Douglas Hofstadter - I Am a Strange Loop

Intricately woven together, then, in my adolescent mind were the study of pattern (mathematics) and the study of paradoxes (metamathematics). I was somehow convinced that all the mysterious secrets with which I was obsessed would become crystal-clear to me once I had deeply mastered these two intertwined disciplines. Although over the course of the next couple of decades I lost essentially all of my faith in the notion that these disciplines contained (even implicitly) the answers to all these questions, one thing I never lost was my intuitive hunch that around the core of the eternal riddle “What am I?”, there swirled the ethereal vortex of Gödel’s elaborately constructed loop.

It is for that reason that in this book, although I am being driven principally by questions about consciousness and self, I will have to devote some pages to the background needed for a (very rough) understanding of Gödel’s ideas — and in particular, this means number theory and logic. There won’t be heavy doses of either one, to be sure, but I do have to paint at least a coarse-grained picture of what these fields are basically about; otherwise, we won’t have any way to proceed. So please fasten your seat belt, dear reader. We’re going to be experiencing a bit of weather for the next two chapters.

Post Scriptum

After completing this chapter to my satisfaction, I recalled that I owned two books about “interesting numbers” — The Penguin Dictionary of Curious and Interesting Numbers by David Wells, an author on mathematics whom I greatly admire, and Les nombres remarquables by François Le Lionnais, one of the two founders of the famous French literary movement Oulipo. I dimly recalled that both of these books listed their “interesting numbers” in order of size, so I decided to check them out to see which was the lowest integer that each of them left out.

As I suspected, both authors made rather heroic efforts to include all the integers that exist, but inevitably, human knowledge being finite and human beings being mortal, each volume sooner or later started having gaps. Wells’ first gap appeared at 43, while Le Lionnais held out a little bit longer, until 49. I personally was not too surprised by 43, but I found 49 surprising; after all, it’s a square, which suggests at least a speck of interest. On the other hand, I admit that squareness gets a bit boring after you’ve already run into it several times, so I could partially understand why that property alone did not suffice to qualify 49 for inclusion in Le Lionnais’s final list. Wells lists several intriguing properties of 49 (but not the fact that it’s a square), and conversely, Le Lionnais points out some very surprising properties of 43.

So then I decided to find the lowest integer that both books considered to be utterly devoid of interest, and this turned out to be 62. For what it’s worth, that will be my age when this book appears in print. Could it be that 62 is interesting, after all?

CHAPTER 9

Pattern and Provability

Principia Mathematica and its Theorems

IN THE early twentieth century, Bertrand Russell, spurred by the maxim “Find and study paradoxes; design and build great ramparts to keep them out!” (my words, not his), resolved that in Principia Mathematica, his new barricaded fortress of mathematical reasoning, no set could ever contain itself, and no sentence could ever turn around and talk about itself. These parallel bans were intended to save Principia Mathematica from the trap that more naïve theories had fallen into. But something truly strange turned up when Kurt Gödel looked closely at what I will call PM — that is, the formal system used in Principia Mathematica for reasoning about sets (and about numbers, but they came later, as they were defined in terms of sets) .

Let me be a little more explicit about this distinction between Principia Mathematica and PM. The former is a set of three hefty tomes, whereas PM is a set of precise symbol-manipulation rules laid out and explored in depth in those tomes, using a rather arcane notation (see the end of this chapter). The distinction is analogous to that between Isaac Newton’s massive tome entitled Principia and the laws of mechanics that he set forth therein.

Although it took many chapters of theorems and derivations before the rather lowly fact that one plus one equals two (written in PM notation as “s0 + s0 = ss0”, where the letter “s” stands for the concept “successor of ”) was rigorously demonstrated using the strict symbol-shunting rules of PM, Gödel nonetheless realized that PM, though terribly cumbersome, had enormous power to talk about whole numbers — in fact, to talk about arbitrarily subtle properties of whole numbers. (By the way, that little phrase “arbitrarily subtle properties” already gives the game away, though the hint is so veiled that almost no one is aware of how much the words imply. It took Gödel to fully see it.)

For instance, as soon as enough set-theoretical machinery had been introduced in Principia Mathematica to allow basic arithmetical notions like addition and multiplication to enter the picture, it became easy to define, within the PM formalism, more interesting notions such as “square” ( i.e., the square of a whole number), “nonsquare”, “prime”, and “composite”.

There could thus be, at least in theory, a volume of Principia Mathematica devoted entirely to exploring the question of which integers are, and which are not, the sum of two squares. For instance, 41 is the sum of 16 and 25, and there are infinitely many other integers that can be made by summing two squares. Call them members of Class A. On the other hand, 43 is not the sum of any pair of squares, and likewise, there are infinitely many other integers that cannot be made by summing two squares. Call them members of Class B. (Which class is 109 in? What about 133?) Fully fathoming this elegant dichotomy of the set of all integers, though a most subtle task, had been accomplished by number theorists long before Gödel’s birth.

Analogously, one could imagine another volume of Principia Mathematica devoted entirely to exploring the question of which integers are, and which are not, the sum of two primes. For instance, 24 is the sum of 5 and 19, whereas 23 is not the sum of any pair of primes. Once again, we can call these two classes of integers “Class C” and “Class D”, respectively. Each class has infinitely many members. Fully fathoming this elegant dichotomy of the set of all integers represents a very deep and, as of today, still unsolved challenge for number theorists, though much progress has been made in the two-plus centuries since the problem was first posed.

Mixing Two Unlikely Ideas: Primes and Squares

Before we look into Gödel’s unexpected twist-based insight into PM, I need to comment first on the profound joy in discovering patterns, and next on the profound joy in understanding what lies behind patterns. It is mathematicians’ relentless search for why that in the end defines the nature of their discipline. One of my favorite facts in number theory will, I hope, allow me to illustrate this in a pleasing fashion.

Let us ask ourselves a simple enough question concerning prime numbers: Which primes are sums of two squares (41, for example), and which primes are not (43, for example)? In other words, let’s go back to Classes A and B, both of which are infinite, and ask which prime numbers lie in each of them. Is it possible that nearly all prime numbers are in one of these classes, and just a few in the other? Or is it about fifty–fifty? Are there infinitely many primes in each class? Given an arbitrary prime number p, is there a quick and simple test to determine which class p belongs to (without trying out all possible additions of two squares smaller than p )? Is there any kind of predictable pattern concerning how primes are distributed in these two classes, or is it just a jumbly chaos?