See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263201015

Our Mathematical Universe?

Article · June 2014

Source: arXiv

CITATION READS

1 12,334

1 author:

Jeremy Butterfield
University of Cambridge
155 PUBLICATIONS   3,495 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Philosophy of Emergence View project

Black Holes View project

All content following this page was uploaded by Jeremy Butterfield on 28 December 2014.

The user has requested enhancement of the downloaded file.

 For  Plus  Magazine,  the  online  magazine  of  the  UK  Mathematics  Millennium  Project:  
http://plus.maths.org/content/  
 
Our  Mathematical  Universe?  
 
Jeremy  Butterfield:  Trinity  College  Cambridge,  CB2  1TQ,  UK  
15  June  2014  
 
Review  of:  Max  Tegmark,  Our  Mathematical  Universe,  Allen  Lane  2014,  £  25,  pp.  425  
 
Abstract:  This  is  a  discussion  of  some  themes  in  Max  Tegmark’s  recent  book,  Our  Mathematical  
Universe.  It  was  written  as  a  review  for  Plus  Magazine,  the  online  magazine  of  the  UK’s  national  
mathematics  education  and  outreach  project,  the  Mathematics  Millennium  Project  
(http://plus.maths.org/content/).  Since  some  of  the  discussion-­‐-­‐-­‐about  symmetry  breaking,  and  
Pythagoreanism  in  the  philosophy  of  mathematics-­‐-­‐-­‐went  beyond  reviewing  Tegmark’s  book,  the  
material  was  divided  into  three  online  articles.  This  version  combines  those  three  articles,  and  
adds  some  other  material,  in  particular  a  brief  defence  of  quidditism  about  properties.  It  also  
adds  some  references,  to  other  Plus  articles  as  well  as  academic  articles.  But  it  retains  the  
informal  style  of  Plus.      
 
 
 
I.  A  bigger,  stranger  cosmos  
The  message  of  this  book  is  that  the  cosmos  is  much  bigger,  and  much  stranger,  than  you  might  
have  thought.  Indeed,  it  is  bigger  and  stranger  in  ways  you  might  not  have  thought  about.    
 
Max  Tegmark  argues  that  it  is  bigger  and  stranger  in  four  main  ways:  each  one  building  on  the  
last.  Each  represents  such  a  vast  extension  of  what  we  usually  call  the  ‘cosmos’  (or  ‘universe’)  
that  he  suggests  we  should  use  the  word  ‘multiverse’  for  these  extensions.  So  he  calls  the  first  
main  way  it  is  bigger-­‐-­‐-­‐the  first  main  way  he  wants  to  surprise  and  unsettle  you-­‐-­‐-­‐the  ‘Level  I  
Multiverse’.  Then  the  second  main  way  is  the  ‘Level  II  Multiverse’;  and  so  on  till  we  get  to  the  
‘Level  IV  Multiverse’.  
 
So  get  ready  for  a  rapid  gallop  through-­‐-­‐-­‐not  just  space,  time  and  spacetime,  as  described  by  
Einstein’s  general  relativity  and  modern  cosmology-­‐-­‐-­‐but  also  the  mysteries  of  how  to  interpret  
superpositions  in  quantum  physics,  and  something  maybe  even  more  mysterious:  the  nature  of  
mathematics  itself.    
 
But  fear  not:    Max  Tegmark  is  not  only  a  gifted  and  distinguished  cosmologist.  He  writes  clearly  
and  wittily  about  all  the  mind-­‐bending  ideas  which  his  book  expounds:  not  just  the  established  
physics  of  relativity,  cosmology  and  quantum  physics,  but  also  the  controversial  possibilities-­‐-­‐-­‐
those  four  main  ways  for  the  cosmos  to  be  bigger  and  stranger  than  you  would  have  thought-­‐-­‐-­‐
that  he  wants  to  convince  you  about.  Besides,  he  combines  all  this  science  with  personal  
anecdotes  about  his  own  life  and  times:  from  his  growing  up  in  Sweden,  to  his  later  career  in  the  
USA.  Thus  the  book  begins  with  a  nearly  fatal  bicycle  accident  which  happened  when  he  was  a  
teenager  (and  the  various  quantum  mechanical  ways  it  might  have  been  fatal!);  and  throughout  
the  book,  there  are  various  anecdotes,  e.g.  about  meeting  the  amazing  Princeton  physicist,  John  
Wheeler.  In  short:  ideal  for  readers  of  Plus  Magazine  .  .  .  So,  here  goes!    
 
 
I.  The  Level  I  Multiverse  
Level  I  is  rather  like  the  idea  that  maybe,  when  you  were  born,  you  were  one  of  two  twins—but  
the  other  twin  got  separated  from  you  at  birth,  to  be  brought  up  far  away,  so  that  you  never  had  
the  chance  to  meet  each  other.  Except  that  for  Tegmark’s  Level  I,  the  birth  in  question  is  not  
yours,  but  the  universe’s!      
 
That  is  to  say,  the  suggestion  is:  at  the  Big  Bang,  there  could  have  been  pieces  of  matter  that  went  
off  in  another  direction  than  did  all  the  matter  that  went  on  to  form  our  galaxy-­‐-­‐-­‐and  moreover,  

  1  
in  a  different  direction  from  all  the  matter  that  makes  the  universe  we  can  now  observe-­‐-­‐-­‐or  
even,  will  ever  observe.  Could  that  be  so?    
 
Well,  the  Big  Bang  is  a  tough  topic  to  understand  in  detail!  But  a  framework  for  understanding  it,  
called  inflationary  cosmology,  answers:  ‘Yes,  it  is  so-­‐-­‐-­‐and  there  are  not  just  a  few  ‘lost  twins’  of  
bits  of  early  matter,  but  countlessly  many:  each  of  them  zooming  off,  never  to  be  seen  by  us,  and  
also  never  to  be  seen  by  almost  all  other  such  bits  of  matter-­‐-­‐-­‐or  rather,  by  whatever  observers  
those  bits  of  matter  might  eventually  aggregate  into’.  So  inflationary  cosmology  proposes  
‘separating  the  children  from  one  another  at  birth’  on  a  truly  cosmic  scale!  
 
But  how  much  should  we  trust  inflationary  cosmology?  There  is  good  news  and  bad  news.  The  
good  news  is  that  in  the  thirty  years  since  the  idea  of  inflation  was  first  invented  (in  about  1980),  
it  has  had  several  theoretical  and  even  observational  successes.  (Cf.  for  example  the  Plus  
Magazine  article  on  the  Planck  satellite:  http://plus.maths.org/content/what-­‐planck-­‐saw.)  So  
much  so,  that  it  is  now  physicists’  orthodox  view  of  the  very  early  universe.  The  bad  news  is  that-­‐
-­‐-­‐as  always-­‐-­‐-­‐we  should  also  listen  to  the  counsel  of  caution,  the  voice  of  scepticism.  It  says:  we  
must  distinguish  between  established  facts  and  speculations.      
 
This  is  the  first  of  several  counsels  of  caution  (‘health  warnings’)  that  I  will  need  to  issue  as  as  we  
proceed  through  Uncle  Max’s  Levels.  Tegmark  himself  of  course  writes  in  a  style  of  enthusiastic  
advocacy.  But  to  his  credit,  he  also  intermittently  pauses  to  emphasize  that  his  stances  are  
controversial,  and  also  to  reply  in  some  detail  to  possible  objections,  or  indeed  to  actual  critics.  
See  for  example,  his  reply  to  George  Ellis  on  pp.  360-­‐363.    
 
I  should  also  note  that  this  controversy  is  widespread  among  very  distinguished  cosmologists.  
For  example,  one  recent  critique  of  inflation  (and  specifically  of  the  recent  BICEP2  experiment)  
by  Paul  Steinhardt  concludes  that  ‘the  inflationary  paradigm  is  fundamentally  untestable,  and  
hence  scientifically  meaningless’  (Nature,  vol  510,  5  June  2014,  p.  9:  cf.          
http://www.nature.com/polopoly_fs/1.15346!/menu/main/topColumns/topLeftColumn/pdf/5
10009a.pdf  ).  
 
 
Agreed:  modern  cosmology  has  established  that  all  the  matter  in  the  universe  we  can  now  
observe  was  once  in  a  hot,  dense  fireball  about  the  size  of  the  solar  system.  It  has  done  this  by  
combining  precision  cosmological  observations,  with  established  physical  theories:  especially  
general  relativity,  and  the  standard  model  of  particle  physics,  which  has  for  forty  years  stood  up  
very  well  to  countless  tests,  for  example  at  the  LHC  in  CERN,  Geneva.  (Cf.  for  example  Plus  
Magazine’s  articles  on  the  Higgs  boson:  http://plus.maths.org/content/hooray-­‐higgs-­‐edit-­‐0,  and  
http://plus.maths.org/content/higgs.)  So  ‘the  standard  model’  means  our  quantum  field  
theoretic  account  of  how  particles  like  electrons  and  quarks  behave  at  energies  we  can  access.  
Those  energies  were  attained  in  the  early  universe  when  it  was  about  the  size  of  the  solar  system,  
at  about  10  -­‐9  seconds  after  the  Big  Bang.  So  this  consilience  of  cosmology  and  quantum  field  
theory  at  high  energies  is  undoubtedly  a  stupendous  scientific  achievement.    
 
Besides,  our  theories  describing  that  hot,  dense  fireball  can  be  extrapolated  back  still  further.  But  
as  the  radius  becomes  smaller,  and  the  temperatures,  densities  and  energies  become  higher,  we  
leave  established  facts  and  enter  the  realms  of  speculation.  And  for  the  epoch  at  which  inflation  is  
supposed  to  have  happened,  we  are  most  certainly  far  beyond  the  realm  of  established  physics.  
That  is:  we  are  far  beyond  the  energies  at  which  we  know  both  general  relativity,  and  the  
standard  model  of  particle  physics,  to  be  accurate.  Indeed,  at  the  end  of  inflationary  epoch  (if  
there  was  one!),  the  radius  of  the  entire  present  observable  universe  was  about  1  metre-­‐-­‐-­‐wow!  
 
But  let  us  be  bold!  Let  us  accept  inflationary  cosmology’s  answer  to  our  question  ‘Could  that  be  
so?’  Let  us  accept  that  there  are  countless  ‘lost  twins’,  separated  into  vastly  many  diverse  
universes,  and  that  observers  in  those  universes  will  never  observe  each  other  directly.  This  is  
the  Level  I  Multiverse.      
 
Before  we  hit  Level  II,  we  should  note  that  Uncle  Max  has  another  mind-­‐bending  suggestion.    
Some  inflationary  models  suggest  there  are  infinitely  many  such  universes,  now  mutually  un-­‐

  2  
detectable.  If  so,  there  seems  nothing  to  stop  any  specific  scenario  which  you  believe  happened-­‐-­‐-­‐
some  Tyrannosaurus  Rex  running  through  some  Jurassic  landscape,  Napoleon  being  defeated  at  
Waterloo,  or  Uncle  Max’s  nearly-­‐fatal  bike  accident-­‐-­‐-­‐being  played  out,  in  replica  form,  in  another  
of  these  universes.  Indeed,  there  seems  nothing  to  stop  the  scenario  being  played  out,  in  replica  
form,  any  number  of  times  across  the  whole  set  of  universes.  Indeed:  even  infinitely  many  times.    
 
And  what  goes  for  a  scenario  which  you  believe  happened,  goes  just  as  well  for  scenarios  which  
you  believe  did  not  happen.  Imagine  Napoleon  winning  at  Waterloo.  Or  rather,  imagine  someone  
very  like  Napoleon,  in  intrinsic  respects  (the  hat,  the  arm  in  the  waistcoat  …)  and  in  historical  
role  (an  artillery  engineer  who  became  a  conquering  general  .  .  .),    winning  at  a  battle  very  like  
Waterloo  (a  flat  landscape,  soldiers  in  red  coats  sounding  like  Englishmen  .  .  .).  Imagine  all  this  
being  filled  out  in  some  totally  specific  way.  Of  course  all  the  myriad  details  vastly  outstrip  your  
powers  of  description:  even  if,  to  tell  your  tale,  you  could  live  much  longer  than  three  score  years  
and  ten.  But  nevermind  the  undescribability  of  it.  Just  imagine  any  such  totally  specific  way  in  
which  (to  put  it  briefly!)  ‘Napoleon  wins  at  Waterloo’.  Then  the  point  is:  there  seems  nothing  to  
stop  that  exact  scenario  being  played  out  in  another  of  these  universes.  And  similarly  for  all  the  
countless  other  variants  of  (again:  to  put  it  briefly!)  ‘Napoleon  winning’;  and  also,  of  course  for  all  
the  countless  variants  of  ‘his  losing’.    
 
This  is  giddy  stuff,  to  be  sure.  But  so  far,  the  alternative  universes  we  have  envisaged  have  been  
modest  in  one  crucial  respect.  Yes,  they  vary  in  myriad  ways:  but  the  variations  concern  
particular  matters  of  fact,  localized  in  space  and  time.  A  dinosaur  runs,  or  doesn’t.  A  man  loses  a  
battle,  or  wins  it.  There  is,  so  far,  no  variation  in  the  physical  laws  that  govern  processes.  That  is:  
assume  for  a  moment  that  general  relativity  gives  the  correct  laws  for  gravity,  and  quantum  
mechanics  gives  the  correct  laws  for  chemical  reactions.  Then  the  point  is:  so  far,  we  have  not  
envisaged  a  different  law  of  gravity  governing  the  dinosaur’s  run,  or  a  different  chemistry  
governing  the  general’s  digestion  of  his  lunch.    
 
But  in  inflationary  cosmology,  many  models  suggest  that  the  physical  laws  do  vary  across  the  set  
of  universes.  If  that  is  right,  what  we  call  ‘physical  laws’  or  ‘laws  of  nature’-­‐-­‐-­‐the  deepest  and  
most  general  patterns  displayed  by  the  ways  in  which  physical  events  play  out  in  our  universe-­‐-­‐-­‐
would  be  better-­‐called  ‘bye-­‐laws’.  For  think  of  how  the  laws  enacted  by  a  municipal,  rather  than  
national,  government  are  called  ‘bye-­‐laws’;  (‘bye’,  like  ‘burgh’  and  ‘borough’,  derives  from  the  old  
Anglo-­‐Saxon  word  for  a  settlement  or  town).  This  panoply  of  universes,  differing  in  their  laws,  as  
well  as  their  particular  matters  of  fact,  Tegmark  calls  the  ‘Level  II  Multiverse’  .  .  .  
 
 
 
II.  The  Level  II  Multiverse  
One  immediately  asks:  exactly  what  kinds  of  variation  (and  how  much  variation  of  each  kind),  are  
being  suggested?  Of  course,  it  all  depends  on  the  model  chosen  .  .  .  
 
Here  we  should  bear  in  mind  that  inflationary  cosmology  is  a  framework  of  ideas  and  results,  not  
a  theory  that  is  defined  with  all  due  mathematical  rigour.  So  neither  are  its  individual  models-­‐-­‐-­‐
and  there  is  a  lot  of  room  for  debate  about  which  models  are  best.    
 
This  is  regrettable,  of  course.  But  the  point  does  not  apply  only  to  inflationary  cosmology.  The  
standard  model-­‐-­‐-­‐indeed  any  of  our  quantum  field  theories  of  interacting  systems-­‐-­‐-­‐is  also  not  
defined  with  all  mathematical  rigour.  Accordingly,  many  brilliant  people,  working  in  the  more  
mathematical  areas  of  theoretical  physics,  are  working  to  provide  more  rigorous  formulations  of  
quantum  field  theories,  and  thus  also  of  inflationary  cosmology.    
 
(In  fact  the  problems  of  rigour  are  deep:  they  originate  in  the  theories’  answers  to  
straightforward  questions  involving  troubling  infinities.  I  will  return  to  these  problems  when  
discussing  renormalization  at  the  end  of  this  Section.  And  you  can  read  more  about  them  in  Plus  
magazine’s  quantum  field  theory  articles,  at:    
http://plus.maths.org/content/brief-­‐history-­‐quantum-­‐field-­‐theory.    
Cf.  especially  http://plus.maths.org/content/problem-­‐infinity)  
 

  3  
So  let  us  bear  this  warning  in  mind,  but  nevertheless  proceed-­‐-­‐-­‐and  ask:  what  kinds  of  variation  
are  being  suggested?  
 
At  first  sight,  there  seem  to  be  two  different  ways  in  which  ‘the  laws  of  physics  could  be  
different’.  I  will  present  them  in  Sections  IIA  and  IIB.  But  as  we  will  see  at  the  end  of  this  Section,  
inflationary  cosmology  suggests  that  the  first  way  in  fact  happens  as  a  special  case  of  the  second  
way.    
 
IIA.  Varying  constants  
The  first  way  is  the  obvious  one:  just  imagine  altering  the  equations!  A  lot  of  attention  has  
focused  on  what  seem  the  most  modest  sorts  of  alteration.  Namely,  not  replacing  ‘the  whole  
shebang’  (like  chucking  the  equations  of  general  relativity  in  the  bin,  in  favour  of  Newton’s  theory  
of  gravity  or  other  modern  rivals  of  general  relativity),  but  altering  the  values  of  parameters  
occurring  in  the  equations,  while  keeping  the  same  form  for  the  equations,  including  which  
quantities  they  mention.  These  parameters  are  often  constants  of  nature.  For  example:  imagine  
altering  the  speed  of  light,  or  the  strength  of  gravitational  attraction  as  encoded  in  Newton’s  
constant  in  his  universal  law  of  gravitation,  or  the  ratio  of  the  strengths  of  the  gravitational  and  
electromagnetic  forces.    
 
In  the  last  several  decades,  there  has  been  a  lot  of  speculation  about  what  the  world  would  be  
like  under  such  alterations  (or  combinations  of  them).  And  these  speculations  have  been  very  
fruitful,  for  two  reasons.  First:  since  one  is  ‘only’  (!)  altering  the  values  of  parameters,  this  means  
that  calculating  what  the  world  would  be  like  is  much  more  tractable  than  it  would  be,  if  one  also  
altered  the  form  of  the  laws-­‐-­‐-­‐for  who  knows  what  alternative  laws  one  should  focus  on?    
 
Second:  it  has  turned  out,  in  countless  such  calculations,  that  even  rather  small  alterations  of  
parameters’  values  from  their  actual  values  would  make  for  a  world  that  is  incompatible  with  life  
as  we  know  it.  For  example,  the  altered  physics  would  be  incompatible  with  the  formation  of  
stars  and  of  planets,  and  thus  of  the  complex  carbon-­‐based  chemistry  on  which  life  depends.    
 
Here,  we  meet  the  idea  of  ‘anthropic  reasoning’.  Roughly  speaking,  this  is  the  idea  that  we  might  
take  this  result-­‐-­‐-­‐that  only  a  narrow  window  around  the  actual  value  of  a  constant  of  nature  is  
compatible  with  life  as  we  know  it-­‐-­‐-­‐to  be  the  best  explanation  of  why  the  value  is  what  it  is.  But  
this  is  not  the  place  to  pursue  this  topic.  (For  more  discussion,  cf.  pp.  143  and  302  of  Tegmark’s  
book;  or  many  articles  in  Plus  magazine,  such  as  http://plus.maths.org/content/planets-­‐
universes-­‐part-­‐I;  and  http://plus.maths.org/content/why-­‐are-­‐we-­‐here;    
Cf.  also  Wilczek  (2008);  the  definitive  book-­‐length  treatment  remains  Barrow  and  Tipler  (1986).)  
 
For  my  purpose  here-­‐-­‐-­‐namely,  to  summarize  Tegmark’s  four-­‐level  multiverse-­‐-­‐-­‐all  I  need  at  the  
moment  is  the  idea  of  varying  the  laws  of  physics  by  making  such  variations  in  the  constants.  
This  is  the  first  of  the  two  ways  of  varying  the  laws;  and  as  I  mentioned,  it  is  the  more  obvious  
one.  
 
II.B.  Varying  laws  by  breaking  a  symmetry  
The  second  way  is  best  introduced  by  an  analogy  with  a  crystal  forming,  and  how  such  a  process  
exhibits  ‘breaking  a  symmetry’.  When  a  crystal  forms,  its  atoms  line  up  in  a  regular  lattice;  for  
example  a  cubical  lattice  like  in  Escher’s  famous  picture.    Such  a  lattice  can  be  tilted  one  way  or  
another:  the  rods  do  not  have  to  be  vertical  and  horizontal,  as  they  are  in  Escher’s  picture.  In  fact,  
in  a  real  crystal  the  lattice  is  not  tilted  in  the  exact  same  way  throughout  the  crystal.  Instead,  the  
crystal  is  made  up  of  tiny  patches,  called  domains.  In  any  one  domain,  comprising  perhaps  
millions  of  atoms,  the  lattice  is  tilted  in  the  same  way.  But  the  lattices  in  different  domains  will  
have  different  tilts.    
 
Agreed:  all  this  is  hardly  surprising.  Typically  the  crystal  forms  by  cooling  a  sample  of  liquid.  
(Remember  the  blue  copper  sulphate  crystals  you  made  in  chemistry  class  at  school,  competing  
for  who  had  the  gentle  touch  to  grow  the  biggest  one!).  The  atoms  settling  down  into  a  lattice  is  
an  example  of  that  ubiquitous  phenomenon  in  physics  and  chemistry:  a  system  settling  down  
into  its  state  of  lowest  energy.  Ideally,  the  lattice  would  have  the  same  tilt  throughout  the  entirety  
of  the  sample.  The  entire  sample  would  form  one  vast  domain,  and  the  energy  would  indeed  be  

  4  
lowest  since  there  would  be  no  domain-­‐walls,  i.e.  lines  or  surfaces  within  the  crystal  where  the  
tilt  changes,  with  an  associated  energy  cost.  But  that  is  of  course  very  much  an  ideal,  i.e.  an  
unrealistic,  situation.  In  reality,  there  will  be  a  vast  number  of  tiny  events  scattered  throughout  
the  cooling  liquid  (e.g.  the  presence  of  a  speck  of  dust,  or  a  passing  cosmic  ray):  each  of  which  can  
result  in  a  shift  in  the  local  orientation  of  the  lattice.    
 
To  sum  up:  we  expect  the  physics  and  chemistry  of  the  crystal’s  atoms  to  determine  the  general  
shape  of  the  lattice,  e.g.  is  it  cubical,  as  in  the  picture?  But  we  expect  the  exact  orientation  to  be  a  
matter  of  sheer  happenstance,  influenced  by  local  events;  and  so  it  is  very  likely  to  be  different,  in  
different  parts  of  the  crystal.    
 
There  is  a  buzz-­‐word  for  the  variety  in  the  tilts,  the  different  tilts  in  the  different  domains:  
symmetry-­‐breaking.  The  idea  is:  the  physics  and  chemistry  of  the  crystal’s  atoms,  and  the  overall  
cooling  process,  does  not  prefer  one  tilt  over  any  others.  These  principles  allow  the  lattice  any  
orientation  in  space:  they  have  full  rotational  symmetry.  But  a  specific  state  of  the  system,  i.e.  a  
specific  domain,  breaks  the  symmetry:  it  picks  an  orientation-­‐-­‐-­‐because  of  microscopic  causes  
that  we  may  never  know  (nor  care!)  about.  
 
This  scenario,  of  symmetric  principles  having  a  non-­‐symmetric  instance,  generalizes.  Namely:  a  
law,  or  collection  of  laws,  of  physics  can  have  a  symmetry,  while  a  solution  to  those  laws  lacks  it-­‐-­‐
-­‐and  again,  not  because  of  some  mystery,  but  simply  as  a  matter  of  sheer  happenstance.  Once  you  
grasp  this  idea,  you  realize  that  examples  are  everywhere:  they  are  two  a  penny.  
 
For  example,  Newton’s  laws  of  mechanics  and  his  law  of  gravitation  are  rotationally  symmetric.  
They  encode  no  preferred  orientation.  Formally  speaking:    if  you  calculate  how  your  equation  
expressing  these  laws  is  transformed  by  rotating  the  coordinate  system  around  any  axis  by  any  
chosen  amount  (e.g.  around  the  axis  passing  through  the  Earth’s  North  Pole,  by  a  33  degree  
angle),  you  will  find  that  the  equation  remains  exactly  the  same.    
 
On  the  other  hand,  almost  all  the  solutions  of  Newton’s  laws,  i.e.  the  instantaneous  states  of  
bodies  (particles,  planets  …)  obeying  Newton’s  laws,  are  not  rotationally  symmetric.  They  are  not  
transformed  into  themselves  by  a  rotation  about  any  axis,  by  any  amount.  That  would  require  
that  the  solution  is  spherically  symmetric,  like  a  perfectly  symmetric  sphere  of  dust.    In  short:  the  
solution  breaks  the  symmetry  enjoyed  by  the  law.  Of  course,  this  is  in  no  way  a  problem:  it  is  
exactly  what  one  would  expect-­‐-­‐-­‐the  instantaneous  state  of  a  collection  of  bodies  is  usually  not  
symmetric,  and  the  lack  of  symmetry  is  no  doubt  usually  a  matter  of  sheer  happenstance,  of  
historical  contingency—of  where  the  bodies  happened  to  be  placed  and  how  they  happened  to  be  
moving.    
 
So  much  by  way  of  explaining  the  idea  of  symmetry-­‐breaking.  The  idea  is  very  important,  quite  
apart  from  Tegmark’s  Level  II  multiverse.  Suitably  developed,  it  is  crucial  in  various  parts  of  
modern  physics.  But  for  present  purposes,  its  importance  is  that  it  engenders  a  second  way  in  
which  one  can  envisage  ‘altering  the  laws  of  physics’.    
 
To  see  this,  imagine  you  are  a  tiny  homunculus  living  in  the  domain  of  a  crystal  with,  say,  a  
cubical  lattice;  and  knowing  nothing  of  the  world  beyond  that  domain.  As  it  happens,  common  
table  salt  has  a  cubic  lattice.  So  imagine  you  live  in  (a  domain  of)  a  crystal  of  salt.    
 
Confined  as  you  are  to  your  tiny  environment,  you  might  well  be  unaware  that  the  tilt  of  the  
lattice  in  your  domain  is  a  matter  of  sheer  happenstance.  You  might  well  imagine  that  the  tilt  of  
your  domain  is  as  much  a  matter  of  law,  as  is  the  cubical  structure  of  the  lattice.    
 
Thus  in  the  jargon  of  ‘bye-­‐laws’  which  I  introduced  at  the  end  of  Section  I:  the  tilt  of  your  domain  
is  in  fact  a  bye-­‐law-­‐-­‐-­‐merely  a  regularity  of  your  local  environment.  But  since  you  know  nothing  
of  the  world  beyond  your  domain,  you  might  well  mistake  it  for  a  universal  law.  
 
The  second  way  of    ‘altering  the  laws  of  physics’  is  now  clear.  The  suggestion  from  inflationary  
cosmology  is:  we  might  be  like  the  homunculus!  What  we  take  to  be  the  laws  of  physics  might  be  

  5  
merely  bye-­‐laws:  with  different  bye-­‐laws  being  true  in  some,  indeed  most,  of  the  other  universes  
of  the  multiverse,  universes  which  we  will  never  observe.    
 
So  in  the  analogy  with  the  crystal,  the  universes  of  the  multiverse  are  like  the  domains  of  the  
crystal.    And  more  specifically:  these  bye-­‐laws  might  involve  symmetry-­‐breaking.  They  might  
break  a  symmetry  of  some  genuine  law-­‐-­‐-­‐a  law  that  holds  good  across  the  multiverse-­‐-­‐-­‐by  
incorporating  some  or  other  matter  of  sheer  happenstance-­‐-­‐-­‐a  matter  that  played  out  differently  
in  different  universes.    
 
Indeed,  the  analogy  with  the  crystal  can  be  pushed  further:  just  as  the  crystal’s  domains  settle  
into  various  different  lower-­‐energy  states  during  a  global  process  of  cooling,  the  multiverse’s  
universes  settle  into  various  different  lower-­‐energy  states  during  a  global  process  of  cooling.    
 
Besides,  it  turns  out  that  in  some  models  of  inflationary  cosmology,  this  second  way  of    ‘altering  
the  laws  of  physics’  incorporates  the  first:  which  (cf.  Section  IIA)  involved  changing  the  constants  
of  nature.  Here,  two  other  great  themes  of  modern  physics  enter  our  story.  One  is  called  
‘renormalization’.  But  it  also  has  a  more  vivid  label:  ‘running  constants’.  The  other  has  only  one  
name,  but  a  very  familiar  one:  string  theory.  
 
Thus  the  first  theme  is  that  a  constant  of  nature  can  be  dependent  on  the  energy  involved  in  the  
process  concerned.  The  jargon  is  that  the  constant  ‘runs’  with  the  energy  scale;  so  people  talk  
about  ‘running  constants’.  Agreed,  this  certainly  sounds  like  weasel-­‐words,  to  avoid  sounding  
self-­‐contradictory  by  saying  ‘varying  constants’!  (Or  maybe  it  sounds  like  statistics  about  your  
training  for  the  marathon,  collated  by  your  athletics  coach!)      
 
But  actually,  it  is  not  self-­‐contradictory.  In  fact,  it  reflects  a  profound  physical  idea:  the  idea  of  
renormalization.  This  is  a  big  topic  which  returns  us  to  quantum  field  theory’s  troubling  
infinities,  which  I  mentioned  at  the  start  of  this  Section.  I  will  not  go  further  into  it.  But,  again,  you  
can  read  more  about  it  in  Plus  magazine  quantum  field  theory  articles.  Among  those  articles,  how  
renormalization  tames  the  troubling  infinities  is  emphasized  in:  
http://plus.maths.org/content/rise-­‐qed  
http://plus.maths.org/content/quantum-­‐pictures    
and  
http://plus.maths.org/content/going-­‐flow-­‐0  
 
So  if  we  accept  that  a  constant  of  nature  can  be  dependent  on  energy,  and  that  the  universe-­‐-­‐-­‐
okay,  the  multiverse!-­‐-­‐-­‐is  cooling  down  after  the  Big  Bang  like  an  enormous-­‐-­‐-­‐okay,  ginormous!-­‐-­‐-­‐
fluid  which  is  slowly  crystalizing,  then  it  makes  sense  that  the  constants  might  also  settle  down,  
as  the  energy  decreases,  to  rather  different  values  in  different  domains  of  the  multiverse:  i.e.  in  
different  universes  of  the  Level  1  multiverse.    
 
It  is  one  thing  to  say  ‘it  makes  sense  that  the  constants  might  also  settle  down  to  different  values  
in  different  domains’.  It  is  another  thing  to  have  a  specific  theory  saying  that  indeed  they  do!  And  
here  enters  the  second  theme:  string  theory.    
 
Again,  we  first  need  a  ‘health  warning’.  Like  the  theories  previously  mentioned-­‐-­‐-­‐    quantum  field  
theories,  inflationary  cosmology-­‐-­‐-­‐string  theory  is  not  yet  defined  in  a  mathematically  rigorous  
way.  (Indeed,  it  is,  broadly  speaking,  more  ill-­‐defined  than  quantum  field  theory:  most  string  
theorists  agree  that  what  we  call  ‘string  theory’  is  really  a  collection  of  insights  into  a  theory  we  
do  not  yet  ‘have’,  in  the  way  we  do  quantum  field  theory.)  But  let  us  again  proceed,  bearing  in  
mind  that  the  voice  of  scepticism  is  again  whispering  in  our  ears.    
 
String  theory  is  of  course  complicated.  (For  some  details,  cf.  the  Plus  article  
http://plus.maths.org/content/string-­‐theory-­‐newton-­‐einstein-­‐and-­‐beyond).  But  for  the  purpose  
of  expounding  Tegmark’s  multiverse,  all  we  need  is  the  fact  that  string  theory  indeed  does  
propose  that  familiar  constants  of  nature,  such  as  the  ratio  of  the  strengths  of  the  gravitational  
and  electromagnetic  forces,  could  settle  down  to  different  values  in  different  domains  of  the  
Level  I  multiverse.  Roughly  speaking,  each  domain  is  associated  with  a  different  lowest-­‐energy  
state  of  string  theory.  

  6  
    
(Here  we  meet  other  bits  of  jargon.  (1)  These  different  lowest-­‐energy  states  are  called  ‘vacua’.  
This  jargon  is  taken  over  from  quantum  field  theory.  To  be  honest,  it  is  misleading,  since  
‘vacuum’  suggests  there  is  nothing,  i.e.  no  system:  while  in  this  usage,  there  is  always  a  system-­‐-­‐-­‐
the  various  fields  and  strings-­‐-­‐-­‐and  ‘vacuum’  denotes  a  state  of    lowest  energy.  (2)  The  
complicated  set  of  possible  states,  with  the  various  different  minima,  is  called  ‘the  landscape’.)  
 
Thus  we  see  how  the  second  way  of    ‘altering  the  laws  of  physics’-­‐-­‐-­‐symmetry-­‐breaking  and  us  
being  like  the  homunculus  in  a  domain  of  a  salt  crystal-­‐-­‐-­‐  incorporates  the  first,  i.e.  altering  the  
constants  of  nature.    
 
So  much  by  way  of  convincing  you  that  the  multiverse  involves  varying  laws.  Welcome  to  Level  
II!  
 
III.  The  Level  III  Multiverse  
I  turn  to  Uncle  Max’s  Level  III.  It  concerns  the  interpretation  of  quantum  theory:  a  controversial  
area  which  we  have  not  yet  had  to  tangle  with,  despite  referring  to  quantum  field  theories.    
 
But  I  will  discuss  it  only  briefly.  For  it  concerns  a  topic  which  readers  of  Plus  magazine  will  be  
very  familiar  with.  Namely:  the  paradox  of  ‘Schroedinger’s  cat’,  and  the  proposal  to  resolve  it  by  
adopting  the  Everettian  interpretation  of  quantum  theory,  also  known  as  the  ‘many  worlds  
interpretation’-­‐-­‐-­‐the  very  phrase  clearly  suggests  a  multiverse!    
 
Recall  that  ‘Schroedinger’s  cat’  is  really  a  parable  for  making  vivid  the  measurement  problem  of  
quantum  theory.  (For  details,  cf.  the  Plus  magazine  article:  
http://plus.maths.org/content/schrodingers-­‐equation-­‐what-­‐does-­‐it-­‐mean).  The  problem  is:  the  
lack  of  values  for  physical  quantities,  such  as  momentum  or  position  or  energy,  that  is  
characteristic  of  quantum  theory's  description  of  microscopic  systems,  should  also  infect  the  
familiar  macroscopic  realm  of  tables,  chairs  etc.    
 
For  example:  measuring  the  momentum  of  an  electron,  which  is  in  a  state  that  is  not  definite  for  
momentum  (called:  ‘a  superposition  of  momentum  eigenstates’)  should  lead  to  the  pointer  of  the  
apparatus  having  no  definite  position.  Rather,  it  should  be  in  a  superposition  of  states  that  are  
definite  for  position  (called  ‘position  eigenstates’).  But  this  is  not  what  we  see!  Witness  the  
success  of  classical  physics  in  describing  countless  physical  systems  as  having  at  all  times  definite  
values  for  all  physical  quantities.    
 
Schroedinger  makes  this  problem  vivid  by  imagining  that  a  cat  is  fatally  poisoned,  or  not,  
according  as  a  microscopic  quantum  system  e.g.  an  electron  has,  or  does  not  have,  some  
prescribed  value  for  some  prescribed  quantity  (e.g.  value  ‘5  units’  for  momentum  in  the  x  
direction).  Thus  Schroedinger  describes  what  he  calls  ‘an  infernal  device’  in  which  the  opening  of  
a  vial  of  poison  registers  a  prescribed  result  ‘5  units’  for  a  measurement  of  an  electron  in  a  
superposition  of  momentum  eigenstates.  So  the  cat  ends  up  in  a  superposition  of  being  fatally  
poisoned,  and  not:    in  a  superposition  of  being  dead  and  alive.  Yikes!  
 
It  is  now  nearly  eighty  years  since  Schroedinger  (in  1935)  wrote  his  ‘cat  paper’;  and  nearly  
ninety  years  since  he  and  Heisenberg  (in  1925-­‐26)  put  quantum  theory  in  its  established  form,  
with  its  idea  of  superpositions  of  eigenstates.  Unsurprisingly,  the  measurement  problem  was  
immediately  recognized  and  vigorously  debated,  both  by  the  new  theory’s  architects  like  
Schroedinger  and  Heisenberg,  and  by  other  luminaries  like  Einstein.  (So  it  was  not  a  matter  of  the  
problem  going  unrecognized  until  the  ‘cat  paper’  ten  years  later.)  What  is  more  surprising  is  that  
to  this  day-­‐-­‐-­‐nearly  ninety  years  later-­‐-­‐-­‐it  remains  controversial.    
 
So  again,  we  need  a  health  warning:  we  are  again  entering  very  contentious  territory.  But  here  it  
suffices  to  say  one  of  the  proposed  solutions  yields  Uncle  Max’s  Level  III  multiverse.  Namely,  
Everett’s  proposal  (in  1957:  so  more  than  fifty  years  ago)  that,  in  short,  a  quantum  superposition  
represents,  not  a  set  of  alternatives  of  which  one  is  real,  but  an  actual  plurality  of  realities.    
 
In  a  bit  more  detail:  the  universe  as  a  whole  has  a  quantum  state,  which  always  evolves  according  

  7  
to  the  Schroedinger  equation:  which  is  continuous  and  indeed  deterministic-­‐-­‐-­‐so  there  is  no  
discontinuous  ‘collapse  of  the  wave  function’.  Thus  the  universe  contains  a  plethora  of  Everettian  
‘worlds’  (also  called  ‘branches’),  where  each  such  ‘world’  is  something  like  the  familiar  
macroscopic  realm,  with  all  tables,  chairs  etc.  in  definite  positions,  with  definite  momenta.  But  
the  worlds  differ  among  themselves  about  these  positions  and  momenta.  Of  course,  each  of  us  
experiences  only  a  single  definite  macrorealm.  So  the  idea  is:  each  of  us  just  happens  to  be  in  one  
world,  rather  than  another  one.  Or  better:  there  are  many  versions  of  each  of  us,  each  version  
happening  to  be  in  one  world  rather  than  another.  So  this  is  indeed  a  proposal  for  a  multiverse.  
Welcome  to  Level  III!  (To  read  more  about  the  Everettian  worlds,  cf.  the  Plus  article:  
http://plus.maths.org/content/parallel-­‐universes)    
 
Notice  that  in  some  ways,  the  same  topics  and  questions-­‐-­‐-­‐even  conundrums-­‐-­‐-­‐will  arise  for  it  as  
arose  for  the  Level  I  multiverse.  In  that  discussion,  we  envisaged  that  there  are-­‐-­‐-­‐very  far  away,  
and  never  observable  by  us-­‐-­‐-­‐all  sorts  of  scenarios  involving  diverse  macroscopic  objects,  some  
of  which  might  well  be  uncannily  similar-­‐-­‐-­‐even  utterly  similar-­‐-­‐-­‐to  familiar  scenarios,  like  
Napoleon  losing  the  battle  of  Waterloo.  And  we  ended  by  envisaging  that  some  of  the  scenarios  
might  be  similar  enough  for  us  to    describe  them  as  containing  a  person  like  Napoleon  and  a  
battle  like  Waterloo-­‐-­‐-­‐but  in  which  that  person  (faux  Napoleon,  as  we  might  call  him)  wins  his  
battle  (which  we  might  call  faux  Waterloo).  
 
Similarly  here.  Just  as  Schroedinger  envisaged  that  an  electron  having  one  value  of  momentum  
rather  than  another  (‘5  units  rather  than  6’)  could  lead  to  a  cat’s  being  dead  or  alive:  so  also,  
Everett  and  his  advocates  envisage  that  the  two  or  more  realities  represented  by  some  
microscopic  quantum  superposition  might  lead  to  a  macroscopic  course  of  events,  a  history,  
‘branching’-­‐-­‐-­‐of  course,  unobservably  for  the  agents  within  it-­‐-­‐-­‐into  two  (or  more)  histories.  For  
example,  a  history  could  branch  into  some  in  which  Napoleon  loses  the  battle,  and  some  in  which  
he  wins.    
 
And  here  again,  we  have  to  be  careful  about  language.  On  this  view,  ‘Napoleon’  really  refers,  not  
to  a  single  person  or  human  body,  but  to  each  of  the  many  ‘descendants’,  in  the  various  branches,  
of  the  original  person  or  body.  Indeed,  the  Everettian  vision  is  that  there  were,  no  doubt,  
countless  microscopic  quantum  superpositions  well  before  the  date  of  the  battle,  appropriately  
correlated  with  the  values  of  macroscopic  quantities,  so  that  the  years  leading  up  to  the  date  of  
the  battle  involved  many  branchings,  and  the  ‘one  Napoleon’  whom  we  at  first  envisaged  starting  
the  battle  (of  whom  some  descendants  would  win,  and  others  would  lose)  was  himself  one  of  
many  descendants  of  earlier  stages  of  a  ‘Napoleon-­‐like  career’.    
 
So  there  are  some  clear  similarities  between  the  ideas  of  Level  I  and  Level  III.  But  notice  that  the  
argument  for  believing  in  Level  III  has  nothing  to  do  with  cosmology.  The  measurement  problem,  
and  the  various  proposals  for  solving  it,  and  the  arguments  back  and  forth  between  these  
proposals,  are  all  concerned  with  local  physics,  not  with  cosmology.    
 
This  difference  is  clear  in  another  way  also,  when  we  recall  Level  II  with  its  variation  in  the  
constants  and  in  the  form  of  the  laws.  The  arguments  for  believing  in  Level  II  concerned  
inflationary  cosmology  and  string  theory,  not  the  measurement  problem.  And  similarly  on  the  
other  side:  nothing  in  the  various  proposals  for  solving  the  measurement  problem  suggests  
variation  in  the  constants,  or  in  the  form  of  the  physical  laws.    
 
 
 
IV.  The  Level  IV  Multiverse  
And  so  to  Tegmark’s  fourth  and  last  dizzying  vision.  Unlike  the  first  three  Levels,  this  proposal  is  
independent  of  the  details  of  physics.  It  is  distinctively  philosophical:  it  is  a  claim  about  how  
mathematics  describes  physical  reality.  It  is  also  a  radical  claim:  namely,  that  physical  reality—
not  merely  is  described  by  mathematics-­‐-­‐-­‐but,  instead,  is  mathematics.  
 
About  this  fourth  Level,  I  propose  to  be  more  critical  than  about  the  three  others.  After  all,  
philosophical  claims  are,  broadly  speaking,  more  controvertible  than  physical  claims.  This  is  no  
doubt  largely  because  the  sorts  of  evidence  counting  for  and  against  them  is  less  sharply  defined  

  8  
than  for  physical  claims.  So  most  philosophers  accept  that  we  make  the  best  progress  in  
assessing  philosophical  claims-­‐-­‐-­‐in  so  far  as  we  ever  make  progress!-­‐-­‐-­‐by  subjecting  them  to  
rigorous  scrutiny.    
 
So  I  will  first  discuss  the  claim  that  physical  reality  is  mathematics,  in  general  terms:  indeed,  in  
partly  historical  terms.  Then  I  will  state  and  assess  Tegmark’s  argument  in  favour  of  it.  But  as  I  
just  warned:  I  will  reject  the  claim.  Sorry,  Uncle  Max!    
 
I  will  proceed  in  five  stages,  giving  each  stage  a  Subsection.  First  I  will  state  the  claim,  adding  a  bit  
of  history  (Section  IV.A).  Then  I  will  state  my  main  reason  for  rejecting  it:  that  reason  turns  on  
introducing  the  distinction  between  pure  and  applied  mathematics  (Section  IV.B).  To  properly  
justify  my  rejection,  I  will  then  give  some  details  about  that  distinction  (Section  IV.C).  In  Section  
IV.D,  I  will  use  this  distinction  to  rebut  Tegmark’s  own  argument  for  the  claim:  he  argues  that  it  
follows  from  something  we  almost  all  accept:  viz.  that  there  is  an  external  reality  completely  
independent  of  us  humans.  Finally,  in  Section  IV.E,  I  add  to  my  rebuttal  by  briefly  discussing  a  
philosophical  doctrine  about  the  nature  of  physical  properties:  a  doctrine  called  ‘quidditism’.        
 
IV.A.  From  Pythagoras  to  Galileo  and  Beyond  
Philosophers  have  long  had  a  name  for  the  claim  that  physical  reality  is  mathematics.  They  call  it  
‘Pythagoreanism’.  This  refers  to  Pythagoras,  the  ancient  Greek  (ca.  572  –  497  BC).  For  millennia,  
he  has  been  regarded  as  a  mystic  mathematician  who,  among  other  accomplishments,  proved  the  
theorem  that  bears  his  name,  advocated  the  idea  of  mathematical  proof,  and  believed  that,  in  
some  obscure  way,  the  world  was  made  of  numbers,  i.e.  of  mathematics-­‐-­‐-­‐hence  the  ‘ism’.    
 
(In  fact,  we  now  know,  thanks  to  modern  scholarship,  that  Pythagoras  did  none  of  these  things.  
As  often  happens,  it  is  a  case  of  the  sober  truth  being  duller  than  the  myth.  For  a  glimpse  of  the  
details,  cf.  the  review  by  Burnyeat  (2007),  citing  the  scholarly  work  of  Walter  Burkett.)    
 
But  setting  aside  the  details  of  the  historical  Pythagoras,  there  is  no  doubt  that  the  history  of  
science,  and  especially  of  physics,  provides  countless  illustrations  of  the  power  of  mathematical  
language  to  describe  natural  phenomena.  Famously,  Galileo  himself-­‐-­‐-­‐the  founding  father  of  the  
mathematical  description  of  motion-­‐-­‐-­‐envisaged  describing  many,  perhaps  all,  phenomena  in  
mathematical  terms.  Thus  in  The  Assayer  (1623),  he  wrote  the  following  (saying  ‘philosophy’  in  
roughly  the  sense  of  our  words  ‘natural  science’  or  ‘physics’):    
 
‘Philosophy  is  written  in  this  grand  book  —  I  mean  the  universe  —  which  stands  continually  
open  to  our  gaze,  but  it  cannot  be  understood  unless  one  first  learns  to  comprehend  the  language  
in  which  it  is  written.  It  is  written  in  the  language  of  mathematics,  and  its  characters  are  
triangles,  circles,  and  other  geometric  figures,  without  which  it  is  humanly  impossible  to  
understand  a  single  word  of  it;  without  these,  one  is  wandering  about  in  a  dark  labyrinth.’  
 
Of  course,  since  Galileo’s  time  the  language  of  mathematics  has  developed  enormously-­‐-­‐-­‐in  ways  
that  even  he,  a  genius,  would  have  found  unimaginable.  After  all,  already  by  the  mid-­‐seventeenth  
century  the  calculus  was  generalizing  vastly  from  Galileo’s  ‘triangles,  circles,  and  other  geometric  
figures’,  so  as  to  consider  arbitrary  curves  and  their  slopes  (derivatives)  and  the  areas  under  
them  (integrals).    
 
(More  precisely:  the  ‘arbitrary’  curves  need  to  satisfy  conditions,  such  as  being  suitably  smooth,  
i.e.  differentiable,  so  that  they  indeed  have  a  derivative-­‐-­‐-­‐conditions  that  are  subtle  to  state,  and  
were  hard  to  discover,  as  the  long  history  of  rigorizing  the  calculus  attests.)    
 
As  I  have  stated  them  so  far,  these  developments  seem  to  support  Pythagoreanism.  If  
mathematics  gives  us  so  powerful  a  language  for  describing  nature,  why  not  conclude  that  nature  
is  mathematics?    
 
As  I  said,  Tegmark  endorses  this  claim.  (In  fact,  this  quotation  from  Galileo  is  one  of  the  mottoes  
of  his  Chapter  10,  where  he  develops  and  defends  his  Pythagoreanism.)  But  he  also  ties  this  claim  
to  his  over-­‐arching  theme  of  a  multiverse.  That  is:  we  should  believe  in  the  equal  reality  of  all  the  
possible  mathematical  structures.  Welcome  to  Uncle  Max’s  Level  IV!  

  9  
  
IV.B.  Beware:  ‘mathematical  structure’  is  ambiguous    
As  I  announced,  my  aim  in  the  rest  of  this  review  is  to  make  sense  of  this  claim  and  criticize  it.  
This  will  of  course  involve  clarifying  the  idea  of  ‘mathematical  structure’  and  thus  the  idea  of  all  
possible  mathematical  structures.  But  I  can  already  state  the  gist  of  what  Tegmark  believes,  and  
of  my  critique  of  it.  
 
 Tegmark  does  not  claim  only  the  following:  
(1)  ‘Nature-­‐-­‐-­‐that  is  the  multiverse,  as  described  so  far,  with  its  levels  I  to  III-­‐-­‐-­‐is  a  vastly  
complicated  mathematical  structure’.  
He  also  claims:  
  (2)  There  is  a  mathematical  multiverse:  all  possible  mathematical  structures  are  equally  
real.    
So,  putting  (1)  and  (2)  together,  he  claims:  
(3)  The  multiverse,  as  described  so  far,  with  its  levels  I  to  III,  is  just  one  of  countlessly  
many,  equally  real,  such  structures.  
 
To  which  I  say:    
(i):  I  am  happy  to  concur  with  (1).  Well,  happy  enough  for  present  purposes.  But  I  think  
we  should  take  seriously  all  those  health  warnings-­‐-­‐-­‐Beware:  here  be  speculations!-­‐-­‐-­‐in  the  three  
previous  Sections.  
(ii):  I  am  happy  to  say  (2):  ‘There  is  a  mathematical  multiverse:  all  possible  mathematical  
structures  are  equally  real.’  But  I  mean  by  this  something  different  from  what  Tegmark  means:  
both  different  and  less  controversial  .  .  .  let  me  explain  .  .  .  
       
In  a  word,  my  message  to  Uncle  Max  is:  Distinguish!  That  is:  we  must  distinguish  between  applied  
mathematics    (also  called:  theoretical  physics)  and  pure  mathematics.  And  once  we  make  this  
distinction,  there  are  three  important  conclusions.  We  see  that:  
 
(a)  It  is  for  applied  mathematics  that  claim  (1)  is  tenable.  That  is:  it  is  tenable  that  the  multiverse,  
as  described  so  far,  with  its  levels  I  to  III  is  an  applied  mathematical  structure.  And  as  I  said:  I  
concur  with  this.      
 
(b)  It  is  for  pure  mathematics  that  I  am  happy  to  say  (2):  ‘There  is  a  mathematical  multiverse:  all  
possible  mathematical  structures  are  equally  real.’    
Here  I  should  note:  I  say  ‘happy’  to  be  vivid,  and  for  rhetorical  simplicity.  I  admit  that  this  
belief  in  a  vast  plethora  of  pure  mathematical  structures  is  controversial.  It  is  controversial  
because  most  philosophers  maintain  that  all  our  concepts,  beliefs  and  knowledge  come  from  our  
experience  of  the  empirical  world:  our  experience  of  nature  in  space  and  time.  That  seems  right  
to  me.  But  it  is  hard  to  reconcile  with  our  having  beliefs,  and  even  knowledge,  about  pure  
mathematical  structures:  especially  infinite  ones,  since  they  certainly  seem  not  to  be  given  to  us  
in  experience.      
Incidentally,  this  belief  in  a  plethora  of  pure  mathematical  structures  is  called  
‘Platonism’.  Note  that  this  is  not  ‘Pythagoreanism’:  because  the  topic  is  pure  mathematics,  not  the  
mathematical  description  of  nature.  Thus  I  call  myself  a  ‘reluctant  Platonist’:  ‘reluctant’  because  
Platonism  is  hard  to  reconcile  with  all  our  knowledge  coming  from  our  empirical  experience.  
 
(c)  Platonism  about  pure  mathematics  (whether  reluctant  or  enthusiastic)  has  nothing  to  do  with  
a  physical  multiverse.  To  put  the  point  in  terms  of  propositions  (1)  to  (3)  above:  if  (1)  says  nature  
is  an  applied  mathematical  structure,  and  (2)  asserts  a  plethora  of  pure  mathematical  structures,  
we  cannot  infer  (3),  that  nature  is  one  of  many  equally  real  structures.  
Indeed,  some  logic  books  classify  various  tempting  fallacies  of  reasoning.  One  of  them  
involves  two  premises  of  an  argument  having  a  word  in  common-­‐-­‐-­‐but  occurring  with  different  
meanings.    Assuming  that  in  each  premise  the  word  has  the  meaning  appropriate  there,  the  
premises  are  true;  or  at  least,  can  be  allowed  to  be  true,  for  the  sake  of  the  argument.  And  the  
argument  looks  valid,  when  you  look  at  the  pattern  of  the  words.  But  the  conclusion  is  false.    This  
is  called  ‘the  fallacy  of  equivocation’.  An  example,  with  apologies  to  Jane  Austen  and  lovers  of  
Pride  and  Prejudice:  ‘Elizabeth  Barrett  is  going  to  the  ball.  To  play  cricket,  we  need  the  ball.  
Therefore:  Elizabeth  Barrett  is  going  to  what  we  need.”  (Note,  incidentally,  that  the  equivocal  

  10  
word,  the  culprit  for  the  deceiving  pattern  of  words-­‐-­‐-­‐here:  ‘the  ball’-­‐-­‐-­‐need  not  occur  in  the  
conclusion.)  
To  sum  up:  in  Tegmark’s  propositions  (1)  and  (2),  ‘mathematical  structure’  is  equivocal  
between  applied  and  pure  structures.  That  in  itself  is  no  error.  The  reader  who  is  aware  of  the  
distinction  naturally  reads  (1)  and  (2)  with  the  appropriate  meaning:  just  as  in  my  example’s  two  
premises,  we  read  ‘the  ball’  with  the  appropriate  meaning.  But  it  is  an  error  to  infer  from  (1)  and  
(2),  read  with  their  respectively  appropriate  meanings,  to  (3).    For  (3)  needs  to  be  read  with  
‘mathematical  structure’  having  one  meaning  or  the  other-­‐-­‐-­‐applied  or  pure.  But  on  either  
reading,  it  doesn’t  follow  from  (1)  and  (2).  (Again:  when  they  are  read  with  their  appropriate  
meanings-­‐-­‐-­‐meanings  that  make  them  true,  or  at  least,  tenable  for  the  sake  of  the  argument.)  
 In  short:  beware  the  fallacy  of  equivocation!  
 
The  upshot  of  this  discussion  is  clear.  One  can  believe  in  the  multiverse  including  all  the  Levels  I  
to  III.  And  one  can  believe  this  multiverse  is  a  structure  in  the  sense  of  applied  mathematics.  And  
one  can  be  a  Platonist  about  pure  mathematics,  i.e.  believe  in  ever  so  many  pure  mathematical  
structures.  One  can  believe  all  this  without  believing  that  the  physical  multiverse-­‐-­‐-­‐that  
‘supreme’  applied-­‐mathematical  structure!-­‐-­‐-­‐is  a  pure  mathematical  structure.    
 
So  much  by  way  of  stating  what  I  called  ‘the  gist  of  what  Tegmark  believes,  and  of  my  critique  of  
it’.  I  apologize:  for  a  gist,  it  was  inordinately  long.  So  after  a  gist  as  long  as  that,  you  may  not  want  
to  hear  further  details!  But  rest  assured:  it  was  the  main  message  of  my  Section  IV.  Thus  what  
follows  is  only  for  afficionados  and  enthusiasts.  But  in  the  spirit  of  friendly  philosophical  debate-­‐-­‐
-­‐let  us  proceed!  
 
There  is  more  to  be  said,  in  three  ways.  First,  I  should  give  more  explanation  of  my  distinction  
between  applied  and  pure  mathematics:  this  I  do  in  Section  IV.C.  Second,  Tegmark  gives  a  specific  
argument  for  the  claim  I  reject:  that  the  physical  multiverse  is  a  (pure)  mathematical  structure.  
So  I  need  to  address  this  argument:  I  will  do  so  in  Section  IV.D.  This  leads  to  my  final  task.  For  in  
the  course  of  his  argument,  Tegmark  does  briefly  register  that  ‘mathematical  structure’  is  
ambiguous  in  the  sort  of  way  I  have  expounded.  So  in  in  Section  IV.E,  I  take  this  up:  it  will  lead  us  
to  a  philosophical  debate  about  the  nature  of  properties.          
 
 
IV.C.  Applied  vs.  Pure  Mathematics  
To  spell  out  the  distinction  between  applied  and  pure  mathematics,  I  will  describe  each  in  turn.    
 
Applied  mathematics  gives-­‐-­‐-­‐or  at  least:  aims  to  give!-­‐-­‐-­‐true  descriptions  of  empirical,  in  
particular  physical,  phenomena  that  are  located  in  space  and  time.  (Here  ‘true’  does  not  mean  
‘complete’:  our  description  can  be  true  even  though  typically  the  phenomena  have-­‐-­‐-­‐and  we  
know  them  to  have-­‐-­‐-­‐myriad  details  that  our  description  omits.)      
 
For  example,  I  spill  my  glass  of  milk  and  it  spreads  across  the  table.  Applied  mathematics  
successfully  describes  how  it  flows,  in  terms  of  relevant  physical  quantities,  such  as  the  positions,  
velocities  and  densities  of  small  volumes  of  the  milk.  So  in  this  example,  the  neglect  of  myriad  
details  is,  in  part,  a  matter  of  neglecting  the  atomic  constitution  of  milk.  Of  course,  this  is  done  by  
modeling  the  milk  as  composed  of  volumes  that  are  large  enough  to  contain  many  atoms  (and  so,  
we  hope,  to  be  unaffected  by  atomic  phenomena);  but  which  are  also  small  by  our  human  
standards,  so  that  the  milk  seems  to  be  continuous  in  its  make-­‐up.    
 
For  my  purposes,  the  crucial  feature  of  this  example  is  the  mention  of  relevant  physical  
quantities:  in  the  example,  the  mention  of  position,  velocity  and  density.  Of  course,  it  is  one  of  the  
great  glories  of  physics  since  Galileo’s  day  that  it  has  introduced  new  quantities-­‐-­‐-­‐sometimes  
very  arcane  ones-­‐-­‐-­‐and  refined  old  quantities-­‐-­‐-­‐sometimes  in  very  subtle  ways;  and  has  
combined  the  new  and  the  old  in  a  collection  of  laws  and  methods  that-­‐-­‐-­‐though  fallible,  and  
indeed  changing  with  the  passing  decades-­‐-­‐-­‐has  gone  from  one  success  to  another,  both  in  
theoretical  understanding  and  in  empirical  quantitative  prediction.      
 
As  examples  of  introducing  new  quantities,  we  might  take  (in  roughly  historical  order):  
momentum,  kinetic  energy,  potential  energy,  electric  charge,  entropy,  spin.      As  examples  of  

  11  
refining  old  quantities  in  a  subtle  way,  we  might  take  (again,  in  roughly  historical  order):  the  
refinement  of  the  concept  of  mass  (and  so  density)  between  the  times  of  Galileo  and  Newton;  
Einstein’s  critique  of    absolute  simultaneity  in  the  Newtonian  notion  of  time;  and  his  unification  
of  mass  and  energy.    
 
Without  doubt,  these  and  similar  examples  of  the  development  of  physical  concepts  are  among  
the  great  glories  of  physics.  And  this  is  so,  wholly  irrespective  of  what  laws  (e.g.  laws  of  motion)  
the  concepts  may  enter  into.  Indeed,  this  is  something  that  every  student  of  physics  knows.  
Grasping  the  ideas  involved  in  quantities  like  energy,  entropy  or  spin,  is  a  struggle  and  an  
achievement-­‐-­‐-­‐and  usually  much  harder  than  remembering  the  general  equations,  i.e.  the  laws,  
which  mention  the  quantities!  
 
So  to  round  off  my  sketch  of  physics  (or  applied  mathematics):-­‐-­‐-­‐  We  have  come  along  way  from  
Galileo’s  ‘triangles,  circles,  and  other  geometric  figures’.  In  his  day,  it  was  indeed  reasonable  to  
hope  that  for  its  concepts,  i.e.  its  quantities,  physics  could  manage  with  just  the  concepts  of  
geometry  (as  inherited  from  the  Greeks:  triangles,  circles,  areas,  volumes  .  .  .)  and  maybe  a  little  
more-­‐-­‐-­‐such  as  notions  of  contact  or  impact,  and  mass  and-­‐or  density.  This  was  precisely  the  
agenda  of  the  seventeenth  century  corpuscularian  philosophy  (also  called:  mechanical  
philosophy)  spearheaded  by  Galileo.  But  it  was  not  to  be.  Nature’s  imagination  outstrips  ours!  So  
as  physics  proceeded  to  examine  successive  new  domains  of  phenomena,  it  has  had  to  introduce  
a  succession  of  new  quantities  (and  has  also  had  to  refine  old  ones).  It  is  these  distinctive  
physical  quantities  (and  of  course,  their  values  for  the  system  described)  that  are  mentioned  by  
the  symbols  in  the  equations  of  physics.    
 
So  to  sum  up:  nowadays,  we  should  revise  Galileo’s  saying.  Namely,  instead  of  saying  ‘Nature  is  a  
book  written  in  the  language  of  mathematics’,  nowadays  we  should  say:    ‘Nature  is  a  book  
written  in  the  syntax  of  mathematics,  but  with  the  semantics  of  physics’.  (This  aphorism  is  due  to  
Ernan  McMullin  in  his  excellent  paper  on  Galilean  idealization  (1985,  p.  252-­‐253).)  
 
I  can  now  be  more  precise  about  what  at  the  end  of  Section  IV.B  I  called  an  ‘applied-­‐mathematical  
structure’.  In  short:  it  is  an  assignment  to  each  of  a  collection  of  objects  of  their  values  for  various  
relevant  physical  quantities.  Or  to  say  it  with  a  bit  more  physics  jargon,  and  in  a  bit  more  detail:-­‐-­‐  
(i) Physics  tends  to  think  of  an  assignment  as  a  function,  to  call  objects  
‘systems’,  and  to  think  of  such  an  assignment  (function)  varying  with  time,  to  
represent  values  changing  over  time.  
(ii) So  the  short  statement  above  applies  to  a  collection  of  systems  at  a  time:  
such  an  assignment  (function)  is  an  ‘instantaneous  state’.  But  the  collection  
of  such  states  as  time  varies-­‐-­‐-­‐i.e.  a  history  of  the  collection  of  systems-­‐-­‐-­‐
would  of  course  also  count  as  an  applied-­‐mathematical  structure.    
(iii) Comments  (i)  and  (ii)  apply  to  field  theories,  where  we  think  of  physical  
quantities  applying  to  spatial  or  spacetime  points:  e.g.  the  electric  of  
magnetic  field  at  points  of  space.  It  is  just  that  here  the  objects  in  question  
are  the  points.  
(iv) Though  I  will  not  go  into  detail:  these  comments  also  apply  to  quantum  
theories,  in  particular  quantum  field  theories.  The  idea  is  that  an  
instantaneous  quantum  state  is  specified  by  the  quantities  for  which  it  is  an  
eigenstate,  i.e.  definite  in  value.  (To  be  strictly  true,  this  statement  requires  
that  we  set  aside  a  subtlety,  called  ‘superselection’;  but  let  that  pass.)    
 
So  much  for  applied  mathematics.  What  of  pure  mathematics?  What  did  I  mean  when  above  I  
shouted  to  my  Uncle  Max:  ‘Distinguish!’  
 
My  point  was  this.  As  every  student  of  mathematics  learns,  in  the  mid-­‐nineteenth  century  there  
arose,  for  several  reasons,  the  idea  of  mathematics  as  the  investigation  of  arbitrary  structures.  In  
other  words:  people  started  to  think  of  mathematics  as  the  investigation  of  the  consequences  of  
arbitrary  rules  that  a  mathematician  postulates  as  governing  some  domain  of  elements.  These  
rules  are  abstract  in  the  sense  that  only  their  structural  behaviour  counts;  and  the  elements  are  
abstract  in  the  sense  that  nothing  is  assumed  about  their  natures,  or  relations  to  one  another,  
except  that  they  obey  the  announced  rules.    

  12  
  
To  clarify  this  development,  it  is  worth  pausing  over  the  history.  I  say  this  happened  ‘for  several  
reasons’  because  one  can  discern  different  motivations  in  different  branches  of  mathematics.  
Three  obvious  ones  are  as  follows.  
   
(1)  In  geometry,  the  rise  of  consistent  axiomatizations  of  non-­‐Euclidean  geometries  
forced  mathematicians  to  distinguish  between  (a)  the  actual  geometry  of  physical  space-­‐-­‐-­‐an  
empirical  matter  that  is  revealed  by  the  behaviour  of  rulers  and  protractors  (‘Do  the  angles  of  
this  triangle  add  up  to  180  degrees?’)-­‐-­‐-­‐and  (b)  the  idea  of  a  pure  system  of  geometry-­‐-­‐-­‐which  
could  be  consistent,  and  worthy  of  investigation,  even  if  it  does  not  describe  physical  space.  
(Again,  you  can  find  out  more  about  non-­‐Euclidean  geometry  in  a  Plus  article.  Go  to:  
http://plus.maths.org/content/non-­‐euclidean-­‐geometry-­‐and-­‐indras-­‐pearls)  
 
(2)  In  algebra,  the  investigations  of  Hamilton  and  Grassmann  into  new  kinds  of  
magnitude  (e.g.  quaternions),  with  new  rules  governing  their  operations  of  addition,  
multiplication  etc.,  liberated  people  in  a  similar  way.    An  algebraic  system  could  be  defined  by  
axioms,  and  consistent,  and  worthy  of  investigation,  even  if  it  violated  the  rules  of  addition  etc.  
familiar  from  arithmetic.  For  a  Plus  article  on  quaternions,  go  to:  
http://plus.maths.org/content/curious-­‐quaternions  
   
(3)  Two  other  factors-­‐-­‐-­‐the  rise  of  set  theory  and  logic,  and  the  concomitant  increasing  
concern  to  make  deduction  utterly  rigorous,  by  formalizing  mathematical  language-­‐-­‐-­‐also  
contributed  to  this  conception  of  pure  mathematics  as  the  investigation  of  structure:  in  principle,  
of  any  structure.  
 (Both  factors  were  themselves  prompted  by  various  paradoxical  arguments  that  could  
be  formulated  in  non-­‐rigorous  mathematical  language,  e.g.  about  sets.  For  more  details  about  
these  paradoxes,  cf.  the  Plus  articles:  
http://plus.maths.org/content/mathematical-­‐mysteries-­‐barbers-­‐paradox  
http://plus.maths.org/content/cantor-­‐and-­‐cohen-­‐infinite-­‐investigators-­‐part-­‐i)  
 Indeed,  set  theory  and  the  formal  languages  introduced  by  the  study  of  logic  provided  
the  perfect  framework  for  this  modern  conception  of  pure  mathematics.  For  the  idea  of  a  set  is  so  
simple,  almost  contentless!  You  just  gather  any  old  things  together  in  a  set  by  a  pair  of  curly  
brackets,  and  you  keep  going,  forming  sets  of  sets,  and  sets  of  sets  of  sets,  as  much  as  you  wish.  
And  you  are  then  free  to  postulate  any  structure  you  like  on  your  sets  (e.g.  any  new  rule  of  
‘addition’  you  like),  and  to  investigate  your  postulates’  consequences-­‐-­‐-­‐by  stating  your  postulates  
and  deductions  in  the  formal  languages  supplied  by  logic.      
 
(For  more  details  about  how  this  conception  of  pure  mathematics  arose,  cf.  Kline’s  magisterial  
history  of  two  millennia  of  mathematics  (1972,  especially  pp.  1036-­‐1038).  Finally:  in  calling  this  
the  ‘modern  conception’  of  mathematics,  I  chose  my  words  deliberately!  For  Gray  (2008)  makes  a  
good  case  that  this  conception  is  ‘modernist’  in  a  similar  sense  to  that  in  which  art  became  
‘modern’  at  about  the  same  time,  i.e.  the  mid-­‐nineteenth  century.)      
 
Thus  we  arrive  at  the  modern  idea  of  a  ‘pure  mathematical  structure’.  In  short:  it  consists  of  a  
collection  of  objects,  equipped  with  some  postulated  properties  and  relations  e.g.  an  operation  of  
multiplication  subject  to  certain  rules.  The  objects,  properties  and  relations  are  all  abstract  in  the  
sense  nothing  is  assumed  about  their  natures,  except  the  behaviour  postulated  by  the  rules.    
 
In  a  bit  more  detail:  as  I  mentioned  above,  set  theory  turned  out  to  be  an  ideal  lingua  franca  for  
articulating  this  idea.  Thus  the  idea  of  ‘a  collection  of  objects,  equipped  with  some  properties  and  
relations’  was  made  precise  as  ‘a  set  of  objects,  with  each  postulated  property  being  a  subset,  
each  postulated  binary  relation  being  a  set  of  ordered  pairs  etc.’.    The  power  of  gathering  things  
together  in  curly  brackets-­‐-­‐-­‐and  iterating!    
 
I  need  not  go  into  further  detail  about  this  abstract  and  structural  conception  of  pure  
mathematics.  What  matters  for  us  is  the  main  contrast  with  applied  mathematics,  and  its  
structures.  Namely:  there  is  no  mention  of  physical  quantities.      
 

  13  
Agreed:  physical  enquiry,  and  more  generally  empirical  enquiry,  may  well  have  played  a  crucial  
heuristic  role  in  leading  a  pure  mathematician  to  articulate  a  structure.  Of  course,  the  history  of  
mathematics  contains  many  examples  of  this.  But  the  resulting  pure  mathematical  structure  
makes  no  mention  of  the  physical  quantities  that  inspired  it.    Being  heuristically  important  does  
not  imply  being  part  of  the  official  content  of  what  results.  
 
I  agree  that  this  ‘lack  of  content’  of  a  pure  mathematical  structure  is  perfectly  compatible  with  
the  idea  that  something  ‘contentful’,  e.g.  something  mentioning  physical  quantities,  is  an  instance  
of  a  pure  mathematical  structure.  For  the  relation  of  instantiation  allows  a  great  deal  of  ‘loss  of  
content’.    
For  example:  Here  I  am  with  my  myriad  properties,  of  size,  shape,  mass,  temperature  etc.  
Here  is  my  family  with  its  myriad  properties  and  relations,  e.g.  our  total  mass,  one  person  being  
more  massive  than  another,  one  person  being  between  two  others  etc.  All  this  is  ‘contentful’.  But  
I  am  an  instance  of  many  abstract  ‘contentless’  properties:  being  less  than  two  in  number,  not  
being  a  set  etc.  And  similarly,  my  family  as  a  collection  of  objects  is  an  instance  of  many  abstract  
‘contentless’  properties  and  relations:  being  more  than  two  in  number,  being  the  base-­‐set  of  a  
group  of  order  6  etc.      
 
But  on  the  other  hand:  this  ‘lack  of  content’  of  a  pure  mathematical  structure  is  incompatible  with  
something  ‘contentful’,  e.g.  something  mentioning  physical  quantities,  being  a  pure  mathematical  
structure.  For  however  we  choose  to  make  ‘content’  precise  (e.g.  in  terms  of  physical  quantities  
or  in  some  other  way):    something  contentful  cannot  be  literally  identical  with  something  
contentless.    
 
To  sum  up:  The  ‘is’  of  identity,  e.g.  in  ‘a  =  b’,  is  NOT  the  ‘is’  of  instantiation,  e.g.  in  ‘Max  is  tall’.  If  
indeed  a=  b,  then  a  and  b  have  the  very  same  content,  if  any  (in  whatever  sense  of  ‘content’  you  
like).  For  there  is  only  one  entity:  a,  which  is  also  called  ‘b’.  But  in  the  case  of  ‘Max  is  tall’,  Max  can  
have  countless  properties,  as  contentful  as  you  like  (in  whatever  sense),  that  are  not  ‘picked  up  
on’  or  encoded  by,  or  part  of  the  meaning  of,  the  predicate  ‘is  tall’.  
 
With  that  discussion  in  hand,  my  critique  of  Tegmark’s  Pythagoreanism-­‐-­‐-­‐his  claim  that  nature  is  
mathematics-­‐-­‐-­‐is  very  simply  stated.  It  all  turns  on  the  simple  but  crucial  distinction  between  the  
‘is’  of  identity  and  the  ‘is’  of  instantiation.    
 
Thus  I  am  willing  to  concede  to  Tegmark:    
(A):  The  physical  multiverse  instantiates  a  pure  mathematical  structure.  Indeed,  being  a  
Platonist  about  pure  mathematics,  though  not  yet  convinced  of  his  multiverse’s  first  three  Levels  
(remember  the  health  warnings!):  I  would  happily  agree  with  Tegmark  to  the  conditional  claim:    
(A’):  If  there  is  a  physical  multiverse  (with  whichever  combination  of  Tegmark’s  first  
three  Levels),  then  it  instantiates  a  pure  mathematical  structure.  
 
But  still  I  deny:    
(B):  The  physical  multiverse  is  a  pure  mathematical  structure.    
 
The  contrast  between  (A)  and  (B),  between  ‘instantiates’  and  ‘is’,  might  seem  a  philosopher’s  
quibble.  But  it  is  not.  For  it  is  claim  (B),  not  (A),  that  is  needed  to  get  from  Platonism  about  pure  
mathematics  to  Pythagoreanism:  from  Platonism  about  pure  mathematics  to  Tegmark’s  Level  IV  
multiverse!  
 
IV.D.  Tegmark’s  argument  from  external  reality  
But  Uncle  Max  has  a  specific  philosophical  argument  for  his  Pythagoreanism,  his  Level  IV  
multiverse:  which  I  have  so  far  not  considered.  So  I  owe  you,  and  him,  an  assessment  of  it.    
 
You  will  not  be  surprised  to  hear  that  I  think  the  argument  fails.  It  is  invalid  in  the  sense  that  it  
has  a  true  premise  but  a  false  conclusion.  Indeed,  it  is  invalid  because  of  the  points  in  Sections  
IVB  and  IVC:  i.e.  the  ambiguity  of  ‘mathematical  structure’  that  arises  from  the  distinction  
between  applied  and  pure  mathematics.    But  at  the  risk  of  flogging  a  dead  horse,  I  should  state  
the  argument  and  say  why  I  think  it  fails.  
 

  14  
Tegmark’s  premise  is  what  he  calls  the  ‘External  Reality  Hypothesis’  (ERH).    
(ERH):  There  exists  an  external  reality  completely  independent  of  us  humans.  
His  argument  is  that  (ERH)  implies  what  he  calls  the  ‘Mathematical  Universe  Hypothesis’  (MUH).  
MUH  is  essentially  (B)  above.  In  his  exact  words:  
(MUH):  Our  external  physical  reality  is  a  mathematical  structure.  
When  you  add  to  (MUH),  Platonism  about  pure  mathematics,  that  all  the  possible  pure  
mathematical  structures  exist—which  I  agree  to-­‐-­‐-­‐you  indeed  get  Tegmark’s  Level  IV  multiverse.  
That  is,  you  get:  our  external  physical  reality  (the  multiverse  with  Levels  I  to  III  taken  as  
accepted)  is  one  of  a  plethora  of  equally  real  mathematical  structures.  
 
So  what  is  the  argument  that  (ERH)  implies  (MUH)?  The  idea  is  that  since  external  reality  is  
completely  independent  of  us  humans,  it  must  have  a  description  which  is  utterly  free  of  
subjective  ingredients:  that  is,  utterly  free  of  factors  arising  from  biological  facts  about  human  
cognition,  or  cultural  facts,  or  facts  about  an  individual  human’s  psychology.    
 
Tegmark  has  a  vivid  metaphorical  name  for  these  subjective  ingredients.  He  calls  them  ‘baggage’:    
the  word  ‘baggage’  connoting  the  burden  or  error  in  our  description  of  nature  due  to  the  biases  
from  our  biological,  cultural  or  individual  history.  He  has  an  even  more  vivid  metaphor  for  the  
effort  to  strip  out  such  subjective  ingredients  from  our  description  of  nature:  the  effort  which  
science,  especially  physics,  has  historically  made,  and  should  continue  to  make,  so  as  to  overcome  
the  biases.  Thus  he  calls  getting  rid  of  the  subjective  ingredients  ‘reducing  the  baggage  allowance’  
(Chapter  10,  especially  pp.  255-­‐265:  the  argument  also  occurs  in  his  earlier  paper  (2008,  pp.  102-­‐
108).      
 
It  is  a  vivid  metaphor:  and  in  an  age  when  we  are  ruining  the  planet  with  too  much  jet  travel,  it  
seems  a  worthy  aim  .  .  .      
 
And  indeed,  I  agree  with  Tegmark  that  science,  especially  physics,  has  historically  made  
successive  efforts  to  overcome  various  cognitive  biases  due  to  our  subjective  constitution  
(biological,  cultural  or  individual).  I  also  agree  that  to  make  progress  in  the  future,  we  must  
expect  physics  to  continue  to  make  that  effort.  Besides,  many  other  scientists,  especially  
physicists,  and  philosophers,  would  agree  with  him.    
 
(In  particular,  many  historians  of  philosophy  would  agree  that  modern  physical  science  from  the  
seventeenth  century  onwards  has  successively  removed  various  subjective  ingredients  from  the  
scientific  world-­‐view.  For  example,  cf.  Williams’  discussion  of  what  he  calls  the  ‘absolute  
conception  of  reality’  (1978,  pp.  65-­‐67,  245-­‐249).)  
 
More  specifically,  concerning  Tegmark’s  argument:  I  also  believe  (ERH).  This  is  often  called  in  
philosophy,  ‘realism’.  As  it  is  usually  put:    the  realist  maintains  that  reality  is  utterly  independent  
of  the  human  mind.  I  also  am  happy  to  agree  with  Tegmark  that  (ERH)  implies  that  reality  must  
have  a  description  which  is  utterly  free  of  subjective  ingredients.  So  Max’s  premise  is  true.  
 
But  obviously,  this  does  not  imply  his  conclusion  (MUH)-­‐-­‐-­‐for  the  reasons  we  saw  in  Sections  IV.B  
and  IV.C.  That  is:  because  of  the  distinction  between  instantiates  and  is,  which  I  emphasized  in  
terms  of  the  physical  quantities.  
   
Thus,  in  terms  of  my  labels  (A)  and  (B)  at  the  end  of  Section  IV.C,  I  reply  to  Tegmark  as  follows:        
(i) (ERH)  and  the  fact  that  reality  must  have  an  utterly  objective  description  
implies  (A):  The  physical  multiverse  instantiates  a  pure  mathematical  
structure.  Or  remembering  the  previous  Sections’  health  warnings,  they  
imply  the  conditional  claim,    (A’):  If  there  is  a  physical  multiverse,  then  it  
instantiates  a  pure  mathematical  structure.  But:  
(ii) They  do  not  imply  (B):  The  physical  multiverse  is  a  pure  mathematical  
structure.  That  is,  they  do  not  imply  (MUH).  
 
 
 
 

  15  
  
IV.E  Is  a  property  ‘more  than’  its  web  of  relations  to  other  properties?    
In  this  final  subsection  (strictly  for  aficionados  and  enthusiasts!),  I  will  raise  some  more  
philosophical  controversy  about  the  distinction  between  instantiates  and  is.    
 
To  his  credit,  Tegmark  does  register  the  distinction,  albeit  briefly:  namely  at  p  260,  footnote.  
There  is  more  discussion  in  a  paper  that  preceded  his  book,  where  he  writes  (2008,  p.107):    
 
‘Whereas  the  customary  terminology  in  physics  textbooks  is  that  the  external  reality  is  described  
by  mathematics,  the  (MUH)  states  that  it  is  mathematics  (more  specfically  a  mathematical  
structure).  This  corresponds  to  the  ‘ontic’  version  of  universal  structural  realism  in  the  
philosophical  terminology  of  Ladyman  (1998)  and  McCabe  (2006)’.    
 
I  reply  as  follows,  with  three  points,  (a)  to  (c).    
(a):  I  agree  that  some  speculative  philosophical  positions  may  collapse  the  distinction,  saying  ‘is’  
where  the  customary  view  says  ‘instantiates’.  But:  
(b):  I  agree  that  McCabe’s  discussion  seems  to  endorse  collapsing  the  distinction.  But  on  my  
understanding  of  Ladyman’s  position—whose  paper  (1998)  named  and  advocated  ‘ontic  
structural  realism’-­‐-­‐-­‐he  does  not  wish  to  collapse  the  distinction.  More  important:    
(c):  Setting  aside  whatever  might  be  the  views  of  these  authors,  we  should  not  collapse  the  
distinction.    
 
My  reason  for  saying  (c)  is  that  to  collapse  the  distinction,  undoubtedly  the  best  strategy  is  to  say  
that  a  property  or  relation,  or  at  least  the  properties  and  relations  mentioned  by  physics-­‐-­‐-­‐such  
as  the  physical  quantities  that  I  used  in  Section  IV.C  to  ram  home  the  distinction  between  
instantiates  and  is-­‐-­‐-­‐is  completely  specified  by  its  web  or  pattern  of  relations  to  other  properties  
and  relations.  The  paradigm  such  “relation  to  other  properties  and  relations”  is  the  relation  of  
being  associated  together  in  a  law.  For  example,  the  quantities  force  and  acceleration  are  related  
by  Newton’s  second  law.    
 
And  so  the  strategy  in  question  maintains  that  everything  there  is  to  know  about  any  physical  
quantity,  e.g.  the  entire  nature  of  force,  can  be  recovered  from  the  web  or  pattern  of  relations  
among  quantities-­‐-­‐-­‐each  of  which  is  thus  specified  only  as  a  node  in  this  web.  In  mathematical  
terms,  this  requires  at  least  that  the  web  is  to  have  only  the  identity  map  as  an  automorphism.  
The  idea  is:  if  this  is  so,  then  the  apparent  ‘contentfulness’  (cf.  the  end  of  Section  IV.C)  of  physical  
quantities  can  be  ‘analysed  away’  as  a  matter  of  structure.  That  is:  we  seem  to  glimpse  how  the  
applied  mathematical  structures,  with  their  ‘contentful’  physical  quantities  could  turn  out  to  be,  
au  fond,  pure  mathematical-­‐-­‐-­‐‘contentless’,  ‘abstract’,  ‘structural’.      
 
But  I  say:  this  strategy  is  wrong  (and  so  the  distinction  between  instantiates  and  is  does  not  
collapse).  For  there  is  every  reason  to  think  that  at  least  some  physical  quantities  each  have  a  
nature  that  is  not  exhausted  by  their  web  or  pattern  of  relations  to  other  properties  and  
relations.    
 
As  I  admitted  in  (a),  I  agree  that  some  philosophical  positions  deny  this.  There  is  even  a  jargon.  
To  say,  as  I  do,  that  at  least  some  properties  (or  relations)  have  a  nature  that  is  not  exhausted  by  
their  web  of  relations  to  others,  is  to  be  a  quidditist.  (This  is  from  ‘quid’,  the  Latin  for  ‘what’:  for  
the  idea  is  that  a  property  has  a  ‘content’  or  ‘whatness’.)    So  in  effect,  I  am  saying  that  for  Uncle  
Max  to  defend  his  argument  that  (ERH)  implies  (MUH),  his  most  promising  strategy  is  to  deny  
quidditism.  But  my  objection  to  him  is:  quidditism  looks  right!  As  the  following  example  suggests.    
 
(Agreed:  my  example  of  electric  charge,  and  the  flip  between  positive  and  negative  values  of  
charge,  is  subtler  than  my  discussion  will  suggest.  (For  it  leads  in  to  the  question  of  the  proper  
treatment  of  time-­‐reversal:  for  which  cf.  e.g.  Malament  (2004).)  So  I  do  not  claim  that  the  case  of  
electric  charge  gives  conclusive  proof  of  quidditism.  But  I  do  claim  that  it  makes  quidditism  
plausible;  and  thus  that  it  shows  Tegmark  should  have  worked  harder  to  justify  anti-­‐quidditism-­‐-­‐
-­‐to  replace  instantiates  by  is.)  
 

  16  
 Take  the  two  properties,  having  negative  electric  charge,  and  having  positive  electric  charge.  Or  if  
you  prefer  some  precise  amount  of  each,  e.g.  having  plus  or  minus  one  Coulomb  of  charge.    
Imagine  them  (or  one  of  them)  occurring  in  an  applied  mathematical  structure,  i.e.  a  
physically  possible  state  of  affairs,  according  to  some  theory.  For  example,  let  the  theory  be  
electromagnetic  theory,  and  let  the  properties  (quantities  or  exact  values  of  them)  occur  in  a  
possible  instantaneous  state  for  a  collection  of  particles.  Thus:  here  is  a  particle  with  charge  +3  
Coulombs,  there  is  another  with  charge  -­‐2  Coulombs.  (Recall  the  sketch  of  applied  mathematical  
structures  in  (i)  t  o  (iv),  in  Section  IV.C.)    
Now  imagine  flipping  positive  and  negative  charge.  Here,  I  do  not  mean  changing  our  
linguistic  convention  about  what  amounts  of  charge  we  call  ‘negative’  and  which  ‘positive’.  I  
mean  an  active  transformation  which  changes  the  state  of  affairs:  in  the  new  state  of  affairs,  here  
is  a  particle  with  charge  -­‐3  Coulombs,  there  is  another  with  charge  +2  Coulombs.  (I  am  using  the  
same  language  as  before!)    
I  agree  that  this  new  state  of  affairs  of  affairs  will  behave,  e.g.  change  over  time,  in  a  
manner  that  mirrors  the  behaviour  of  the  original  state  of  affairs.  Here,  the  ‘mirroring’  can  be  
spelt  out  in  detail  using  the  idea  of  a  symmetry  of  the  laws.  I  discussed  this  in  Section  II.B.  In  
terms  of  the  current  idea  of  the  laws  as  a  web  of  relations  between  quantities,  a  symmetry  is  a  
transformation  of  the  web,  i.e.  a  function  sending  nodes  to  nodes  etc.,  that  preserves  the  web’s  
structure  e.g.  the  connectivity  relations  between  nodes.  But  I  do  not  need  to  go  into  details  about  
the  mirroring  of  the  original  state  of  affairs’  behaviour.    
For  the  relevant  point-­‐-­‐-­‐the  point  I  want  to  emphasize-­‐-­‐-­‐is  a  different,  and  a  simpler,  one.  
It  is  just  that  the  new  state  of  affairs  is  different  from  the  original.  
And  just  that  point  is  enough  to  imply  that  there  are  facts-­‐-­‐-­‐‘content’  or  ‘nature’-­‐-­‐-­‐about,  
say,  negative  charge,  or  having  an  exact  value  of  negative  charge  (e.g.    -­‐2  Coulombs)  that  outstrip  
its  web  of  relations  to  other  quantities:  that  cannot  be  ‘analysed  away’  in  terms  of  the  quantity’s  
position  as  a  node  in  the  web.  For  the  ‘flip’  being  a  symmetry  of  the  laws  implies  that  this  web  of  
relations  is  shared  with  positive  charge  (or  the  corresponding  exact  value:  in  my  example,  +2  
Coulombs):  but  the  states  of  affairs  are  different!  
 
 
Acknowledgments  
I  am  very  grateful  to  Marianne  Freiberger,  the  editor  at  Plus,  for  the  invitation  to  review  the  book,  and  for  
dividing  and  editing  the  bulk  of  this  discussion  into  three  separate  articles  appropriate  for  Plus.    They  can  be  
found  at  these  URLs:  
The review: http://plus.maths.org/content/mathematical-universe-0
Discussion of symmetry breaking: http://plus.maths.org/content/breaking-symmetry
Discussion of the world being mathematics: http://plus.maths.org/content/world-made-maths
I  am  also  very  grateful  to  Harvey  Brown,  George  Ellis,  Huw  Price,  Joe  Silk,  David  Sloan  and  David  Wallace  for  
comments  on  a  previous  version:  I  wish  I  could  have  incorporated  all  their  wisdom!  
 
 
References  
Barrow,  J.  and  Tipler,  F.  (1986),  The  Anthropic  Cosmological  Principle,  Oxford  University  Press    
Burnyeat,  M.  (2007),  ‘Other  Lives’  [a  review  of  two  books  about  Pythagoras],  The  London  Review  
of  Books,  22  February  2007,  pp.  3-­‐6.  
Gray,  J  (2008),  Plato’s  Ghost:  The  Modernist  Transformation  of  Mathematics.  Princeton  University  
Press.  
Kline,  M.  (1972),  Mathematical  Thought  from  Ancient  to  Modern  Times,  Oxford  University  Press    
Ladyman  J.  (1998),  ‘What  is  Structural  Realism?’,  Studies  in  History  and  Philosophy  of  Science,  
volume  29,  pp.  409-­‐424.  
Malament,  D.  (2004),  ‘On  the  time  reversal  invariance  of  classical  electromagnetic  theory’,  Studies  
in  the  History  and  Philosophy  of  Modern  Physics,  volume  35B,  pp.  295-­‐315.    
McCabe,  G  (2006),  ‘Possible  physical  universes’,  arxiv:  gr-­‐qc/0601073  (2006)  
McMullin,  E.  (1985),  ‘Galilean  idealization’,  Studies  in  the  History  and  Philosophy  of  Science  volume  
16,  pp.  247-­‐273.  
Tegmark,  M.  (2008),  ‘The  mathematical  universe’,  Foundations  of  Physics  volume  38,  pp.  101-­‐150.    
Wilczek,  F.  (2008),  `Enlightenment,  knowledge,  ignorance,  temptation’,  in  B.  Carr  ed.,  Universe  or  
Multiverse?,  Cambridge  University  Press;  pp.  43-­‐54.  
Williams,  B.  (1978),  Descartes:  the  project  or  pure  enquiry,  Penguin.    

  17  

View publication stats