Features

Wiki-Government

How open-source technology can make government decision-making more expert and more democratic.

By Beth Simone Noveck

Tagged GovernancePublic Policy

George Bernard Shaw once wrote, “All professions are conspiracies against the laity,” and nowhere is this more the case than in a democracy. Although political legitimacy demands accountability to an electoral process, those living in a democracy readily submit to what sociologist Michael Schudson calls the permanent embarrassment of expertise. We believe that administrative governance by a professional elite is the best way to organize decision-making in the public interest. Experts decide on acceptable levels of mercury emissions in the air, anti-discrimination rules in education and the workplace, and the standards for cross-ownership of newspapers and broadcasting stations.

The justification for this professional decision-making,
articulated by theorists ranging from Max Weber to Walter Lippmann, is
that while citizens can express personal opinions based on values, they
are incapable of making fact-based decisions on matters of policy. For
Weber, the complexities of modern governance call for "the personally
detached and strictly objective expert." Only institutionalized and
governmental professionals possess the expertise, resources,
discipline, and time to make public-policy decisions. And citizen
participation is hard to organize and administer, and even harder to
scale. It is one thing for 10 bureaucrats to debate a policy and come
to an informed consensus; try getting the same result with 10,000
people–or 10 million.

Now, however, new technology may be changing the relationship
between democracy and expertise, affording an opportunity to improve
competence by making good information available for better governance.
Large-scale knowledge-sharing projects, such as the Wikipedia online
encyclopedia, and volunteer software-programming initiatives, such as
the Apache Webserver (which runs two-thirds of the websites in the
world), demonstrate the inadequacy of our assumptions about expertise
in the twenty-first century. Ordinary people, regardless of
institutional affiliation or professional status, possess
information–serious, expert, fact-based, scientific information–to
enhance decision-making, information not otherwise available to
isolated bureaucrats. Partly as a result of the simple tools now
available for collaboration and partly as a result of a highly mobile
labor market of "knowledge workers," people are ready and willing to
share that information across geographic, disciplinary, and
institutional boundaries.

Consider how communal pooling of knowledge is already at work. An
individual may submit the first "stub" about the history of the Ming
Dynasty or the biography of Winston Churchill in Wikipedia, and a wider
community of millions collaborates on writing, editing, and refining
every article. Wikipedia is open enough to allow expertise to emerge,
but it is also structured enough, with outlines and to-do lists, to set
the rules for a certain kind of group collaboration–and that
collaboration is producing high-quality results.

Or take sites that utilize self-reinforcing "reputation" systems to
improve quality and reliability. These so-called social-networking
sites–like Dopplr for travelers, LinkedIn for business professionals,
or Facebook and MySpace–use community rating and friend-of-a-friend
(FOAF) accreditation mechanisms to build the reputation and trust
necessary to form knowledge groups and communities. Making expertise
relevant for the complex processes of policy-making also requires
forming communities that can collaborate, but it goes beyond that. It
demands "civic networking," tools designed for groups to transform data
into knowledge useful to decision-makers, as well as the concomitant
institutional practices designed to make use of that knowledge.

Political philosophers from Aristotle to Rousseau to Rawls have
suggested that when groups engage in the public exchange of reason,
they produce better ideas. In practice, however, more talk usually
slows decision-making and comes with the attendant problem of
groupthink. Increasingly, however, we are discovering how to use
computers to enable deliberation without endless talk and without
having to be in the same room. And those structures–enforced through
software–are what transform the subjective, free-wheeling, dynamic
expertise of amateurs into effective communities of experts.

For example, the Omidyar Network, the philanthropy launched by eBay
founder Pierre Omidyar, asks the public to participate in awarding its
grants. Rather than invite submissions from thousands of
individuals–which would have strained resources to review–the Omidyar
Network created an online framework for the interested community to
deliberate and winnow the proposals first. Or consider New Assignment,
which was launched to demonstrate that "open collaboration over the
Internet among reporters, editors and large groups of users can produce
high-quality work that serves the public interest, holds up under
scrutiny, and builds trust." The site set forth the social practices to
elicit collaborative reporting (instead of collaborative
gossip-mongering), resulting in the publication of seven original
essays and 80 interviews, as well as a series of stories about
collaborative journalism for Wired magazine. While this fell short of the number of pieces the organizers
had wanted, New Assignment still enabled the "crowd" to produce stories
as good as any found in a national magazine and demonstrated how to
organize (and how not) the process. Similarly, the Sense.us program at
the University of California–Berkeley provides public mechanisms to
allow people across disciplinary boundaries to collaborate in making,
and thereby making sense of, census data graphs and charts. And the
United States Patent and Trademark Office (USPTO) is taking this idea
to the next level: connecting experts directly to actual
decision-making in the "Peer-to-Patent" project.

In each case, we are beginning to see how, if designed with clear,
simple tasks to that help create more open and collaborative yet
organized practices, the computer screen can shift power from
professional sources of authoritative knowledge to new kinds of
knowledge networks. Speaking truth to power is easiest to do–and more
accurate–when spoken not as an individual, but as a group. This has
particular application to policymaking: Non-governmental participants
have something more to offer than voting once a year–namely, good
information. In much the same way that we devise legal procedures to
ensure fairness in the courtroom or open deliberation in Congress, we
can design technology–and the legal and policy framework to support
it–that elicits specific, structured, and manageable input, not from
individuals, but from collaborative groups. If we can harness the
enthusiasm and knowledge of "netizens" to the legal and political
processes generally reserved for citizens, we can produce government
decision-making that is both more expert and, at the same time, more
democratic.

The Problem with Experts

In his award-winning book On Political Judgment, social
psychologist Philip Tetlock analyzed the predictions of those
professionals who advise government about political and economic
trends. Pitting these professional pundits against minimalist
performance benchmarks, he found "few signs that expertise translates
into greater ability to make either ‘well-calibrated’ or
‘discriminating’ forecasts." It turns out that professional status has
much less bearing on the quality of information than we might assume,
and that professionals–whether in politics or other domains–are
notoriously unsuccessful at making informed predictions.

Moreover, the traditional reliance on institutionalized expertise is
fraught with political controversy. Sometimes these pre-selected
scientists and outside experts are simply lobbyists passing by another
name. The current administration, for example, regularly replaces
experts on agency advisory panels with ideologues and political allies.
In a published statement titled Restoring Scientific Integrity in Policy Making, over 60 preeminent scientists, including Nobel laureates and National Medal of
Science recipients, lambasted the Bush Administration for "manipulation of the process through which science enters into its decisions." But if Bush is among the more egregious violators of the presumed wall between politics and institutionalized expertise, his actions only go to show how easy it is for any executive to abuse his or her power of appointment to disrupt experts’ advisory function.

The problem of relying only on professionals in our bureaucracy has
been compounded by institutionalized practices of confidential
decision-making, even in areas where public input could have a decisive
impact, such as the review of patent applications. Whereas a first- or
second-year civil servant at another agency might be drafting memos, a
patent examiner with limited supervision is doling out a patent that
could impact the fate of an industry or fundamental scientific
research. Yet, until 1999, the USPTO, in order to protect the
confidential trade secrets of inventors, did not publish applications.
Even today, most (but not all) applications are only published after 18
months, and even then the examiner cannot communicate with the public.
The USPTO allows third parties to submit publications with no
commentary, annotations, or markings, and only for a fee, effectively
eliminating participation by all but the most die-hard corporate
competitors. As a result, despite a backlog of 700,000 applications
last year (by now it is 800,000 and counting), the USPTO received only
40–100 such submissions, none of which were necessarily used in
decision-making.

But if excessive reliance on governmental or selected professionals
is the problem, direct public participation, as we have typically known
it, is no panacea. Public participation in regulatory decision-making
has been a part of the federal landscape since the end of World War II,
when it was enshrined as a right in the Administrative Procedure Act of
1946. This may sound good in the abstract, but there is such a thing as
too much public participation. Whereas democracy is to be practiced at
the polling booth, bureaucrats are not supposed to feel the pressures
of direct democratic involvement. It can corrupt the rationality of the
process with the distortions of private interests. And more
participation does not mean better participation. How should
administrators deal with individuals who carp but offer little useful
information to improve decision-making? Or interest groups that
electronically submit tens of thousands of identical "postcard
comments"?

Scientific peer review provides an alternative mechanism for
oversight and quality control. The practice is widespread in government
grant-making: The National Science Foundation relies on a network of
more than 50,000 reviewers, while the National Institutes of Health
relies on outside review groups and advisory councils from the
scientific community to review over 70 percent of its grant
applications. And the Environmental Protection Agency’s grant-selection
process relies heavily on "science review panels," peer review groups
chosen and managed by an outside scientist.

But peer-review panels, like agency advisory boards and other
professional commissions, are empanelled, not self-selected. Membership
is closed. Even if their deliberations are posted online or made
available after the fact, the public has no say over who participates
and no voice in the process. People have no option to self-select on
the basis of enthusiasm, rather than being chosen on the basis of
profession. Even when not unduly subject to political influence, the
decision as to who participates is based on institutional status. Those
who may have meaningful contributions to make–graduate students,
independent scientists, avid hobbyists, retired executives, and other
consultants in the "free agent nation"–fall outside the boundaries of
professional institutions and status and will of necessity be excluded,
regardless of their intellectual assets.

Peer review is also time-consuming to organize and run. The group
has to be selected, vetted, and approved, and because fights can arise
over membership, conflicts of interest have to be identified and sorted
out. Participants must also be convinced to join. It is, perhaps, in
part because of the work that must go into maintaining a peer-review
system that review generally happens only in limited contexts, and too
late in the process to have the maximum impact on regulatory
decision-making.

Nonprofessional Expertise

In contrast to what we see in government, in many other fields there
is a move away from the preeminence of institutionalized professionals.
Instead, technology aggregates and refines the knowledge of
distributed, non-institutional experts. Patients, not doctors, are
providing medical information to each other about cancer via the
Association of Online Cancer Resources website and its 159 associated
electronic mailing lists. Almost 30,000 citizen-journalists report on
stories for OhMyNews.com. Amazon’s Mechanical Turk project outsources
the work of answering simple questions or doing basic tasks to a
distributed network. YouTube depends on amateurs to post video content.
The Internet Movie Database (IMDb), which offers information about
close to one million movie titles and more than two million
entertainment professionals, started as a collection of movie trivia
submitted by members of two online newsgroups. In South Korea, the
Naver search engine, where Korean speakers answer each other’s queries,
far outpaces Yahoo and Google as the most popular.

But how might this open, online collaboration improve governmental
decision-making? How do we square the expertise and talent we see
emerging via Wikipedia or YouTube with the professional standards of
science to which we expect governmental decisions to conform? After
all, Wikipedia has been known to contain errors and defamation. For
every brilliant art film or newsworthy clip, there are thousands of
pieces of video junk on YouTube. To put it bluntly, information quality
may not be a matter of national security when it comes to a fifth-grade
book report, but it is essential to nuclear regulations or
environmental safety standards or the quality of issued patents. While
we know that excessive reliance on professionals is problematic, we do
not want to replace one set of abuses with another by eliminating the
professionals and replacing them with direct, popular decision-makers.

Rather, we want to design practices for "collaborative governance,"
shared processes of responsibility in information-gathering and
decision-making that combine the technical expertise of public experts
with the legal standards of professional decision-makers. There are
plenty of people with expertise to share if their knowledge can
successfully be connected to those decision-makers who need it. It is
not necessary to pre-select authenticated and known professionals when
structures can be put in place to ensure that informational inputs are
discernable, specific, well-labeled, and easy to search, sort, and use.
An online system will not be without its own problems and abuses, but
the assumption is that greater public participation, not in setting
values but in supplying information or making sense of and connections
between informational sources supplied by others, can substantively
improve decision-making.

Peer-to-Patent

On June 15, 2007, the USPTO launched an experiment, the
"Peer-to-Patent: Community Patent Review," which could become a model
for precisely this sort of collaborative governance. The program
solicits public participation in the patent examination process via the
Web. This system (the design and implementation of which I direct in
cooperation with the USPTO) allows the public to research and upload
publications–known in patent law as "prior art"–that will inform the
patent examiner about the novelty and obviousness of the invention and
enable her to decide whether it deserves a patent. This is truly
revolutionary: In the 200 years since Thomas Jefferson founded the
patent office, there has been no direct communication between the
patent examiner and the public.

"For the first time in history," David J. Kappos, vice president and assistant general counsel at IBM, says in the Washington Post, patent-office examiners will be able "to open up their cubicles and get access to a whole world of technical experts." With the consent of participating inventors, this USPTO pilot allows the self-selecting public to review 250 software-patent applications from such companies as CA, Hewlett-Packard, General Electric, IBM, Intel, Microsoft, Oracle, Red Hat, Yahoo, and several smaller firms. The community not only submits information, but it also annotates and comments on the publications, explaining how the prior art is relevant to the claims of the patent application. The community rates the submitted prior art and decides whether or not it deserves to be shared with the USPTO. Only the 10 best submitted prior-art references, as judged on the basis of their relevance to the claims of the patent applications by the online review community, will be forwarded to the patent examiner.

The USPTO pilot program is neither a blog nor a wiki. It is not a
free-for-all for "software patents SUCK!" comments. Rather, the
software is designed to provide a structured environment that solicits
specific information targeted to the decision-making process. Eligible
applications are posted online for review for up to four months. There
is a shared discussion space, but it requires registration and joining
the group committed to reviewing that application. The group can
deliberate about the application’s quality, decide what research needs
to be done, discuss where prior art may be found, and even divvy up the
work of finding it. At the same time, the group has tools to "filter"
comments in the discussion, identifying those that are most important,
contain a request for follow-up research, or are low-quality "noise,"
thus mitigating the influence of the person who simply talks the
loudest or the most. Private-sector websites–from the book reviews on
Amazon to the movie reviews on IMDb to the news postings on Slashdot,
the technology news site–have shown that these community ratings can be
aggregated with surprisingly accurate results to sort and filter
discussion.

While the discussion is designed to foster belonging and community
and offer a space where ideas can be refined collaboratively, none of
the discussion is forwarded to the agency, limiting the opportunity for
undue influence. Instead, only the work that will be most helpful to
the decision-making process, namely the community’s research and the
prior art, are forwarded. Those who join an application community
research the background to the application and share that research,
helping to guide the governmental professional in her work. The
research may help the examiner identify fruitful avenues for her
search, decreasing the work done in that limited 20-hour window or, at
the very least, shortening wild goose chases.

To be sure, the patent examiner still conducts a search. She has all
the same information available to her as before. But now she also has
the results of this "human database." By structuring the request for
feedback, the agency avoids inviting participation it cannot use. And
the public has an opportunity to participate in a way that is directly
relevant to and will shape decision-making.

What’s more, through the software administrators can measure the
level of expertise of public reviewers and thus better understand how
this online participation process shapes that expertise. They can also
measure the impact of public participation on examiner decision-making
and on the resulting quality of the issued patent. Over time, with the
benefit of greater experience and more data, it may become possible to
introduce more refined algorithms for assessing the quality of
information on the basis of the past performance of
citizen-participants. We can also better understand and anticipate the
ways in which those with an interest to do so will attempt to "game the
system" (an unavoidable part of the process) and improve the practice
accordingly.

Driving this pilot program is a combination of public attention to
patent reform and the availability of the technology to do open peer
review online. Patent reform has become the subject of intense, recent
legislative debate in the United States and in Europe. The problem is
not so much with the patent office itself as it is with the explosion
in the number of applications, which has put enormous strain on an old
system, resulting in the issuance of low-quality patents that
subsequently become the subject of expensive and wasteful litigation.
Given the doubling of the number of patent applications in the last
decade, examiners currently have less than 20 hours to review an
application about the most cutting-edge nanotechnology, the latest
genetic bio-science, or the most controversial financial business
method. In that time, they have to search the USPTO’s limited databases
to determine if there is a prior publication that would reveal that the
application lacks the requisite novelty or significance required. Yet
publications–which go beyond traditional scientific journal articles
and include websites, software code, and products–are not all to be
found in this closed database. Examiners must also contend with poorly
drafted applications. There is no legal duty incumbent upon inventors
to do a thorough search of the prior art and submit it to the agency.

The problem is as obvious as it is intractable: It is hard to find
information that is not there. In other words, the patent examiner has
to scour the scientific literature to figure out if a particular
chemical compound has already been invented by someone else, lest she
award the patent to the wrong person who turns it into the next
billion-dollar pharmaceutical blockbuster. And she can’t delay her
research, as it would risk exacerbating a backlog that will soon
approach one million applications.

Rather than going to the experts, the USPTO realized that it is
easier to let the experts find it–and that is precisely what the new
program does. The process is open to anyone. At the same time, just as
Wikipedia’s two million entries are actually maintained by a few
thousand knowledgeable and dedicated die-hards, this patent public
participation program does not need to attract everyone, just everyone
enthusiastic enough to do the hard work of sharing in the burdens of
decision-making. Of course, for those who work or study in a particular
area of innovation, participation should be anything but hard.

It is assumed that competitive self-interest will be one of the
drivers causing people to get involved. Far from distorting the
process, when someone from IBM or Microsoft (both of which have
publicly announced that employees may participate in the program while
on the job) has relevant information to contribute, ultimately the
public benefits. Graduate students and those with deep knowledge but
little status may also be inclined to participate, whether out of moral
indignation that a patent is being applied for in an area of research
that they know or out of the desire to gain status in, and professional
recognition from, their community of interest. Incentives could
eventually be created through prizes and rewards, such as a monetary
bounty from an inventor for those who find the best prior art, or
public recognition (an "Oscar" for public participation) to the
reviewer and the team who contribute the most useful information. It is
also an opportunity to generate interest in and a market for the
invention.

The UK patent office has announced plans to do its own, much-wider
pilot for 10,000 to 15,000 patent applications, adapted to its legal
regime and cultural practices. This will make it possible to begin to
build a global community with expertise and interest and, at the same
time, to test different practices, gather real data about how expertise
can best be obtained in response to this complex subject matter, and
refine the process for patents. It will also provide valuable learning
about how such practices could be adapted for other areas of
governance.

Evolving Institutions

Our institutions of governance are characterized by a longstanding
culture of professionalism in which bureaucrats–not citizens–are the
experts. Until recently, we have viewed this arrangement as legitimate
because we have not practically been able to argue otherwise. Now we
have a chance to do government differently. We have the know-how to
create "civic software" that will help us form groups and communities
who, working together, can be more effective at informing
decision-making than individuals working alone. We know from James
Surowiecki’s book, The Wisdom of Crowds, as well as Simon and Schuster’s new MediaPredict project (which encourages readers to guess which manuscripts will become best-sellers), that technology can be used to aggregate predictions. But Peer-to-Patent is teaching us that we can go beyond the tallying of votes. While the general public has good instincts about value-based decisions and could be engaged betterto identify "big mistakes" (such as egregiously unfair media ownership rules), there are specific people out there who possess specific information about patents or trucks or chemicals whom we can now incorporate into our decision-making.

Nor does it have to stop at patents. Scientists who currently give
their time to review grant applications might be just as willing to
contribute their knowledge to decision-making about the environment,
transportation, nuclear power, and agriculture. Frequent travelers have
useful information to share with homeland security officials about how
to best organize security at airports. Economists, businesspeople, and
lawyers know a great deal about financial markets, securities, and
consumer protection. State Department officials do not possess better
information than select graduate students in computer science about
RFID chips for passports. Immigrants and welfare recipients have
information based on lived as well as learned experience to contribute.

In order to make it possible for ordinary, busy people to
participate, and for government to make use of their knowledge,
government must design the practices of public participation to enable
groups, not just individuals, to participate. Agencies must crystallize
the questions they ask of the public and embed those targeted practices
into software. Of course, not even the best-designed civic software is
going to stop government from making bad decisions or
ignoring–willfully or otherwise–good information dropped in its lap.
And there will be manipulation and gaming, to be sure. But these are
not reasons to shy away from opening up the practices of governance,
especially when those practices can evolve over time to respond to
problems that may arise.

Opening up closed decision-making also introduces a greater degree
of transparency and accountability than we have had before. For
example, even if Peer-to-Patent does not yield good research or prior
art every time, many eyeballs on the application still encourage the
inventor to do a better job of writing it and produce public debate and
discussion about the application. It drives more information into the
open and encourages the "liberation" of government data, not for its
own sake, but as an enabler for engagement. And public involvement
reminds the agency official that he is working for (and being watched
by) the public.

To bring about the new revolution in governance, the next president
ought to issue an executive order requiring that every government
agency begin to pilot new strategies for improved decision-making. For
example, he or she could require that each agency, as part of their
Semi-Annual Regulatory Agenda delivered to Congress and as part of a
new collaborative governance report to the Office of Management and
Budget (OMB), set forth at least one "Peer-to-Policy" experiment to see
how it could make its decision-making practices more collaborative.

Experimentation with community feedback would provide the impetus
for independent rulemaking agencies, such as the Environmental
Protection Agency, to post questions to the scientific community before
enacting regulations. It would encourage the Department of Labor to
create an online network to solicit the expertise of those with
disabilities. These new opportunities for participation are not limited
to rulemaking activity. The FCC can seek targeted but open advice on
the economics of spectrum policymaking as part of its efforts to decide
which rules to draft. The National Oceanic and Atmospheric
Administration might create an online marketplace for the collaborative
creation of weather maps. The National Institute for Standards and
Technology could award venture funding for scientific innovation and
research, which would benefit from public collaboration and input. With
public collaboration, decisions about everything from health care
initiatives to housing programs to efforts to attract foreign direct
investment could be improved by collaboration with public experts.

And imagine if this executive order were backed by an industry
"X-Prize," a large monetary reward for the private or public sector
entity that designs the best new civic software tools to support more
open decision-making. This X-Prize–or Peer-to-X-Prize–would provide a
matching grant to support agency pilot programs and allow OMB to give
this software to agencies. It might also go to support the development
of tools to be disbursed on a state and local level. The aim would be
to create turn-key technologies that can be deployed by any government
decision-maker to enlist the help of her "public staff" and give her
the tools she needs to send out a short questionnaire, enable
collaborative drafting on a document, invite proposals and suggestions,
or perform any number of useful activities to engage the interested and
expert community. By enabling agencies to try different practices–while
bringing down the costs of procuring technology–we can learn a great
deal about how to construct new and better practices of public
participation.

By being explicitly experimental with new forms of digital
institution-building, we have an opportunity to increase the legitimacy
of governmental decisions. The tools–increasingly cheap, sometimes
free–will not replace the professionals. Technology will not, by
itself, make complex regulatory problems any more tractable, or
eliminate partisan disputes about values. What this next generation of
civic software can do, however, is introduce better information by
enabling the expert public to contribute targeted information. In doing
so, it can make possible practices of governance that are, at once,
more expert and more democratic.

Read more about GovernancePublic Policy

Beth Simone Noveck is the Jerry M. Hultin Global Network Visiting Professor at New York University's School of Engineering and a Professor of Law at New York Law School. She was the first U.S. deputy chief technology officer under President Obama. Her book Smart Citizens, Smarter State will be published by Harvard University Press in 2015. This essay is adapted from "If We Only Knew What We Know: Open Regulatory Review at the FDA," Yale Law & Policy Review (Summer 2014).

Also by this author

Bridging the Knowledge Gap: In Search of Expertise

Click to

View Comments

blog comments powered by Disqus