December 25, 2007
You Can and Must Understand Giant Brains Now!
A Review of Giant Brains
or Machines That Think
by Edmund Callis Berkeley
John Wiley & Sons, Inc.
(DjVu reader required)
Let me begin by admitting that I trudged to this book like a zombie, simply hoping that I might be amused to encounter frequent, non-ironic mentions of computers as “giant brains.” What I discovered was, while not a masterwork, a clearly-written and evidently very influential volume that is fascinating at times. It provides a view that goes beyond a single computing machine but is situated very early in computing history, before the first working stored-program computer (the EDSAC) had been completed. Giant Brains enlarged my thinking about the history of computing, tickling my digital media senses with its speculations about how individuals might use computers.
In Computer: A History of the Information Machine by Martin Campbell-Kelly and William Aspray I found a reference to this book, which was apparently the first “semipopular” book about computers. Berkeley’s book expresses hopes and concerns from its own historical moment, collects a useful bibliography of important early writing on computers, and issues the first call for popular engagement with and understanding of computers. With its simple but thorough descriptions of how computer systems work, it is an ancestor of The Elements of Computer Systems, but the book is also notable as a predecessor of the populist and manifesto-like Computer Lib / Dream Machines.
The book covers many of the important mathematical and logical concepts that are essential to computing — including differential equations, since one of the brains discussed is an analog computer, MIT’s Differential Analyzer No. 2. The other machines discussed are Harvard’s IBM Aiken Mark I, the University of Pennsylvania’s ENIAC, Bell Labs’ General-Purpose Relay Calculator, and the Kalin-Burkhart Logical-Truth Calculator. This is an interesting variety of computing machinery, ranging from electromechanical to fully electronic and even including one which cannot count. The concepts behind these computers are introduced and the inputs, outputs, and internal workings of each of the machines – from architecture to components – is described in surprising detail. The way people operate computers is also discussed in some sections. There is also a chapter, preceding the ones on the more sophisticated brains, on how punched card systems calculate. It describes in an introductory but rather detailed way how numbers and letters are coded on cards, how they are read, and what machines carry out what functions. A census tabulation example is used to explain how large amounts of data can be counted and summarized. While this sort of information about punched card systems is available in plenty of other places, the presentation of it in this book, written almost 60 years ago, happens to be very apt for those interested in the material history of computing today.
Before any of the discussion of the real giant brains, Berkeley introduces a highly simplified computer that he has specified, Simon. The description of Simon in the book is detailed and adequate for a complete conceptual understanding of the essentials of electronic computing. While Simon was only a concept when Giant Brains was published, a working version of the machine was built by a mechanic and two Columbia engineering students in 1950 for about $600; Berkeley followed up his book with a series of 13 articles in Radio-Electronics covering the machine’s construction in even more detail, so that people would be better able to build their own. It seems like a stretch to consider Simon the first PC. The machine could only compute using the numbers 0, 1, 2, and 3, and so couldn’t do much useful work. Certainly, Simon couldn’t manage the useful personal tasks that were outlined in Giant Brains as examples of how a computer could help an individual, such as keeping an address book. But the device wasn’t supposed to compute usefully; it was supposed to be instructional, like today’s LC-3, a computer implemented in software for teaching purposes. While perhaps not a PC, Simon was clearly also an ancestor of the Altair 8800 – it was a computer that could be assembled by a hobbyist.
The three chapters that look to the future have some particularly interesting discussion. The chapter on the future of particular brains and brain components is not the richest of these. Details of new-fangled technologies such as magnetic tape and electrostatic storage tubes are provided. A section about a page long, “New Ideas in Programming,” is about all the discussion there is of the possibility of software. And some notable late-breaking and under-construction brains, including the IBM Selective-Sequence Electronic Calculator, are described. But chapter 11 (“The Future”) suggests several interesting applications for computers. The automatic address book with mail-merge capability is one; another is the “automatic library,” equipped with a search engine:
You will be able to dial into the catalogue machine ‘making biscuits.’ There will be a flutter of movie film in the machine. Soon it will stop, and, in front of you on the screen, will be projected the part of the catalogue which shows the names of three or four books containing recipes for biscuits.
Berkeley foresaw that the pages of the book would be available online, too. At this point there are two important leaps from the vision of Vannevar Bush’s “As We May Think,” which is cited in the bibliography of Giant Brains. First, although the information is still captured photographically, on film, the process described by Berkeley seems to be governed by a digital computer, given the overall context of the book. Second, we have gone from the bow and arrow in Bush’s essay to biscuits. The automatic library as Berkeley described it is for personal use, not just for scientific research. Not satisfied to arrive at this point, Berkeley goes on to speculate that the system could deliver not the only a recipe but a program to allow the biscuits to be cooked automatically by machine.
Translation, handwriting recognition and speech recognition are also discussed as possible applications for computers. On the individual level, Berkeley notes that people should be able to use computers for tax preparation, keeping personal accounts, “to remember many things we need to know, and perhaps even to give us more information.” He goes on to speculate about a pocket-sized computer with voice recognition that could be used to store and retrieve information.
In the final, overly wide-ranging chapter, the vision and specter of the robot machine – a giant brain that thinks combined with a machine that acts – appears. A cautionary discussion ranges through Frankenstein and R.U.R. to touch on concerns about robot-induced unemployment and more direct robot attacks.
Few will want to study all of the sections on specific giant brains in detail, but those with some real interest in the history of computing are sure to find a great deal of information and an interesting perspective in this book. It offers material, technical details that are hard to come by elsewhere and goes beyond those to discuss the computer’s place in society.