The legacy of British mathematician Alan Turing (1912–1954) has been widely acknowledged and discussed in the last twenty years or so in the history-of-science community. The main focus of the debate is usually on his concept of the ‘universal machine’ (also referred to as the ‘Turing machine’), his famous paper on the limits of computability from 1936, his wartime work on deciphering the German Enigma codes, and his postwar work on artificial intelligence – and his tragic death, of course.
Little is known about his time at Princeton in 1937 and 1938 where he met with the champions of mathematical logics such as Kurt Gödel, John von Neumann and Alonzo Church. In 1938 Turing wrote a PhD thesis under the supervision of the latter on ‘ordinal logics’ in Princeton. The present book is a facsimile of the typescript with two introductory chapters by editor Andrew W. Appel (pp. 1–12) and Solomon Feferman (pp. 13–26). The facsimile is the typescript that was submitted to the London Mathematical Society in 1939 in their Proceedings, with markers in the margin (and a list at the beginning) where the reviewers of the journal requested changes to the manuscript.
Given that most books on Turing lack any information on Turing's PhD work – Andrew Hodges's path-breaking biography Alan Turing: The Enigma (1983) only deals on two pages with his actual doctoral work, giving much more space to his social life while in the USA – the present book could have closed an important lacuna in our knowledge of Turing. However, the volume falls somewhat short on delivering just that.
The Austrian logician Kurt Gödel published a now-famous paper in 1931 presenting his ‘incompleteness theorem’: mathematics, he stated, can never be ‘complete’ (i.e. free of contradictions). He left open, however, the question whether there is a method to decide if a given theorem is provable or not. Alonzo Church of Princeton took up this question. Shortly after Church had published a paper on this issue on 15 April 1936, he was contacted by the editor of the Proceedings of the London Mathematical Society, who told him that a young mathematician from Cambridge – Alan Turing – had sent in a paper dealing with the same topic, and coming to the same conclusion as Church's, but with a very different approach, and that this approach was so original that this paper should also be published. Church reviewed the paper and confirmed the editor's assessment; furthermore it convinced him that Turing should come to Princeton, where – not least due to the exodus of scientists from Germany after January 1933 – the elite of mathematics, logics and physics had gathered.
After a stay in Princeton beginning in September 1936, Church persuaded Turing that he should write a PhD thesis under his supervision. Turing came back to England in the summer of 1937 and managed to get funding for this venture, returning in the autumn of the same year. He took Gödel's incompleteness theorem as the point of departure and essentially devised an ‘iterative’ construction of logical systems. Starting with a basic set of ‘ordinal’ formulae L1, which need to be checked by intuition, one would then add on top of that another logic L2, and so on, arriving at a logical system that at every step would be more complete than the previous one. Turing's own findings in his 1936 paper basically reaffirmed Gödel's position that there will always be at least one statement that cannot be proven. His PhD work then introduced an approach to successively narrow down the set of provable logical systems to as few unprovable statements as possible by iteratively joining axioms ‘seen intuitively to be correct’ (p. 91). He thus tried to mitigate his own findings on the limits of computability by offering an approach to compute as much as possible, hopefully leaving only the matter of self-referential statements as uncomputable (or rather, undecidable). In his PhD thesis Turing also presented the concept of an ‘oracle – or just ‘oracle’ (pp. 52 ff.) – a black box kind of machine that can solve any kind of problem, even undecidable ones. Whenever a given machine fails to find a solution to a problem it can consult a more powerful machine (the oracle) which will produce an answer. The oracle functions as a placeholder, i.e. as a partial problem to be solved at a later stage. The more ordinal formulae have been joined, the less often the oracle needs to be consulted, ideally leaving only the undecidable problems for the oracle to solve.
As mentioned, little has been published on Turing's PhD thesis beside some internalistic articles in computer-science journals. Turing's thesis could not only serve as an important episode in the history of computer science, it could also help in understanding the philosophy of it. However, Appel's introductory chapter gives only a glimpse of this, and by itself the facsimile of the thesis will only be accessible to readers with a good knowledge of mathematical logics. The book's usefulness for the wider history-of-science community will thus be rather small. It must also be asked how much sense the reproduction of the thesis makes, given that the submitted typescript is not so very different from the printed version in the Proceedings of the London Mathematical Society (the changes requested by the editors were rather minimal), copies of this journal are available in many libraries, and it has been reprinted in the volume of Turing's Collected Works (2001) devoted to mathematical logic, to which Feferman also provided an introduction. It would have been much more interesting had Appel or Feferman discussed more of Turing's interactions with the other scientists present in Princeton. Most of what is presented in the book is available elsewhere, making one wonder if the purchase of the book is really worthwhile.
