Google Poses a Serious Threat Constitutional Representative Republic Governance Rooted in free speech and press

Google Poses a  Serious  Threat  Constitutional Representative Republic Governance Rooted in free speech and press

 AIBRT American Institute for Behavioral  Research  and  Technology Why Google Poses a  Serious  Threat  to Democracy,   and How  to  End That  Threat Testimony  by Robert Epstein, Ph.D.  (re@aibrt.org)   Senior Research Psychologist, American  Institute  for  Behavioral Research  and Technology Before  the United States  Senate Judiciary  Subcommittee  on the Constitution Tuesday, June  16, 2019, 2:30 p.m. I  am  Dr.  Robert  Epstein,  the  proud  father  of  five  children,  a  resident  of  California,  and Senior  Research  Psychologist  at  the  American  Institute  for  Behavioral  Research  and Technology. I  love  America  and  democracy,  and  I  am  also  not  a  conservative. I  have  been center/center-left  my  whole  adult  life.  You’ll  see  in  moment  why  this  fact  is  relevant  to  my testimony. I  am  here  today  for  three  reasons:  to  explain  why  Google  presents  a  serious  threat  to democracy  and  human  autonomy,  to  explain  how  passive  monitoring  systems  can  protect us  both  now  and  in  the  future  from  companies  like  Google,  and  to  tell  you  how  Congress can  immediately  end  Google’s  worldwide  monopoly  on  search.  My  plan  for  ending  that monopoly  was  published  just  yesterday  (Monday,  July  15,  2019)  by  Bloomberg Businessweek  (Epstein,  2019d).  I  am  attaching  a  copy  of  my  article  to  my  testimony  and respectfully  request that it be entered into the Congressional Record. I  have  been  a  research  psychologist  for  nearly  40  years  and  have  also  served  in  various editorial  positions  at  Psychology  Today  magazine  and  Scientific  American  MIND.  I received  my  Ph.D.  at  Harvard  University  in  1981  and  have  since  published  15  books  and more  than  300  scientific  and  mainstream  articles  on  artificial  intelligence  and  other  topics. Since  2012,  some  of  my  research  and  writings  have  focused  on  Google  LLC,  specifically on  the  company’s  power  to  suppress  content  –  the  censorship  problem,  if  you  will  –  as  well as  on  the  massive  surveillance  the  company  conducts,  and  also  on  the  company’s unprecedented  ability  to  manipulate  the  thoughts  and  behavior  of  more  than  2.5  billion people worldwide. Data  I’ve  collected  since  2016  show  that  Google  displays  content  to  the  American  public that  is  biased  in  favor  on  one  political  party  (Epstein  &  Williams,  2019)  –  a  party  I  happen to  like,  but that’s irrelevant.  No private  company  should have  either the  right or the  power to  manipulate  large  populations  without  their  knowledge.   I’ve  published  articles  about  my  research  on  Google  in  both  scientific  publications  and  a wide  array  of  mainstream  news  sources:  in  TIME  magazine,  U.S.  News  &  World  Report, USA  Today,  Dissent,  The  Hill,  and  Huffington  Post,  for  example,  but  also  in  The  Daily Caller  and even  in  Russia’s  Sputnik  News. 1 I  reach  out  to  diverse  different  audiences  because  the  threats  posed  by  Google,  and,  to  a lesser  extent,  Facebook,  are  so  serious  that  I  think  everyone  in  the  world  needs  to  know about  them.  I  put  my  own  political  leanings  aside  when  I  report  my  data  and  concerns because  the  problems  these  companies  present  eclipse  personal  politics.  To  put  this  another way, I  love  humanity,  my  country, and  democracy  more  than I  love  any  particular  party  or candidate.  And  democracy  as  originally  conceived  cannot  survive  Big  Tech  as  currently empowered. If  you  were  to  examine  the  data  I  have  been  collecting  over  the  past  6-and-a-half  years, every  one  of you  would  put  partisanship  aside  and  collaborate  to  reign  in  the  extraordinary power that Google  and Facebook now  wield with unabashed arrogance. Here  are  five  disturbing  findings  from  my  research,  which  adheres,  I  believe,  to  the  highest possible scientific  standards in  all  respects: 1.  In  2016,  biased  search  results  generated  by  Google’s  search  algorithm  likely impacted  undecided  voters  in  a  way  that  gave  at  least  2.6  million  votes  to  Hillary Clinton  (whom  I  supported).  I  know  this  because  I  preserved  more  than  13,000 election-related  searches  conducted  by  a  diverse  group  of  Americans  on  Google, Bing,  and  Yahoo  in  the  weeks  leading  up  to  the  election,  and  Google  search  results –  which  dominate  search  in  the  U.S.  and  worldwide  –  were  significantly  biased  in favor  of  Secretary  Clinton  in  all  10  positions  on  the  first  page  of  search  results  in both blue  states and red states.   I  know  the  number  of  votes  that  shifted  because  I  have  conducted  dozens  of controlled  experiments  in  the  U.S.  and  other  countries  that  measure  precisely  how opinions  and  votes  shift  when  search  results  favor  one  candidate,  cause,  or company.  I  call  this  shift  “SEME”  –  the  Search  Engine  Manipulation  Effect.  My first  scientific  paper  on  SEME  was  published  in  the  Proceedings  of  the  National Academy  of Sciences  (PNAS) in 2015  (https://is.gd/p0li8V)  (Epstein &  Robertson, 2015a)  and  has  since  been  accessed  or  downloaded  from  PNAS’s  website  more than  200,000  times.  SEME  has  also  been  replicated  by  a  research  team  at  one  of the Max  Planck  Institutes in Germany.   SEME  is  one  of  the  most  powerful  forms  of  influence  ever  discovered  in  the behavioral  sciences,  and  it  is  especially  dangerous  because  it  is  invisible  to  people –  “subliminal,”  in  effect.  It  leaves  people  thinking  they  have  made  up  their  own minds,  which  is  very  much  an  illusion.  It  also  leaves  no  paper  trail  for  authorities to trace.  Worse  still,  the  very  few people who can  detect bias in search results shift even  farther  in  the  direction  of  the  bias,  so  merely  being  able  to  see  the  bias  doesn’t protect  you  from  it.   Bottom  line:  biased  search  results  can  easily  produce  shifts  in the  opinions  and  voting  preference  of  undecided  voters  by  20  percent  or  more  –  up to 80 percent in some  demographic  groups.   2 Bear  in  mind  here  that  all  Google  search  results  are,  in  a  sense,  biased.  There  are no  equal-time  rules  built  into  Google  algorithm.  It  always  puts  one  widget  ahead  of another  –  and one  candidate  ahead of another. SEME is an example  of  an  “ephemeral experience,”  and that’s  a  phrase  you’ll  find in  internal  emails  that  have  leaked  from  Google  recently.  A  growing  body  of evidence  suggests  that  Google  employees  deliberately  engineer  ephemeral experiences  to  change  people’s  thinking.    (For  details  about  the  methodology  used in SEME experiments, please  see  the Appendix  at the  end of  this  testimony.) Since  2013,  I  have  discovered  about  a  dozen  subliminal  effects  like  SEME,  and  I am  currently  studying  and  quantifying  seven  of  them  (https://is.gd/DbIhZw) (Epstein, 2018i). 2.  On  Election  Day  in  2018,  the  “Go  Vote”  reminder  Google  displayed  on  its  home page  gave  one  political  party  between  800,000  and  4.6  million  more  votes  than  it gave  the  other  party.  Those  numbers  might  seem  impossible,  but  I  published  my analysis  in  January  2019  (https://is.gd/WCdslm)  (Epstein,  2019a),  and  it  is  quite conservative.  Google’s  data  analysts  presumably  performed  the  same  calculations I  did  before  the  company  decided  to  post  its  prompt.  In  other  words,  Google’s  “Go Vote”  prompt was  not  a  public  service; it was a  vote manipulation. 3.  In  the  weeks  leading  up  to  the  2018  election,  bias  in  Google’s  search  results  may have  shifted  upwards  of  78.2  million  votes  to  the  candidates  of  one  political  party (spread  across  hundreds  of  local  and  regional  races).  This  number  is  based  on  data captured  by  my  2018  monitoring  system,  which  preserved  more  than  47,000 election-related  searches  on  Google,  Bing,  and  Yahoo,  along  with  the  nearly 400,000  web  pages  to  which  the  search  results  linked.  Strong  political  bias  toward one  party  was  evident,  once  again,  in  Google  searches  (Epstein  &  Williams,  2019). 4.  My recent  research  demonstrates  that  Google’s  “autocomplete”  search  suggestions can  turn  a  50/50  split  among  undecided  voters  into  a  90/10  split  without  people's awareness  (http://bit.ly/2EcYnYI)  (Epstein,  Mohr,  &  Martinez,  2018).  A  growing body  of  evidence  suggests  that  Google  is  manipulating  people’s  thinking  and behavior from the very  first character  people type  into the search box. 5.  Google  has  likely  been  determining  the  outcomes  of  upwards  of  25  percent  of  the national  elections  worldwide  since  at  least  2015.  This  is  because  many  races  are very  close  and  because  Google’s  persuasive  technologies  are  very  powerful (Epstein &  Robertson, 2015a).   These  effects  are  nothing  like  Russian-placed  ads  or  fake  news  stories.  Russian interference,  although  troubling  and  unacceptable,  does  not,  in  my  opinion,  shift  many votes  (Epstein,  2017d,  2018a).  Ads  and  news  stories  are  competitive  and  visible,  like billboards.  The  kinds  of  ephemeral effects  I  am  studying, however,  are  invisible  and  non3 competitive.  They  are  controlled  entirely  by  Big  Tech  companies,  and  there  is  no  way  to counteract them. I  have  also  studied  and  written  about  Google’s  massive  surveillance  operations  –  most  of which  people  are  completely  unaware  of  –  and  Google’s  pervasive  and  unpredictable pattern  of  censorship,  but  time  does  not  not  permit  me  to  discuss  my  work  in  these  areas today.  On  the  issue  of  censorship,  I  refer  the  Committee  Members  to  a  2016  report  I published  in  U.S.  News  &  World  Report  called  “The  New  Censorship” (http://bit.ly/28PgBmW)  (Epstein,  2016d),  which  described  nine  different  blacklists Google  maintains  to  suppress  information  worldwide.    We  are  all  aware  that  Google  deletes or  blocks  access  to  videos  on  YouTube,  which  it  owns,  but  few  people  are  aware  that Google  blocks  access  to  millions  of  websites.  On  January  31,  2009,  Google  blocked  access to virtually  the entire  internet for  40 minutes. By the  way,  it  is  not  just  conservative  content  that  gets  censored  (Epstein,  2018h).  At  times, Google  also  censors  progressive  and  socialist  content.  The  problem  with  Google  is  not  that it  censors  conservatives;  the  problem  is  that  it  has  the  power  to  determine  what  content billions  of  people  worldwide  will  or  not  see.  No  single  entity  –  especially  a  private company  that is  not accountable  to the  public  –  should  have  such power  (Epstein, 2016d). I  know  how  to  stop  Big  Tech  companies  dead  in  their  tracks,  and  that  brings  me,  finally, to  monitoring  systems and the proposal  I  published  yesterday.   Back  in  2015,  a  telephone  call  from  Jim  Hood,  the  attorney  general  of  Mississippi, prompted  me  to  start  a  years-long  project  in  which  I  learned  to  monitor  what  Big  Tech companies  are  showing  real  users.  In  early  2016,  I  launched  my  first  large-scale,  Neilsentype  monitoring  system  that  allowed  my  team  to  look  over  people’s  shoulders  and,  with user  permission,  to  capture  the  search  results  they  were  seeing  on  their  computer  screens before  those  results  disappeared  (Epstein,  2018d).    I  successfully  deployed  such  systems in  2016  and  2018,  and  I’m  raising  funds  now  to  build  a  much  larger  and  more comprehensive  system in early  2020  –  one  that will allow us  to  catch  Big  Tech companies in  the  act  –  to  instantly  spot  when  Google  is  showing  people  politically  biased  search results;  when  Twitter  is  suppressing  tweets  sent  by  the  President,  Ann  Coulter,  or  Elizabeth Warren;  when  Facebook  is  sending  out  “Register  to  Vote”  reminders  only  to  members  of one  party. This  system  must  be  built  to  keep  an  eye  on  Big  Tech  in  2020  because  if  these  companies all  support  the  same  candidate  –  and  that’s  likely,  needless  to  say  –  they  will  be  able  to shift  upwards  of  15  million  votes  to  that  candidate  with  no  one  knowing  and  without leaving  a  paper trail. To  let  Big  Tech companies  get away  with  invisible  manipulation  on  this  scale  would  be  to abandon  the  free-and-fair  election,  a  cornerstone  of  democracy.    It  would  make  democracy meaningless, even if  your chosen candidate  prevailed. A worldwide  network  of  passive  monitoring  systems  must  be  built  to  protect  humanity  and democracy from  manipulations  by  today’s  Google  and  the  Googles  of  tomorrow.  Only  tech can fight tech; laws  and regulations will never keep up  (Epstein, 2018d). Finally,  yesterday  I  published  an  article  explaining  how  Congress  can  quickly  end  Google’s worldwide  monopoly  on  search  (Epstein,  2019d).    The  solution  to  The  Google  Problem  is 4 to  declare  Google’s  massive  search  index  –  the  database  the  company  uses  to  generate search  results  –  to  be  a  public  commons,  accessible  by  all,  just  as  a  1956  consent  decree forced  AT&T to  share  all  its patents.   There  is  precedent  in  both law  and  in  Google’s own business practices to  justify  taking  this step. Declaring  Google’s  index  a  commons  will  quickly  give  rise  to  thousands  of  search platforms  like  Google.com,  each  competing  with  Google,  each  providing  excellent  search results,  each  serving  niche  audiences,  large  and  small,  exactly  like  newspapers  and television  networks  and  websites  do  now.  Search  will  become  competitive,  as  it  was  during its early  years,  and  democracy  will  be  protected  from Google’s  secretive  machinations. In  his  famous  departing  speech  in  January,  1961,  President  Eisenhower  warned  about  the possible  rise  of  a  “technological  elite”  that  could  control  public  policy  without  people’s awareness  (Epstein,  2016a,  2018c).  That  elite  now  exists,  and  they  have  more  power  than you think.  It’s up to Congress to determine  where  we  go from here. Chairman  Cruz,  Ranking  Member  Hirono,  other  Members  of  the  Committee,  thank  you  for the opportunity  to testify  today.  I  look  forward to  your questions. _____________________ APPENDIX: The  Methodology  of SEME Experiments The  methodology  of  SEME  experiments  adheres  to  the  highest  standards  of  research  in  the social  and  behavioral  sciences.    All  experiments  are  randomized,  controlled,  double-blind, and  counterbalanced  (Epstein  and  Robertson,  2015a).    Multiple  SEME  experiments conducted  over  a  period  of  more  than  five  years  have  involved  more  than  10,000 participants  and  five  national  elections  in  four  countries.    Reasonable  efforts  have  been made  to  assure  that  participants  are  diverse  across  multiple  demographic  characteristics, and,  when  possible,  representative  of  the  voting  population.  When  samples  are  not representative  of  the  voting  population,  adjustments  are  made  statistically  or  by  examining subsamples. In  most  experiments,  participants  are  selected  who  are  “undecided,”  by  which  I  mean  either that  they  haven’t  yet  made  up  their  minds,  or,  in  some  cases,  that  we  are  deliberately showing  them  materials  from  an  election  they  are  not  familiar  with  (for  example,  when  we show people from the  U.S. materials from an election in Australia). All search  results and web pages used in the  experiments are  real, drawn from the  internet and  from  Google’s  search  engine.    The  elections  we  have  examined  are  also  real:    the  2010 election  for  Prime  Minister  of  Australia;  the  2014  Lok  Sabha  election  in  India;  the  2015 national election in the  UK, and the 2016 and 2018 elections in the U.S. Search  results  are  presented  to  participants  using  a  mock  search  engine  called  Kadoodle, which  looks  and  functions  almost  exactly  like  Google.    The  difference  between  Google and  Kadoodle  is  that  with  Kadoodle,  we  control  what  search  results  we  show  and  the  order in  which  those  results  are  shown.    Our  search  results  link  to  copies  of  real  web  pages,  but links  on  those  pages  have  been  disabled  so  we  can  keep  our  research  participants  in  a  closed online  environment.    5 In  the  basic  procedure,  participants  are  randomly  assigned  to  one  of  three  groups:    a  group in  which  search  results  favor  Candidate  A  –  which  means  that  high-ranking  results  link  to web  pages  that  make  Candidate  A  look  better  that  his  or  her  opponent  –  a  group  favoring Candidate  B,  and  a  group  in  which  neither  candidate  is  favored  in  search  results  (the  control group). Participants  are  told  they  will  be  asked  to  use  our  custom  search  engine,  Kadoodle,  to conduct  research  on  political  candidates.    They  are  first  asked  to  read  short  paragraphs about  each  candidate  and  then  asked  several  questions  about  each  candidate:    How  much they  like  each  candidate,  trust  each  candidate,  and  so  on.    They  are  also  asked,  both  in  a binary  fashion  and  on  a  scale,  which  candidate  they  would  vote  for  if  they  had  to  vote today. These  are  all  “pre-search questions.” Then,  typically,  they  are  given  up  to  fifteen  minutes  in  which  to  use  the  Kadoodle  search engine  to  conduct  further  research  about  the  candidates.  They  are  typically  given  access  to five  pages  of  search  results,  with  six  results  per  page  (30  in  total),  and  they  can  navigate through  the  search  results  and  the  web  pages  exactly  as  they  would  on  Google.    They  can stop searching  when they  please. Then  they  are  asked  those  same  questions  about  the  candidates;  now  these  are  “post-search questions.” Remember  that  the  only  difference  between  the  three  groups  is  the  order  in  which  the search  results  are  shown.    All  participants  in  all  three  groups  have  full  access  to  all  the search  results and all  the  web pages. The  typical findings are  as follows:   Prior  to  search,  all  three groups  tend  to  answer  the pre-search questions  the same  way.   After  the  search,  the  opinions  and  voting  preferences  of  people  in  the  control  group  shift very  little or  not  at  all.   After  the  search,  both  the  opinions  and  the  voting  preferences  of  people  in  the  two  bias groups  shift  fairly  dramatically  in  the  direction  of  the  favored  candidate.    In  other  words, opinions  and  votes  shift  in  opposite  directions  in the two groups.   A shift  of  20  percent  or  more  is  typical.  In  large  studies  in  which  we  have  enough participants  to  look  at  demographic  differences,  we  have  found  shifts  in  the  60-to-80 percent  range  in  some  demographic  groups.    In  other  words,  some  people  are  especially trusting  of  search results.   Typically,  very  few  people  show  any  awareness  of  the  bias  they  have  seen.  In  a  large  study we  conducted  in  India  in  2014,  for  example  –  a  study  with  more  than  2,000  undecided voters  throughout  India  in  the  midst  of  an  intense  election  –  99.5  percent  of  our  participants showed no  awareness of  bias  in  the  search  results we  showed them.   The very  few  people  who  do  detect  the  bias  tend,  on  average,  to  shift  even  farther  in  the direction of  the bias. Some of  my  SEME research attempts to explain  why  the effect is so large. One  reason appears  to  be  that  people  trust  algorithmic  output,  believing  that  because  it  is  computergenerated, it  is inherently  objective  and unbiased. 6 Research  I  have  conducted  also  suggests  that  SEME  is  a  large  effect  because  people are  conditioned  –  very  much  like  rats  in  a  Skinner  box  –  to  believe  that  results  at  the top  of  the  list  are  better  and  truer  than  results  farther  down  the  list  (Epstein  &  Robertson (2016b).    This  is  because  most  searches  we  conduct  are  for  simple  facts,  such  as  “Who is  the  governor  of  Texas?”    The  correct  answer  always  turns  up  at  the  top  of  the  list, which is one  reason  50 percent of all  clicks  go to the  top two search positions. But  then  that  day  comes  when  we  search  for  something  with  a  less  certain  answer:  What  is  the  best  sushi  restaurant  in  town?    Who  is  the  best  candidate?    Again,  we  are most  likely  to believe  the  highest-ranking  answers. When,  in  one  experiment,  we  changed  people’s  beliefs  about  high-ranking  search results  by  placing  answers  to  simple  questions  in  random  positions  in  lists  of  search results, politically-biased  search  results has  less impact on them. 7 REFERENCES Attached Article  (Cited in the  Testimony) Epstein, R. (2019d, July  15). To break  Google’s  monopoly  on search, make  its index public.  Bloomberg  Businessweek.  https://www.bloomberg.com/news/articles/2019-0715/to-break-google-s-monopoly-on-search-make-its-index-public Other  Relevant Publications  and Peer-Reviewed Conference  Papers  (Some  Cited in the  Testimony) 2019: Epstein, R. (2019c, April  3).  Zucked  again:  Zuckerberg’s proposal for regulating  the internet is self-serving.  Epoch Times.  https://www.theepochtimes.com/opinionzuckerbergs-proposal-for-regulating-the-internet-is-self-serving_2864920.html   Epstein, R. (2019b, March 22). Google, Facebook,  Amazon: Warren's toothless break-up plan ignores real Big  Tech threats.  USA  Today. https://www.usatoday.com/story/opinion/2019/03/22/elizabeth-warren-plan-missesdangers-facebook-amazon-google-surveillance-column/3205451002/ Epstein, R.  (2019a, January  2). How  Google shifts votes: A “go vote”  reminder is not always what  you think it  is.  Epoch Times.  https://www.theepochtimes.com/another-waygoogle-manipulates-votes-without-us-knowing-a-go-vote-reminder-is-not-what-youthink-it-is_2754073.html Epstein, R., &  Williams,  E.  (2019, April).  Evidence  of systematic  political bias in online search results in the  10 days leading up to the 2018 U.S. midterm elections. Paper presented  at the  99th  annual meeting  of the  Western Psychological Association, Pasadena, CA. 2018: Epstein, R. (2018i, September 26).  Ten  ways  Big  Tech can shift millions of votes in the November  elections—without  anyone  knowing.  Epoch Times. https://www.theepochtimes.com/10-ways-big-tech-can-shift-millions-of-votes-in-thenovember-elections-without-anyone-knowing_2671195.html   Epstein, R. (2018h, September 13). Not just conservatives: Google  and  Big  Tech can shift millions of votes in  any  direction.  USA  Today.  https://www.usatoday.com/story/opinion/2018/09/13/google-big-tech-bias-hurtsdemocracy-not-just-conservatives-column/1265020002/   Epstein, R. (2018g, August 27). How  major  news  organizations, universities and businesses surrender their privacy  to Google.  The  Daily Caller. https://dailycaller.com/2018/08/27/surrender-privacy-google/   8 Epstein, R. (2018f, June  25).  Zuck off: Six  reasons Mark  Zuckerberg  should quit Facebook now.  The  Daily  Caller.  https://dailycaller.com/2018/06/25/mark-zuckerbergshould-quit-facebook/   Epstein, R. (2018e, May  25). Transcript to  Google's internal video, "The  Selfish  Ledger." https://aibrt.org/downloads/GOOGLE-Selfish_Ledger-TRANSCRIPT.pdf    Epstein, R. (2018d, May  17). Taming  Big  Tech: The  case  for  monitoring.  Hacker Noon. https://hackernoon.com/taming-big-tech-5fef0df0f00d  (Archived version at https://web.archive.org/web/20190323031918/https://hackernoon.com/taming-big-tech5fef0df0f00d?gi=7d11d84801a2) Epstein. R. (2018c). Manipulating  minds:  The  power of  search engines to  influence  votes and opinions.  In M. Moore  &  D. Tambini (Eds.),  Digital dominance: The power of Google, Amazon, Facebook, and Apple  (pp. 294-319).  Oxford, UK: Oxford University Press.  https://aibrt.org/downloads/EPSTEIN_2018-Manipulating_minds-Thepower_of_search_engines_to_influence_votes_and_opinionsUNCORRECTED_PROOFS.pdf   Epstein, R. (2018b). The  unprecedented power of  digital platforms to control opinions and votes.  In G. Rolnik (Ed.),  Digital platforms and concentration: Second annual antitrust and competition  conference  (pp. 31-33).  Chicago,  IL:  University  of Chicago Booth School of Business.  https://promarket.org/wp-content/uploads/2018/04/DigitalPlatforms-and-Concentration.pdf   Epstein, R. (2018a, March 22).  Cambridge  Analytica  is not the  problem: Google  and Facebook are  the problem.  The  Daily Caller.  https://dailycaller.com/2018/03/22/googleand-facebook-are-problem-not-cambridge-analytica/   Epstein, R., &  Mohr, R., Jr. (2018, April).  The  Answer Bot Effect (ABE): Another surprising way  search engines can impact opinions.  Paper presented at the  98th annual meeting  of the  Western Psychological Association, Portland, OR. https://aibrt.org/downloads/EPSTEIN_&_MOHR_2018-WPA-The_Answer_Bot_EffectABE-WP_17_04.pdf   Epstein, R., Mohr., R., Jr., &  Martinez, J. (2018, April).  The  Search Suggestion Effect (SSE):  How search suggestions can be  used to shift opinions  and voting preferences dramatically.  Paper presented at the  98th annual meeting  of the  Western Psychological Association, Portland, OR.  https://aibrt.org/downloads/EPSTEIN_MOHR_&_MARTINEZ_2018-WPAThe_Search_Suggestion_Effect-SSE-WP-17-03.pdf   2017: Epstein, R. (2017h, August 31). Google’s fighting  hate and trolls with a dangerously mindless AI.  Fast  Company.  https://www.fastcompany.com/40459339/googleperspective-fighting-hate-and-trolls-with-a-mindless-a-i   Epstein, R. (2017g, July  17). Hacking  Google.   Huffington Post. https://www.huffingtonpost.com/entry/596d0670e4b0376db8b659fd   9 Epstein, R. (2017f, May  28).  Is it  still  possible to  stop ‘Big  Tech’ from killing democracy?  [Review  of Jonathan Taplin’s  Move  Fast and Break  Things].  The  Hill. https://thehill.com/blogs/pundits-blog/technology/335507-is-it-still-possible-to-stop-bigtech-from-killing-democracy Epstein, R. (2017e, May  10).  The  potential for  Google  and Facebook to manipulate millions of voters poses a  bigger threat than  fake  news  [Letter to the  editor].  Los Angeles Times. Epstein, R.  (2017d, April  10).  Fake  news is a fake  problem.  Medium.  https://medium.com/@re_53711/fake-news-is-a-fake-problem-914d7ffc7a91    Epstein, R. (2017c, March 16). Seven simple  steps toward online  privacy.  Medium. https://medium.com/@re_53711/seven-simple-steps-toward-online-privacy-20dcbb9fa82    Epstein, R. (2017b, April).  Can search suggestions impact what we  search for  online? The  role  of negativity bias.  Paper presented at the  97th annual meeting  of the  Western Psychological Association, Sacramento, CA. https://aibrt.org/downloads/EPSTEIN_2017-WPACan_Search_Suggestions_Impact_What_We_Search_for_Online.pdf Epstein, R. (2017a, March). The  Search Suggestion Effect  (SSE): How  autocomplete can be  used to impact votes and opinions.  Paper presented at the  2nd biennial  meeting  of the International Convention  of Psychological Science, Vienna, Austria. https://aibrt.org/downloads/EPSTEIN_2017-The_Search_Suggestion_Effect-SSEICPS_Vienna-March_2017.pdf   Epstein, R., Ding., M., Mourani, C., Olson, E., Robertson, R.E., &  Tran, F.  (2017, April). Multiple  searches increase  the impact of  the Search Engine Manipulation Effect  (SEME). Paper presented at the  97th annual meeting  of the  Western Psychological Association, Sacramento, CA.  https://aibrt.org/downloads/EPSTEIN_et_al._2017-WPAMultiple_Searches_Increase_the_Impact_of%20_the_Search_Engine_Manipulation_Effe ct.pdf Epstein, R., Mourani, C., Olson, E., &  Robertson,  R.E. (2017, April).  Biased search rankings can shift opinions on a wide range  of topics.  Paper presented at the  97th annual meeting  of the  Western Psychological Association, Sacramento, CA. https://aibrt.org/downloads/EPSTEIN_et_al._2017-WPABiased_Search_Rankings_Can_Shift_Opinions_on_a_Wide_Range_of_Topics.pdf   Epstein, R., &  Robertson, R.E. (2017c, June  1).  A method for detecting bias  in search rankings,  with evidence  of systematic  bias  related to the  2016 presidential election.  Vista, CA: American  Institute  for Behavioral Research  and Technology, White  Paper no.  WP17-02. Retrieved from  https://aibrt.org/downloads/EPSTEIN_&_ROBERTSON_2017A_Method_for_Detecting_Bias_in_Search_Rankings-AIBRT_WP-17-02_6-1-17.pdf   Epstein, R., &  Robertson, R.E. (2017b, March).  The  Search Engine  Manipulation Effect (SEME):  Understanding its power to change  opinions and votes. Paper presented at the 2nd biennial meeting  of the  International Convention of  Psychological Science, Vienna, Austria. 10 DC.   Epstein, R., &  Robertson, R. (2017a).  Suppressing  the Search Engine  Manipulation Effect  (SEME).  Proceedings of  the ACM: Human-Computer Interaction, 1(2), Article 42. https://aibrt.org/downloads/EPSTEIN_et_al-2017Suppressing_the_Search_Engine_Manipulation_Effect_(SEME).pdf   Epstein, R., Robertson, R., Shepherd, S., &  Zhang,  S. (2017, April).  A method for detecting bias  in search rankings,  with evidence  of  systematic  bias  related to the  2016 presidential election. Paper presented at the  97th annual meeting  of the  Western Psychological Association, Sacramento, CA. https://aibrt.org/downloads/EPSTEIN_et_al_2017-SUMMARY-WPAA_Method_for_Detecting_Bias_in_Search_Rankings.pdf   2016: Epstein, R. (2016l, November  17).  Fake  news and  Facebook: There  are  far  more pernicious ways social media can sway  elections [Letter to the  Editor].  Los Angeles Times.  https://www.latimes.com/opinion/readersreact/la-ol-le-fake-news-facebook20161117-story.html   Epstein, R. (2016k). Subtle new forms of  internet influence  are  putting  democracy  at risk worldwide.  In N.  Lee  (Ed.),  Google  it: Total information awareness  (pp. 253-259). Springer.  https://aibrt.org/downloads/EPSTEIN_2016Suble_New_Forms_of_Internet_Influence-In_N.Lee_Ed.-Google_It-Springer.pdf   Epstein, R. (2016j, October  14).    Breaking  news:  Google to donate its  search engine to the American public.  Huffington Post.  https://www.huffingtonpost.com/dr-robertepstein/breaking-news-google-to-d_b_12446856.html   Epstein, R. (2016i, September 12). Are  we  being  manipulated by  Google’s  autocomplete? Sputnik  International.  https://sputniknews.com/us/20160912/1045214398/google-clintonmanipulation-election.html   Epstein, R. (2016h, September 6). Free  isn’t freedom:  How  Silicon Valley  tricks us. Motherboard.  https://motherboard.vice.com/read/free-isnt-freedom-epstein-essay   Epstein, R. (2016g, September). Cyber sway: The  new mind control.  Ladybeard. https://aibrt.org/downloads/EPSTEIN_2016-Cyber_Sway-The_New_Mind_ControlLADYBEARD-9-16-16-FINAL_flat.pdf   Epstein, R. (2016f, July  12). Five subtle  ways Facebook could influence  the  US presidential election this  fall.  Quartz.  https://qz.com/703680/five-subtle-ways-facebookcould-influence-the-us-presidential-election-this-fall/   Epstein, R. (2016e, July).  Can search  engine rankings swing  elections?  New Internationalist.  https://newint.org/features/2016/07/01/can-search-engine-rankingsswing-elections/   11 Epstein. R. (2016d, June  22). The  new  censorship.  U.S. News  &  World Report. https://www.usnews.com/opinion/articles/2016-06-22/google-is-the-worlds-biggestcensor-and-its-power-must-be-regulated   Epstein, R. (2016c, May  4). Bigger brother: Microsoft and Google’s new pact could signal the  beginning  of the  end for personal privacy.  Quartz. https://qz.com/676184/microsoft-and-googles-pact-is-the-end-of-personal-privacy/    Epstein, R. (2016b, April  27).  Google knows:  In the  future,  Big  Data  will make  actual voting  obsolete.  Quartz.  https://qz.com/669983/maybe-we-should-let-google-vote-for-us/   Epstein, R. (2016a, February  18).  The  new mind  control.  Aeon. https://aeon.co/essays/how-the-internet-flips-elections-and-alters-our-thoughts  [Russian translation here:  http://mnenia.zahav.ru/Articles/7302/soznanie_pod_controlem]   Epstein, R., &  Edelman,  B. (2016, November  5).  The  other  elephant in the  voting  booth: Big  Tech  could rig  the election.  The  Daily Caller.  https://dailycaller.com/2016/11/04/theother-elephant-in-the-voting-booth-big-tech-could-rig-the-election/   Epstein, R., &  Robertson, R.E. (2016b, April).  Why  is the  search engine  manipulation effect  (SEME) so large?  A test of  an operant  conditioning hypothesis.  Paper presented  at the 96th annual meeting  of the  Western Psychological Association,  Long  Beach, CA. https://aibrt.org/downloads/EPSTEIN_&_ROBERTSON_2016SEME_Testing_an_operant_conditioning_hypothesis-WPA.pdf   Epstein, R., &  Robertson, R.E. (2016a, April).  A replication of the  search engine manipulation effect (SEME), plus  methods  for suppressing the effect.  Paper presented  at the 96th annual meeting  of the  Western Psychological Association,  Long  Beach, CA. https://aibrt.org/downloads/EPSTEIN_et_al-2017Suppressing_the_Search_Engine_Manipulation_Effect_(SEME).pdf    2015: Epstein, R. (2015d).  Wie  Google Wahlen beeinflussen kann [How  Google  influences opinions]. In T.  Fricke  & U.  Novak (Eds.),  Die  Akte Google:  Wie der US-Konzern Daten missbraucht, die  Welt  manipuliert und Jobs vernichtet, Mit  Beiträgen von Prof. Dr. Robert Epstein und Dr. Thomas Höppner  (pp. 165-173).  München,  Deutschland: F.A. Herbig  Verlagsbuchhandlung  GmbH. Epstein, R. (2015c, October 6). Google’s hypocrisy.  Huffington Post.  Retrieved from https://www.huffingtonpost.com/dr-robert-epstein/googles-hypocrisy_b_8253332.html   Epstein, R. (2015b, September 6). Google’s vote  counts  more  than  yours.  Pittsburgh Post-Gazette.  https://www.post-gazette.com/opinion/Op-Ed/2015/09/06/Google-s-votecounts-more-than-yours-because-its-search-engine-is-determining-the-outcomes-ofelections/stories/201509060042   Epstein, R. (2015a, August 19). How Google could rig  the 2016  election.  Politico. https://www.politico.com/magazine/story/2015/08/how-google-could-rig-the-2016election-121548.html   12 Epstein, R., &  Robertson, R.E. (2015b, April).  The  Search Engine  Manipulation Effect (SEME): Large-scale replications in two countries.  Paper presented at the  95th annual meeting  of the  Western Psychological Association,  Las  Vegas,  Nevada. Epstein, R., &  Robertson, R. E. (2015a, August 4). The  search engine manipulation effect (SEME)  and its possible  impact on the  outcomes  of elections.  Proceedings  of the National Academy  of Sciences USA, 112(33), E4512-E4521. https://www.pnas.org/content/112/33/E4512.full.pdf?with-ds=yes 2014: Epstein, R. (2014d). Democracy  at risk from new forms of  internet influence.  EMMA Magazine.  https://aibrt.org/downloads/EPSTEIN_2014New_Forms_of_Internet_Influence-EMMA_Magazine.pdf   Epstein, R. (2014c, June). How Google could end  democracy.  U.S. News  &  World Report.  https://www.usnews.com/opinion/articles/2014/06/09/how-googles-searchrankings-could-manipulate-elections-and-end-democracy?src=usn_tw   Epstein, R. (2014b, May). Google  critic killed in “ironic”  car accident: Struck by  Google Street View  vehicle.  Huffington Post.  https://www.huffingtonpost.com/dr-robertepstein/google-critic-killed-in-i_b_5351458.html   Epstein, R. (2014a, May). Google’s snoops: Mining  our private data  for  profit and pleasure.  Dissent.  https://www.dissentmagazine.org/online_articles/googles-snoopsmining-our-data-for-profit-and-pleasure   Epstein, R., &  Robertson, R.E. (2014, April).   Helping people  preserve  their privacy online: The surprising power  of a click  requirement.  Paper presented at the  94th annual meeting  of the  Western Psychological Association, Portland, Oregon. https://aibrt.org/downloads/EPSTEIN&ROBERTSON_2014Helping_People_Be_More_Cautious_Online-ABSTRACT-WPA-April_2014.pdf   2013: Epstein, R. (2013b, May). Google’s  gotcha.  U.S. News  &  World Report. https://www.usnews.com/opinion/articles/2013/05/10/15-ways-google-monitors-you   Epstein, R. (2013a, March). Google’s dance.  TIME. https://techland.time.com/2013/03/27/googles-dance/   Epstein, R., &  Robertson, R. E. (2013, May).  Democracy  at risk: Search rankings can shift voter preferences substantially.  Paper  presented at the  25th annual meeting  of the Association for  Psychological Science, Washington, D.C., May  2013. 2012: Epstein, R. (2012e, November  5). Why  Google  should be regulated  (Part 4  -  End). Huffington Post.  https://www.huffpost.com/entry/why-google-should-beregu_b_2069223   13 Epstein, R. (2012d, November  2). Why  Google  should be regulated  (Part 3).  Huffington Post.  https://www.huffpost.com/entry/why-google-should-be-regu_b_2054111  6-29-19 update:  Link to Part 4  at the  end of  the  article is broken. Use: https://www.huffpost.com/entry/why-google-should-be-regu_b_2069223    Epstein, R. (2012c, October 31). Why  Google should be regulated  (Part 2).  Huffington Post.  https://www.huffpost.com/entry/online-privacy_b_2013583  6-29-19 update:  Link to Part 3 at the  end of  the  article is broken. Use:  https://www.huffpost.com/entry/whygoogle-should-be-regu_b_2054111   Epstein, R. (2012b, October  23).  Why  Google should be regulated  (Part 1).  Huffington Post.  https://www.huffpost.com/entry/google-privacy_b_1962827  6-29-19  update:  Link to Part 2 at the  end of  the  article is broken. Use:  https://www.huffpost.com/entry/onlineprivacy_b_2013583  (Note: An edited version of  the  entire  article first appeared in  The Kernel  [UK]  on September 5, 2012. See  below.) Epstein, R. (2012a, September  12).   Google: The  case  for  hawkish regulation.   The Kernel.  https://web.archive.org/web/20121016112820/https://www.kernelmag.com/features/repor t/3281/google-the-case-for-hawkish-regulation  (Expanded version appeared in four  parts in  The  Huffington Post  beginning  on October  23,  2012. See  above.) 14 https://www.bloomberg.com/news/articles/2019-- 0715/tobreak googlesmonopolyonsearchmakeitsindexpublic Bloomberg  Businessweek To  Break  Google’s Monopoly  on  Search, Make  Its  Index  Public The  tech  giant  doesn’t  have  to  be  dismantled.  Sharing  its  crown jewel  might  reshape  the  internet. By   Robert  Epstein July  15,  2019,  3:00  AM  PDT PHOTO ILLUSTRATION:  731;  PHOTO:  GETTY  IMAGES 1 Recognition  is growing  worldwide that something  big  needs to  be done  about Big Tech,  and  fast. More than  $8  billion  in  fines  have  been levied  against Google by the European  Union since  2017.  Facebook  Inc.,  facing  an  onslaught of  investigations,  has  dropped in  reputation  to almost  rock bottom  among  the 100  most visible companies in the U.S.  Former employees  of Google and  Facebook have warned  that these companies  are  “ripping  apart the  social  fabric”  and  can  “hijack  the  mind.” Adding  substance  to the concerns, documents and  videos have  been leaking  from Big  Tech  companies, supporting  fears—most often  expressed  by  conservatives— about  political  manipulations  and  even aspirations  to  engineer  human  values. Fixes on  the  table include forcing  the tech titans to  divest  themselves  of  some  of the companies they’ve  bought  (more  than  250  by Google and  Facebook alone) and  guaranteeing that user  data  are transportable. But these  and  a  dozen  other  proposals never  get  to  the  heart of  the  problem,  and  that is that Google’s  search  engine and  Facebook’s 2 social  network platform  have  value only if they are intact. Breaking  up Google’s search  engine would  give  us  a  smattering of  search  engines that yield  inferior results (the  larger  the  search engine,  the  wider  the range of results  it  can give  you),  and  breaking  up  Facebook’s platform  would  be like building  an  immensely long  Berlin  Wall  that would  splinter millions of  relationships. With those  basic  platforms intact,  the three biggest  threats that  Google  and  Facebook  pose to societies worldwide are barely  affected  by almost  any intervention:  the  aggressive surveillance,  the  suppression  of  content,  and the  subtle  manipulation  of  the  thinking  and behavior of  more than  2.5 billion people. Different  tech  companies pose different kinds of  threats.  I’m focused here  on  Google, which I’ve  been  studying  for  more than  six years through  both  experimental research  and  monitoring  projects.  (Google is  well aware  of  my work and  not  entirely  happy  with me. The company did  not respond  to requests for comment.) Google is especially worrisome because it has  maintained  an  unopposed monopoly  on search  worldwide for  nearly  a 3 decade.  It  controls  92  percent  of  search,  with the  next  largest  competitor,  Microsoft’s  Bing, drawing  only 2.5%. Fortunately,  there is  a  simple  way to end the  company’s  monopoly without breaking  up its  search engine,  and  that is to  turn  its “index”—the  mammoth  and  ever-growing database it  maintains of internet  content—into a  kind  of  public commons. There is precedent for  this both in  law and in  Google’s business  practices. When  private ownership of essential  resources and services—water,  electricity, telecommunications,  and  so on—no  longer serves the  public  interest, governments  often step in  to control  them.  One particular government intervention  is especially relevant to the Big  Tech dilemma:  the  1956  consent decree  in  the U.S.  in  which AT&T  agreed  to share  all  its  patents with other  companies free of  charge.  As  tech  investor  Roger  McNamee  and others have pointed  out,  that sharing reverberated  around the world,  leading  to a significant increase in technological competition  and  innovation. 4 Doesn’t Google  already share its index with everyone in  the  world?  Yes,  but only  for  single searches.  I’m  talking  about requiring  Google to share its  entire index with outside entities— businesses, nonprofit organizations,  even individuals—through  what  programmers  call an  application  programming  interface,  or API. Google already allows this kind  of  sharing with a  chosen  few, most  notably  a  small  but ingenious company called  Startpage,  which is based in  the Netherlands. In  2009,  Google granted  Startpage access to  its index  in  return for fees generated  by  ads  placed  near Startpage search  results. With  access  to Google’s index—the most extensive  in  the world,  by far—Startpage gives you  great  search  results,  but  with  a difference. Google tracks  your  searches and  also  monitors you  in  other  ways,  so it gives you  personalized results.  Startpage doesn’t track you—it respects  and  guarantees  your  privacy—so it gives you  generic results. Some  people like customized results;  others treasure  their privacy. (You  might have  heard of another privacy-oriented  alternative to  Google.com called DuckDuckGo,  which  aggregates 5 information  obtained from 400 other  nonGoogle sources,  including  its own  modest crawler.) If entities  worldwide were given  unlimited access to  Google’s index, dozens of Startpage variants  would  turn  up within  months; within a  year  or  two,  thousands  of  new  search platforms might  emerge,  each with  different strengths  and  weaknesses.  Many  would  target niche audiences—some small,  perhaps,  like high-end  shoppers,  and  some huge,  like all  the world’s  women,  and  most  of  these platforms would  do  a better job  of  serving  their constituencies  than  Google ever  could. These aren’t just  alternatives  to  Google, they  are  competitors—thousands  of  search platforms, each  with its  special  focus and emphasis,  each  drawing  on  different  subsets of information  from Google’s  ever-expanding index, and  each using different  rules  to  decide how to  organize  the  search results they display.  Different platforms would  likely have different business models, too,  and  business models that have never  been  tried  before would  quickly be  tested. 6 This system  replicates the  competitive ecology we now  have of both  traditional  and online  media sources—newspapers, magazines, television  channels, and  so on— each  drawing on  roughly the same body of knowledge,  serving  niche  audiences,  and prioritizing  information  as it sees  fit. But what  about those  nasty filter  bubbles that trap people  in  narrow worlds of information? Making  Google’s index public doesn’t  solve  that problem,  but it shrinks it to nonthreatening  proportions.  At  the moment, it’s  entirely up  to  Google to  determine which bubble you’re  in,  which search suggestions  you receive,  and  which search  results  appear  at  the top of the list;  that’s the stuff of  worldwide  mind control.  But with  thousands of search platforms vying  for your  attention,  the power is back in  your  hands. You  pick your  platform or platforms  and  shift  to  others  when  they draw your  attention,  as  they will  all  be trying to do  continuously. If that happens, what becomes of  Google? At  first, not  much.  It should  be allowed,  I believe, to  retain  ownership  and  control  of its index. That will  assure it continues  to  do  a 7 great job  maintaining  and  updating it. And even  with  competition  looming,  change will take time. Serious competitors  will  need months to  gather  resources  and  generate traffic. Eventually,  though,  Google will  likely become  a  smaller,  leaner,  more diversified company,  especially if some of the other proposals out there for  taming  Big  Tech  are eventually implemented.  If, over  time, Google wants  to continue to  spy on  people through  its search engine, it  will  have  to work like hell  to keep them.  It  will  no  longer  be able to  rest on its  laurels,  as it has for  most of the past 20 years; it’s  going  to  have to  hustle,  and  we  will all  benefit from  its  energy. My kids  think  Google was the  world’s first search  engine, but it was  actually  the  21st.  I  can remember  when  search  was highly competitive—when  Yahoo!  was the big  kid  on the  block  and  engines such  as  Ask Jeeves and Lycos  were hot commodities. Founded in  1998 amid  a  crowded  field  of competitors,  Google didn’t begin  to  dominate search  until  2003,  by which time it  still  handled only about  a  third  of searches  in  the U.S.  Search can  be  competitive again—this time  with a  massive,  authoritative, 8 rapidly expanding  index  available to  all parties. The alternative  is frightening.  If  Google retains its  monopoly on  search,  or even if a government  steps in and  makes  Google a public utility,  the obscene power  to  decide what information  humanity  can  see and  how that information  should  be ordered will remain  in  the hands of  a  single  authority. Democracy  will  be an  illusion,  human autonomy  will be compromised, and competition  in  search—with  all  the innovation that implies—might  never  emerge. With internet penetration  increasing  rapidly worldwide, do  we really want  a  single player, no  matter  how benign  it appears  to be, to control  the gateway to all  information? For  the  system I  propose  to work  fairly  and efficiently,  we’ll  need rules. Here  are some obvious ones  to think about: Access.  There  might have to  be  limits  on who  can  access the API. We might not  want every high  school  hacker  to be able to  build  his or her  own  search  platform.  On  the other hand,  imagine  thousands of  Mark Zuckerbergs 9 battling  each other  to find  better  ways  of organizing  the world’s information. Speed.  Google  must not be  allowed  to throttle  access  to  its  index, especially in  ways that give  it a  performance  advantage or  that favor one  search  platform  over  another. Content.  To  prevent Google from engineering  humanity  by being  selective  about what content it  adds to  its index,  all  parties with API access  must be  able to  add  content. Visibility.  For  people using Google to  seek information  about other  search platforms, Google must be forbidden  from  driving  people to itself  or its affiliated  platforms. Removal.  Google must be  prohibited  from removing  content from  its  index.  The only exception  will  be when  a  web  page no  longer exists.  An  accurate,  up-to-date record of  such deletions  must be accessible through  the  API. Logging.  Google must  log  all  visits  to  its index, and  that log  must be accessible through the  API. Fees.  Low-volume  external  platforms (think:  high  school  hackers)  should  be able to access the index  free of charge.  High-volume users  (think:  Microsoft  Corp.’s Bing)  should 10 pay  Google nominal  fees set  by  regulators. That  gives  Google another  incentive  for maintaining  a  superior index. Can  we  really justify bludgeoning  one of the world’s biggest  and  most  successful companies?  When  governments  have regulated,  dismembered, or,  in  some  cases, taken  ownership  of private  water  or electricity companies, they  have done so  to serve the public interest, even when  the company in question  has developed  new technologies  or resources at great  expense. The  rationale is straightforward:  You  may have  built the pipelines,  but water  is a  “common”  resource that belongs to  everyone,  as David  Bollier reminded us in  his seminal  book,  Silent  Theft: The  Private  Plunder  of  Our  Common  Wealth. In  Google’s case,  it would  be absurd  for  the company to claim  ownership  rights  over  the contents  of its index  for  the simple reason  that it  gathered  almost all  those contents. Google scraped  the  content by  roaming  the internet, examining  webpages, and  copying  both the address  of  a  page and  language used  on  that page. None of those websites or  any external 11 authority  ever  gave Google permission  to do this copying. Did  any  external  authority  give Google permission  to  demote a  website in its  search results  or  to  remove  a  website from  its index? No,  which is  why both  individuals  and  even  top business  leaders  are  sometimes  traumatized when  Google demotes  or  delists  a  website. But when  Google’s index  becomes  public, people  won’t care as much about its machinations. If  conservatives  think Google is messing with them,  they’ll  soon  switch to other  search  platforms,  where  they’ll  still  get potentially excellent  results. Given  the possibility  of  a  mass migration,  Google will likely stop  playing  God,  treating users  and constituencies  with new respect and  humility. Who  will  implement this  plan?  In  the  U.S., Congress, the  Federal  Trade  Commission,  and the  Department  of  Justice all  have the  power to make this happen.  Because  Google is a global  company  with,  at this  writing,  16  data centers—eight in  the  U.S.,  one in  Chile, five in the  EU,  one in Taiwan,  and  one in Singapore— countries  outside the U.S.  could  also declare 12 its  index  to be a  public commons.  The EU  is a  prime  candidate  for taking  such action. But there is another  possibility—namely, that Google itself will  step up.  This isn’t as crazy  as you  might think.  Likely  prompted  by the  EU  antitrust  investigations, the company has  quietly gone  through  two  corporate reorganizations  since 2015, and  experts I’ve talked to in  both  the U.S.  and  the  U.K.  say the main  effect of these reorganizations has been to distance Google’s major  shareholders from any calamities  that  might befall  the  Google search  engine. The company’s lawyers have also undoubtedly been taking  a  close look  at the  turbulent  years  during  which  Microsoft unsuccessfully fought U.S.  antitrust investigators. Google’s leaders  have been  preparing  for  an uncertain  future  in  which the search  engine might be  made  a  public utility,  fined  into bankruptcy,  frozen  by  court orders, or even  seized  by  governments.  It  might be able to avoid  ugly scenarios simply  by  posting  the specs for its  new  public API  and  inviting people  and  companies around  the world  to compete  with its  search platform.  Google 13 could  do  this tomorrow—and  generate glowing headlines  worldwide.  Google’s data  analysts know how to  run  numbers better  than  anyone. If the models predict  that the company  will make  more  money,  minimize risk,  and optimize its  brand  in  coming  years by  making its  index  public, Google will  make this happen long  before the roof caves in. Epstein  (@DrREpstein),  a  former  editor-inchief  of  Psychology Today,  is  senior  research psychologist  at  the  American  Institute  for Behavioral  Research  and  Technology.  He  has published  15  books  and  more  than  300  articles on  AI  and other  topics. 14