As a boy, like many of my contemporaries, I was fascinated by cryptography and spent many happy hours composing ciphers and cracking encrypted texts. This interest largely lapsed over the years but didn't entirely disappear, so I was pleased to find Singh's new book, which provides people like me with a convenient and readable way of familiarising ourselves with the present state of the art.
Singh's book is partly historical but also brings the story fully up to date, with a discussion of quantum cryptography. The development of the subject has followed a classic Darwinian sequence, with continual warfare between cipher inventors and cipher breakers. Throughout history, there have been repeated claims, always disproved, that this or that cipher was unbreakable. Mary Queen of Scots eventually met her death because of her misplaced faith in such an 'unbreakable' cipher; this story is related in considerable detail (perhaps too much).
Mary's cipher was susceptible to analysis on the basis of letter frequency. The Vignère cipher, using multiple cipher alphabets, was supposed to remove this weakness, but Charles Babbage, the inventor of the Difference Engine, showed how the Vignère cipher could be broken, though he did not publish his work and his paper came to light only in the twentieth century. His method was independently discovered by a Prussian, who therefore received all the credit. Singh explains and illustrates the method in some detail.
The book includes a number of historical asides. In one of the most interesting of these detours, Singh relates the story of the Beale papers, which purportedly gave details of a buried treasure in Virginia, deposited in the nineteenth century. There are three cipher texts, one of which has been cracked; unfortunately, the text giving the location of the treasure has proved impenetrable. though not from want of trying; some people have ruined themselves trying to crack it. Singh provides the text in case anyone feels inclined to try.
Coming into more modern times, Singh goes into a fair amount of detail about the Enigma machine, which, during the Second World War, was thought by the Germans to produce unbreakable cipher texts; however, a team of code breakers in England proved them wrong and in so doing made a substantial contribution to the war effort. In the process they built the world's first electronic computer, although this was later dismantled for security reasons. The Americans, meanwhile, did find a secure method of communicating that defeated the Japanese; they recruited Navajo speakers as wireless operators, but this was not a cipher in the ordinary sense of the word.
Ciphers are, by definition, constructed so as to baffle eavesdroppers, but cipher breakers have sometimes applied their skills to a different but related kind of challenge: deciphering ancient scripts. The two outstanding examples are the decipherments of Egyptian hieroglyphics and of Linear B, the Cretan alphabet. The methods that were used are explained in adequate but not excessive detail.
The development of the modern computer has produced dramatic changes in cryptography. The invention of PGP has given us a form of encipherment that is practically unbreakable. However, if quantum computers are ever built they would be so much faster than present-day machines that ciphers such as PGP would be easy to break. On the other hand, quantum ciphers are, in principle, unbreakable. Quantum cryptography does work, but the problem here is to find a satisfactory method of sending photons between machines; this has so far been achieved only over fairly modest distances, but that may soon change. The social and political implications of all this are already with us and are likely to become more pressing as time goes by.
It is traditional in books of this kind to provide some texts for readers to try out their skills. Singh is no exception; in fact, he goes one better, by offering a 10,000 pound prize for finding the solution. However, you'd better hurry: a glance at Singh's website (22 August 2000) shows that 9 of the 10 stages have already been solved.