See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/248049934

The Universal Computer: The Road from Leibniz to Turing

Article  in  The American Mathematical Monthly · June 2002

DOI: 10.2307/2695463

CITATIONS READS
193 7,335

1 author:

Martin Davis
University of California, Berkeley
127 PUBLICATIONS   11,243 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Martin Davis on 10 February 2015.

The user has requested enhancement of the downloaded file.

rev-blank.qxp 3/20/01 10:35 AM Page 498

Book Review

The Universal Computer: The

Road from Leibniz to Turing
Reviewed by Brian E. Blank

The Universal Computer: The Road from Leibniz 1939 the German en-
to Turing gineer Konrad Zuse
Martin Davis designed and con-
W. W. Norton and Company, 2000 structed two experi-
ISBN 0-393-04785-7 mental electro-
$26.95, 257 pages mechanical digital
computers, the Z1
If you teach a course on number theory nowa- and Z2. In 1937
days, chances are it will generate more interest Howard Aiken sub-
among computer science majors than among mitted to IBM a for-
mathematics majors. Many will care little about mal proposal titled
integers that can be expressed as the sum of two Proposed Automatic
squares. They will prefer instead to learn how Calculating Machine.
Alice can send a message to Bob without fear of The product of
eavesdropper Eve deciphering it. No doubt they Aiken’s initiative, the
would be surprised to see the theory of numbers Harvard Mark I (also
described as a “purely theoretical science without known as the IBM Au-
practical applications” or, even more bluntly, as tomatic Sequence
“useless”. Yet, those are exactly the assessments Controlled Calculator) was placed in service in the
of number theory that were given by Uspensky spring of 1944. It is considered the first electro-
and Heaslet in 1939 and by Hardy in 1940. It is with mechanical number-crunching computer. Mechan-
a sense of irony that we read these pronounce- ical it certainly was. The 750,000 moving parts of
ments now, knowing that the seeds of their Aiken’s machine are said to have produced a roar like
contradiction had already been sown. Work that that of a textile mill. Less than two years later, in
would lead to the modern digital computer was February 1946, a computer known as the ENIAC was
already under way. fully operational. This 30-ton behemoth, conceived
The great theoretical advance that led to the and constructed by John Presper Eckert and John
modern computer may be traced to 1936 when Alan William Mauchly, is considered to be the first elec-
Turing formulated a highly original concept that tronic computer. Electronic it certainly was. When the
would eventually be called the Turing machine. At ENIAC went online, its 17,468 vacuum tubes are said
the time, projects to build simpler computing to have dimmed lights throughout Philadelphia.
devices were just about to begin. Between 1936 and
The Mark I and the ENIAC were both funded
Brian E. Blank is professor of mathematics at Washing- by the military for the purpose of doing numeri-
ton University, St. Louis, Missouri. His e-mail address is cal calculations vital to the war effort. With the
brian@math.wustl.edu. conclusion of the war, seminumerical commercial

498 NOTICES OF THE AMS VOLUME 48, NUMBER 5

rev-blank.qxp 3/20/01 10:35 AM Page 499

applications such as accounting, scheduling, calculation. As he explains in his introduction, “A

record-keeping, and billing were developed. As the computing machine is really a logic machine.
computer rapidly evolved from its eponymous Its circuits embody the distilled insights of a
function, the list of tasks assigned to it swelled. remarkable collection of logicians, developed over
Even tasks that do not involve a single computa- centuries. Nowadays, as computer technology
tion have been taken over by the computer. Nowa- advances with such breathtaking rapidity, as we
days a book review, for example, is likely to be admire the truly remarkable accomplishments of
solicited by computer communication, composed, the engineers, it is all too easy to overlook the
researched, spell-checked, and typeset on a logicians whose ideas made it all possible. This
computer, submitted by computer, and posted book tells their story.”
for access by a worldwide network of computers. One cannot imagine an author more qualified
Even the “printer’s proofs” might arrive in the than Martin Davis for such an endeavor. Many
form of a computer file. Notices readers will be familiar with Davis from his
The conversion of number theory from a “use- contributions, both in research and exposition, to
less” pursuit to an applied science has been due Hilbert’s tenth problem. Others will know him
in large part to an espe- from his excellent text-
cially ironic consequence books, which have become
of the computer’s evolu-
tion: in order that we may
Davis’s perspective standard references of the-
oretical computer science
securely rely on the com- is unique: he is [3], [6]. Those who keep
puter for such noncompu- track of awards will recog-
tational tasks as com- concerned with the nize him as the recipient
merce, communication, of the Chauvenet, the
and archiving, we must development of the Lester R. Ford, and the
first enlist the theory of Leroy P. Steele Prizes. In
numbers to foil the com-
computer as an addition to his credentials
putational power of the
computer to decrypt. Like
engine of logic as distinguished logician
and honored expositor,
the sea change in number rather than as an Davis is also a pioneer pro-
theory that it occasioned, grammer who wrote code
the metamorphosis of the instrument of for both the Institute for
computer from number Advanced Study computer,
cruncher to all-purpose calculation. a historic machine that has
logic machine has been a been in the collection of
profound transformation the Smithsonian Institution
that is now taken for since 1960, and for one of
granted but was not origi- its clones, an Army Ord-
nally transparent. Aiken, for example, did not rec- nance “johnniac-class” computer known as the
ognize the transition in progress. “If it should turn ORDVAC. His engaging autobiographical sketch
out,” he wrote in 1956, “that the basic logics of a [5] offers a rare glimpse of the programmer’s craft
machine designed for the numerical solution of dif- as it existed in 1951, when the state of the art
ferential equations coincide with the logics of a ma- amounted to five kilobytes of random access mem-
chine intended to make bills for a department ory tenuously implemented as static charge on
store, I would regard this as the most amazing co- the surfaces of cathode-ray tubes.
incidence I have ever encountered.” In The Universal Computer Davis begins his tale
Although the electronic digital computer is with Leibniz, whose proposal for an algebra of
barely more than half a century old, its history has logic is the point of departure on the road to the
attracted a devoted following. For more than twenty universal Turing machine. It is indicative of the
years a scholarly journal, the Annals of the History enthusiasm with which Davis infuses his writing
of Computing, has chronicled the development of that where others see “fragmentary anticipations
computing in minute detail. A steady stream of of modern logic” [12], Davis perceives “a vision
books—some erudite, some popular—has allowed of amazing scope and grandeur.” As Davis tells
engineers, historians, and journalists to delve into the story, Leibniz “dreamt of an encyclopedic
nearly every facet of the computer revolution. compilation, of a universal artificial mathemati-
Martin Davis’s new book, The Universal Computer: cal language in which each facet of knowledge
The Road from Leibniz to Turing, is not like any of could be expressed, of calculational rules which
them. would reveal all the logical interrelationships among
Davis’s perspective is unique: he is concerned these propositions. Finally, he dreamed of ma-
with the development of the computer as an chines capable of carrying out calculations, freeing
engine of logic rather than as an instrument of the mind for creative thought.” The chapter is

MAY 2001 NOTICES OF THE AMS 499

rev-blank.qxp 3/20/01 10:35 AM Page 500

called “Leibniz’s Dream”, and that dream is a In keeping with the chronology, Davis interrupts
sort of North Star toward which the axis of each Turing’s biography to direct his attention to the
subsequent chapter points. engineers who would take the next steps toward
Following the style of “Leibniz’s Dream”, Davis the fulfillment of Leibniz’s dream. He begins his
devotes each of the next six chapters to the life eighth chapter, “Making the First Universal Com-
and contributions of a leading logician: the list puters”, with thumbnail summaries of the contri-
comprises Boole, Frege, Cantor, Hilbert, Gödel, and butions of the hardware pioneers Aiken, Atanasoff,
Turing. In making these choices, Davis has taken Eckert, and Mauchly. It may be argued that these
great care not to stray from the road to Turing. Lo- sketches are too brief, but in fact these hardware
gicians such as Brouwer and Russell are discussed implementations fall outside the scope of Davis’s
in a fitting amount of detail, but De Morgan, Peano, book. That said, I do find it surprising that Davis
and Skolem are mentioned only in passing, while accords only one paragraph to Claude Shannon,
Peirce, Schröder, Löwenheim, and Zermelo are not whose 1938 master’s thesis in electrical engineer-
mentioned at all. So coherent is the narrative, how- ing showed how to apply Boole’s algebra of logic
ever, that one has the illusion that one is reading to electronic switching circuits. The complete omis-
the entire history of mathematical logic without any sion of Konrad Zuse is even more puzzling. In any
discontinuity in its evolution. (The reader who is event, the early history of computing is well pro-
inspired to seek out a more conventional history vided for: readers who wish to learn more may con-
of logic may turn to [11] as sult [1], [2], [7], [8], [9], and
well as to the references in [13].
[12, page 1].)
Through the first seven
Not only does Davis The historian Tom Set-
tle has used the death of
chapters the principal log-
ical concepts of each pro-
captivate us with a Galileo to illustrate how
elusive historical truth can
tagonist are presented at a fascinating story, be. Despite an authentic
level that is appropriate for death certificate that cites
a general audience. It was a he caps it with a the evening of January 8,
shrewd idea to embed these 1642, calendrical variation
discussions inside capsule moral as well. renders uncertain which
biographies of the logicians. one of four days is actu-
This stratagem serves both ally being specified. It is
to lighten the load of the tempting to believe that
reader who has no prior training in mathematical more recent events must prove less troublesome.
logic and to maintain the interest of the more ex- Indeed, the authors of a new book [10] about com-
perienced reader who is already familiar with the puter scientists assert that “in most sciences the
logical theories. It is true that standard biogra- seminal thinkers lived in the remote past.
phies exist, and, with few exceptions, Davis does To uncover what they did…we must scavenge in
not go beyond them. Nevertheless, most readers the historical record, picking among scraps of in-
will welcome his lively, informal synopses, replete formation, trying to separate facts from mythol-
as they are with amusing anecdotage. Perhaps the ogy. Computer science is different.” Regrettably,
best of these involves Davis himself. Driving in this plausible claim is not true. Above all, priority
Princeton with his wife, Virginia, he happened to for one of the indispensable principles of modern
pass the town’s most famous denizen, dressed computing, the stored program concept, has
like a tramp, walking with Gödel, nattily attired in proved to be hopelessly and bitterly controver-
suit and tie, briefcase in hand. “Einstein and his sial.
lawyer,” quipped Virginia. Naturally Gödel and In a nutshell, John von Neumann, who worked
Turing provide ample grist for the raconteur’s with Eckert and Mauchly, has often been given full
mill, but the fact is, every one of the featured lo- credit for the stored program concept because he
gicians, the dusty Victorian pedant George Boole advanced the idea in a widely circulated report that
included, makes for a fascinating character study. he released under his name alone. Later both Eck-
By the end of the seventh chapter, Davis’s ert and Mauchly disputed the importance of von
readers will have learned about Boole’s algebra Neumann’s contribution. Their position is argued
of logic, Frege’s Begriffsschrift, the Continuum eloquently in a recent book about the ENIAC [7]. Al-
Hypothesis, Gödel’s theorem on undecidable though Davis admits that the question of von Neu-
propositions, Hilbert’s Entscheidungsproblem, and mann’s personal contribution “will probably never
Turing machines. At this point the timeline of the fully be resolved,” he seems to come down squarely
narrative has reached the end of World War II: all on von Neumann’s side. His analysis is interesting,
the developments in logic that are needed for the but in the big picture this acrimonious squabble
universal computer are in place, and their physi- lacks significance. For one thing, Zuse has a real
cal realizations are literally on the drawing boards. claim to priority: he unmistakably proposed the

500 NOTICES OF THE AMS VOLUME 48, NUMBER 5

rev-blank.qxp 3/20/01 10:35 AM Page 501

stored program concept as early as 1936 (but did [7] S C O T T M C C A RT N E Y , ENIAC: The Triumphs and
not pursue it, since it would have been of little use Tragedies of the World’s First Computer, Walker and
on his slow, mechanical memory machines). More Company, New York, 1999.
[8] BRIAN RANDELL, The Origins of Digital Computers,
importantly, the issue is something of a red herring.
third ed., Springer-Verlag, New York, 1982.
Davis himself first advanced this point of view in a [9] RAUL ROJAS and ULF HASHAGEN (eds.), The First Com-
1987 article [4] that may be regarded as a skeleton puters: History and Architectures, The MIT Press,
of the book under review. “What was really revolu- Cambridge, MA, 2000.
tionary about these machines,” Davis points out, [10] DENNIS SHASHA and CATHY LAZERE, Out of Their Minds:
“was their universal all-purpose character, while The Lives and Discoveries of Fifteen Great Computer
the stored program aspect was only a means to an Scientists, Springer-Verlag, New York, 1998.
[11] N. I. STYAZHKIN, History of Mathematical Logic from
end.”
Leibniz to Peano, The MIT Press, Cambridge, MA,
That Turing had nailed the future of comput- 1969.
ing before all the others may be seen from several [12] HAO WANG, Popular Lectures on Mathematical Logic,
of his statements, of which the following from Van Nostrand Reinhold Company, New York, 1981.
1945 is typical: “There will positively be no inter- Enlarged republication by Dover Publications, New
nal alterations to be made even if we wish suddenly York, 1993.
to switch from calculating the energy levels of the [13] KONRAD ZUSE, The Computer—My Life, Springer-
Verlag, New York, 1993.
neon atom to the enumeration of groups of order
720.” In 1948 he put it this way: “We do not need
to have an infinity of different machines doing
different jobs. A single one will suffice.” Turing did
not refer to this single machine by the misnomer
that others with narrower visions were already
using: he called it the universal machine, and, as
Davis compellingly demonstrates, it was Turing’s
conception of the universal machine that influ-
enced von Neumann.
When a distinguished expert offers a popular
exposition of his subject, we greet the effort with
keen anticipation. That is all the more true when
the writer is as skilled as Martin Davis. It is a plea-
sure to report that in this case our anticipation is
richly rewarded. Not only does Davis captivate us
with a fascinating story, he caps it with a moral as
well. I have echoed this moral at the beginning of
this review, but it is worth repeating in the author’s
own words: “This book underscores the power of
ideas and the futility of predicting where they will
lead.” Seldom has this point been made so well.
Read this book and enjoy.

References
[1] PAUL E. CERUZZI, A History of Modern Computing, The
MIT Press, Cambridge, MA, 1999.
[2] I. BERNARD COHEN, Howard Aiken: Portrait of a Com-
puter Pioneer, The MIT Press, Cambridge, MA, 1999.
[3] M ARTIN D AVIS , Computability and Unsolvability,
McGraw-Hill, 1958. Reprinted in an enlarged edi-
tion by Dover Publications, 1982.
[4] ——— , Mathematical logic and the origin of modern
computers, Studies in the History of Mathematics
(Esther R. Phillips, ed.), Math. Assoc. Amer., Wash-
ington, DC, 1987, pp. 137–165.
[5] ——— , From logic to computer science and back,
People and Ideas in Theoretical Computer Science
(C. S. Calude, ed.), Springer-Verlag, Singapore, 1999,
pp. 53–85.
[6] MARTIN DAVIS, RON SIGAL, and ELAINE WEYUKER, Com-
putability, Complexity, and Languages, second ed.,
Academic Press, New York, 1994.

MAY 2001 NOTICES OF THE AMS 501

View publication stats