by Lawrence M. Sanger
The Digital Universe Foundation
100 Enterprise Way, Suite G-370
Scotts Valley, CA 95066
The ideal information resource would feature high quality of content
(i.e., be accurate and complete) as well as high accessibility
(i.e., excel in availability, ease of use, and interactivity). This
very programmatic paper first describes these various features and
their implications. Then it applies the set of features to some
extant resources, arguing that the ideal information resource does
not yet exist.
The paper speculates that, in the
future, there will be little debate whether a startlingly new and
better information resource is possible, because that much will be
taken for granted; the debate will concern what the resource’s main
features should be.
Aiming to foster this debate, the paper
concludes with a list of topics needed to be addressed to fully
justify an answer to the question,
“What would the ideal information
resource look like?”
Introduction: The Ideal Information Resource
What would the ideal information resource look like?2
Let me begin by clarifying what I mean by this question. By
“information resource” I mean the sort of thing from which one can
get a piece of human-communicated information that has some
reasonable presumption of reliability.3
So consider some main examples of what various information resources
there are: human beings such as librarians and experts; nonfiction
books and all sorts of media, visual, auditory, and otherwise; and
various repositories of these informative people, books, and media,
such as libraries and the Internet.4
When I ask, then, “What would the ideal information resource look
like?” I am asking you to imagine the finest feasible single5
information resource (surely no such resource now exists). It is
not, of course, a single person, because a single person can reach
only so many people, and knows only so much. Nor is it a traditional
library, much less any single book, because only a limited number of
people can access any single library.
Running over these few ideas, it seems clear enough that there are
several conditions that tend to make an information resource better,
whether or not they are jointly possible—although I see no reason to
think they are not jointly possible. The more that an information
resource is complete, accurate, available, easy to use, and
interactive, the better it is, or so I will argue.
These five characteristics of an ideal information resource can, I
think, be usefully grouped under two more general heads, quality and
accessibility, which are individually necessary and jointly
sufficient to capture the notion of an ideal information resource
that I have in mind.
High quality of content is essential because we want an information
resource to provide us the means of gaining knowledge—which is what
I take an information resource to be for. So I say that an
information resource is of a higher quality the more conducive it is
to the gaining of knowledge6
rather than mere uninformed opinion or, say, random unintegrated
data and unassimilated facts.7
Accessibility is essential because an information resource can
fulfill its function of facilitating knowledge only so far as people
can and want to use it to get knowledge. We can say that while
quality ensures that there is something to learn from, accessibility
ensures that people actually learn it. So I am using “accessibility”
in a very broad sense, to include even such features as coherent
organization and engaging presentation. Mere availability is not
While I am reasonably sure of these two broad heads, I am less sure
about precisely what belongs under them. Next, then, I will
elaborate on the above-listed five subheads, or characteristics of
an ideal information resource, with even more characteristics
marshalled beneath the subheads.
Afterwards I will evaluate a few current
information resources in terms of how well they square with this
ideal, and conclude with a list of some issues that need to be
resolved in order to adjudicate the question of what an ideal
information resource would look like.
2. Features of
the Ideal Information Resource: High Quality
There are two features that determine the quality of an information
resource: its accuracy and its completeness.
Accuracy is perhaps the
most obvious requirement, but what is to be counted as
accurate is a thorny issue that would, no doubt, distinguish
different conceptions of an ideal information resource.
free, user-created encyclopedia
Wikipedia has defenders, for
example, who claim that one can do little better than give
all of humanity equal editorial access over the same body of
information—thereby repudiating the notion that expert8
approval is necessary.
But it hardly needs argument (at least not in a paper of
this size) to maintain that some sort of robust expert
involvement and leadership will increase the accuracy of a
resource. It would be more precise (and less controversial)
to say, however, that expert involvement will increase a
resource’s faithfulness to expert opinion. How accurate
expert opinion is as a body is a distinct question—and an
issue, to be sure, that may inform the debate about what the
ideal information resource looks like, as the case of
Wikipedia makes all too clear. But, again, this paper is not
the place to adjudicate that issue.
Still, all hands will admit that experts (under any useful
description) can be mistaken, and this is one reason to
think that a maximally accurate information resource should
be reviewed. Indeed, the more widely reviewed it is—so long
as there is a robust mechanism for the reviews to improve
the end result—the greater the accuracy. We might say, then,
that accuracy increases with distributed editorship.
Finally, even large bodies of experts reviewing a work can
be systematically inaccurate, not only with respect to the
facts themselves (which is obvious enough—consider any
number of once-popular and now-debunked intellectual fads)
but with respect to the full range of expert opinion.9
The ideal information resource,
to represent expert opinion faithfully, would have to be
independent both of any particular group in a given field
and of corporate (including both commercial and
governmental) interests. In short, it would have to be
neutral and fair to the broad range of opinion within a
Otherwise, non-experts could not
trust the resource, even if it were faithful to the facts;
from an outsider’s point of view, there is just a puzzling
disagreement among experts, and a resource that favors one
group and omits the views of another will appear less
valuable than a resource that treats the competing views
more neutrally, fairly, or with equal sympathy.
necessary for high quality.
It seems one needs only add some
missing piece of (“accurate”) information to a general
information resource, and one increases its quality.
Completeness is desirable in an information resource because
a high degree of completeness is necessary for certain kinds
of research. Sometimes only a truly exhaustive resource will
do—as in the case of a patent database, where the decision
to grant a patent depends on the conclusion that no similar
prior art already exists. Without an exhaustive database,
such a conclusion is not possible.
In discussions about Wikipedia
online, some have taken the view that Wikipedia contains too
much information, e.g., about obscure celebrities, video
games, and science fiction universes. I find this view
puzzling; so long as the information really is accurate,
then there seems to be no advantage in excluding it.
I suppose the view reflects an
attitude that seems to be a holdover from the days in which
the length or size of a resource was determined by
publishing and other space concerns. But if the resource is
digital, then of course there is no reason to truncate it
artificially simply on grounds that a subject is rather more
obscure than what can ordinarily be found in a general
encyclopedia. Disk space is cheap.
There is some information, of course, that might be thought
to be harmful to be included in an information resource,
such as libel, pornography, speeches inciting violence, and
instructions for making bombs or weapons of mass
destruction. Another question is whether there ought to be a
database listing every human being on the planet—if such a
thing were possible.
It seems that at some point, the
interest of ordinary persons to their privacy might trump
the interests of humanity in having a truly exhaustive
database. For, even now, if the will of humanity and legal
opportunity were there for it, we could have a public
database listing every human being on the planet; there are
no technological or logistical impediments.
But, as no clear advantage would
come of it, it is very doubtful to me that there will ever
be sufficient will for such a thing.
When I say that the ideal information resource should be
complete, then, I mean complete with regard to at least
responsible (not harmful) and general information. I take it
that this might exclude whole types of information that it
would be an infringement on privacy to include—such as a
complete roster of living humanity—or that it would be
impractical to include.
There are two other things I would like to say about the aim
or goal of completeness: the fulfillment of this aim
requires both many people and many different types of
information. The former first.
Completeness in the
above-described sense clearly requires maximum participation
by both experts and the general public. Editors of
specialized volumes and academic hiring committees know that
frequently only a specific type of person will do, and very
few of that type are available (sometimes it is a type that
has one, or no, members).
Furthermore, as Wikipedia well
demonstrates, the general public far outstrips the expert
community alone in terms of its available time, motivation,
ability to keep much information up-to date, and ability to
write about, say, Star Trek. What we might say is that,
until there is a quorum of expert and public participation,
a general information resource simply cannot aspire to being
nearly as complete as it could be, with vast participation
from the entire spectrum of the educated population and from
across the globe.
The ideal information resource would also be complete with
respect to the range of information types available. It
would be, of course, more than an encyclopedia; it would
include books, website data, curricula (educational
materials of various kinds), raw scientific data, photos,
video, audio, software, and more. Libraries and the Internet
are two possible examples of resources containing such
diverse types of information.
But, as I will explain,
libraries lack accessibility, and the Internet at present
lacks adequate quality.
3. Features of
the Ideal Information Resource: Accessibility
There are at least three distinct features of an information
resource that increase its “accessibility” (in the sense I mean):
availability, ease of use, and an interactive community.
The more widely available a
high-quality resource is, the better it is in terms of the
impact it has. Availability is best achieved by making the
resource both as widespread and as inexpensive as possible.
If the resource is digital and accessible on the Internet,
that will make it as nearly widespread as it can be
(although, given the so-called digital divide, not as
widespread as we would like it to be).
Moreover, if it is open
content—free “as in freedom,” as the phrase among open
source advocates goes11—then
not only is it free of cost, it has the opportunity to be
developed further and to become a self-perpetuating
institution of free knowledge.12
An open content license guarantees availability not only
across income levels and physical space, but across time as
Outside of a few corporate talking heads and curmudgeons,
there has been little opposition to open source and
content, probably because there is no good reason to be
opposed to making freely-distributed information as widely
available as possible—but also because the profits of
proprietary projects have not yet seen much threat from
these projects. It seems unlikely that all of the world’s
information will be open content in the future; as long as
authors, artists, and coders perceive no other viable model
but traditional intellectual property to support their work,
many of them will be opposed to simply “giving away” their
But increasingly large segments
of academe, government, and the general public, whose
livelihoods do not depend on payment for specific pieces of
work, have shown themselves to be perfectly willing to
release their work under a license that makes it as widely
available as possible. This trend is thriving.
Widespread availability imposes another sort of requirement,
namely, that the ideal information resource be made
available in the widest possible range of languages that are
being read online.
Wikipedia has, again, demonstrated one
way how this might be done.
Ease of use.
No matter how high-quality and
widely available it is, an information resource can easily
slide into disuse or obscurity if it is hard to use. (For
example, this is one reason that Google’s simple, effective
search became more popular than Yahoo’s directory
Consider: commitment to quality can create a
valuable resource; making it available lowers the barrier
between the resource and the user; but the ideal resource
must take another step and as it were bring the user toward
One might well say that some
relatively “complete” information resources—large university
libraries and the Internet—are not particularly easy to use.
But precisely the massiveness of an information resource is
at odds with its ease of use, and this surely places some
constraints on how easy an information resource can be to
use. Of course, design that improves based on public (i.e.,
user) feedback can solve some of the more obvious problems.
This is another reason to have an open project: it will give
the public more incentive to help designers with usability
A further difficulty with nearly
all information resources is the fact that they lump
together material for various levels of educational
attainment. This is not the case, of course, with some
encyclopedic and especially some educational resources, but
it is certainly a difficulty with libraries and the Internet
in general. Many a parent has wished for a reliable and
vibrant “Internet for children,” but no such thing exists
In any event, a resource can be
made more useful if it marks off what level(s) of background
a user needs to have in order to appreciate its offerings
“Hidden” information becomes easier to use as it becomes
more findable. Google and other advanced search engines, as
well as online library search tools, have solved this
problem to a certain extent. Moreover, libraries use
cataloging systems, such as the
U.S. Library of Congress system, which allow readers to find
related information serendipitously while browsing the
stacks. But the Web directories of Yahoo! and the Open
Directory Project (www.dmoz.org)
notwithstanding, and with deep respect for their creators’
achievements, there simply is no very useful only taxonomic
Moreover, what does not yet
exist is an information tool that places all different types
of information about a single topic in a single “place” or
“portal.” A new free, collaborative information project, the
Digital Universe (www.digitaluniverse.net),
will attempt to improve on previous resources by placing the
most authoritative information of all sorts in
topic-oriented portals which are, in turn, arranged
taxonomically. Whether the project will succeed at this
ambitious aim remains to be seen.
In addition, information becomes easier to use when its
presentation is made more engaging or attractive. This can
be facilitated by a simple yet well-designed interface, but
what might really enhance the presentation of information is
full integration with multimedia. The easier it is to see
and hear what one would otherwise only be reading about, the
Furthermore, the modeling of
information in explorable 3D spaces—long anticipated, and
now brought to a popular audience by NASA’s
World Wind and
Google Earth—has the potential to completely change the way
we interface with information.14
Building communities around
information resources can grease the wheels of interaction
between users and the resource. Librarians play this role
with respect to their libraries, and, if a school or
university can be considered an information resource,15
instructors play this role with respect to their texts and
other material. Sometimes, having more or less direct
contact with an authority who as it were “stands behind” the
resource is the best way to drive home a point.
As Hubert Dreyfus (2001) well
points out, direct human contact has a way of imprinting
information and values that information imparted at a
distance, over the Internet, cannot. Moreover, digital
communities formed around an online resource can teach each
other enormous amounts. But to ensure that the interaction
is helpful, as befits an ideal information resource, there
must be safeguards against abuse.16
Perhaps the most salutary aspect of interactivity is the
fact that the resource itself can be improved through
interaction, through what I have elsewhere (2005a) called
“radical collaboration.” As the Web goes from “read-only” to
becomes more interactive. This redounds to the benefit not
only of the participants doing the interaction, but also of
the accuracy and completeness of the information in the
resource—and thus benefits future users as well.
Wikipedia is, at least in
theory, probably the best example of this online; but
universities, again if they can be considered information
resources, are perhaps the best illustration of all, as
interaction between faculty and students, and faculty with
each other, leads to the addition to and improvement of the
teaching resources created by the faculty.
4. The Ideal
Information Resource, Summarized
An ideal information resource would, on this account, be an
Internet-mediated project (including but not necessarily limited to
a website) featuring maximum involvement by both experts and the
general public, working together to create the widest range of
information tools—an online library, web directory, encyclopedia,
and so forth—enhanced with much multimedia content.
The project would be free of commercial
influence, and the content would be neutral, maximally free (open
content), arranged into taxonomically-sorted portals, and available
in multiple languages and education levels. Experts would be leading
the effort, but editorship would be widely distributed, and the
public would have ample opportunity for both contribution and
The experts themselves would be made
available to the public, and everyone would be organized into a
community of mutual aid, but with many safeguards against abuse.
I advance this just as a suggestion of an answer, by way of setting
out what I mean by the question,
“What would the ideal information
resource look like?”
A full elaboration and defense of my
answer would require many more pages, and I cannot pretend to have
done any more than simply sketch out and barely motivate one
But I hope that, in having done at least
this much, I have not only elaborated what the question means, I
have gone some way toward explaining why it is important and worth
This paper’s considerations are not merely an impractical
philosophical exercise. Having long since become aware of the
potential of the Internet to work an information revolution, and,
more recently, having become exposed to the power of asynchronous,
distributed collaboration, various individuals and groups have
started a wide variety of collaborative content creation projects.
The media finally discovered this trend
during the 2004 U.S. national elections, as blogs were touted as
significant political tools, and as Wikipedia’s success became hard
to ignore. As a result, more and more people are not just getting
online, but getting together online, to create things.
Increasingly I see evidence that people
are asking themselves:
“If we collaborate online, what is
And more specifically, they ask,18
“How can we use online community
tools, like wikis, to solve our problems or to achieve our
This is a good question to ask, but it
is not the best; there is a question that excites me much more.
The better question is the one addressed in this paper, viz., what
the ideal information system would look like. This is a
philosophical question, and the best answer is apt to take the form
of a set of principles. But in designing their projects, it seems
that many people, rather than follow principles where they lead,
prefer to take their cues from what they regard as models, or
prototypes; then they apply those prototypes to their own problems
or goals, with varying success.
So what I want to do is to consider a
few such prototypes and why they are less than ideal (according to
the principles laid down earlier). The point here is not merely to
be critical, but to illustrate those principles further and justify
why, perhaps, the model described in Section IV above is a natural
development of all of these prototypes.
Let me begin with some more traditional information resources.
Traditional proprietary encyclopedias, such as Encyclopedia
Britannica, are passably accurate but very far from complete.
Indeed, to compete with the amount of information available on the
Internet—albeit, granted, often of dubious provenance—one would need
to add the entire reference section of the library. Furthermore,
subscriptions to digital encyclopedias require an investment that
some households will find prohibitive, and hence such encyclopedias
are not as readily available as one might like.
Traditional multimedia libraries and
archives are considerably larger (usually containing multiple
encyclopedias) and are free of charge. But their primary failing is
that they exist in particular places, and so they are not as
widespread as one might like: they again fail the availability test.
The Internet as a whole is enormous, mostly free of charge, and
available wherever there are computers with Internet connections.
Its main problems concern quality.
Though it is so diverse that one dares
not make generalizations, I will anyway: on average one finds
relatively poor accuracy, and even weakness in terms of
completeness, since (depending on the field) one must sometimes
search very hard indeed to find much of the very specialized
information one can find with ease in a good university library.
Commercial search engines and Web portals, like Google and Yahoo!,
ameliorate these problems only a little. They do often place
high-quality websites on the first page of a search. But Google’s
algorithms— based on the insight that, if a website is linked more
often, then it is better—are a measure of a kind of popularity,
which must not be confused with authority or reliability.
Yahoo! now has a similar search engine,
but it built its reputation on its Web directory, and it is now
becoming increasingly focused on community-building. To my
knowledge, neither in its directory services nor in its
community-building are many relevant subject area experts given any
special role. I see little evidence of such roles being played,
anyway, even if they are.
Furthermore, even with the success of
Google’s search algorithms, if one is looking for a very specific
piece of information, search engines and Web portals may not be able
to find it quickly, easily, or at all. After all, they can find only
what exists online, and very far from all of the information we are
looking for is online.
The Internet makes a rather poor showing
compared to good university libraries (which, naturally, include
terminals connected to the Internet, and thus as it were contain the
Internet), simply because there is more specialized information in
the library. For researchers and diligent students, living at the
library is still the best way to put one in touch with the best
This brings us to Wikipedia, which is perhaps the most instructive
example for us to study when thinking about the question what the
ideal information resource would look like. Wikipedia is a free
online encyclopedia and has recently added its one millionth article
in English, and it boasts over three million in all languages
Its sheer size allows it to make a
credible claim to be solving the specialization problem: a frequent
observation in blogs and news reports about Wikipedia is that, for
some topics, one can find specific information more quickly in it
than through any other source, Google included.19
It is remarkable in that it is written
by its users via a Web technology that allows anyone to edit any
page on the spot—it is a so-called wiki website—and thus it has
developed an interactive community. Wikipedia is also open content,
which allows anyone to use and further develop Wikipedia’s content
free of charge, thus increasing its availability. Because so many
people in the world want to teach each other, and because they are
motivated to do so especially if their work does not profit any
person or business, in this system in which contributing to an
encyclopedia is so easy and quick, Wikipedia has grown its millions
of entries in just five years.
Wikipedia’s parent organization, the Wikimedia Foundation, also
manages a number of other information projects, making use of wikis—such
as collaborative book-writing projects and a collaborative
dictionary. Thus, one might say, the Wikimedia projects are aiming
collectively at something at least like the ideal information
In an incident now well known, however, the distinguished journalist
John Seigenthaler Sr. discovered that defamatory claims were made
about him in his Wikipedia biography. He publicly denounced
Wikipedia in a column (Seigenthaler, 2005) that sparked a firestorm
of controversy. The much-debated question was whether a system so
open to defamation and error could be trusted. In the aftermath, a
news article from the leading scientific journal Nature (Giles,
2005) found that, when 42 articles on scientific topics from
Encyclopedia Britannica and Wikipedia were compared, the Britannica
articles had an average of three errors, and Wikipedia articles,
This investigative news report—for it was not a peer-reviewed
study—was then used by Wikipedia’s defenders to rebut the
increasingly shrill charges that the user-built encyclopedia was
unreliable. Nature had confirmed that Wikipedia was almost as good
as the Britannica, Wikipedia’s defenders said; so all the talk of
its unreliability was just elitist bigotry.
The difficulty with drawing this
conclusion, however, is that there was one simple metric in the
study—“number of errors”—and, more importantly, the articles
examined were all on scientific subjects. It is not the slightest
bit surprising that Wikipedia’s articles on scientific and technical
subjects are relatively good, for the obvious reasons that more
technically-minded people are apt to write for an Internet
encyclopedia and that there are relatively well-agreed facts in
science and technology.
With virtually all other subject
matters—the arts and humanities, for instance—one justifiably
suspects that matters are quite different.
Wikipedia’s entries are indeed, on the whole, quite a bit better
than what one might expect from such a wide-open project. Whatever
one might think of its reliability, it is clear that Wikipedia can
serve as an example to the world of what can be achieved by a very
open worldwide collaboration on a free information resource. Wikipedia’s success demonstrates that much of the ideal outlined
above is possible.
Consider: Wikipedia features extensive
involvement by the general public (and some experts are involved),
and they are working together to create a wide range of information
tools, and first and foremost, an encyclopedia. Editorship is widely
distributed, and the public has ample opportunity for both
contribution and feedback. The project is free of commercial
influence, and the content is or at least aims to be neutral,
maximally free (open content), and available in multiple languages.
This is not the whole ideal; but it is
much of it. So it is not at all surprising that Wikipedia should
have such strong defenders.
Even excellent things can often be improved. To create the ideal
information resource, the Wikipedia model (I do not say Wikipedia
itself 20) would have to be
extended in several ways.
feature expert leadership and
much-expanded expert participation
extend participation to an even
larger cross-section of the educated public than now feels
extend the scope of the projects
to the contents of traditional libraries and archives (with
the Number 2006–1 © 2006 Digital Universe Foundation
contents frozen in reliable versions, not left on wikis to
be further edited)
more closely integrate the
different information projects so that information is sorted
by topic into taxonomically-arranged “portals” about the
feature content at and sorted
usefully into multiple educational levels.
Some of these problems (particularly (1)
and (3)) are addressed by a very exciting, relatively recent
development: the Open Content Alliance.
This appears to be the first serious
effort by a consortium of major libraries, corporations, and other
institutions and organizations to bring the contents of (largely
copyright-free) library contents to a much broader audience. The
current collection—found on
www.archive.org—is wonderful in
point of availability, but not (at present) so much in point of
accuracy, ease of use, completeness: parts of works are missing, the
collection is not quite as deeply searchable or well organized as
one would like, and there are still many holes in the collection.
But surely fixing these problems is only
a matter of time.
What might turn out to be a more serious
(long term) problem with the OCA is its relative lack of
interactivity—i.e., it does not seem there are any plans to allow
either experts or the public to interact with and improve upon the
information in the database in a collaborative way. Although it is
possible to contribute materials to it, it is for the most part
still a read-only Web. It is, after all, an archive—something very
useful indeed, but not the ideal described above.
Making up another sort of information resource are the many
professional and academic Web projects that have sprung up in the
last decade or so, such as:
These are peer-reviewed resources that
excel in representing expert opinion and research. Within their
fields and missions, they are more or less complete (PubMed has a
reputation for comprehensiveness, while the Stanford Encyclopedia of
Philosophy still has many gaps), but virtually none is complete with
regard to the types of information available in their disciplines.
It seems unlikely that a compilation of all professional or academic
Web project contents, taken all together,21
would be any more complete, and would probably be much less so, than
Web content that is not part of any such academic project.
Moreover, as any active researcher
knows, such online resources simply are not a substitute for library
and archive research. Furthermore, while much academic information
online is free of charge and likely to remain so, and thus
adequately available, it is very frequently prepared by academics
for their peers, not for the general public.
Even further, the presentation of
information on academic websites is frequently unexciting and not
particularly easy to navigate; some of these websites even require
registration for access. Hence the information is not particularly
easy to use for most people.
Table 1 summarizes these observations of Part 5 so far. This table
must not be taken too seriously; it is personal and impressionistic.
Certain ratings are explained in the text above and would likely
require significant discussion to explain and justify. For example,
I rate university libraries and academic Web projects a “3” for
accuracy when, of course, very many of the books found in libraries
are full of inaccuracies.
My point in giving these resources these
high ratings is not to say they are sources of objective truth, but
rather to say that they provide the best representation of expert
opinion (some of which is, of course, wrong).
A more informative table would actually
break the five features listed into subcomponents.
An assortment of information resources
and their fit
with the ideal information resource as
Key: 3 = near ideal; 2 = good,
but needs improvement; 1 = poor
As personal and impressionistic as this
table is, it does illustrate a useful point: none of these
information resources is in fact ideal across the board. What the
world needs is a single information resource that has “straight 3s.”
Moreover, I assert—without any attempt
to defend this assertion—that an information resource with straight
3s is feasible. When and if the institutional will to create such a
resource is found, it will come into existence. I am aware of no
compelling technical, economic, or social reason to think that such
a resource could not come into being. In fact, I think that it is
very likely that such a resource will come into being as soon as the
idea of the thing is propounded compellingly to those who can make
In fact, there is an Internet project that has a strong claim to
being at least aimed at this ideal, though it is far from having
reached it: the Digital Universe (www.digitaluniverse.net).
It aims at accuracy in the sense
described above, because it aims to amass large and diverse
governing bodies from every field, led by genuine experts, including
some academic luminaries. Its principals have committed publicly to
being neutral and the project’s managing organizations are, or will
be, nonprofit and noncommercial. It is also committed to building a
large body of contributors, both expert and from the general public,
and to host or link “deeply” to reliable information of every type
that exists online.
So the Digital Universe aims to be a
very high-quality affair.
The Digital Universe also aims at high accessibility. Most of the
information and supporting software will be free (open content and
open source), although there will be some premium services to help
pay for expert labor and for the cost of the platform. There will,
in time, be versions in all major languages online, and at every
level of educational attainment.
Due to its heavy public involvement,
there will be ample opportunity to perfect the user interface, to
make it as easy to use as possible. Information will of course be
searchable, but in addition, in portals devoted to particular
topics, all types of information will be cataloged, with the topics,
or portals, further arranged taxonomically. This Web experience will
be married as much as possible to cutting-edge 3D and multimedia
ways of navigating information.22
Finally, the Digital Universe aims to
build an interactive community that both engages the public directly
and brings experts and expert-vetted information to the public. The
hope in general is that the Digital Universe will be very easy to
use, and more generally, very accessible.
Whether the Digital Universe will be able to bring off such a
high-minded and ambitious affair remains to be seen.
What I think is likely, in any case, is
that some such information resource will come into existence, as its
broad outlines and its desirability become increasingly obvious to
the leaders of academia, education, the Internet, and publishing.
The Coming Debate About the Future of Free Information
I conclude with some speculations about a coming debate over the
future of free information.
As the world, increasingly interconnected, increasingly aware of the
power of digitization, open content, and radical collaboration,
awakens to the dumfounding possibilities before it, I believe the
central question of this paper—“What would the ideal information
resource look like?”—will come to the fore.
Ultimately, the debate will not concern
whether a startlingly new and better information resource is
possible, because in time (perhaps very soon) that much will be
taken for granted; the debate will concern what its main features
should be. Furthermore, I think this debate will turn out to be both
important and global, because humanity in concert, with its relative
shortage of expertise in many specialized subjects, will be able to
support only a small number of “super-projects” aiming at anything
like the ideal information resource envisioned here.
Ultimately the question will become:
what sort of massively collaborative information projects deserve
the support of the intellectuals of the world?
In the discussion above, I rather quickly passed over a number of
extremely difficult issues—not, of course, because I thought they
could be settled so quickly. Indeed, these issues need some careful,
sober, well-informed philosophical thinking.
If this essay has any
lasting value, other than in identifying an interesting
question—what the ideal information resource would look like—I hope
it might be in the identification of the set of issues that need to
be adjudicated in order to justify an answer fully.
This is surely only a partial list:
For purposes of developing an
ideal information resource, should experts be given any
special authority? And, if so, who counts as an expert?
Should we simply try to get clear on what the conventional
notion of what an expert is, or instead attempt to apply
some new conception of expertise? What sort of conception,
if a new one?
Is neutrality or fairness
required for an information resource that aspires to be
ideal? If so, in what does neutrality consist, and how can
it be safeguarded? How widely should the net be cast in
order to fulfill the requirements of neutrality? Should the
full range of expert opinion be represented, or the full
range of all opinion—or something more subtle than either of
To be ideal, should an
information resource be absolutely exhaustive or complete?
Or should it exclude any information on grounds that it is
harmful in some way? If so, harmful in what way(s)? Where,
exactly, should the line be drawn? Also, is there some
information that is simply too trivial or poor quality to be
Are schools and universities to
be considered as information resources for purposes of this
discussion? That is, should our notion of the ideal
information resource extend to education—not just
educational materials, but education itself?
Given both that the ideal
information resource would be free and that the world’s
“knowledge workers” need to be paid for their work, what
financial model can be found for it? Should governments
Should there be a single global
information resource for all languages managed by a single
organization, or should different projects be started in
different languages, with the best examples being
“franchised” under independent management in other
What features in general make an
information resource very easy to use?
Would the ideal information
resource include a “walled garden” of trustworthy,
responsible information for children?
How is information that differs
by subject, type, educational level, and quality best sorted
Are 3D and multimedia “spaces”
improvements on more traditional methods of information
presentation? How important is it that we begin to navigate
information in these ways?
What sorts of safeguards against
abuse are necessary to keep “healthy” a community that is
organizing an ideal information resource?
What concerns about user rights
are relevant, and how are those concerns to be adjudicated?
DiBona, Chris, Danese Cooper,
and Mark Stone, eds. (2005). Open
Sources 2.0: The Continuing
Evolution. O’Reilly Media. Dreyfus, Hubert L. (2001). On the
Internet. Routledge. Giles, Jim (2005). “Internet
encyclopedias go head to head.”
Nature. Published online Dec.
Gillmor, Dan (2004). We the
Media: Grassroots Journalism by the People, for the People.
Jones, Pamela (2005). “Extending
Open Source Principles Beyond Software Development.” In
DiBona, et al., 273–80. Moore, Adam D. (2005). Information
Ethics: Privacy, Property, and Power. University of
Washington Press. Newman, John Henry (1873). The Idea of a
University. Accessible at
Sanger, Lawrence M. (2001).
“Neutral Point of View–Draft.” meta.wikipedia.org.
Sanger, Lawrence M. (2005a).
“Why Collaborative Free Works Should Be Protected by the
Law.” In Moore (2005), 191–206. Accessible at
Sanger, Lawrence M. (2005b.)
“The Early History of Nupedia and Wikipedia: A Memoir.” In
DiBona, et al. (2005), 307–38. A slightly different version
is accessible at
Seigenthaler, John (2005). “A
false Wikipedia ‘biography’.” USA Today. Published online
Nov. 29, 2005. Accessible at
I would like to thank Eric Saudete,
Tereza Sena, and their colleagues and the attendees
of the 2005 Macau Ricci Institute Symposium: “History and
Memory—Present Reflections on the Past to Build Our Future.”
The theme of the day was “Preserving Memory and Teaching
History.” Some ideas from an earlier version of this paper
were presented in a panel discussion on “History and New
Technology,” and this paper is forthcoming in Chinese Cross
A very slightly updated version
of the paper was presented at Purdue University on March 28,
2006. I would also like to thank my colleagues with the
Digital Universe project (with which I am currently
employed) and with Nupedia and Wikipedia for much insight
over the years, which has found its way into this paper.
This problem was relevant to the
topic of the Macau Ricci Institute Symposium in a perhaps
roundabout way. The problem of how best to preserve the
memory of the human race to a large extent overlaps the
problem of what sort of resource is best suited to organize
and access information.
For, after all, the information
that humans produce in a sense constitutes the “memory” of
the human race. The activity of cataloguing, archiving,
chronicling, organizing, and otherwise aggregating
human-generated information thus has the function (among
others) of preserving human memory. If, then, one takes as a
starting-point the question how best to preserve the memory
of the human race, one might well begin by examining what
the ideal information resource would look like.
Thus, thermostats are not
information resources because their data is not communicated
by humans, nor are (arguably) novels taken alone simply
because they make no claim to accuracy. But a database
containing the precise published texts of novels would be an
information resource because there would be some presumption
that it contained reliable versions of texts.
And perhaps schools and
universities, but I will be discounting this possibility in
I do not comment here on how to
individuate information resources, although perhaps I
should. I could say that the limits of an information
resource are determined by where or how it is accessed. Note
that I am willing to consider as “single” information
resources both libraries and encyclopedias (contained in
libraries), and both the entire Internet and specific
Internet websites and projects. But I would not consider
certain disjunctive sets of resources as “single” resources.
For example, I would not say that the set containing the
Britannica and Wikipedia is a “single” resource.
This is not to say that an ideal
information resource would contain only objective truth.
That would not be feasible, and in speaking of the ideal
information resource, I do mean something feasible. Perhaps
the highest practical goal at which an information resource
can aim is the neutral, fair presentation of the entire
spectrum of expert opinion. Note also that an information
resource could be “conducive” to getting knowledge in
various ways—not just in the straightforward way of, for
example, reading and believing.
In drawing the distinction
between a resource that is conducive to knowledge, or even
“wisdom,” and one that is conducive more to unjustified
opinion and unintegrated memorization of facts, I suppose I
am saying that an ideal information resource has an aim
similar in that respect to the aim of university education,
according to John Henry Newman in The Idea of a University
(Newman, 1873), Part I, Discourse 6 (and elsewhere). This is
a distinction that makes a difference; it has various
possible implications for the design of an information
I do not here intend to offer an
account of expertise, but I will say this. The notion of
expertise I have in mind here is a very conventional one,
and is measured by degrees, certifications, and other solid
evidence of attainment—not necessarily in that order. In
fact, more important for the old-fashioned notion of
expertise I am working with is long, focused study,
experience, professional-level conversations that comes
after an in-depth but broad grounding in a discipline. It
hardly needs saying that experts can be wrong. The more
interesting question is to what extent expertise in this
conventional sense improves the probability of a person’s
testimony being true.
One might go further to say that
an ideal information resource should represent the full
range of opinion, period, about a subject, whether “expert”
(on any conception of expertise) or not. This is yet another
fascinating possibility that I do not have space to discuss
The sense of “neutral” I mean
here is unobvious; it is the one I articulated on behalf of
Wikipedia (see Sanger, 2001).
Free software advocate Richard
Stallman is particularly well-known for this locution.
In Sanger, 2005a, I argue that “shopworks,”
or free, radically collaborative works, form the basis for a
new sort of institution that is so valuable that it should
be protected by the law.
There are, of course, “walled
gardens,” websites or services that specifically restrict
access to other websites or services, as well as filters.
What does not exist, however, is an entire network anything
near to the size of the Internet, made specifically for
children. Perhaps that, I am saying, should exist. But I do
not mean to claim that such a thing should replace the
This is another feature of the
Digital Universe planned by cofounder Joseph P. Firmage.
The trouble with such a claim,
however, is that it blurs a useful distinction between
persons qua persons and persons qua information resources.
Persons (qua persons) are essential to a school or
university. A person is not information, although one can
get information from a person. It is more helpful, and does
not blur this distinction, to say that various educational
material would be contained in the ideal information
resource, such as lectures and discussions, or recordings
The role of persons in an
information resource qua information resource is to help
bring people to the information. This might include some
functions very like teaching, but insofar as, in a system, a
person him- or herself is the focus rather than the guide of
learning, the system includes an element of “schooling”
beyond the mission of an information resource. Note that if
a university were to be considered an information resource,
then one might want to offer a much-expanded answer to the
question, “What would the ideal information resource look
And then the question really
goes beyond philosophy of information to philosophy of
This essay cannot hope to
explore the issues that must be addressed to set up a really
healthy community. Some pitfalls to avoid can be found in
the experience with, e.g., Wikipedia (see Sanger, 2005b) and
Groklaw (see Jones, 2005).
See Gillmor, 2004, Chapter 2.
I have been approached many
times over the past few years by people asking just this.
Although, speaking for myself,
when I expect some information to be most quickly located on
Wikipedia, I use Google to search Wikipedia. So in my case,
the claim more precisely stated is that it is more difficult
to find some specific piece of information using Google
without Wikipedia than by using Google to search Wikipedia.
A good search engine, in any case, is essential to finding
Bear in mind that I am not
suggesting that Wikipedia itself actually change in these
ways. I am not sure that, given the self-selecting nature of
its community and its consequent strong commitment to a sort
of intellectual egalitarianism, its community could agree to
And it would be dubious to
consider this a single information resource anyway: such a
consortium of academic projects would have to be created to
make a unified entry point.
As of this writing one must use
a special Web browser to view the Digital Universe, and
registration is required. These are admittedly barriers to
access. Consequently, in the Spring of 2006 a
“browser-neutral” version of the Digital Universe—i.e., a
specially-designed website that will work in any
browser—will launch. 3D capabilities, which now require the
browser, will probably be enabled in the future using
browser “plugins” (software add-ons).