[77] in Information Retrieval
Ethics of Digital Librarianship
daemon@ATHENA.MIT.EDU (sao@Athena.MIT.EDU)
Thu Mar 19 13:15:06 1992
From: sao@Athena.MIT.EDU
To: ir-mtg@menelaus.mit.edu
Date: Thu, 19 Mar 92 13:13:58 EST
The following essay was originally from the "Computers and Academic
Freedom" mailing list.
:Andy Oakland
sao@athena.mit.edu
------- Forwarded Message
Date: Wed, 18 Mar 1992 22:29:06 -0500
X-Digest-Sender: "Carl M. Kadie" <kadie>
Message-Id: <199203190329.AA11464@eff.org>
Subject: Computers and Academic Freedom News 02.09 (Digest)
Apparently-To: cafn-mail@eff.org
Computers and Academic Freedom News
Vol. 02, No. 09
- ----------------------------------------------------------------------
From caf-talk Caf Feb 19 00:00:00 1992
Date: Tue, 18 Feb 92 12:51:23 PST
From: Brewster Kahle <brewster@Think.COM>
Subject: Article 9--Ethics of Digital Librarianship
Message-ID: <fyi.199202191413.AA07128@eff.org>
Ethics of Digital Librarianship
Brewster Kahle
Thinking Machines
February 1992
"As digital librarian, you should serve and protect
each patron as if she is your only employer."
As more of us become involved in serving information electronically
to other users, we so-called "digital librarians" must become
conscious of our ethical responsibilities to protect the privacy of
our the users being served. Since computers are being used by many
more people to find answers from diverse information sources, we
librarians that operate these servers are coming exposed to the exact
questions and interests of people we do not know. This information
has power, a power that can be abused and thereby thwart the
usefulness of the tools we promote. In this essay, I will use the
Wide Area Information Server system as an example of a system of
digital librarians to show what information is collected and used.
With this example, I hope to illustrate some of the dangers and help
list some of the rules of etiquette for this emerging class of
information providers.
The Wide Area Information Server (WAIS) system is an electronic
publishing system that allows end-users to ask questions of remote
information sources. The system encourages people to ask questions in
natural language so that the server system can try its best to find
appropriate documents. Therefore the operator of the server can
collect the questions, and importantly, collect what documents the
users thought were worth looking at. This combines to portray exact
interests of the users. While the identity of the user is not trivial
to determine since only the machine that the query came from is
accessible from the server logs, as personal computers become
networked, the identity of the machine will approximate the identity
of the user.
On the positive side, this means that the server operator (the
"digital librarian") can use that data to refine the database and the
search techniques used in the system. On the negative side, this is
exposing many remote operators to private information that may not be
consciously given by the users.
This surrender of information is not new to librarians; and the
responsibility is taken very seriously by the professionals in the
field. Through training in library schools and by an intuitive sense
of ethics, reference librarians do not betray their patron's interests
to others that are curious or devious. This ethical code is not coded
in law as it is with psychiatrists, so these records can be extracted
through subpoena, but this level of demand is usually required to pry the
information from librarians. From the patron's point of view, having
a librarian know what she is interested in can be a great value
because the librarian can help select and route useful information in
the future.
The same type of information is available to the digital librarians of
the WAIS system. I operate the directory of servers in the WAIS
system, and as such, I know what users are requesting access to what
what type of servers. I know, for instance, every time Mitch Kapor
uses the system, and what he asks for (he specifically allowed me to
include his name here). At this point this is not a problem since
few servers are of a personal nature yet, but as the system
grows to include entertainment, employment, health and other servers,
it is easy to imagine the types of information that will be accessible
through operating such a server. Furthermore, I know when particular
users are at their machines, and therefore know where they are and
when.
The abuses possible with this information are often not as direct as
other offenses, but should not be discounted. People will act
differently if they think they are being watched. Most people will
try not to look silly or ignorant in public, and therefore might be
less willing to try something new, to learn about a subject that they
know nothing about. If using a WAIS server feels like raising one's
hand in school, then people will craft their questions more carefully
than if it felt more like browsing through a new book. Often people
say "I have nothing to hide," which may be true, but if a stranger
approaches on the street and knows quite a bit of personal
information, then the innocent will likely take that person more
seriously than if a cold stranger approached. Even with nothing to
hide, most people feel they should who knows what about them. The
personal nature of information access makes distributing collected
questions a bit unnerving.
The information collected by the digital librarians have some different
characteristics from physical librarians which can make abuse easier
and more widespread: more people can be served, these people are often
in other organizations, and the digital librarians rarely have personal
contact with these users. Therefore, the patrons seem further away
and therefore less real as human beings. Since the computer networks
that are being used with WAIS span the globe and span company
boundaries, the information collected can be useful in knowing what is
important to a distant, and possibly competitive group. The lack of
human contact can lead to the decay in social relations as has been
documented in studies of electronic mail where the language and nature
of relations tend to be stripped of grace, etiquette, and often respect
[cite Sherry Terkle]. This detached nature of electronic
interaction might lead librarians to not respect their patrons
interests where they would if they knew them personally.
On the other hand, the information collected from patrons can be very
useful to the digital librarian to refine and enhance the server. An
example of this is a reporter at a financial newspaper. She is in the
business of collecting information from corporate contacts, finding
the trends in that information, throwing out the proprietary details,
and selling it back to that same population. If the reporter
published too many details, then her contacts would not be forthcoming
the next time, and if she sanitized the information to the point of
uselessness, similarly, her contacts would not invest the time.
Therefore, it is precisely the interaction with the users that builds
the information that is sold. This example shows another facet, and
that is value of the contacts invest in the reporter for their own
benefit. The digital librarian is a less extreme case, but still she
is being invested and entrusted with what the users want, and if this
information is misused or not used, then the users will not be as well
served as could be. Thus, the users will want to be able to be
served better by the librarian through feedback on services rendered.
While there are some technological mechanisms to obscure the identity
of the patron, such as encryption and redirection, hopefully these
will only be used in extreme cases. Encryption can be used to protect
packets in transmission and also be used to sign packets so that they
can not be forged [cite Whitfield Diffie]. This can be useful in a
system where the transport media is insecure, such as radio
transmission. Redirection is a server forwarding technique that would
concentrate all the requests from one trusted host so that the
individual requesters are more difficult to determine. Combinations
of these techniques have been contemplated to provably obscure
requesters while still providing accountability for charges, but
hopefully these techniques will not be the norm if most server
operators will act in good faith towards their patrons.
To try to list a code of ethics for this field is difficult since the
technology keeps changing, but I will offer a principle that can be used to
test a code. As digital librarian, you should serve and protect each
patron as if she is your only employer. Therefore each patron should be
served and protected individually. In terms of WAIS, I feel it is safe to
suggest:
* Dont give away user logs except for scholarly use. Consider
sanitizing the records before any transfer is undertaken.
* Take the job of information serving seriously. This means to
provide a consistent, reliable service and represent the service
provided accurately.
* Count on wide use of the information served, for good
uses and bad, so be proud of the information and the collection.
* Completeness is important. Users learn as much from a question
that has no answer as from the ones with answers. This requires a
complete and up-to-date collection.
* Assume that the patron will not know the your affiliations, and
therefore do not tempt patrons to use a service they would regret if
they new more about you.
* Respect your patrons. The opinion that users are "rocks with
arms", as said by a colleague years ago, will not lead you to become a
very helpful digital librarian.
In conclusion, the rewards from being a digital librarian are numerous
and can be evident from notes from users from remote countries and
companies. This electronic publishing revolution allows anyone with a
personal computer and a modem to be a publisher will have far reaching
effects on the structure of our society. Being a good digital
librarian is a concrete way to create a future we all want to live in.
- --
Carl Kadie -- I do not represent EFF; this is just me.
=kadie@eff.org, kadie@cs.uiuc.edu, or (anonymous) ap.3619@layout.berkeley.edu=
------- End of Forwarded Message
