. 35
( 53 .)


more likely to distill some important point; and because documents
were limited in number, each could more easily be put forth as an
de¬nitive expression of a bureau™s or of¬cial™s point of view. Obvi-
ously the view of bureaucratic life that was available through paper
¬les was partial and biased; technological limitations meant that the
full back-and-forth within organizations was not recorded. On the
other hand, the task of sorting through the documentary record was
simpler, as was the job of holding of¬cials accountable for the content
of any one document. Because the document was dif¬cult to produce,
the statement within it was presumed to be important; because the
stock of documents was limited, an of¬cial™s capacity to dismiss it
as one part of a long chain of communications, or to rebut it with
another contradictory document, was also constrained.
In the digital world, we may have more facts about the internal life
of organizations, but the burden of sense making “ of deciding what
all the back-and-forth means “ is also increased, and put on the shoul-
ders of individuals and organizations outside of government. Some
outside observers may have the capacity to sift through a large pool of
digitized and paper documents, but many do not. Furthermore, the
capacity to use any single document as a “smoking gun”95 “ author-
itative evidence of governmental predispositions “ is also weakened.
How much can one e-mail mean, in a sea of millions? How much
weight can be put on the wording in one draft of a memorandum or
report, if there are a dozen other variations of the same ¬le?

Metadata: a new surveillance tool?
The growth of digitized information is creating comparable problems
for of¬cials working within government agencies. Specialists respon-
sible for managing government documents have spent the last decade
struggling to ¬nd new techniques for preserving this deepening pool
of information. They have been compromised by the unwillingness
of many governments, facing overall spending constraints, to allow
substantial new investment in records management. The risk posed
by inadequate record keeping is now being aggravated by a looming
wave of retirements: As older workers leave, so too will the personal

Liquid Paper

knowledge that helped to overcome inadequacies in document man-
agement. The risk of so-called “knowledge bleed” is substantial.
Within the U.S. civil service, over half of the total number of pro-
gram managers are expected to be eligible for retirement by 2006.96
To improve their capacity to organize and access inventories
of internal documents, many public organizations have begun to
implement Electronic Document and Records Management Systems
(EDRMS). An EDRM system is designed to give structure to unstruc-
tured data. At the core of any EDRM system is a database that is
intended to house any form of unstructured data that is important to
an agency “ any draft report, memorandum, presentation ¬le, spread-
sheet, or e-mail message. When a new document is added to the
database, basic information about the document “ for example, its
title, subject, author, and date of creation “ is also added. This basic
information is known as metadata, and is roughly equivalent to the
data recorded for each book in a library catalog. An EDRM system
ful¬lls two roles. It allows better day-to-day management of docu-
ments for the ongoing business of the agency (a function known
to specialists as “document management”), and it ensures that the
archival tasks of preserving important documents and pruning tran-
sitory documents are handled properly (a function known as “records
EDRM systems are being adopted widely in the advanced democ-
racies. The Canadian government, repeatedly ranked by the consult-
ing ¬rm Accenture as the world leader in “eGovernment maturity,”97
has deployed EDRM systems in over thirty departments and agen-
cies. The British government aimed to establish EDRM systems
in major government departments by 2004.98 The United States
has been slower in implementing EDRM systems, although stan-
dards have been adopted and implementation was made a priority
within the Bush administration™s management reform program.99
The European Union has also developed standards to guide the devel-
opment of electronic records management systems by its member
Some observers have already suggested that EDRM systems can
help to improve transparency in government by making it easier
for agencies to ¬nd and retrieve documents that relate to a request
for information. The Information Commissioner of the Canadian

Blacked Out

province of Ontario says that EDRM systems could reduce wide-
spread problems of delay in responding to information requests.101
Canada™s federal government also says that EDRM systems will help it
to be “truly open and transparent” by providing better tools for locat-
ing documents citizens are entitled to receive under its FOI law.102
In fact, EDRM systems could provide a level of transparency in
government operations that might startle the agencies that are now
deploying the systems. Because EDRM systems encourage the pro-
duction of standardized metadata about documents, and consolidate
that metadata in a central database, they will improve the capacity of
workers within agencies to search the stockpile of agency documents.
What may not be appreciated is that individuals outside the agency
may also be able to access metadata in bulk form, just as they have
done with older government databases.
The revolutionary potential that this may create is best illustrated
by a historical example. One of the long-standing frustrations of open-
ness advocates has been their inability to know whether there are
documents held within a government agency that might be relevant
to their interests “ or to know where those documents might be.
Those advocates point to a practice of the Swedish government that
seems to remedy this problem. Under Swedish law, all public author-
ities must maintain a publicly accessible register of all of¬cial doc-
uments in their possession, including documents that might them-
selves be inaccessible: “It is therefore possible to keep the contents of
a document secret, but rarely its existence.”103 The practice of main-
taining public registers is often cited as evidence of Sweden™s deep
commitment to transparency. There is, however, a critical limitation.
The class of “of¬cial documents” is de¬ned to exclude many inter-
nal working documents “ draft reports, memoranda, and e-mails “
that are often sought through requests under FOI laws in other
By using FOI laws to extract EDRM metadata in bulk elec-
tronic format, and using that data to create their own web-accessible
databases, nongovernmental organizations might go as far as Swe-
den “ and further. An EDRM-derived database could include every
signi¬cant document generated within a government agency “ includ-
ing e-mail traf¬c, internal memoranda, and draft reports. The docu-
ments themselves may not be accessible under FOI law, but the fact

Liquid Paper

of their existence would be publicly known. Outside actors could gain
an unprecedented view of the ¬‚ow of information within the arteries
of the organization. Even rudimentary information about the volume
and subject of newly generated documents might reveal secrets about
agency priorities.
This is not a hypothetical possibility. In 2004, I asked the Treasury
Board Secretariat of the Canadian government “ the agency respon-
sible for oversight of management and expenditure within the federal
government (roughly comparable to the U.S. Of¬ce of Management
and Budget) “ for a download of metadata out of its EDRMS for doc-
uments entered into the database in July of that year. TBS employs an
internally developed database, the Records/Document/Information
Management System (RDIMS), which has become the EDRM stan-
dard for Canadian government departments. TBS is a small agency,
with only about 400 employees. Nevertheless, the volume of docu-
ment production is substantial, with over 4,000 documents being
added to the RDIMS database in July 2004 alone. (The variety of
documents is also broad, comprising eight types of ¬le in twelve pro-
prietary formats.)
The metadata allows a detailed analysis of emerging priorities
within TBS. For example, it is possible to look only at brie¬ngs logged
into the RDIMS and determine what subjects appeared most fre-
quently. Topping the list was a TBS review of “federal institutional
governance,” a subject on which 15 brie¬ngs had been prepared
that month. The brie¬ngs related to a review the Canadian govern-
ment promised would be undertaken by a committee of ministers
in response to scandals over waste and mismanagement. RDIMS
data shows that all documents on this subject were produced by a
select team of analysts operating within the of¬ce of TBS™s Associate
An analysis of e-mail logged into RDIMS for the same month sug-
gests another pressing topic in TBS: a study of government spending
on biotechnology programs being undertaken by the same team of
analysts. A closer look at documents entered into the RDIMS database
by TBS staff who have been publicly identi¬ed (in the government™s
telephone directory) as members of this high-pro¬le review team
tends to con¬rm this view. A large majority of their “operational”
documents “ a catch-all for internal working papers “ relate to the

Blacked Out

studies on “institutional governance” and biotechnology programs.
Insight can also be gleaned from data about the subjects that do not
appear in RDIMS. The team™s entries into the database suggest that
many areas that had been publicly identi¬ed as government priorities
were not occupying its attention that month.105
EDRMS metadata does not simply allow outside observers to track
trends within government agencies. It also allows individuals to pin-
point exactly the government documents that they want to request.
This helps to resolve two basic problems that confront users of dis-
closure laws in many countries: a lack of knowledge about precisely
what documents a government agency might have in its possession;
and a suspicion (sometimes well founded) that of¬cials will contort
vaguely worded requests to exclude sensitive material. Requesters
often attempt to deal with these problems by making very general
requests, like this one sent by a nongovernmental organization to
Canada™s Department of Fisheries and Oceans in 2004:

All brie¬ng notes, handwritten notes, meeting minutes, correspon-
dence and email, or any other communications or documents
whatsoever sent to or received from the Department of Foreign
Affairs and International Trade since June 28, 2000 on topics
related to Alaskan transboundary issues.106

This is (¬guratively, and perhaps literally) a ¬shing expedition “
the sort of request that often exasperates government of¬cials. Obvi-
ously EDRM metadata cannot help with handwritten notes. But with
other kinds of data, it can provide nongovernmental organizations
with a precise inventory of relevant documents “ as well as the unique
identi¬er that can be used to request each document. (See the follow-
ing compilation, “Browsing an ERDM Library.”)
As more EDRM systems come online, government agencies will
begin to grapple with the unexpected potential for heightened trans-
parency through the release of bulk metadata. Of¬cials are not likely
to view the results with equanimity, because the capacity to have
any conversation in con¬dence could be seriously eroded. The fact
that exploratory conversations on particular subjects have begun
would be quickly revealed, and outside actors would be able to
make rapid and precise requests for the documents generated during
that conversation. Of¬cials will be put on the defensive much more

Liquid Paper

Browsing an EDRM Library

By studying EDRM metadata, it is possible for nongovernmental organizations to
track the production of documents within a government agency. In this case, I
sorted metadata for the month of July 2004 from the EDRM system of Treasury
Board Secretariat, a central agency in the Canadian government. The following
shows a partial list of documents in the system that related to the subject “Social
Insurance Number [SIN] and Data Matching.” The trustee is the TBS employee
responsible for the document. In this example, all documents were generated by
employees of TBS™s Chief Information Of¬cer (CIO) Branch. The documents
reveal a bureaucratic effort to revise government policy on the use of Social
Insurance Numbers as a tool for combining personal information in different
government databases. The subject of database integration had aroused intense
public controversy four years earlier. Groups wishing to probe further could make
a precise request for these documents by specifying the EDRM library number on
the left.
Number Title Trustee Application Date
236106 Presentation for the CIO Murray, Terry POWERPOINT 07/05/04
on the Status of the SIN
& Data Matching Review.
237077 Agenda for July 14 Taillefer, MS WORD 07/09/04
Meeting of the Charles
Committee on SIN and
Data Matching.
238454 Notes for discussion with Taillefer, MS WORD 07/16/04
Of¬ce of the Privacy Charles
Commissioner regarding
new uses of SIN by
several institutions
238733 Presentation for the Murray, Terry POWERPOINT 07/19/04
Rescheduled Meeting of
the Committee on SIN
and Data Matching.
237924 Presentation on the SIN & Murray, Terry MS WORD 07/14/04
Data Matching Review for
the Interdepartmental
Privacy and Service
238894 SIN and Data Matching: Taillefer, MS WORD 07/20/04
Questions and Answers. Charles
239651 SIN and Data Matching: Murray, Terry MS WORD 07/26/04
Revised Policy.

Blacked Out


. 35
( 53 .)