Peter Smith’s Gödel Without (Too Many) Tears - Or Not?

Page last updated 09 Nov 2022

The convicted paedophile ex-professor Peter Smith is a prominent apologist for Gödel’s incompleteness proof. He has written several versions of his own “proof ” of incompleteness. All of his accounts include the same fundamental error. The most detailed account is to be found in his book, ‘An Introduction to Gödel’s Theorems’, (Footnote: Peter Smith. An Introduction to Gödel’s Theorems. Cambridge University Press, 2006.
ISBN: 9780521857840 See An Introduction to Gödel’s Theorems: Details.) which he has now made freely available as a downloadable PDF at PDF Smith’s An Introduction to Gödel’s Theorems. I have analyzed the “proof ” Smith presents in that book in depth in my paper PDF A Fundamental Flaw in an Incompleteness Proof by Peter Smith. If you want to investigate Smith’s claims in depth, I would recommend that you follow the book and my analysis of it.

Smith also used to have a PDF freely available on his website entitled Gödel Without (Too Many) Tears. I see that now he has taken it off his site, and he wants you to pay for it at Amazon. I have a copy of it, if you want it for educational purposes (the copyright specifically permits copying for educational use) I can send you a copy, just ask through my email or in a comment with your email address (I will then remove that address). Smith says that it is a cut-down version of the book, and it covers much of the same material as the book, but it skips over the crucial technical details, so it cannot really be recommended as a substitute for the book. Many people will want to use this version rather than pay for the book. This is the reason I have decided to make an analysis of that version available here.

Readers should be aware, that while Smith’s argument in this cut-down version is essentially the same as that expounded in his book, in the cut-down version a lot of details are omitted, or glossed over. However, this lack of detail is not the cause of the error shown below - it is also evident in the detailed version in Smith’s book.

Smith’s “diag ” function

The key place where the crucial error occurs is in Smith’s “proof ” is in his “Theorem” 32, which Smith states as follows: (Footnote: Smith sometimes changes his text and in doing so, may change the theorem numbers, so in a future edition of Smith’s text, the actual theorem number may be different. The terms that Smith uses are as follows:
p.r. function:   a primitive recursive number-theoretic function.
L:   A language system that can express p.r. functions.
wff:   A well formed formula in that language.
g.n:   A Gödel number of an expression of the language L.
n:   The number n in the format of the language L (Note: n should appear with a bar over it where it occurs in φ( n ). If no bar is showing, it may be due to your browser).)

So what Smith is saying in his “proof ” above is this:

If n is the Gödel number of a valid formula φ with one free variable, then diag(n) is the Gödel number of φ( n ). In other words, Smith’s claim is that:

diag(n) = g.n.[φ( n )]

Yes, this is a recursive function. But it is most certainly not a primitive recursive function, since a primitive recursive function is necessarily number-theoretic. And regardless of the terminology, the entire point of the reference to primitive recursion is supposedly that any primitive recursive expression can be expressed in the language of the formal system L which can only deal with numerical expressions.

As defined by Smith, diag(n) cannot be a primitive recursive function, since it is defined in terms of the Gödel numbering function, which is a function of the meta-language and cannot be expressed in the formal language (see PDF The Impossibility of Representation of a Gödel Numbering Function by a Formula of the Formal System).

Does it matter? Yes, it does. Because further on, in the “proof ” of Theorem 33, Smith refers to an expression:

Prf (m, diag(n))

But although he has already defined Prf as a primitive recursive number-theoretic function, his “proof ” requires that the entire expression Prf (m, diag(n)) can be expressed in the language L, which it clearly can’t, since it is not a number-theoretic expression.

Smith’s fudge asks you to believe that one can have two valid definitions of a function diag(n) , where one definition defines the function as having a particular property (primitive recursive) and the other definition defines the function as not having that property. It’s not what one might expect from someone who has written a book entitled “An Introduction to Formal Logic”.⁠ (Footnote: Peter Smith, An introduction to Formal Logic, Cambridge University Press, 2003.)

Note that I will not attempt to discuss Smith’s error in his cut-down version in any more detail, simply because that version omits quite a lot of detail anyway. Since the error is essentially the same as in Smith’s book, if you want to see a fully detailed analysis of Smith’s error, please see the paper PDF A Fundamental Flaw in an Incompleteness Proof by Peter Smith.

Conclusion

Smith’s “proof ” is fatally flawed, since the error of assuming that an expression is a number-theoretic expression when it clearly isn’t completely undermines the entire argument. The error is the same error as in Smith’s book An Introduction to Gödel’s Theorems, and there is a much more detailed analysis of that error in the paper PDF A Fundamental Flaw in an Incompleteness Proof by Peter Smith.

In fact this type of error crops up in numerous attempted incompleteness proofs, see Errors in Incompleteness Proofs and Analyses of Incompleteness Proofs. If you want to ask Peter Smith about the error in his proof, send him an email at peter [email protected]. If you get an interesting reply, please let me know.