FOM: {n: n notin f(n)}

Harvey Friedman friedman at
Tue Sep 3 15:48:09 EDT 2002

>There is little I can add to my previous arguments except to clarify some
>points, and to put them in context.  I don't see any substantial objection
>so far to my claim that we can put natural numbers and reals in 1-1
>correspondence, without dropping Axiom of Separation.
>1.  I am of course committed to a claim of the sort Richard Heck mentions
>     (ER)(S)[(x)(Sx --> Px) --> (Ey)(Py & (z)(Sz <--> yRz))].
>but, since I have specified that S be a singular concept, and P a predicate,
>it does not follow we can substitute "x is not in f(x)" for S, and so it
>does not follow, as Richard argues, that I have to reject Axiom of

You are setting up some restriction that is not standard in the 
accepted foundation for mathematics. You need to show that this way 
of looking at things provides a coherent alternative foundation for 
mathematics. I don't see anything so far that indicates how such 
ideas work for even the foundation of very elementary mathematics.

Cantor's theorem about no one-one correspondence is intended as a 
theorem in set theory, and therefore appears unimpeachable. You must 
be talking about some unusual interpretation of his theorem. It is 
not clear what this unusual interpretation is. Even more so, it is 
not clear whether this unusual interpretation, if it exists, is part 
of any interesting reinterpretation of mathematics.

>The set theoretic interpretation

This is the usual interpretation of a theorem of set theory such as 
Cantor's - i.e., the set theoretic interpretation. You may wish to 
suggest a new interpretation of set theory. If so, what?

>makes it seem more complicated than it is.
>All we're saying, is that we can map 1 to {1,2,3}, for example, 2 to {3,4,5}
>and so on, always mapping a known identifiable object to a known
>identifiable bunch of things.  But "all x that are not in f(x)" does not
>such a known identifiable bunch of things.

This has nothing to do with the existence of a function from N onto 
S(N). (That is another form of Cantor's theorem, that there is no 
function from N onto S(N)). It appears that you are not talking about 
set theory. The obvious question is: if you are not talking about set 
theory, then what are you talking about?

Here are some other theorems of Cantor (and earlier!). What do your 
unusual ideas have to say about them?

There is no function from the empty set onto S(emptyset).

There is no function from {0} onto S({0}).

There is no function from {0,1} onto S({0,1}).

There is no function from {0,1,2,...,2^1000000} onto S({0,1,2,...,2^1000000}).

There is no function from {0,1,2,...,2^1000000} onto {0,1,2,...,1 + 2^1000000}.

The image of any sequence of real numbers has Lebesque measure zero.

The image of any sequence of real numbers is disjoint from at least 
one of its translates by a real number.

Given an image of a sequence of real numbers that forms a subfield of 
the field of real numbers, there is another image of a sequence of 
real numbers that forms a subfield of the field of real numbers, 
where the intersection of the two images is exactly the rational 

>On Harvey's points:
>3.  That the set theoretical interpretation of mathematics is coherent, and
>natural.  "The lingering life maintained by the old Aristotelian and
>scholastic logic .. is an extraordinary fact in the history of philosophy; I
>believe it can be accounted for only by supposing that the syllogism is
>substantially the correct analysis of the process which passes through the
>mind in reasoning".  This was written by a professor from Princeton in 1870,
>on the dawn of an apocalypse for that system.  The fact that ZFC is the
>currently accepted standard, i.e. is established, could equally well have
>been written in 1870, of the syllogistic logic.  I.e. being established or
>apparently natural or whatever is not really an argument for anything.
>Similarly for historical status.  The syllogistic did after all survive
>about 2,300 years.

This argument is fallacious. You can take any development in science 
that achieves something far beyond earlier developments, whose 
success comes partly from overhauling earlier attempts, and make your 
same remark.

The syllogistic approach, with its restrictions to universal monadic 
reasoning, was obviously grossly inadequate for the foundations of 
even the mathematics of the time it was invented. I.e., reasoning 
with inequalities are binary relations not covered by syllogisms. 
Survival is not the issue. It did not even remotely form a foundation 
for very elementary mathematics.

The current foundation for mathematics is obviously grossly adequate 
for the foundations of current mathematics. Of course, I spend a 
great deal of time trying to punch holes in this, but everything that 
I do suggests that the current foundation needs to be augmented with 
additional ideas, and not criticized for having wrong ideas.

>4. " The progress of mathematics and science has necessarily lead to the
>consideration of concepts that do not coincide with any that are
>commonly used in ordinary informal discourse. For instance,
>physicists and chemists do not convey their discoveries in terms of
>"fire" and "hot" and "wet" and "dry". "
>(a) Aristotelian physics survived as long as it did, because of the time to
>develop a framework of enquiry, and a climate that allowed people (in a
>reasonable way) to question the existing assumptions.  This was the real
>intellectual achievement of the Western tradition.   (b) However, are any of
>the concepts embedded in ZFC different  from any of those used informally,
>ot in the logical systems that preceded it?

Yes. E.g., the cumulative hierarchy has a ranking idea, or at least a 
smallness idea, that was not recognized properly before Russell's 
Paradox. E.g., not recognized by Frege. I've heard some people claim 
that Cantor was sensitive to it.

>5.  "The kind of attitudes and objections being raised in the FOM discussion
>...are merely incoherent, useless, vague, and arbitrary - at least in their
>present form."
>It's nice of Harvey to add the last rider.  I would say: a discussion group
>like FOM is like a workshop, not a shop window.  So don't expect to see any
>finished goods!

I'm waiting to see a plan for an alternative foundation for 
mathematics, in light of your criticism of the present one.

>That said, what is incoherent about 1-2 above?  It consists of an idea that
>we have managed to shoehorn into a set-theoretic (actually second order)
>formulation, plus the ideas of "singular" and "general".  The latter are
>well-established philosophical concepts, though maybe not so familiar in
>mathematics.  If they are not clear, then someone speak up, please.
Since you seem to be talking about foundations of mathematics 
(focusing on Cantor's theorem), you can start by showing how ideas 
like "singular" and "general" connect with (foundations of) 

For example, you could propose some new interesting formal systems 
and argue that, in some appropriate sense, they form a foundation for 
at least some elementary mathematics, and also are insufficient to 
prove things like Cantor's theorem.

More information about the FOM mailing list