1   The  Pace  of  Academic  Prototyping:  The  Case  of  the  DTOC   Dr.  Stan  Ruecker  /  Associate  Professor  /  IIT  Institute  of  Design   Nadine  Adelaar  /  Researcher  /  University  of  Alberta   Dr.  Susan  Brown  /  Professor  /  University  of  Guelph   Dr.  Teresa  Dobson  /  Associate  Professor  /  University  of  British  Columbia   Dr.  Ruth  Knechtel  /  Researcher  /  University  of  Alberta   Andrew  MacDonald  /  Researcher  /  McMaster  University   Ernesto  Peña  /  PhD  student  /  University  of  British  Columbia   Milena  Radzikowska  /  Associate  Professor  /  Mount  Royal  University   Geoff  G.  Roeder  /  MA  Student  /  University  of  British  Columbia   Dr.  Stéfan  Sinclair  /  Associate  Professor  /  McGill  University   Jennifer  Windsor  /  MA  Student  /  University  of  Alberta   and  the  INKE  Research  Group     Abstract   It  is  sometimes  said  that  the  academy  today  is  undergoing  rapid  change,  and  that  one  of  the  changes   we  have  been  experiencing  is  in  the  rate  of  dissemination  of  research  results.  Our  previous   generation  could  comfortably  spend  10  years  on  a  project,  and  produce  a  monograph  at  the  end,   confident  that  they  were  meeting  the  expectations  of  everyone  involved.  Contemporary  academics,   on  the  other  hand,  typically  publish  and  present  on  an  annual  cycle;  research  professors  spending  10   years  on  a  project  without  publishing  would  be  considered  to  be  underperforming,  if  not  outright   failing  in  their  duties.  However,  in  some  respects  the  rapid  cycle  of  dissemination  masks  the  actual   pace  of  academic  understanding:  people  still  need  time  to  work,  to  write,  to  reflect.  In  this  paper,  we   argue  that  researchers  who  routinely  design,  prototype,  and  test  experimental  systems,  far  from   being  the  exception  to  the  rule,  may  actually  require  more  time  to  build  and  make  their  cases.  As  an   example,  we  discuss  the  Dynamic  Table  of  Contexts,  first  conceived  of  in  2003  and  described  as  a   concept  at  a  DH  conference  in  2005.  Ten  years  later,  in  2013,  it  is  still  moving  forward  as  a   constellation  of  design,  programming,  and  testing,  but  more  importantly,  as  a  locus  of  ideas  about   what  it  means  to  remediate  a  venerable  print  tradition  in  such  a  way  as  to  make  it  acceptable  to  print   readers  while  at  the  same  time  increasing  its  affordances.     Introduction   It  is  commonly  recognized  that  academics  in  2013  tend  to  produce  more  publications  more  quickly   than  was  the  case  for  academics  in,  for  instance,  1953.  Part  of  the  reason  is  the  current  state  of  the   academy,  especially  in  the  arts  and  humanities,  where  competition  for  academic  jobs  has  resulted  in   earlier  and  more  aggressive  publication.  It  is  not  uncommon  to  see  people  applying  for  junior   positions  with  a  dozen  or  more  papers  in  good  journals,  and  in  some  cases  a  book  or  more.  The   funding  agencies  also  have  a  role  to  play,  with  an  emphasis  on  evidence  of  dissemination.  Finally,   there  continue  to  be  changes  in  publication  models,  which  can  provide  more  opportunities  for  a   wider  range  of  publication,  although  alas  at  this  point  we  still  have  no  way  to  recognize  experimental   prototypes  directly  through  a  process  of  peer  review.     What  we  typically  have  instead  is  a  situation  where  people  interested  in  working  with  experimental   prototypes  will  propose  some  relevant  research  questions,  often  involving  user  study,  then  write  and   publish  about  the  design  ideas  and  the  user  tests,  with  the  prototypes  arising  as  a  kind  of  side  effect   of  the  process.  There  have  occasionally  been  efforts  to  address  prototypes  more  directly,  but  what   has  tended  to  happen  is  that  a  brief  discussion  of  the  prototype  turns  out  to  be  insufficient  for  people   attempting  to  assess  their  value  as  instances  of  new  knowledge,  and  the  length  of  the  discussion   increases  over  time  until  we  are  back  to  writing  a  paper  for  evaluation  rather  than  having  the   prototype  directly  reviewed  in  some  way.     We  have  previously  suggested  that  the  review  of  prototypes  might  leverage  experience  in  reviewing   other  kinds  of  scholarly  output  (Galey  and  Ruecker  2010),  but  so  far,  despite  a  growing  interest,   nothing  of  note  has  emerged.  All  of  which  is  not  to  say  that  experimental  prototypes  are  not  of  value   in  the  pursuit  of  new  knowledge,  and  in  fact,  it  is  possible  to  claim  that  a  kind  of  intellectual     2   trajectory  can  be  recognized  in  looking  at  the  changes  made  to  prototypes  over  time.  Our  claim  here   is  that  this  trajectory  shows  reasonably  clearly  that  at  least  some  academic  prototypes  have  a   relatively  long  lifespan  as  prototypes,  with  each  successive  cycle  of  design  and  development   extending  our  understanding.  As  an  example  of  the  process,  we  will  look  at  the  conception  and   history  of  a  relatively  simple  prototype  system  called  Dynamic  Table  of  Contexts  (DToC).     Research  Question:  what  are  the  possible  new  affordances  of  digital  text?   The  concept  of  the  DToC  arose  as  one  answer  to  a  much  larger  question:  what  are  the  possible  new   affordances  of  digital  text?  In  general,  these  involve  research  areas  such  as  text  analysis  and   visualization.  In  more  specific  terms,  we  have  distant  reading,  algorithmic  criticism,  text   visualization,  and  interactive  visualization.  Our  research  projects  in  response  to  this  question  have   included:     • the  Bi  Sheng  electronic  book   • watching  the  script   • the  simulated  environment  for  theatre  (SET)   • No  One  Remembers  Acronyms  (NORA)   • Metadata  Offer  New  Knowledge  (MONK)   • Implementing  New  Knowledge  Environments  (INKE)   • the  paper  drill   • citelens   • the  multitouch  variorum   • the  text  as  a  string  of  words   • Just  in  Time  Research  (JiTR)   • repetition  loops   • dialR   • the  novel  as  slot  machine   • the  magic  circle   • the  Mandala  Browser   • texttiles   • bubblelines   • conversational  modeling   • the  reorganizable  textbook     This  list  is  significant  in  two  ways.  First,  it  shows  the  diversity  of  subtopics  within  a  single  research   question  that  can  be  addressed  through  an  approach  that  may  include  the  design,  prototyping,  and   testing  of  concepts.  Second,  for  anyone  familiar  with  the  list,  it  may  bring  to  mind  the  complex   interactions  that  occur  between  prototypes  and  projects.  To  take  one  example,  the  magic  circle  (Fig.   1)  is  a  visualization  that  shows  a  breakdown  of  various  contributions  to  some  sort  of  whole,  provided   by  a  number  of  designated  parts  of  that  whole.  It  was  originally  sketched  as  a  possible  component  of   the  MONK  project,  where  it  was  going  to  serve  as  a  way  of  showing  how  vocabulary  use  varied  across   different  works,  either  by  the  same  author  or  else  across  multiple  authors.  By  extension,  it  could  also   be  used  to  look  at  varying  use  of  lemmatized  vocabulary,  stemmed  vocabulary,  or  even  parts  of   speech.       3     Fig.  1:  The  original  magic  circle  design,  showing  search  results  for  the  word  “love”  across  multiple  books  by   multiple  authors.     However,  the  magic  circle  did  not  find  its  way  into  the  final  production  of  MONK,  and  was  shelved  for   a  couple  of  years.  It  next  emerged  in  a  collaborative  project  on  wiki  authorship,  where  the  team  was   looking  for  some  means  of  conveying  the  results  of  an  algorithm  that  could  give  people  credit  for   their  contributions  to  wiki  pages.  Somewhat  at  odds  with  the  typically  anonymous  approach  to  wiki   authorship,  the  magic  circle  was  an  innovation  intended  to  accommodate  writing  environments  like   industry  and  the  classroom,  where  anonymous  contribution  is  deprecated  in  favour  of  knowing  who   has  been  working  on  what  (Arazy  et  al.  2010).       Taking  it  one  step  further,  the  magic  circle  was  included  as  an  example  of  the  kinds  of  output  that   could  be  anticipated  from  the  interface  design  research  team  in  the  Implementing  New  Knowledge   Environments  (INKE)  project.  In  this  context,  our  intention  was  to  apply  it  more  broadly  to  questions   of  co-­‐authorship,  not  just  on  wikis,  but  also  across  a  number  of  media.     Finally,  extending  beyond  INKE,  the  magic  circle  was  adopted  for  production  by  the  Canadian  Writing   and  Research  Collaboratory  (CWRC),  where  it  is  intended  for  use  in  helping  to  manage  and  to  provide   appropriate  credit  for  collaborative  authorship  across  a  range  of  projects.  This  one  fairly  small  visual   idea  has  therefore  served  as  a  component  of  three  major  projects,  helping  researchers  to  consider   some  of  the  complexities  inherent  in  the  activities  they  are  facilitating.     A  Brief  Visual  History  of  the  Dynamic  Table  of  Contexts   In  the  case  of  the  DToC,  we  were  interested  in  the  design  of  the  electronic  book,  and  how  the  people   who  want  to  read  and  study  books  might  have  their  experience  improved  over  what  is  possible  with   a  print  book.  Our  contention  was  not  that  digital  books  would  replace  print  books,  since  we  had   previously  carried  out  a  study  that  suggested  that  they  would  not,  at  least  for  dedicated  readers  in   this  lifetime  (Ruecker  2002).  Instead,  we  wanted  to  identify  growth  points,  where  what  we  would   now  call  the  skeuomorphic  design,  and  what  we  then  spoke  of  as  the  remediation  of  the  print  book   into  digital  form,  could  be  extended  in  ways  that  would  be  beneficial  to  the  reader.       Our  initial  concept  and  conference  paper  (Ruecker  2005)  asked  how  the  print  table  of  contents  might   serve  as  a  rich-­‐prospect  browser.  We  identified  which  of  the  principles  of  rich-­‐prospect  browsing  it   met,  and  which  it  failed  to  meet,  then  posited  a  digital  system  that  would  meet  all  of  the  criteria.  The   purpose  of  meeting  these  criteria  was  that  the  TOC  could  then  become  a  more  robust  tool  for   researchers.  Our  example  used  the  dynamic  insertion  by  the  reader  into  the  TOC  of  characters,   dialog,  and  locations  in  The  History  of  Tom  Jones,  a  Foundling.  At  this  stage,  we  were  looking  at  some   early  sketches  (Fig.  2)  but  had  not  begun  to  think  about  the  details  of  programming  a  working   prototype.       4   TABLE OF CONTENTS DEDICATION BOOK I Chapter I no characters present 29 Chapter II Squire Allworthy Miss Bridget Allworthy, his sister 31 Chapter III Mr Allworthy The infant Tom Jones Mrs Deborah Wilkins 33 Chapter IV Mr Allworthy Miss Bridget Mrs Deborah Wilkins The infant Tom Jones 36 Chapter V Miss Bridget Mrs Deborah Wilkins The infant Tom Jones 39 Chapter VI Mrs Deborah An elderly maiden Jenny Jones Miss Bridget Mr Allworthy 40   Fig.  2:  Characters  in  the  first  few  chapters  of  Tom  Jones  •  where  in  the  book  do  they  appear?  •  what  are  they   called?  •  what  are  they  doing?  •  how  are  they  characterized?  •  what  is  their  ontological  status?  (c.f.  Willard   McCarty  on  the  Metamorphosis)     This  might  be  an  appropriate  place  to  mention  one  of  the  virtues  of  the  digital  humanities  as  a  field:   we  are  at  least  as  interested  in  the  ideas  behind  the  prototype  as  in  the  prototype  itself.  In  practical   terms,  this  means  that  we  can  discuss  design  concepts  before  we  build  them,  allowing  us  to  put   extensive  work  into  the  design  thinking  rather  than  needing  to  push  forward  quickly  into  a  software   version.  It  is  possible  in  DH  to  present  and  publish  on  ideas,  sketches,  interactive  sketches,   prototypes,  and  development  systems,  as  well  as  user  studies  at  practically  every  stage.     The  next  phase  of  the  DToC  took  place  in  association  with  the  Orlando  Project,  which  is  an  online   history  of  women  writers  in  the  British  Isles,  developed  at  the  University  of  Alberta  and  the   University  of  Guelph,  and  first  published  by  Cambridge  University  Press  in  2006.  Our  question  in  this   case  was  whether  an  interactive  TOC  could  be  embedded  somehow  in  the  existing  Orlando  interface   to  accommodate  more  conventional  period  histories  (Fig.  3).       Fig  3:  An  interactive  TOC  for  the  Orlando  Project,  showing  a  collapsed  first  volume  and  expanded  volume  2.     5       Since  there  were  three  proposed  books,  the  combined  tables  of  contents  in  the  Orlando  site  would  be   quite  complicated.  We  therefore  added  the  possibility  for  the  reader  to  collapse  or  expand  selected   sections  of  the  TOC.  It  also  became  clear  from  the  pattern  of  use  of  Orlando  that  it  would  be  useful  to   accommodate  not  only  XML-­‐encoded  items  but  also  free-­‐text  searches.       An  additional  complication  in  this  case  was  that  the  TOC  would  be  situated  in  the  current  interface.  It   would  therefore  become  another  option  in  the  panel  on  the  left  (shown  as  the  first  selection  under   “Histories”.  Other  options  related  to  the  proposed  volumes  would  include  direct  reading  access  to   any  of  the  three  books,  a  display  showing  just  their  XML  tags,  or  a  conventional  index.  Note  that  in   this  version  of  the  design,  the  tags  and  index  were  assumed  to  be  aggregated  for  all  three  volumes,   rather  than  broken  out  by  volume.     Placing  the  history  volumes  as  an  extra  item  in  the  navigation  panel  is  one  possible  strategy,  based  in   part  on  the  principle  that  readers  would  want  to  have  all  the  navigation  items  available  at  all  times.   However,  this  approach  was  also  potentially  confusing  for  readers,  in  that  the  existing  four  options  in   the  navigation  panel  (i.e.  People,  Chronologies,  Tag  Search,  and  Links)  provided  access  to  the   biographical  material  in  the  project,  rather  than  to  the  more  conventional  historical  volumes.       At  this  stage,  we  had  done  some  preliminary  thinking  about  how  the  mechanism  of  adding  and   subtracting  material  to  the  TOC  would  actually  look  and  work.  The  reader  would  be  provided  with  a   list  of  possible  items  to  include,  and  could  toggle  them  off  and  on.  We  recognized  that  we  could  use   the  XML  encoding,  and  leverage  the  human-­‐readable  versions  of  the  tags  that  were  already  in  place   for  the  Tag  Search  option.  We  tried  to  condense  the  list  as  much  as  possible  by  using  bullets  as   separators,  since  there  could  conceivably  be  on  the  order  of  200  tags  to  consider.  Since  the  contents   would  already  be  indexed,  we  were  also  able  to  show  a  count  next  to  each  tag,  indicating  how  many   instances  would  be  inserted  when  the  reader  toggled  that  tag.         6   These  designs  combined  existing  Orlando  material  with  the  prospective  material  about  the  proposed   volumes.  In  practice,  the  tagset  for  the  volumes  would  most  likely  differ  in  some  respects  from  the   tagset  previously  used  for  the  biographies.  However,  since  the  sketches  were  based  on  the  existing   Orlando  tagset,  and  at  this  point  that  tagset  had  not  been  publicly  released,  we  removed  it  from  the   published  version  of  the  design  (Ruecker  et  al.  2007).     Working  with  Orlando  therefore  provided  us  with  a  variety  of  ideas  about  ways  in  which  a  relatively   simple  original  concept  could  be  complicated  by  embedding  it  within  a  specific  pre-­‐existing  interface.   This  led  us  in  the  next  iteration  to  ask  how  the  system  could  be  designed  to  act  in  a  more  general  way   as  a  standalone  interface  for  any  digital  text  contents  (Fig.  3).     Fig  3:  Design  sketches  by  Milena  Radzikowska  for  a  generic  DToC.  The  process  goes  from  choices  made  on  either   of  the  two  panels  to  the  left  and  results  in  a  display  on  the  right,  which  would  then  be  used  to  access  the  text   itself.     In  the  generic  design,  the  reader  has  the  opportunity  to  choose  from  two  different  interfaces  in  order   to  select  the  items  to  be  inserted  in  the  TOC.  The  list  view  appears  on  the  left;  in  the  center  is  a  tag   cloud.  Clicking  on  an  item  in  either  would  insert  the  instances  of  that  tag  into  the  TOC  on  the  right.   Clicking  on  an  item  from  the  TOC  would  take  the  reader  into  the  text.     A  significant  addition  to  the  concept  in  this  iteration  is  the  inclusion  on  the  left  of  more  than  just  a  list   of  tags.  There  is  a  Scope  panel  that  allows  the  reader  to  choose  which  chapter  to  display.  There  is   also,  associated  with  each  tag,  a  list  of  its  attributes,  and  associated  with  each  attribute,  the  values  it   contains.  With  this  mechanism,  the  reader  is  provided  with  a  finer  grain  of  selections  for  which  items   to  insert  in  the  TOC.  Since  it  also  quite  complicated,  choices  made  in  this  panel  can  also  be  used  to   generate  a  much  simpler  tag  cloud.     We  subsequently  had  the  opportunity  to  have  the  generic  design  built  into  a  prototype  by  a  group  of   senior  undergraduate  computing  science  students1,  who  chose  this  project  for  a  team  programming   class  (Fig.  4).  Their  primary  interest  was  in  figuring  out  how  to  provide  the  necessary  functionality,   so  attention  to  the  design  was  left  for  another  phase.  This  prototype  was  built  using  Google  Web   Toolkit  (GWT).                                                                                                                     1  The  students  were  Mark  Bieber,  Jamie  Czerwinski,  Xuefeng  Ding,  Matt  Gooding,  and  Mike  Packer.   The  instructor  was  Dr.  Ken  Wong.         7     Fig  4:  First  prototype  programmed  in  GWT  by  senior  students  in  computing  science  at  the  University  of  Alberta.     The  prototype  was  useful  in  examining  how  the  TOC  interacted  with  the  XML-­‐encoding.  At  this  point,   it  became  clear  that  we  would  need  a  curator  mode,  where  someone  would  manage  the  collection,   choose  which  tags  should  be  displayed  to  the  reader,  and  rename  those  tags  to  make  them   understandable  by  people  who  were  not  involved  in  the  encoding  project.       Our  XML-­‐encoded  content  for  this  prototype  was  Frances  Burney’s  novel  Cecilia,  provided  by  the   Brown  Women  Writer’s  Project.  In  conversations  with  Syd  Bauman,  it  became  clear  that  a  useful   feature  to  provide  for  the  curator  would  be  the  ability  to  apply  the  same  choices  of  tags  and  human-­‐ readable  tagnames  to  any  documents  that  shared  a  tagset.     Since  Cecilia  has  more  chapters  than  could  appear  in  a  list  on  a  conventional  screen  at  the  time,   inserting  material  of  any  length  quickly  reduced  the  number  of  chapters  that  were  visible.  We   therefore  inserted  only  a  small  amount  of  text,  but  added  a  popup  that  would  appear  on  mouseover.   The  popup  showed  a  reasonably  large  section  of  the  text  containing  the  tagged  material.       To  address  the  same  issue  of  screen  real  estate,  readers  also  had  the  option  of  removing  chapters   from  the  display;  they  could  be  added  back  with  a  “restore”  switch  at  the  top  of  the  panel.     We  also  became  aware  at  this  stage  that  people  may  want  to  insert  more  than  one  tag  at  a  time.  A   consequence  of  providing  this  affordance  is  that  they  would  need  to  be  able  to  keep  track  of  which   tag  selected  from  the  panel  on  the  right  had  resulted  in  the  contents  being  inserted  on  the  left.  Our   solution  was  to  preface  the  inserted  text  with  the  name  of  the  tag.     Although  we  now  had  a  working  prototype,  after  the  class  was  over  we  had  no  ready  means  of   modifying  the  interface.  In  talking  further  with  one  of  the  student  programmers  (Mark  Bieber),  we   also  became  convinced  that  rewriting  it  from  scratch  would  provide  significant  advantages.       Fig.  5:  Second  prototype  programmed  in  Flash  by  Mark  Bieber;  design  by  Milena  Radzikowska.   8       Our  opportunity  to  attempt  this  next  iteration  arose  in  connection  with  the  INKE  project,  where  our   mandate  was  to  look  at  improved  interfaces  for  people  working  with  electronic  text.  As  opposed  to   the  previous  goal  of  creating  a  standalone  reader  with  improved  navigation,  in  this  phase  we  were   beginning  to  consider  how  the  reader  might  interact  with  other  features  of  an  entire  “new  knowledge   environment”.       Working  again  with  Cecilia,  we  produced  a  second  prototype  that  incorporated  facing  pages  in  the   reading  panel,  as  well  as  additional  text  analysis  and  visualization  tools.  In  particular,  we  added  a   search  function,  as  well  as  a  Bubblelines  visualization  for  comparing  either  tagging  or  search  results   across  multiple  chapters.  It  was  also  possible  to  swap  out  the  reading  panel  for  a  search  term   frequency  graph.     This  was  the  first  prototype  where  we  had  the  opportunity  to  run  a  user  study  (Dobson  et  al.  2012).   Among  other  things,  we  learned  that  it  would  be  useful  to  carry  out  user  studies  earlier  in  the   process,  preferably  using  sketches,  so  that  we  could  more  quickly  narrow  our  design  target  before   programming.  For  instance,  our  assumption  that  facing  pages  would  be  a  skeuomorphic  feature  of   importance  to  readers,  the  user  study  suggested  that  it  was  more  important  to  accommodate  people   with  smaller  screens.  Another  important  finding  was  that  people  were  familiar  with  the  idea  that  any   encoding  is  a  kind  of  interpretation,  and  they  were  interested  in  understanding  more  about  the   rationale  behind  the  available  tags  they  were  suddenly  using  to  navigate  the  text.     Simultaneously  with  the  user  study,  we  began  to  involve  researchers  who  had  been  working  in  the   history  of  book  design,  thinking  about  ways  in  which  print  experiments  might  be  brought  back  into   play  within  a  digital  environment  (e.g.  Nelson  et  al.  2011).  The  second  prototype  was  also  polished   enough  that  it  could  be  used  in  conversation  with  potential  INKE  project  partners  to  suggest  the   usefulness  and  practicality  of  the  approach.       Building  on  the  insights  from  Nelson  et  al.  and  the  findings  of  the  study  by  Dobson  et  al,  we  therefore   began  a  third  prototype  (Fig.  5),  in  conjunction  this  time  with  two  partners:  the  Canadian  Writing   Research  Collaboratory  (CWRC)  and  the  University  of  Alberta  Press.  This  third  prototype  was  also     9   built  using  the  Voyant  tools  platform,  meaning  that  it  could  potentially  be  robust  enough  to  work  as  a   development  version  for  release  to  other  researchers  and  projects.  Close  integration  with  Voyant   also  meant  that  we  could  conceivably  find  ways  to  leverage  the  many  existing  text  analysis  and   visualization  tools  to  work  in  conjunction  with  the  DTOC.     Fig  6:  Third  prototype,  programmed  by  Andrew  MacDonald  for  CWRC  in  Voyant  using  HTML5;  design  by   Jennifer  Windsor.       This  version  of  the  system  adds  several  significant  features.  First  is  the  index  panel,  which  parallels   the  tag  panel  and  serves  a  similar  function  with  dissimilar  contents  (Brown  et  al.  2013).  The   traditional  index  is  intended  to  help  the  reader  locate  specific  information  that  in  many  cases  will   only  appear  once  or  twice  within  a  book.  Semantic  encoding,  on  the  other  hand,  has  tended  to  be   applied  to  larger  concepts  that  occur  frequently  enough  to  make  including  them  in  the  tagset  seem   worthwhile.  It  would  be  normal,  for  example,  to  expect  to  see  semantic  tags  for  material  such  as   people’s  names,  significant  dates,  and  geographical  locations.  In  an  index,  however,  what  would   typically  appear  are  the  actual  names,  dates,  and  places.  There  is  no  easy  mechanism  for  finding  in  an   index  all  of  the  names,  dates,  or  places  at  once;  using  an  encoded  file,  however,  will  allow  the  reader   to  find  all  of  them  easily  by  searching  for  the  presence  of  the  appropriate  tag.     The  second  addition  to  this  prototype  is  the  document  model,  which  appears  as  a  thin  column  of  lines   between  the  navigation  panels  and  the  reading  panel.  This  model  provides  a  prospect  view  on  the   entire  document  (Ruecker  et  al.  2005b),  allowing  the  reader  to  see  at  a  glance  where  the  currently   selected  tags,  index  items,  or  search  terms  occur,  since  they  appear  as  coloured  lines  at   approximately  the  right  position.  The  lines  are  also  interactive,  so  they  can  be  used  to  jump  to  the   next  instance  in  the  text.     The  document  we  were  using  for  the  third  prototype  is  a  frankentext  that  combines  book  chapters   out  of  a  recent  edited  collection  with  other  book  chapters  that  have  not  yet  appeared  in  print,  but   have  already  been  subjected  to  encoding  and  indexing.  One  advantage  of  this  content  is  that  it  also   contained  footnotes,  reminding  us  to  accommodate  them  in  the  reading  panel  with  popup  balloons.     10   The  balloons  in  turn  required  some  careful  thinking,  since  we  wanted  the  contents  to  be  available  to   cut  and  paste,  while  at  the  same  time  we  wanted  to  avoid  having  them  clutter  up  the  reading  panel.   Our  solution  was  to  make  them  appear  when  the  reader  rolls  the  cursor  over  the  footnote  number  in   the  text,  but  disappear  when  the  cursor  leaves  the  number—unless  the  user  clicks  on  the  number,  at   which  point  the  footnote  bubble  becomes  persistent  until  closed  by  clicking  on  a  standard  X  in  the   top  right  corner.     Fig.  7:  Voyant  DTOC  with  panes  collapsed  for  reading  on  an  iPad.       Given  the  current  prevalence  of  smaller  reading  devices  such  as  the  iPad  and  Kindle,  the  question   also  arose  as  to  how  difficult  it  might  be  to  adjust  the  Voyant  DTOC  so  that  it  can  serve  with  the   smaller  screen  real  estate.  As  it  turns  out,  since  the  panes  are  collapsible,  it  is  possible  to  set  the   navigation  panels  to  the  side  while  reading  (Fig.  7),  then  expand  them  when  necessary.     Conclusions   The  DTOC  is  one  example  of  many  of  the  perhaps  surprisingly  slow  pace  at  which  an  academic   prototype  can  proceed.  It  has  been  an  intermittent  focus  of  attention  for  a  dozen  or  more  researchers   over  a  ten-­‐year  period.  It  is  of  course  possible  to  interrupt  a  research  trajectory  involving  a  series  of   prototypes.  In  fact,  given  the  exigencies  of  academic  research  funding,  it  is  sometimes  the  case  that  a   single  prototype  is  all  that  is  possible.       However,  provided  that  a  trajectory  can  be  followed,  it  can  be  argued  that  pursuing  it  over  an   extended  period,  perhaps  as  in  this  case  with  a  changing  constellation  of  researchers,  can  result  in  an   increasing  degree  of  understanding  of  a  variety  of  topics  that  are  brought  into  focus  by  thinking   through  prototyping.  In  addition  to  the  logistics  of  assembling  the  necessary  people  and  resources,   what  the  pace  allows  is  time  for  reflection.       11   Perhaps  taking  a  few  liberties  with  the  details  of  the  timeline,  we  have  essentially  addressed  the   following  sequence  of  questions:     2005:   how  does  a  table  of  contents  need  to  be  augmented  to  serve  as  a  rich-­‐prospect  browser?   2006:   what  issues  may  arise  in  attempting  to  embed  a  table  of  contexts  in  an  existing  project   interface?   2007:   how  can  an  electronic  reading  environment  that  includes  a  dynamic  table  of  contexts  be   designed  to  act  in  a  more  general  way  as  a  standalone  interface  for  any  XML-­‐encoded  texts?   2008:   what  other  text  analysis  and  visualization  tools  might  be  useful  to  associate  with  the  DTOC?   2009:   what  can  we  learn  from  a  user  study  of  a  standalone  DTOC  prototype?   2010:   what  aspects  of  particularly  interesting  print  TOCs  and  Indices  can  inform  the  DTOC?   2011:   how  can  the  DTOC  be  modified  to  accommodate  scholarly  editions  as  opposed  to  normal   monographs  or  collections?   2012:   using  the  DTOC  as  a  case  study,  what  can  we  learn  by  comparing  semantic  XML  encoding   with  a  conventional  book  index?   2013:   what  changes  would  need  to  be  made  to  the  dynamic  table  of  contexts  in  order  for  it  to   potentially  work  on  a  smaller  reading  device  such  as  an  iPad  or  Kindle?     The  immediate  future  of  the  DTOC  will  see  further  developments  of  it  as  a  reading  environment  that   supports  other  forms  of  visualization  and  text  analysis.  We  would  like  it  to  easily  access  texts  from  a   content  management  system.  We  are  wondering  if  it  would  be  useful  for  the  reader  to  be  able  to   dynamically  reorganize  the  TOC  panel,  and  whether  editors  would  want  to  provide  more  than  one   form  of  TOC  if  the  system  were  to  make  it  possible.  Then  if  multiple  TOCs  are  available,  should  more   than  one  be  visible  at  a  time?  We  have  not  even  begun  to  consider  what  it  might  mean  to   accommodate  other  media,  such  as  images  or  videos,  but  dealing  with  them  seems  like  a  logical  next   step.  No  doubt  other  possibilities  will  present  themselves,  as  we  ponder,  build,  and  test  our  way  into   the  future.     References   Arazy,  Ofer,  Eleni  Stroulia,  Stan  Ruecker,  Cristina  Arias,  Carlos  Fiorentino,  Veselin  Ganev,  and   Terence  Yau.  “Recognizing  Contributions  in  Wikis:  Authorship  Categories,  Algorithms,  and   Visualizations.”  Journal  of  the  American  Society  for  Information  Science  and  Technology   (JASIST).  61(6),  1166-­‐1179.  2010.   Brown,  Susan,  Brent  Nelson,  Stan  Ruecker,  Stéfan  Sinclair,  Nadine  Adelaar,  Ruth  Knechtel,  Jennifer   Windsor  and  the  INKE  Research  Group.  “Text  Encoding,  the  Index,  and  the  Dynamic  Table  of   Contexts.”  Paper  presented  at  the  annual  Digital  Humanities  conference  (DH2013),  Lincoln,   Nebraska.  July  16-­‐19,  2013.   Dobson,  Teresa,  Brooke  Heller,  Stan  Ruecker,  Milena  Radzikowska,  Mark  Bieber,  Susan  Brown,  and   the  INKE  Research  Group.  “The  Dynamic  Table  of  Contexts:  User  Experience  and  Future   Directions.”  Paper  presented  in  the  panel  “Designing  Interactive  Reading  Environments  for   the  Online  Scholarly  Edition”  at  the  DH2012  conference,  Hamburg,  Germany.  July  16-­‐20,   2012.   Galey,  Alan,  Stan  Ruecker,  and  the  INKE  Research  Group.  “How  a  Prototype  Argues.”  Literary  and   Linguistic  Computing.  25(3).  2010.   Nelson,  Brent,  Stan  Ruecker,  Milena  Radzikowska,  Stéfan  Sinclair,  Susan  Brown,  Mark  Bieber,  and  the   INKE  Research  Group.  “A  Short  History  and  Demonstration  of  the  Dynamic  Table  of   Contexts.”  Paper  presented  at  INKE:  Research  Foundations  for  Understanding  Books  and   Reading  in  a  Digital  Age:  Text  and  Beyond.  November  18,  2011.  Ritsumeikan  University,   Kyoto,  Japan.   Ruecker,  Stan.  “Carrying  the  pleasure  of  books  into  the  design  of  the  electronic  book.”  In  Pleasure   with  Products:  Beyond  Usability.  William  S.  Green  and  Patrick  W.  Jordan,  Eds.  London:   Taylor  and  Francis,  2002.  pp.  135-­‐146.   Ruecker,  Stan.  “The  Electronic  Book  Table  of  Contents  as  a  Research  Tool.”  Paper  presented  at  the   President’s  Panel.  Congress  of  the  Humanities  and  Social  Sciences:  Consortium  for     12   Computers  in  the  Humanities  /  Consortium  pour  Ordinateurs  en  Sciences  Humaines   (COCH/COSH)  Annual  Conference.  London,  Ontario.  May  27-­‐  30,  2005a.   Ruecker,  Stan,  Eric  Homich  and  Stéfan  Sinclair.  “Multi-­‐level  Document  Visualization.”  Visible   Language.  39(1),  33-­‐41.  Jan/Feb  2005b.   Ruecker,  Stan  and  the  INKE  Research  Group.  “Introducing  the  Dynamic  Table  of  Contexts  for   Scholarly  Editions.”  Paper  presented  at  the  Modern  Language  Association  (MLA)  Conference.   Los  Angeles,  CA.  Jan  6-­‐8,  2011.   Ruecker,  Stan,  Milena  Radzikowska,  Susan  Brown,  Thomas  M.  Nelson,  Isobel  Grundy,  Patricia   Clements,  Sharon  Balasz,  Jeff  Antoniuk,  and  Stéfan  Sinclair.  “The  Dynamic  Table  of  Contents:   Extending  a  Venerable  List  in  a  Digital  Context.”  The  Potential  and  Limitations  of  a  List:  an   International  Transdisciplinary  Workshop.  At  the  Center  for  Theoretical  Study,  the  Charles   University  and  the  Philosophical  Institute  of  the  Academy  of  the  Sciences  of  the  Czech   Republic,  Prague.  Nov  8-­‐9,  2007.   Ruecker,  Stan,  Susan  Brown,  Milena  Radzikowska,  Stéfan  Sinclair,  Thomas  M.  Nelson,  Patricia   Clements,  Isobel  Grundy,  Sharon  Balasz,  and  Jeff  Antoniuk.  “The  Table  of  Contexts:  A   Dynamic  Browsing  Tool  for  Digitally  Encoded  Texts.”  In  The  Charm  of  a  List:  From  the   Sumerians  to  Computerised  Data  Processing.  Ed.  Lucie  Dolezalova.  Cambridge:  Cambridge   Scholars  Publishing,  2009.  pp.  177-­‐187.