their speed. It would be shortsighted indeed to sneer at a computer made of Tinkertoys merely because it is not electronic. After all, even electrons and wires may not be the best materials for quick computer processing. Photons and fibers are gaining on them fast.

   Actually, Tinkertoys are well suited to digital computing. For example, the memory spindles use a binary principle: the presence or absence of spools denotes the status of a particular square on a tic-tac-toe board. The core piece exhibits digital logic: it can turn only if all its fingers miss corresponding spools on a memory spindle. Such an operation is called "and." One can trace the logic for the core piece in the illustration on the opposite page: if the first spool is absent from the first section of the memory spindle and the second spool is absent from the second section and the third spool is absent from the third section and so on­only if all nine conditions are met will the core piece turn. The beauty of the Tinkertoy computer is not just its clever mechanics but its subtle logic.

   Tinkertoy purists will be happy to know that the M.I.T. students stuck to the original wooden sticks and spools with only a few exceptions. An occasional aluminum rod runs through the framework to strengthen it. Two wire cables, an axle and a crank transmit

motive power to the awesome machine for its next move. Finally, the very joints of sticks and spools were made firm by glue and escutcheon pins­pieces of hardware that commonly hold commemorative plaques in place. The team inserted the pins in holes drilled through the rim of the spool down to the original, central hole and through its stick­a task they had to repeat more than 1,000 times. (When Hillis walked into a hardware store to obtain several thousand escutcheon pins, the manager looked bewildered. "We have," Hillis said with a straight face, "a lot of escutcheons.")

   The Tinkertoy tic-tac-toe computer suffered the fate of most museum exhibits. It was taken apart and crated. It sits in storage at the Mid-America Center, waiting to reemerge, perhaps, into the limelight. It may yet click its way to victory after victory, a monument to the Tinkertoy dreams of childhood.

   Well into my sixth year of "Computer Recreations," I am as painfully aware as ever that there are many things the department cannot do. It cannot, for example, teach readers how to program, nor can it mention the hundreds of fascinating programs and the many computer stories and ideas that readers

send in, given the limitations of space and time. It took six years to discover a remedy to these and other needs: a newsletter. Its name is Algorithm: The Personal Programming Newsletter, and the first issue is now available.

The newsletter will appear bimonthly. It seeks to pack a lot of information between its covers. In particular it will have two columns for people who like to program. One will be for beginners and the other for more experienced practitioners. A "bulletin board" at the back of the newsletter will make some of the world's underground programs public for the first time. Letters, stateof-the-art-icles and speculative pieces will aim to lead the mind into unexplored territory. I shall be delighted to send a free sample of the first issue to anyone who writes to me in care of Scientific American.

CHARLES BABBAGE: ON THE PRINCIPLES AND DEVELOPMENT OF THE CALCULATOR AND OTHER SEMINAL WRITINGS. Charles Babbage et al. Edited by Philip Morrison and Emily Morrison. Dover Publications, 1961.

OPTICAL COMPUTING. Special issue edited by Sing H. Lee and Ravindra A. Athale in Optical Engineering, Vol. 28, No. 4; April, 1989.