Notes on Rosenzweig, Can history by open source?

In his article Can history be open source? Roy Rosenzweig takes up the question of whether the Wikipedia model can lead to good history.

Contents

Reading Notes

Background Facts

  • Only 6% of history papers are co-authored.
  • Currently Wikipedia has 2.2 million in English.
  • The W entry on Franklin Roosevelt was written over four years has 500 authors who made 1000 edits.
  • W gets more than 1 million visits a day, way more that NY Times, Library of Congress, Britannica.
  • Started by Jimmy Wales with help from Larry Sanger and funds from the Sales' Bomis Babe Report.
  • Nupedia had expert editors. Sanger eventually split from the project over its spurning of experts and the tolerance of "trolls."
  • Based on WikiWikiWeb software, developed by Ward Cunningham in the 1990s.
  • Encyclopedia: no personal essays, dictionary entries, propaganda or original research. It summarize and reports the conventional wisdom.
  • Articles should have a NPOV (Neutral Point of View).
  • Licensed under the GNU Free Documentation License (GFDL), designed for manuals and textbooks. As such Wikipedia content may be copied and distributed in any medium provided "you add not other conditions to...this License." ("Copyleft").

Are Wikipedians good historians?

  • Articles on synthetic topics (U.S. immigration history) are often incomplete and unevenly written.
  • Certain topics are missing, reflecting lack of interest of Wikipedians.
  • Biographies are more favorable terrain.
  • Rosenzweig compared 25 biographies with those in Encarta (Microsoft) and American National Biography Online (Oxford), both of which have mulitmillion dollar budgets.
  • Widipedia lags behind ANBO but exceeds Encarta in coverage. W has found unpaid volunteers "to write suprisingly detailed and reliable portraits of relatively obscure historical figures."
  • Article lengths vary (3500 word on Isaac Asimov, 3200 on Woodrow Wilson). W's authors are not uniformly distributed demographically--English-speaking males and Internet denizens.
  • See Why Wikipedia is Not So Great page: "geek priorities."

RR's Observations

  • RR found factual errors in only 4 of the 25 articles, mostly small and inconsequential. By comparison he found 3 such errors in Encarta and 1 in ANBO.
  • W beats Encarta in coverage and roughly matches it in accuracy. These observations are consistent with more systematic studies.
  • THUS: W offers a "formidable challenge" to Britannica and Encarta. But, ANBO, with its reliance on professional historians, easily outperforms W because good historical writing requires a "command of the scholarly literature, persuasive analysis and interpretations, and clear and engaging prose."
  • The NPOV policy leads to tepid conclusions on controversial topics.
  • Writing is the "achilles heel", with the newest and least edited articles being worst.
  • Watch lists keep vandalism to a minimum.
  • Wade Roush, Technology Review: "The community-editing models gives us an newfound power to create wrongs--but also to reverse wrongs."

Implications for Historians

  • Students use W--this is not cause for alarm.
  • Echo-chamber effect--inaccurate conventinal wisdom in W are reinforced by search engines.
  • It's an encyclopedia--spend more time teaching about the limitations of information sources.
  • Historians should join the open source movement and make their journals accessible.
  • W's open production model fosters good sourcing skills.
  • RR's own "tentative" answer is the professional historians houls join popular history makers in writing history in Wikipedia. This may require W to modify its policy on "no original research".
  • Can the wiki way foster the collaborative creation of historical knowledge? Perhaps by leveraging the volunteer labor of amateurs?
  • Commons-based peer production has many potential strengths. But it also has potential difficulties for professional scholarship, which is based on credit. Do we reward people for collaborative work?