Bypassing Interlibrary Loan Via Twitter.pdf

(308 KB) Pobierz

Bypassing Interlibrary Loan

Via Twitter: An Exploration of

#icanhazpdf Requests

Carolyn Caffrey Gardner and Gabriel J. Gardner

Introduction

Twitter has emerged as a popular social media plat-

form for many scientists and scholars.

Priem and Costello estimate that one out of every

forty scholars, defined as faculty, postdoc, or doctoral

student, in the United States and the United Kingdom

is a registered Twitter user.

What’s more, they are using the platform to share

scholarly material. Moriano et al. have demonstrated

that the number of tweets containing links to scholarly

publications increased substantially from 2011 to 2013;

similarly, they also increased as a percentage of overall

twitter traffic.

Their analysis also noted that certain

publishers’ content is shared more than others. The top

domain names tweeted in their sample were: nature.

com, arxiv.org, sciencemag.org, wiley.com, and science-

direct.com. While the interdisciplinary nature of these

domains makes it difficult to draw conclusions about

Twitter use rates by specific academic discipline, Priem

and Costello concluded that no one academic discipline

was significantly overrepresented on Twitter.

Apart from everyday social use of the micro-

blogging service, scholars are clearly using Twitter to

increase their professional networks, organize pre-

publication review of working papers and manuscript

drafts, offer post-publication critique,

disseminate

published research,

and share pre-prints.

Twitter is also used to facilitate access to schol-

arly articles that would otherwise be denied to users

behind a paywall or obtained using interlibrary loan.

Like peer-to-peer sharing in the music industry, this

peer-to-peer access to scholarly material is ethically

dubious, and may run afoul of copyright laws, but it is

easy to accomplish. The Twitter user simply appends

the metadata label, or “hashtag”, #icanhazPDF in the

tweet rendering it discoverable through traditional

linking and search functions.

This peer-to-peer access, while ethically dubious,

is coordinated by the use of the Twitter hashtag #ican-

hazPDF. The hashtag is included as a tag in the tweet,

which renders it discoverable through traditional

linking and search functions.

Literature Review

The modern interlibrary loan (ILL) office has been

likened to that of a detective’s office, assisting schol-

ars of all stripes track down materials not held in the

library. ILL offices can often find materials when only

partial citation information is available.

In spite of al-

ternatives and freely available content online, the av-

erage ILL request per ARL library has increased in 31

of the 35 years ARL has kept such statistics.

Users ex-

pect that these ILL requests will happen as easily and

instantaneously as they do using popular interfaces,

such as Google

and many ILL services have made

strides, sometimes filling requests in as little as 24

hours. Also, as electronic journals have become an es-

tablished part of the scholarly communications land-

Carolyn Caffrey Gardner is Information Literacy & Educational Technology Librarian, University of Southern California, e-

mail: ccgardne@usc.edu; Gabriel J. Gardner is Librarian for Romance, German, Russian Languages & Literatures, and Lin-

guistics, California State University, Long Beach, e-mail: gabriel.gardner@csulb.edu

scape, fewer lending agreements contain ILL restric-

tions. Lamoreux and Stemper found that at University

of Minnesota, 89% of licenses allowed for lending and

it was primarily small scholarly associations that re-

stricted lending.

The scholarship practices that have

resulted in increased ILL requests are bleeding over

into non-library spaces.

Usage of the #icanhazPDF hashtag to facilitate

the sharing of scholarly articles dates back to 2011

with the suggestion from Andrea Kuszewski on the

hashtag language, a riff on a popular internet cat

meme.

The hashtag and the social sharing networks

it facilitates have received passing mentions in some

medical literature

12,13

but little direct study. Dunn et al.

characterize #icanhazPDF as a form of “guerrilla open

access” through subversion of publisher agreements,

in the tradition of internet activist, Aaron Swartz’s,

manifesto of the same name.

14,15

Since its inception,

the hashtag has been a controversial topic of discus-

sion on science blogs. Michael Eisen, co-founder of

the Public Library of Science (PLoS), portrayed the

use of the hashtag as an act of civil disobedience in

opposition to the current copyright regime that gov-

erns scientific publishing.

The sharing mechanics follow a simple protocol.

First, a requestor tweets a link or partial citation to

a pay-walled article with the hashtag #icanhazPDF

and their e-mail address. Second, sympathetic users

then use their institutional subscriptions or personal

memberships to download the desired PDF and email

it to the requestor, off of Twitter. Once in possession

of the desired PDF, diligent requestors delete their

tweet containing the original request. Thanking a user

who fulfills the request is discouraged.

This allows

the fulfilling user, who likely violated a copyright or

license agreement, to maintain anonymity.

The small amount of research on #icanhazPDF

has focused on demographic data. In a blog post,

Jean Liu collected tweets using the hashtag over the

course of a year beginning in May 2012. Her analy-

sis revealed that overall use of the hashtag slowly in-

creased over the period of study to an average of 3.6

#icanhazPDF tweets per day.

Using #icanhazPDF

ACRL

2015

Carolyn Caffrey Gardner and Gabriel J. Gardner

represents a small percentage of Twitter-user behav-

ior compared to sharing scholarly research by send-

ing links to pay-walled papers. In addition to initial

quantitative analysis, Liu also sampled location and

profile data. These were then used to determine oc-

cupation and country location of #icanhazPDF us-

ers. Usage was overwhelmingly an Anglophone phe-

nomenon with the almost half of the tweets sampled

coming from the United States; the United Kingdom

produced the second highest number of tweets. Occu-

pation data revealed that academics and students, de-

spite being the most likely to have institutional access

to scholarly research, were the most frequent users of

the hashtag. Finally, Liu’s category of communicators

which encompassed journalists and bloggers, had the

third highest use of #icanhazPDF.

Though scholars and scientists have been the

primary focus regarding #icanhazPDF, librarians

have also taken note of the phenomenon and begun

to grapple with how it might affect their institutions

and workflows. Greenhill and Wiebrands argue that

libraries should view #icanhazPDF and other copy-

right-violating (or license-breaching) methods of

content sharing as competition and not ignore the

black market transactions.

When such peer-to-peer

access is viewed as a competitor, libraries are at a dis-

advantage because they must adhere to copyright and

intellectual property laws, which may take more time

and/or financial resources. To differentiate themselves

from crowdsourced methods they might emphasize

the local or niche content they provide, and the physi-

cal space they provide, while advocating for “more

open and fair” publishing models.

One apparent impact that #icanhazPDF shar-

ing has on libraries, is in the area of interlibrary loan

(ILL). Each request fulfilled through Twitter repre-

sents one side of a possible ILL transaction.

The in-

stitutions of the fulfilling users will record downloads

of the requested files. Any libraries that requestors

might have used however, are left without any re-

cord of user demand. Thus, #icanhazPDF and other

methods of peer-to-peer sharing distort library use

statistics: libraries serving users who fulfill requests

Bypassing Interlibrary Loan Via Twitter: An Exploration of #icanhazpdf Requests

via #icanhazPDF have artificially inflated download

statistics, while libraries whose (potential) users ob-

tain articles over Twitter have artificially deflated ILL

statistics. The magnitude of these errors is unknown

and an area for further research. Therefore, Jill Emery,

reacting to Liu and Bond, urged librarians involved

with collection development and technical services

to treat #icanhazPDF as an impetus to improve our

services, specifically in the area of document deliv-

ery.

Users bypassing interlibrary loan, particularly

students and professors who have institutional access,

reveal their preferences for a different method of ful-

fillment that is simpler and often faster.

This study

seeks to analyze our competitor and take a closer look

at the prevalence of #icanhazPDF requests and ana-

lyze them in order to understand user demand and

improve library services.

ics of #icanhazpdf to suggested rules for #icanhazpdf

request structure. Of those 824 requests—74 were

retweets by the original user or other accounts and

were thus excluded from the pool for further analy-

sis. The authors tracked down the full citation infor-

mation for each item requested, and recorded it in a

shared spreadsheet. While they made every attempt

using the limited information available, 14 requests

were unable to be fully captured. In most cases these

requests were links with no other information, and

the links were parsed through university proxy serv-

ers that the authors could not access. For Tweets in

which the authors were able to determine the correct

citation, they recorded the title of the material, jour-

nal title if applicable, publication date, publisher, and

content format (journal article, book chapter, etc.).

While some users did not supply location infor-

mation, 378 of the 475 users who had requested items

had entered in an identifiable geographic location

such as a city, state, province, or country associated

with their Twitter profile. Many more users included

location information that was facetious such as “the

internet” or “everywhere” and these results were dis-

carded from country analysis.

As Priem and Costello concluded that Twitter

use was not dominated by any one discipline, the

authors were interested in seeing if similar patterns

exist among #icanhazPDF requests. Web of Science

subject categories were chosen as a level of analysis

because they are reviewed regularly by experts and

are non-hierarchical.

Web of Science’s classification

of journals with multiple subject categories provides

a clearer picture of what disciplines are represented

in #icanhazPDF requests by mirroring the often in-

terdisciplinary nature of scholarship. The authors

searched for the journal titles in the Web of Science

publication index, and if available collected the “Re-

search Domain” information for the title. Since some

journals were not indexed in Web of Science, the au-

thors also searched Ulrichsweb.com and recorded the

Ulrich’s subject classifications. With only 4 months’

worth of #icanhazPDF requests it was also necessary

to group the disciplines into larger categories to get

Methods

The deletion of the original requesting tweet after

fulfillment makes gathering information difficult. To

solve this issue of data collection, the authors explored

several tweet archiving services and settled upon using

Tweet Archivist (https://www.tweetarchivist.com/).

Tweet archivist is relatively affordable, and most im-

portantly, it captures tweets in real-time automatically

by hashtag, solving the deletion conundrum. The au-

thors activated Tweet Archivist and captured 1,238

publically available tweets using #icanhazPDF from

the end of April through the beginning of August

2014 and did not include other less common varia-

tions of the hashtag (such as #icanhasPDF). Private

tweets are only accessible to friends of that user, so

while there are also likely private tweets using #ican-

hazPDF during this time frame, the authors did not

have access to that data. Twitter Archivist captures the

full text of the tweet, user name, geographic location,

the number of followers the user has, time stamp, and

language of the tweet.

Of the 1,238 tweets collected, 824 made a request

for material either through partial citation informa-

tion or a link to the original source. The remaining

tweets ran the gamut from discussions over the eth-

March 25–28, 2015, Portland, Oregon

a clear picture of the disciplines represented. The au-

thors then categorized the collected subject catego-

ries into the four larger categories of: Life Sciences &

Biomedicine, Physical Sciences, Technology, Arts &

Humanities, and Social Sciences using the Research

Areas chart in the Web of Science help pages.

Books,

chapters, conference papers, and other miscellaneous

items were not further analyzed by subject in either

Ulrich’s or Web of Science.

Carolyn Caffrey Gardner and Gabriel J. Gardner

Most users (76%) only used #icanhazPDF once

during the data collection dates. However, there were

some repeat users. Not counting retweets, the most

items requested by any one user was 12. This suggests

that for most users #icanhazPDF is not their primary

method of access to scholarly material, but is instead

used for those hard-to-access publications in a one-off

request.

Results & Discussion

Who is Using icanhazPDF?

There were 475 unique users requesting items through

#icanhazPDF during the data collection period. 80%

(378) of users had a Twitter profile with an identifiable

location. This sample is nearly 4 times larger than Liu’s

and confirms her results. The top two countries with

requests were the United States and Great Britain,

with other countries contributing marginally. Taking

Canada and Australia into account demonstrates that

#icanhazPDF is overwhelmingly an Anglophone phe-

nomenon as indicated in Table 1.

TABLE 1

#icanhazPDF Requests by Countries

Countries

United States

Great Britain

Germany

Canada

Australia

France

Netherlands

Brazil

Sweden

Chile

Number of Requests

128

110

When Were the Requested Articles

Published?

The authors hypothesized that #icanhazPDF requests

might be originating because of publisher embar-

gos for the most recent items at a user’s institution.

While the most requested publication date for items

was the current year, 2014, there were more histori-

cal materials than anticipated. Items from the cur-

rent year, 2014, only made up 34.5% of all requests

with publication dates. Of further surprise, the past

five years (2009-2014), only brings the percentage up

to 56.4% of requests (Figure 1).

FIGURE 1

Publication Year by Percentage

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

Publication Year

1921-1989

1990-1999

2000-2008

2009-2013

2014

34.5

13.1

13.9

17.6

20.9

While many #icanhazPDF requests are filled by

devoted followers of the hashtag, these requests reach

much farther than one might expect. The mean fol-

lower count was 1,207, which does not account for

when these tweets are retweeted to ever expanding

networks.

ACRL

2015

Bypassing Interlibrary Loan Via Twitter: An Exploration of #icanhazpdf Requests

FIGURE

#icanhazPDF Request Frequencies

Publication Year

#icanhazPDF Request Frequencies by

by Publication Year

300

250

Number of Requests

200

150

100

The earliest published item requested was from

1921, and item requests were scattered across the 20th

century (Figure 2).

More research which takes into consideration oth-

er forms of “guerrilla open access”, such as article shar-

ing over email or discussion boards, is required before

explicitly endorsing the embargo hypothesis. However,

the data collected for this study suggest that embargos

are not the only reason users turn to #icanhazPDF.

What is Being Requested?

The majority of requests, 89.86% (674) were for in-

dividual journal articles. The remaining 73 requests

were for other materials, which included conference

papers, book chapters, entire books, business reports,

music scores, and ISO standards. The journal article

requests came from 493 unique titles. While most

journals only had one #icanhazPDF request, there

were some outliers.

Nature

had the most requests,

with 16 unique article titles;

Science

followed with 14,

and

Proceedings of the Royal Society B: Biological Sci-

ences

had 7. Large science-focused publishers were

well represented in the requests. Table 2 identifies the

10 most represented publishers:

1921

1933

1951

1953

1956

1962

1964

1966

1968

1971

1974

1976

1978

1980

1982

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

Years of Publication

A majority of the articles requested, 87.4% (589),

were indexed in Web of Science. There were 188 dif-

ferent research domains represented and 34% (78) of

them occurred only once. Likewise, the majority of

requests came for journals that fell within ISI’s cat-

egory for Life Sciences & Biomedicine journals (Fig-

ure 3). Conversely, the Arts & Humanities made up a

TABLE 2

#icanhazPDF Requests by Publisher

Plik z chomika:

hailee

Inne pliki z tego folderu:

CHŁOPI.avi (305080 KB)
DOKTOR JUDYM - LUDZIE BEZDOMNI.avi (825107 KB)
DZIADY - MICKIEWICZ.avi (1345959 KB)
DŻUMA.avi (1214877 KB)
FARAON.avi (1033253 KB)

Bypassing Interlibrary Loan Via Twitter.pdf

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: