[PREV - IMPOSSIBLE_GEOMETRIES]    [TOP]

MATHISM


                                                       January      16, 2019
                                                       September 20-30, 2019
                                                  Rev: October      25, 2019
_Principia Mathematica_                             
was published 1910-1913:
                                After 1910 or
Whitehead and Russell made a    so, I gather it
game attempt at rebuilding      was mostly
all of mathematics from         Russell working
scratch, logically deriving     on this-- Alfred
everything from a minimal       Whitehead
number of postulates.           dropped out of
                                the project,
    THE_DOCTRINE_OF_POSTULATES
                                because Russell
                                had an affair
                                with his wife.

                                Russell was a
                                big advocate of
                                free love in
                                those days-- his     And so that's all
                                personal life        another piece of the
                                was complicated.     scene, at the beginning
                                                     of the 20th century,
  This is now widely                                 just before WWI.
  regarded as a dead end,
  having been shot down by                               DUSTY_MYSTERY
  Goedel's Incompleteness
  Theorems in 1931.              No formal system
                                 can be complete,
  I'm beginning to wonder if     or so they say.     It's not clear that
  it might not have been                             Russell ever really
  fundamentally silly for                            understood Goedel's
  other reasons: a lot of                            Incompleteness (for
  effort for little benefit.                         that matter, I'm not
  Perhaps: indulging in an                           sure that I do).
  intellectual vice.
                                                     Russell's comment on
                                                     hearing about it was
  Even before Goedel there                           something along the lines
  were troubles, they                                of "I'm glad I'm not
  needed some way to deal                            working on that project
  with "Russell's Paradox":                          right now."

    Is the set of all sets           Perhaps you could defend the _Principia_
    that do not belong to a          effort in the usual way that pure math is
    set a member of itself?          always defended: fields or mathematical
                                     research often begin just as intellectual
  Setting up that kind of self-      curiosities but have practical applications
  contradiction is obviously         that are discover only later.
  bad, and the solution is, of
  course, "don't do that"-- but             EXCURSIONS_IN_THEORY
  don't do what?
                                            Re-factoring all of mathematics
  They experimented with various            might not have an obvious point, but
  kinds of what I would call logical        it doesn't mean it has no point.
  hacks to try to cover for this;
  they were trying to contrive a set                       I gather part of the
  of rules that would forbid these     I don't             idea was to develop
  contradictory self-reference         think they          a new foundation of
  situations in a way that didn't      got to              math that would be
  seem too ugly or unreasonable...     inventing           easier to extend:
                                       "weaken".           algorithmic
                                                           generation of new
                                                           theorems and
                                                           automated proofs.
Over the years, Russell changed his
mind about which approach was best, and
it seems that there are people still
playing around with these hacks, though       I have seen a remark by
I gather that ZFC, a variant of               Goedel that claimed that
"axiomatic set theory" is essentially         once you understood the
the standard view.  But then, one also        the solution it wasn't
hears about "type theory", "category          any big deal.
theory", and so on, which seem to me to
be competing theories (or at least                        I don't really know
related fields?).                                         which solution
                                                          Goedel had in mind.
  At present, I only have the vaguest                     "Axiomatic Set
  sense of how these work, but I would                    Theory", perhaps.
  gather the idea is something like you
  treat "meta-sets" as a different order                  [link]
  of thing than a "set"...  but that
  would need a few more pieces to get it                  Wildberger says the
  to work...                                              consensus settled on
                                                          Zermelo-Frankel plus
  The metaset of all sets that do not belong              the Axiom of Choice:
  to some other metaset is not a member            ZFC.
  of itself, because it's a metaset not a set.
                                                          WILDBERGER_MATH
  The metaset of all sets that do not belong
  to any metaset would not be a member
  of itself for the same reason, but there's         There are sets that should
  the question of whether any set could       be added to it, but if
  belong to it without contradiction.                they're added, they should
                                                     be removed.
    Some rule about sets not being
    allowed to refer to themselves in
    their definition might do it-- but
    I would guess that could rule out     ZFC has a rule against sets
    other things we'd rather see          being members of themselves--
    allowed.
                                               The notation it makes it look a
    But then, if metasets are allowed          little more complex than that,
    to be members of other metasets,           though: I think they were going
    then you've got a parallel to              after indirect self-reference
    Russell's paradox at a different           as well.
    level-- you'd need the concept of
    meta-meta-* sets ad infinitum.

                          So much for my naive
                          attempt at fixing
                          naive set theory.


           There are also other paradoxes kicking        [link]
           around, such as the Burali-Forti. Patrick
           Suppes, the author of "Axiomatic Set
           Theory" (1960) makes the point that the
           hacks that cover one paradox don't
           *always* cover the others.


In any case, as I was looking over Russell's         MODEST_PROPOSAL
philosophy of mathematics, I started
developing a funny feeling-- familiar from
other fields-- that I was looking at a smart     It's a pretty common pattern
guy plunging off into some highly intelligent    for a field to make great
madness...                                       success early on using a
                                                 heavily mathematical approach,
I'm used to feeling this way about               only to hit a wall and
"computer science" which seems to be             reluctantly fall back on
dominated by people who want it to be            empirical approaches.
all about math, though to most
practicing programmers the work seems                        Fluid Mechanics
like something much different...                      Crystal Growth
Mathematical elegance just doesn't                    Computer Science...?
matter all that much for the practical
problems of getting software working.                    MATH_SIMPLE

Now I found myself wondering if                   There's often a lingering
"mathematical elegance" really made               sense of mathism in these
all that much sense for mathematics               fields-- papers with masses of
itself.                                           impenetrable equations are
                                                  always awarded respect.

There's a habit of mind of
mathematicians that leads them to
start thinking things like "hm,                                 GENERAL_RUSSELL
Euclid's fifth postulate seems
kind-of clunky, wouldn't if be cool
if we could prove it in terms of the         "... in mathematics, every new
first four?  Then there'd be only            axiom diminishes the generality of
four postulates!"                           the resulting theorems, and the
                                             greatest possible generality is
                                             before all things to be sought."
But what exactly would that
get you? They're "playing golf" with               Bertrand Russell,
Euclidean geometry, sinking the hole               "The Study of Mathematics"
in the fewest number of strokes, but
still getting to the same place.

   Now, as it happens, Russell            But showing that the 5th axiom is
   had some issues with Euclid--          redundant would hardly increase the
   evidently there are some               range of application of what's been
   implicit assumptions scattered         derived.
   throughout the work outside
   of the explicit 5 postulates.

   But that shows exactly the               I can see how hypothetically that a
   sort of thing I'm talking                reduction in the number of premises
   about...  why wouldn't you               could reduce the number of things
   just tack the additional                 you'd need to check to make sure the
   postulates on to the list of             math is applicable to a given
   givens and call it done?                 situation.... but it'd need to be a
                                            big reduction to make a practical
   And that indeed has                      difference (to be precise: a big
   been done by modern                      reduction in the difficulty of
   mathematicians--                         checking the postulates, not
                                            necessarily their total number).
   [link]

     Russell had an                             And in practice, I doubt even
     extremly dismissive                        that really matters.  When a
     attitude toward                            scientist or engineer tries to
     Euclid, something like                     use some math, they hardly pay
     "Oh that guy was such                      any attention to the
     a *poseur*, he was so                      mathematician in the back
     overrated-- no way      Those were         waving their hands and shouting
     should we be teaching   the days,          "But you haven't established
     Euclid to school        when you           continuity!".  The physicists
     children.               could gripe        Just Try It, and if it seems to
                             about              work they keep using it, and
     Russell's take          schools            get someone to stuff a gag in
     was that the            that had           the mathematician's mouth.
     first *real*            kids read
     math book was           original
     Boole's "Laws           texts like
     of Thought"             Euclid.
     from 1854.
                                               Actually, I can think of one way
                                               this could matter in mathematical
                                               research: if you could show that
           Could it be that there's a          one of your postulates was
           certain nuttiness about this        redundant, you'd stop worrying
           entire drive to minimalism?         about examining alternative
                                               possibilities.

                                                  In the case of Euclid's 5th,
                                                  it actually isn't redundant,
                                                  and there are variant forms
                                                  of geometry that drop it
                                                  ("hyperbolic geometry").


Bertrand Russell, in "The Study of Mathematics"
objects to justifying the study of math on           And as is not unusual,
practical grounds ("facilitates the making of        whenever you find someone
machines") or on pedagogic ones ("trains the         promoting a weirdly
reasoning faculties"), instead he passionately       idealized view of the
declares it should be studied for it's own sake:     world, there you will find
                                                     a fan of Plato:
   "Mathematics, rightly viewed, possesses
   not only truth, but supreme beauty-- a            "Plato, we know, regarded
   beauty cold and austere, like that of             the contemplation of
   sculpture, without appeal to any part of          mathematical truths as
   our weaker nature, without the gorgeous           worthy of the Deity; and
   trappings of painting or music, yet               Plato realised, more
   sublimely pure, and capable of a stern            perhaps than any other
   perfection such as only the greatest art          single man, what those
   can show."                                        elements are in human
                                                     life which merit a place
This passionate aesthetic fascination--               in heaven."
and an explicit denial that practical
applications matter-- starts
to look like evidence of mania...


Now, going after the entire discipline of
mathematics, is shall we say, ambitious,
but let's look this over:

Russell's ideas aside, the usual
defense of "pure math"-- itself a
revealing name, perhaps-- is that it
often later turns out to be useful.

    The classic example is
    non-euclidean geometry: an                              EXCURSIONS_IN_THEORY
    intellectual curiosity before
    general relativity convinced
    us it was reality.

                   An example I came across recently:
                   Number theory and digital encryption       CERTAINTY

                       This one is interesting because
                       it's a technological application,
                       not a scientific one.


      There's a famous essay by          I think some have trouble seeing
      the physicist Eugene               what Wigner was getting at, I've
      Wigner "The Unreasonable           heard remarks like:
      Effectiveness of
      Mathematics", which                  "Math theories are conceptual tools
      ponders why it is that               created by humans so why would it
      understanding physical               seem strange that they're useful?"
      reality is so tractable
      to mathematic approaches.          The point is that Math is often
                                         invented with *no expectation* that
      There's an apropos                 it will be of any use.  It's often
      quip by Russell himself,           literally a matter of playing
      to the effect that                 around with abstract concepts.
      math is simple, but
      then so are we.

           MATH_SIMPLE



Now, granted that these logical games
sometimes turn out to be useful in
retrospect, we don't really know if there
might be some other approach to math that
would work as well or better.
                                                       SCARCE_RIGOR
  Pure Math might be compared to flinging stones
  blindly over a wall, and demanding respect because
  they occasionally land in useful configurations.


In a few places Russell speaks dismissively
about Kant's philosophy of mathematics,            WILDBERGER_MATH
because it allows room for something like
intuition, and a path toward understanding
besides logical deduction.

Russell would have none of this, of course:
his attitude was something like "if you're
going to allow things like that you might
as well not bother at all".


But Russell's critique of Kant implicitly
acknowledges that there was a time when
mathematics was done differently-- that's
what Kant was describing.
                                           Though for some, that contradicts
The fact that we changed once always       their faith that we always advance:
suggests to me that we may change the      the new ways *must* be better or
way we do things again.                    else why did we switch?

                                                          WHIG_OUT

We're already in an era where
math differs from the way it was
done in Russell's day:
                                       The first of these was the proof
Modern mathematical proofs are often   of the four color theorem.
complex, elaborate things that may
even be software assisted--

And it's not unusual for a long        There's an interview with Conway
proof to be more-or-less accepted,     where he complains about this
even though it's acknowledged it       attitude-- he evidently think it's
probably has mistakes in it, and       a justification for intellectual
further work will be needed to         sloppiness that a good mathematician
bridge the gaps.                       would be able to avoid.




--------
[NEXT - THE_DOCTRINE_OF_POSTULATES]