The multiple institutional logics of innovation
Department of Political Science
Department of Public Administration
Department of Politics and History
Liverpool Hope University
Department of Political Science
University of California, Riverside
Department of Political Science
The Ohio State University
How do decentralized systems deal with innovation? In particular, how do they aggregate the
myriad experiences of their component parts, facilitate diffusion of information, and encourage
investments in innovation? This is a classic problem in the study of human institutions. It is also
one of the biggest challenges that exists in the governance of decentralized systems: how do
institutions shape individual behavior around solving problems and sharing information in a
fashion that is reasonably compatible with collective well being? We use a particular
decentralized institution (the House of Representatives), wrestling with a novel problem (how to
utilize the Internet) to explore the implications of three archetypical principles for organizing
collective problem solving: market, network, and hierarchy.
Institutional Entrepreneurship, Collaborative Learning, Information Technology, Congress, Diffusion, intra-, inter-
and extra-organizational influence, online practices.
Acknowledgements: We gratefully acknowledge the support of NSF grant No. 0429452. Any opinions, findings, and
conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of
the National Science Foundation (NSF).
How do decentralized systems deal with innovation? In particular, how do they aggregate
the myriad experiences of their component parts? This is a classic problem in the study of
human institutions— at all scales, from small groups of individuals wrestling with common
problems to policy learning across national boundaries. It is one of the biggest issues in collective
action and thus governance: how do institutions shape individual behavior around solving
problems and sharing information in a fashion that is reasonably compatible with collective well
being? In fact, much of human history can be viewed through the lens of information production
and sharing: how different ways of organizing humans provided reasons for individuals to solve
other people‟s problems (Diamond 1998). Innovation presents a particular governance challenge
within the public sector because the market-based mechanisms (e.g., intellectual property rights)
that have evolved to encourage innovation seemingly do not apply to the public sector. There is
no way for an innovative jurisdiction to profit from the successful adoption by another of its
We use a particular decentralized institution (the House of Representatives), wrestling
with a novel problem (how to utilize the Internet) to explore the implications of three
archetypical principles for organizing collective problem solving: market, network, and
hierarchy (Powell 1990). We find that these three institutional logics intertwine in this particular
case. The market turns out to be a powerful mechanism for fueling innovation, because an array
of vendors has emerged that supplies web services to House offices. These vendors provide the
scale necessary for the development of cutting edge features to House websites. Information
about vendors, in turn, flows through interpersonal networks, although there is relatively little
consultation among House offices about actual features on websites work. There are multiple
hierarchical mechanisms that direct and limit what Members do with their websites. There are
House rules that limit the content of websites, and parties play a limited role in subsidizing
innovative practices. We do find, however, evidence of underinvestment in innovative practice
within offices, with little critical examination of what constituents want and need from Member
websites. That is, while there are powerful drivers of conformity (DiMaggio and Powell 1983)
with respect to Internet practices, the system fails at harnessing the collective capacity of these
offices for problem solving.
PROBLEM SOLVING IN DECENTRALIZED SYSTEMS
Imagine a system where the agents in the system are each playing the same game against
the environment. Because these games are similar to one another, the lessons learned in one
game have implications for the best strategy in the other games. These potential informational
externalities create a governance challenge: how do different principles for organizing collective
human efforts affect the pooling of experiences, create incentives to innovate, and to share the
lessons learned from a particular experiment?
Here we explore the role that three institutional mechanisms for organizing collective
human effort play in the aggregation of problem solving: market, hierarchy, and network (cf
Powell 1990). By market we mean that the problem is solved through an arms-length transaction
between actors with a problem (consumers) and actors who can address that problem
(producers). By hierarchy we mean that the problem is solved through authoritative fiat or
through concentration of power and resources in a system. And we define problem solving in
networks to refer to the informal exchange and flow of favors and information within a given
system, outside the existing hierarchical reporting structures.
The governance challenges with respect to decentralized innovation revolve around the
production, sharing, and aggregation of information:
Challenge 1—Information production: The very possibility that the lessons learned by
one actor would be useful to another means that there may be a mismatch between private
incentives and public benefits to experimentation. Innovations that will diffuse and benefit
others may be underproduced or not shared, because actors do not incorporate those benefits to
others in their calculations. The incentive issue creates a subsidiary issue of scale—where there
may be some innovations that only produce net benefits if there are multiple adopters. To take an
extreme example, imagine an actor could invest c to yield benefits less than c, but where many
other actors could then receive those benefits for free. In the absence of some type of return from
the benefits others receive, it will not be in the interest of any single actor to produce them.
Challenge 2—Information sharing: The fact that innovative practices can be public
goods—where the innovative practices of one actor do not reduce the value those practices offer
other actors, means that communicating those practices can greatly increase collective welfare.
However, sharing sometimes comes with a cost—perhaps quite high, depending on the quality of
information being conveyed. A key question, then, is whether actors share private information
when it would increase overall welfare.
Challenge 3—Information aggregation: Even if information is being shared, it does not
mean that there will be a convergence to optimal array of practices. For any given domain, there
are likely many alternative practices, and a key question will be whether the system of
communication facilitates a convergence toward optimal practices. As the information cascade
literature highlights (Bikhchandani et al. 1992), it is eminently possible that “thin”
communication processes, that do not convey all private information, may result in the
convergence on practices that, if all private information were properly pooled, would be
understood to be suboptimal.
A key question, then, is how do these various institutional mechanisms succeed or fail at
addressing these governance issues. We consider each institutional archetype in turn.
The issue of market failure and information production has been well explored in
economics. Economic models emphasize a few mechanisms to address the issue of
informational externalities. The first is intellectual property protection. That is, if an actor
produces an innovation that produces value for others, intellectual property rights provide a
mechanism for the innovator to gain some rents from other beneficiaries. In principle,
intellectual property protection addresses much of the issue of information production and
sharing, because actors internalize some of the benefits other actors receive from an innovation.
Apple invested in the iPhone because it gained substantial profits from its sales; similarly, it has
not been shy about telling the world about the benefits of the iPhone. Finally, one of the
potentially compelling features of markets is that they aggregate information through price
signals (Hayek 1945).
These mechanisms of governance are seemingly inapplicable to the public sector, because
there is limited capacity to gain rents from successful policies that are emulated elsewhere (e.g.,
Massachusetts cannot charge other states for lessons learned from its health coverage reforms),
and constitutional limits on consolidation (e.g., New York cannot launch a hostile takeover of
New Jersey). However, as the case study below illustrates, for some policies, there is significant
potential for market actors to “join” the informational ecosystem that includes the relevant
government actors, developing and offering innovations that can then be transferred to other
government actors through payment. Indeed, much production of various kinds in the public
sector has been outsourced (sometimes called the hollowing out of the state — Milward &
Provan 2000), where the same private firm serves multiple jurisdictions. The potential of this
market mechanism depends on:
(1) The capacity of actors to contract out production. Certain types of policies can be
contracted out and others not, e.g., because of their complexity (Brown, Potoski, and Van Slyke
forthcoming). For example, implementation of IT policy, to some extent, can be outsourced,
which means that lessons learned from implementation in one location can be utilized by the
contractor elsewhere, providing the incentive to the contractor to innovate if they can capture
some of the resulting rents. This example also highlights that non-government consumers on the
demand side are also part of the broader information ecosystem, because the same contractors are
providing services to both government and nongovernment entities.
(2) Formal or informal mechanisms of intellectual property rights. There will only be an
incentive for contractors to innovate if those innovations are not easily emulated. If another actor
could easily copy what was done in one location, then it is not possible for the original innovator
to capture rents from successful innovations. Such emulation can be blocked through formal
property protection (e.g., through the development of proprietary systems), or through informal
mechanisms (the development of systems that are difficult to copy because of their complexity).
Market-based approaches create their own dysfunctions. The existence of intellectual
property rights creates market power, and necessarily shifts surpluses from consumers to
producers (contractors). An exclusive market arrangement also means that innovation will be
biased toward the development of systems that are protected by formal or informal intellectual
property rights. Some of the support for the development of open source software for
government use has derived from this dual critique (Hamel and Schweik 2009). That is, the open
source software movement may be seen as support for outsourcing to the “commons” rather than
to private sector entities.
As the US system of dual sovereignty (and the case study below) highlights, systems can
have strands of both hierarchy and decentralization. That is, one can embed certain elements of
hierarchy in an essentially decentralized system. Hierarchical governance resolves the scale issue
by, essentially, removing (at least partly) “decentralized” from decentralized decision-making.
There are no informational externalities for a centralized decision maker (Strumpf 1999). Thus, a
hierarchy can consolidate systemic production, potentially eliminating the issues of externalities.
A hierarchy can also subsidize innovation, providing resources to subsidiary actors for
experimentation. Alternatively, a centralized decision maker can authoritatively constrain or
mandate particular behaviors—e.g., mandating that particular units attempt innovative policies
that might not be in their self-interest to pursue. In the US many policy domains have a blend of
these relationships between the federal and state governments, where, for example, with
Medicaid there is a mix of some flexibility with federal mandates and dollars.1
Hierarchies, of course, have their own failings in the creation and dissemination of
information. First, if production is consolidated, the hierarchy is creating an internal monopolist
that may have the same issues around pricing (or efficiency) that a monopolist in the market
would have (Williamson 1971). Second, even in a world where agents confront identical
problems, they might have different information (or perspectives) on what are promising
solutions, thus potentially reducing the exploration of alternative policies. As Brandeis pointed
out long ago, one of the presumed benefits of a decentralized system is that it promotes
experimentation, and subsequent dissemination of successful policies.2 Constraints on the
behaviors of actors would, from this perspective, reduce innovation.3 Third, the central actor
may confront vast informational overload in terms of its capacity to aggregate different
information, especially if peripheral actors have subtly different problems. In fact, this was the
primary Hayekian critique of central planning.
1 Of course, federal-state relationships entail an array of interdependencies beyond policy information (Esterling
2 New State Ice Company v. Liebmann (285 U.S. 262, 311).
3 Analytically, this is a tricky question, since the central actor would presumably have greater capacity to extensively
evaluate policy alternatives than peripheral actors.
The premise of a network approach to innovation is that agents have differentiated
connections and/or finite attention. That is, A can emulate B only if it sees what B is doing.
Visibility depends on the nature of the innovation. Some “innovations,” by their nature, are
publicly visible. This is part of a larger pattern of systems where units can cheaply observe or
refer to particular other units, where that link requires no reciprocal effort.
The literature on networks and governance has emerged in the breach—where market or
hierarchy are clearly missing and an informal system has emerged that apparently addresses the
underlying governance problem. Notably, Ostrom (1990, 2009), for example, has focused on the
role that networks (among other things) regulate behavior in the provision of collective goods in
the absence of a hierarchy. This research has focused on the provision of common pool
resources (e.g. Ostrom et al. 1994), but, as discussed below, much of the logic could apply to the
creation and dissemination of innovative information. The emergence of enforced norms is
likely one key component of information production and sharing—e.g., recognition of helpful
behavior, punishment of actors who do not share information (Mergel et al. 2008).
The economic sociology literature, in contrast, has focused on the role of networks where
markets fail. Much of the work on social capital (in particular, the vein of work following from
Bourdieu (2001) and Coleman (1988) is really about how networks govern dimensions of
exchange that would otherwise fail because of asymmetric information. A substantial literature
has emerged around “network industries”—e.g., Broadway, construction, apparel— where
economic actors are repeatedly reconfigured around existing projects (Uzzi 1999; Jones 1996;
Jones et al. 1997). This literature has focused on the role that (1) relational embeddedness
(repeat interactions with the focal actor); and (2) structural embeddedness (reputation vis-à-vis
third parties) play in regulating the behavior of individuals (Uzzi 1996).
There is also a rapidly emerging literature on public organizations, which has focused, for
example, on the role that networks play in the success of public managers (Meier & O'Toole
2001); as well as the role that network structure plays in the provision of public services (Provan
& Kenis 2008; Milward et al. 2010; Provan & Milward 2001) . Most of this literature, we would
note, focuses on formal networks, rather than emergent, informal networks (Isett et al. 2011) on
which we focus on in this paper.
How do informal networks address the three governance challenges outlined above? Part
of the answer may lie in reciprocity (Axelrod & Hamilton 1981), where dyads of actors with long
run relationships develop cooperative relationships. However, at most, reciprocity can be a small
part of the answer, because the benefits of exchange within a dyad will not reflect the broader
systemic benefits resulting from information sharing. It is not clear how dyadic reciprocity will
encourage investments in innovation that reflect the benefits to the whole system. It is unclear
whether dyadic reciprocity will also reflect the benefits that third parties (tied to one or both
members of the dyad) get when that information is subsequently shared with them. It is possible
that information exchanged within a dyad offers currency for exchange in other dyads, but it is
not at all obvious that this should lead to a healthy equilibrium of information sharing. Finally,
as the information cascade literature highlights, it is not clear that dyadic informational exchange
should yield effective information aggregation. Rather than reciprocity, more promising, is that
informal networks might foster the emergence of pro-social norms that encourages information
sharing (Mergel et al. 2008). Norms that, for example, reward risk taking and innovation with
status; or where information sharing is publicly recognized and seen as appropriate; or where
sharing otherwise private information regarding failure as well as success is encouraged, would
certainly increase innovation, information sharing, and information aggregation, respectively.
As a general proposition, all of these organizational forms coexist, in different degrees.
For example, if you look at one of the classic studies of diffusion—tetracycline (Coleman, Katz,
and Menzel 1957)—all three mechanisms were at work within the system. At its core, this was a
study of the decentralized decision-making of doctors, whose objective was to take care of their
patients. Coleman and collaborators focused on how doctors learned through their network
whether tetracycline was effective. However, not too far in the background was the fact that
there was a market for drugs, with powerful intellectual property rights, which facilitated the
creation of the drug in the first place, and which created an incentive to share information about
the drug (in fact, a recent re-analysis—Van den Bulte and Lilien (2001) — of the Coleman data
suggest that the network had little effect on behavior once one controlled for who was targeted
for marketing). Further, in the background was a legal and regulatory regime (i.e., hierarchy) that
(1) protected those intellectual property rights, and (2) approved of the drug for circulation.
Our objective is to use the emergence of a particular “problem”, the use of the Internet by
members of Congress, to explore how these mechanisms for organizing collective problem
solving coexist. We now turn to a discussion of our methodology.
Case selection and context
The empirical focus of this paper is on the use of the official websites by congressional
offices. There are two primary reasons that we chose this as a setting in which to study
innovation. First, is the novelty of the medium, and second is the decentralized management of
Congressional offices. We discuss each in turn.
Novelty of the medium: Every member of the House currently has an official website.
These websites present a virtual representation of the member to the world, and most notably, to
his/her constituents. The existing rules, such as the Franking rules, focus on the use of paper-
based communication with constituents, so that the Internet is still a relatively novel medium for
offices with a lot of uncertainty to what is acceptable practice.
Given the wide reach of the Internet, these official websites offer a potentially powerful
means to communicate with constituents and the public in general. However, it is also a
relatively novel medium for communication—as of the late 1990s, only about half of the
Members even had a website, and a few years before that the World Wide Web did not even
exist. A casual perusal of websites suggests a substantial degree of convergence, but also
nontrivial variation and experimentation. From our perspective, learning about the power of the
medium is still fresh. This provides the opportunity for us to examine this learning process.
The following screenshots show a) an example of one of the first websites of a Senator
(in this case Senator Edward Kennedy in 1993) and b) an example of a highly developed website
by Representative Mike Honda who used a crowdsourcing approach to include his constituents‟
feedback into the web-design process:
Edward Kennedy: First senator with a website 1993
Figure 1: Sample Congressional websites
Mike Honda‟s crowd-sourced website in 2009
Decentralized nature of Congress: Congressional offices can be considered as 440 small,
functionally identical, and independent public organizations (Salisbury and Shepsle 1981). As
one staffer we interviewed stated: “There‟re 435 small businesses here, and each „CEO‟ can do
what they want.”4 There are some House rules that limit how offices can use their websites, but
within those constraints, Congressional offices collectively come close to matching the
assumption that decision-making is truly decentralized.
There is clearly some interdependence of payoffs in the success of Members—e.g.,
Democrats have a stake in the success of other Democrats and in the failure of Republicans, and
vice versa. However, the payoff a particular Member gets out of the effective use of the official
website by a specific other Member is surely fairly tiny.
We would also note that Members confront somewhat different challenges in
communicating with constituents. The communication needs of a Member from a rural Colorado
district differs from the needs of a Member representing Manhattan. However, all offices are
operating under similar resource constraints, and similar desires to satisfy constituents and
portray a positive image of the Member. Further, much of this heterogeneity is visible both to
the offices and to the researcher; and one key subsidiary question we pursue is how this
heterogeneity affects the search process.
4 There are 435 voting Members, and five non-voting delegates.
We conducted interviews with the Congressional staff person who had primary
responsibility for the official Member website of 99 Members of Congress in the summer of
2006. We were assisted in recruitment by the Congressional Management Foundation, a small
nonprofit organization dedicated to helping Members of Congress better manage their offices.
The sample was constructed purposively to be roughly reflective of the body as a whole, but, due
to the vagaries of who was willing to cooperate, is biased toward affluent urban districts,
Democrats, and offices with above average websites (Table A1 in the appendix compares key
descriptives for the body as a whole to the sample). In our analyses below we examine whether
our observations are robust across subsamples, focusing, in particular, on under-represented
strata. We note where we find significant deviations across subsamples.
The semi-structured interviews each lasted about 45 minutes. The interviews were
transcribed and each statement in every interview coded in an iterative process using the
qualitative data analysis program NVivo 2 (QSR 2002).
At approximately the same time we conducted the interviews we also conducted a survey
of the Communications Directors in House offices. This survey was more broadly about
communication strategy, but did include a few items regarding where offices got information
about what to do with their websites. We received 100 responses from the 440 offices, for a
response rate of 23%.5
Our core research question is to understand the ways in which Congress pools the
experiences of Congressional offices vis-à-vis the use of the Internet. We break this down into
three subsidiary research questions, matching the three governance challenges laid out above:
(1) How do offices learn about which of their practices are successful and which are
How do offices learn from other offices? In particular, how do they learn which
of the practices of other offices are successful and which are not successful?
What institutions outside of offices have a major impact on aggregating
information, provide incentives for certain practices, and set the normative
5 There was an overlap of 53 offices between our two samples.
The first question focuses on feedback from the environment—e.g., do offices gather data
on constituent preferences? Do they track what information web-surfers look at, and what
proportion of hits are from the district? This information has possible positive externalities,
because what is learned by one office potentially has relevance for others. For example, if an
office finds that particular content viewed more than others, this could be beneficial to other
offices if this observation has relevance to what issues they might place on the website. This
information might be transmitted through interpersonal communication, or through the simple
observation of what that office is and is not doing with its website. Finally, what role in
aggregating information do the institutional structures within Congress play? The two obvious
institutions with an interest in the successful aggregation of information are (a) the parties, and
and (b) the administrative infrastructure of Congress.
Our analytic interest is how the system pools together the experiences of various offices‟
use of their official websites. Viewing the Member‟s office as the relevant locus of decisions
about how to use the website, we split our analysis up into (1) environmental feedback; (2)
market-driven influences on decisions regarding websites; (3) hierarchical drivers of practices;
and (4) inter-office (network) flows of experiences.
Innovation: the role of environmental feedback
The Internet is a communication medium. The primary function of an official website is
to facilitate communication with constituents. A key part of viewing the innovations around the
official websites is to understand how offices assess whether a particular intervention was
effective. In this case, do offices find out what works and does not work through feedback from
their constituents? The Internet offers a particular promise for learning from the environment,
because it is possible to track with some detail what visitors to the website are doing (e.g.,
number of hits, pages visited, referrers, etc.).
We therefore evaluate two questions: to what extent do offices (1) proactively assess
what their target audience (generally, constituents) wanted from the website; and (2) ex post
assess whether their website was hitting the target?
The data definitively show that there was remarkably little effort by offices to proactively
assess what constituents wanted. In fact, only two out of 99 offices (both representing districts
with above average educations and incomes) indicated any type of research into what their
constituents wanted, as one of them describes:
We … sent out a survey to about 40,000 constituents… asking whether podcasting was a
feature they‟d use, and whether tele-video conferences, online town halls, all [that] stuff,
what did they want, what should our very limited resources be devoted to. The website was
one of the top ones, without a doubt. Podcasting, in contrast, had a very limited response…
The lack of active proactive research on what constituents want is, perhaps, unsurprising,
given the expense of surveys, focus groups, etc. The monitoring of the use of the website,
however, is much cheaper. Readily available data which are usually automatically collected
include: number of unique visitors and page views; what parts of the website are viewed; the
approximate geographic location of each visitor to the website; who is linking to the website;
where traffic to the website is coming from. It should therefore be easy to produce regular
reports indicating how much traffic the website is getting, and from what part of the district,
looking at which web pages. It is therefore surprising that few offices reported looking regularly
(or even recently) at these data (although the offices, when asked about this, very often noted this
lack with regret). Only four of the interviewees specifically stated that they used web traffic
reports on a regular basis to help determine how they should operate their websites, and two
other interviewees indicated that they collected information about their web traffic but did not
state how they used this information. All other offices stated that they did not get the
information, did not use the information, had not looked at it recently, or did not know the
current specifics about their website‟s traffic. Even those offices that reported tracking what
parts of the website got hits did not examine the data carefully, as this one exchange highlights:
[Q: Do you look at what pages on your website get the most hits?]
[Q: Do you do anything with that?]
We used to do it more, I actually need to find out. (Sigh)
From another office:
We don‟t monitor as closely as I think would be helpful in terms of the Web
traffic. We do an email program that runs separately that I can tell what links
people are clicking on, but we don‟t often go back to the website to see, okay, this
page was the most visited.
In short, assessment of practices with respect to these websites was quite limited. There
was certainly evidence over time of the emergence of novel practices—e.g., new features to a
website—but there was little assessment of whether those practices were effective, even in the
minimal sense of a count of how many times they were used.
As noted above, markets resolve collective problem solving through the development of
proprietary information, which is then monetized and sold (often times in the form of services).
In this case, we did observe that a population of small companies that caters to the web needs of
Congressional offices has emerged during the past decade. Congressional offices are provided
with a fixed budget, with which they have a fair degree of discretion. For some services, they
therefore confront a build or buy decision—do they use staff to build a website, or purchase
expertise to do so? Many offices have pursued the latter route, where a small cottage industry of
firms have sprung up to offer web services to Members, where those firms are divided along
Democratic and Republican (e.g., one Republican oriented firm was called “Right click”) lines.
Fifty-eight percent of the offices in our survey reported hiring an outside consultant, where the
following were typical comments from interviews:
We had a great vendor that allowed us to change our whole front page. We could
do a lot of things internally.
[O]ur vendor came up with the ideas for how exactly to make the tour pages, a lot
of pages, automated. Because when we did that, that was almost five years ago,
so that was when nobody was doing it. Now it‟s pretty standard to have some of
The essential market logic behind the outsourcing of the website is that there are
economies of scale in the problem solving process, where the unit cost of production for many
websites is lower than the cost of producing a single website. In principle, the resulting surplus
can be divided among profit, quality of website, and the ability to offer lower prices to offices.
The effectiveness of the market in part rests on how well information can be protected.
That is, if one vendor comes up with an innovative solution at some cost, and either offices or
other vendors are able to copy that solution cheaply, there will be an underinvestment in
innovation. In this particular domain, protection likely does not come in the form of intellectual
property rights, but through the development of expertise that cannot be easily copied or built
internally. However, that may still leave much relevant information unprotected, as highlighted
by the statement of one clever staffer made use of vendor services in another way:
One of the things that I‟ve seen several external consultants do to pursue our
business is send us a review and a report on our website, sort of analyzed by their
staff and saying this really should be updated, this really should be changed to
meet the standard government configuration or whatever… I then see that and I
say, okay, I think I‟m going to implement these things myself.
In this case, the staffer is essentially free riding on the insights provided by a vendor.
An additional issue with respect to vendors is that there will be a natural push from
vendors to homogenize their products, because it is cheaper to offer a limited set of options, and
in part because they are using off-the-shelf software designed for commercial purposes.
Individual solutions for the different offices are then adapted from the standard set of solutions
plus a personalized configuration. As one individual noted:
Looking at some of the other companies, it sounded like they were trying to sell us
more of like a template of like, what a website would look like, instead of… just
completely designed to us, and how we wanted it.
In short, vendors play a critical role in aggregating experiences and standardizing practice
through their provision of services to multiple offices.
There are two layers of hierarchy within the House; one based on authority, and one based
on power and resources. With respect to the former, the Committee on House Administration
(CHA) sets budgets and rules constraining off-budget behavior (e.g., postings to official websites
were constrained in the last 30 days before the election). They also run the technical
infrastructure of the House, House Information Resources (HIR).6 HIR provides essential IT
support to the House. A part of that support is the hosting of Member websites, as well as basic
consulting on design, providing a number of templates for offices to choose from (at no cost) and
security checks of content that goes up on the website. It is clear, however, that those templates
are seen as somewhat constraining for some offices.
Part of the logic behind a hierarchical model of problem solving, like the market model,
is that economies of scale in the provision of services can emerge. Production for offices may
thus be consolidated at significantly lower cost. However, there is not a mandate that offices use
HIR, creating a mixed model of production, through HIR, vendors, and self-production, where
only 15% of offices in our survey reported using an HIR template.
As a point of comparison, interview responses suggest that while HIR is cheaper (free)
private vendors generally provide more customized service and innovate faster than HIR is able
[HIR] are more involved than we‟d like them to be… for instance, someone found
a typo on her [Member‟s] biography this weekend and I have to email them and
wait for them to change it on that site. Also they just updated our content editor,
which we have yet to make work… also, any time you want to put streaming
media up, or, we can‟t actually put streaming media up, but floor speeches, it has
to go through them.
In short, the hierarchical model, in this setting, does not offer enough flexibility for many
offices, and adds an additional layer of transaction costs.
The second layer of hierarchy is based on the party organization of Congress. Parties
have no direct authority over what Members do with their websites. They do, however, have
what we would label “problem-solving” resources. Earlier analysis of websites from 2003
suggested that Republicans had systematically better websites than Democrats (Esterling, Lazer,
and Neblo 2005). This may have reflected a residual of the high priority that Newt Gingrich put
on technology. Our interviews suggest that while the perception (especially among Democrats)
persists that Republicans put a high priority on Member websites, in fact the Democratic party
made websites a high priority after Pelosi became minority leader. Thus, for example, many
6 See the following website for additional information on CHA (http://www.cha.house.gov).
Democratic interviewees said that under Pelosi Democrats put a lot of energy into developing the
website, HouseDemocrats.gov. Further, in an example of the resources of the parties, the
Democratic caucus provided a customized analysis of every Democrat‟s website. One staff
member described the consequences of this formal analysis as follows:
[A] couple of months ago the Democratic caucus… audited everyone‟s website.
And told you what they thought was wrong with formatting, or telling you what
rules you might have accidentally broken…. So that was a very… individualized
One office that was just starting out used the report to help them get their website off the
ground and followed all the rules laid out by the Party:
[T]hey also gave a document of the highest scoring Democratic websites… the
highest scoring was, [specific MC], and so I looked at his website and I thought it
was very clear. And, as you can tell, ours is really similar. So I just worked with
our HIR consultant, [name], a very good guy, and we created this.
The caucus thus plays a particular role in conveying to Members how to utilize new web-
related innovations. The following interviewee works in the leadership, talks about their role in
conveying lessons about online video and e-newsletters.
I was one of the first to get video up and running on my website…. I took the
initiative…. We held a large meeting…, and I spoke and told them… it‟s really
easy and everyone needs to do it…. And… at six months I was helping offices
left and right copy that function…. We did it more recently with… e-
newsletters… saying… you‟ve got to do them. Such an easy, great, cheap way to
reach all your constituents.
In 2006, the Democrats also played a key role in pushing online town halls, in part by
centralizing and subsidizing online town halls, in part by educating Democratic offices on how to
do town halls:
[T]he online town hall was done through the Democratic office. It‟s cost
prohibitive for every office to have their own online town hall run on their own….
So… that‟s all centralized through the Democratic office.
Yeah, the e-town hall was pretty neat. We worked with the House Democratic
Caucus and they helped us set up the program that made it easy for us to post it.
The Democratic caucus also played an important role in conveying the party‟s message via the
web. Thus, for example, predictably, the audit that the Democrats conducted in part had a party-
[The] survey was more focused on the party message than it was on the individual
site. … So… if we had a link to the Appropriation‟s Committee website, but not
the Democratic Appropriations website, then we had points taken off because we
should have been linking to the Democratic site…
The Democrats also more directly pushed for certain content and messages on Member
websites. Unsurprisingly, the election was playing a big role during this period; a large number
of Democratic offices expressed similar experiences with party leadership encouraging a unified
message for the Fall election:
[T]he Democrats in July and August rolled out their message platform. It‟s called
A New Direction for America…. The main thing was one sheet, a description of
what the new direction for America was. And the leader‟s office encouraged
Members to put it up on the website….
[T]he leadership office sent out a template on the New Direction for America,
and they made an announcement at our press secretary‟s meeting on Mondays,
asking us to immediately put this up on our websites, because Leader Pelosi
wanted all Democrats on message.
I thought that her [Pelosi‟s] office would just keep calling and telling me that I
had to put it [New Direction template] up. (Laugh) But also, you know, it‟s party
unity and people should see what the Democrats are standing for.
More generally, leadership plays a role in providing content to offices, as these comments
I think it‟s gotten better over the past year. The Democratic leadership will offer
suggestions, templates, links…
If you look at our website, there‟s something about veterans on there right now
which came basically straight from Pelosi‟s office. There‟s also something about
cuts to college loan programs, which is another email that was sent to us.
The Republican leadership also played a role in disseminating information regarding Internet
related practices to their caucus, as this example highlights:
The House Republican Conference has a blog seminar, probably twice a year, and I
attended that. and they have bloggers come in from all the different popular websites and
sit down and talk about what they like to see and what they can do. And then, there‟s also
a segment on how to create and start your own blog, which is very helpful.
This example aside, it is clear that at this time, the Republican leadership played a far
less important role vis-a-vis their Members than did the Democratic leadership. Not a single
Republican staffer indicated that the support from their party was an important driver of any
feature on their website. Most staff were not negative about the conference, but even those who
had apparent confidence that there was useful information to be found there did not look, as these
two Republican staff, among the most positive about the support supplied by the leadership,
I never once spoke with [the Republican conference] about developing our
website… I‟m trying to think. I‟m sure at some point in time they… provided
examples of different types. I can‟t remember any specifics… but I‟m sure if I did
they would have given me great tips. But I didn‟t.
The] conference does a really good job… putting out… best practices…. [Q: To
what extent have you gotten insights?] Not much.
Other staffers were less positive regarding what the party had to offer:
[T]he Republican Conference will send out stuff electronically to us…. I‟ve had
this conversation with [them], like ninety-five percent of what you send me is
useless. It‟s too broad. It wouldn‟t work in my district anyway.
I think the Republican party‟s a little bit behind the Democratic party as far as…
their commitment to technology… I don‟t know if they‟ve been the leader that
maybe [other staffer] and I feel that they should be. You know the Republican
party being the leader and having its members really devote resources to internet
communication… in my opinion, I think [other staffer] and I have really had to
take it on ourselves, and it‟s our prior commitment to it, and our commitment to it
that has made us really work hard on this, and not really our party‟s commitment
In short, during the period of our interviews, it is clear that the Democrats were more
proactive in providing both form and content for Member websites than Republicans, and that
they had a real impact on Democratic websites. We would note that the relative importance of
the party to Democrats as compared to Republicans in part reflects the relative popularity of
Democrats as compared to Republicans during this period. In particular, while Democrats were
trying to nationalize the election of 2006, Republicans were trying to keep races local. That is,
identification with the Democratic Party was clearly a plus and with the Republicans a minus in
2006, and this almost certainly affected party strategy and Member receptiveness during this
Interpersonal versus attentional networks
Do congressional offices learn website practices from each other? The relative similarity
of the situations of Members creates substantial potential for informational externalities. This
potential of cheaply copying the practices of other Members is not lost on harried staff, as one
I was not in the business of trying to reinvent the wheel. I just wanted to… go out
and steal the best ideas I could from other people.
There are two pathways by which offices may directly affect each other: interpersonal
communication and passive observation. The particular practices we are studying are, by their
very nature, in significant part public. Member websites are observable to all other offices, and
viewing what particular other offices are doing is quite cheap and convenient. This is a contrast
to many other practices, such as the use of particular types of databases to manage
correspondence, which are essentially invisible to outsiders. We also note that the capacity to
observe is finite—for example, in this case it is only (generally) possible to observe some subset
of other Member websites. What is interesting about Member websites is that while one
component is easily observable (the actual website), there is much information that is not visible,
such as the experiences of the office (e.g., failed experiments that are no longer on the website)
and the management processes underlying what is on the website. These experiences are only
accessible interpersonally—by talking to someone in another office, and engaging in a give and
take about their management practices and experiments. The following interviewee illustrates
searching both through observation and communication:
We looked at every single website…. We probably had [a list of] the top 30 sites.
I individually contacted every single one of those offices, and found out, who do
you use for your website, who does the upkeep of it, who designed it, got all of the
specifics of it…
Unsurprisingly, the vast majority of offices reported looking at other websites for insight.
However, most interviews suggested that interpersonal interactions with other offices were pretty
minimal; the following represents the typical response:
I didn‟t talk to other staff members. I definitely looked at other Members‟ sites
very carefully and tried to see what they were doing that might work for us.
Notably, in the interviews, only one office reported talking extensively to other offices
about internal processes (e.g., website design; how to get content on the website, etc.).
Interpersonal communication was largely focused on identifying vendors, where the following
two responses are typical:
We‟ve looked at basically every other member site there is…. [W]e picked out
the ones we liked the best, and I contacted their staff, and we found out who
their… vendor was.
I mean, I just talked to a bunch of my friends that are press secretaries basically. I
just called them and said hey, who do you guys use, or, you know, is there
anything you‟d recommend? I really like your site, who did you use? Yeah, you
know, basically word of mouth. [Name] did the same, our chief of staff. And he
talked to some other chiefs of staff, in terms of cost and benefits, and customer
These last two quotes illustrate two different strategies for identifying who to talk to. The
first relied on an extensive search of what was observable to guide personal contact. That is, the
passive observation guided interpersonal information seeking. The second indicates a reliance on
friends, that is, a particular type of interpersonal network, friendship, guided interpersonal
In short, search largely takes place with respect to things that are publicly available—
offices looking at each others‟ web practices, but not talking to each other frequently, and rarely
delving into issues beyond vendor-related issues. We do not have a definitive explanation for
this. This may be partly because offices do not look at each other as prime sources of best
practices, as two individuals explain:
[Q: Have you talked to other people around the Hill who are running their
I haven‟t really, because, to be honest with you, there aren‟t a lot of people who
really know web design that well.
I looked around at most of, of the congressional websites on the web and most of
them I kind of found to be relatively unimpressive. You know, I‟ve often heard
the complaint from web savvy people that congressional websites all look the
same. And basically they look very average. They might have some information
on them about what the member is doing, but as far as being visually appealing
they just can‟t hack it with some of the big private sector sites.
Further, there was also evidence of a “not built here” mentality within offices, that a
particular office is so idiosyncratic that lessons regarding management practices are unlikely be
successful (Katz and Allen 1982). As one individual states, “[E]very congressional office just
runs… their organization… in their own way.”
Unsurprisingly, only a tiny fraction of the individuals we spoke with reported looking at
every other website in Congress. Where did offices focus their attention? Some staff reported
randomly browsing other Members‟ websites. Many were more purposive, looking at websites
that had received an award from the Congressional Management Foundation (see discussion
below). A smaller group reported looking at salient other Members, e.g., from nearby districts, or
of prominent Members:
We [looked] at all the neighboring congressional districts… as well as leaders in
Congress...: Nancy Pelosi, the Hoyer site, Denny Hastert…. So we copied a lot
of… the best ideas from other websites.
Another subset explored non-Congressional or non-governmental websites:
I researched other sites, both commercial, Hill, other government entities, other
things in life and you know jotted down ideas… what I liked about certain
things... There are two [that stand out]… One was Senator Kennedy‟s website. In
fact his website was pretty solid at the time… the other was Wine Inspector,
wineinspector.com. They don‟t have a lot of information on their homepage, it‟s
Figure 1 summarizes the distribution of network strategies we observed among the people
we interviewed (based on a coding of their responses to our structured interviews). The heavy
reliance on passive observation—with 62% of interviewees indicating that they looked at other
websites but did not talk to other offices about web practices—is particularly striking.
Figure 3: Search strategies
The relative unimportance of interpersonal networks is striking, where only 26% of
interviewees indicated that they talked to other offices about their websites. This seemed like a
particularly favorable environment for interpersonal networks to matter: the obvious potential
for lessons that would cross offices, the physical proximity of offices, and the potential for
perceptions of solidarity, at least for offices affiliated with the same party. However, there were
other powerful factors that clearly undermined the importance of networks. First, the position of
communication director/press secretary had remarkable turnover—e.g., the median tenure of our
communications directors is about two years (Congressional Management Foundation 2004).7
Second, we saw little evidence that staff saw their responsibilities as transcending the office—
7 These data were gathered by the Congressional Management Foundation's 2004 House Staff Employment Study
(2004). The Congressional Management Foundation sampled 211 offices in the House of Representatives. The
survey gathered data on characteristics of the personal staffers of members of Congress, including the time they
have spent in their current position.
i.e., making the Member look good was vastly more important than making the party look good.
Third, as noted above, many staff felt there was little potential in learning from other offices.
What were the drivers of the attentional networks? The interviews suggest that proximity
of district plays a role, and that the websites of particular Members, either because of the
prominence of the Member, or because the website had been identified as particularly good. In
the survey, we queried respondents about websites that they thought were “particularly good”.
We would view this sociometric construct as capturing the views of respondents as to which
websites were worth paying attention to for their design. That is, the selection of a website as
“particularly good” suggests that the respondent is aware of that website, and views it as a good
model. Below are the websites that were identified more than once.
Table 1: Congressional Members’ websites viewed by others as particularly good (mentioned more
In Table 2 we present the results of a rare events logistic regression to assess the key
determinants of a Member‟s website being mentioned by respondents as „particularly good.‟ As
possible determinants, we included: whether the Member in question was in leadership, the
tenure of each Member mentioned, whether the Member was in the Democratic party, and
whether the target and the subject were from the same state and party. We also included a proxy
for quality, the quantitative grade that each website received from the Congressional
Management Foundation (CMF—see discussion below). We included whether Members were
part of party leadership and tenure in House, because senior Members, and especially Members
in leadership are generally more salient. We also include a dummy for “Democrat” in case there
is a difference in how much Democratic versus Republican websites are paid attention to.
Table 2: Determinants of respondents mentioning a particular Member’s website as
Webgrade of website mentioned
Member mentioned was in party leadership
Tenure of Member mentioned
Member mentioned was a Democrat
Respondent and Member mentioned were of the same party
Respondent and Member mentioned were from the same state
Rare events logistic regression coefficients. Standard errors are included in parentheses.
N=44239 instead of the expected 54400 (all potential website mentions of the 100 survey respondents) because not all
respondents answered the question and a few Members‟ websites were not graded by the Congressional Management
* significant at p<0.001
The above analysis suggests that while quality of the website was quite important in
driving the structure of the attentional network, that a number of institutional factors were
extremely important as well. Unsurprisingly, the websites of Members who were in leadership
were also far more likely to be mentioned (p < .001). State and partisan boundaries were
extremely important (p < .001) in which websites were evaluated as “particularly good”—i.e., the
bulk of websites identified were from the same state and party of the respondent. This is
consistent with a related, quasi-experimental paper, in which we found that the quality of
websites was strongly correlated with the quality of other websites from the same state, and not
proximate congressional districts.
We also have some hints at what the structure of the interpersonal network might look.
We asked in the survey about what other offices they communicated with.8 Only 52 offices were
named, with some respondents naming multiple offices, so caution needs to be exercised in
interpreting the data, but of those 52, 86% were within the same party, and (when combined with
11 responses indicating the state delegation), 60% were within the same state delegation.9 That
is, interpersonal networks were largely determined by party and state.
The emulation process begs the question: How do you know when you are looking at a
good website? How do you know what is good practice on any website? There are virtually no
available objective data on “success” or “failure”—e.g., approval/disapproval numbers from
constituents, traffic etc. The general criteria used by staff were based on their own subjective
experience of a particular website—as one individual stated: “[Y]ou know a bad one when you
see it, and you know a good one when you see it too.” Further, the very presence of a feature on
websites was an endorsement, as reflected in the reasoning of one staffer, “It was obvious
everyone had a kids page, so we should definitely have a kids page...”
We would also note that sometimes practices were emulated simply because they solved a
technical problem that an office was trying to solve, where presumably it was clear whether the
8 The specific item was: “For those Congressional staff that you talk with frequently from offices other than your
own, what offices are they from?” [Emphasis in original question].
9 We also asked about who the Member was friends with, with similar results: of 90 “friends” named, 87% were
same party, and 44% were same state.
solution in question worked or not. How this particular staffer, who was trying to figure out how
to podcast, learned from another website is a case in point:
[T]here was no … huge database of website to explain how to do [podcasting] on
the Internet… The Committee on Government Reform figured out how to do it….
I just looked at their source code and… deciphered how that‟s done just by
studying the code and understanding… the way it‟s structured.
Obviously, in the case of technical issues like this, it is easy to evaluate whether the
solution of a particular website works (e.g., is it possible to play the file on an iPod).
The Congressional Management Foundation was also an important actor in the
informational network, in particular, as a norm setter. Beginning in 2002, CMF began issuing the
Gold Mouse Report, in which it identified best practices, as well as those websites best
complying with those practices.10 The Gold Mouse awards were based on a detailed set of
criteria that CMF developed, in part based on focus groups it conducted, in part in consultation
with staff on the Hill. The influence came through the construction of a set of norms as to what
Members were supposed to do on their websites—the explicit identification of recommended
practices. Thus, for example, many staff, given the responsibility for their Member‟s website,
began by consulting the CMF report:
I basically took the CMF book from a couple years ago [to] see what the Gold
Mouse winners were doing, and used those techniques.
[A]bsolutely. I‟ve been to the CMF website and I‟ve looked to see what their best
practices are or what they say these members are doing correctly… to me, CMF,
they‟re experts in what they recommend, and they look at every single house…
website and they critique them. And to have that information that‟s available
online for free, in my mind you‟d be nuts not to look at it and use it and try to
integrate that into your own website design.
The second theme that came up repeatedly was that the Gold Mouse award created a
competitive environment among the offices to win the award in 2006, as the statements by the
following three staff make clear:
I think, if we don‟t get at least a silver, it‟s off with my head. [laughter]…
Powerful people, they like to compete with other old powerful people, and then
show their little awards.
10 All of the reports are available at www.cmfweb.org.
[T]he Congressman really wants to have a Gold Mouse…. That was part of the
deal when he was going to hire me.
[K]nowing that, the whole Gold Mouse thing was going be coming around again I
wanted to make sure that anything that we did for the Congressman was going to
be very, very content rich, very, very content driven, and as one or two click
friendly for a visitor as possible.
CMF thus has played a key role in influencing the direction of what is perceived as good
practice of congressional websites. CMF set the normative environment, creating an
environment that rewarded or punished staff that performed well based on CMF‟s standards.
How do institutions structure how people and organizations collectively solve problems?
We have examined the multiple types of institutional processes that govern the production and
dissemination of a particular innovation among Congressional offices. Market, network, and
hierarchy have their distinctive logics in organizing human activity, and each plays an important
role in this setting. Offices can be seen as consumers of a service, where a small industry has
arisen to supply their needs. Information about the innovation flows through the network— and
within the multiple networks that exist, more the attentional than interpersonal. Multiple
hierarchies govern office behavior, with both the administration of the House (through CHA) and
the powers within the House (through the parties) playing key roles. The following table
summarizes these findings:
Market Hierarchy Network
capacity in private
actors, from whom
Authority and power Relational influences
Examples Surveys and focus
Tracking usage of
CHA directives and
Party advice/ subsidies
Norms set by CMF;
Gold Mouse Award
evaluation of the
There was very
heavy reliance on
CHA was important in
Democratic party was
more important in
affecting practice than
were very important
CMF played a critical
role in setting the
Table 3: Factors influencing Congressional online practices
Notable as the key drivers of problem solving aggregation were: (1) the emergence of an
array of small firms to serve the specialized market niche of web services for Members of
Congress; (2) the parties—especially the Democratic party—in pushing particular practices; (3)
passive observation in spreading practices among the websites of other Members; and (4) the
informal norm setting role of the Congressional Management Foundation. These institutional
logics interplay, with the interpersonal networks, for example, playing a key role in creating
market share (through reputation) for the relevant firms; and the formal structure of the House
clearly playing a key role in the emergent informal network. Particularly surprising to us was the
dog that did not bark: the relative unimportance of direct interpersonal exchange networks. The
relative importance of the attentional network—where ties are directed from the actor who pays
attention to actors that receive attention—has fairly broad applicability. That is, in a world where
everything is visible, individual capacity to process information may be quite limited relative to
the amount of information available. The pattern of who is paying attention to whom may thus
be viewed as a network structure.11 The social science literature on networks has typically
focused on ties where the information being transmitted is private. Examples of private
informational networks include, for example, sharing of information about employment
opportunities (Granovetter  1995), or lobbyists sharing information with each other about
political or policy information (Carpenter, Esterling, and Lazer 1998). As compared to private
informational networks, the incentive issue discussed above may be exacerbated in attentional
networks, because the innovator has no control over the information created by the innovation,
and therefore reciprocity cannot sustain information development and exchange. However, there
may be particular benefits to an actor providing a model which subsequently receives substantial
attention. For example, in academia, citation patterns may be viewed as an attentional network,
and receiving many citations is considered a mark of success; on the World Wide Web,
similarly, incoming links drive traffic. Thus, the role of norms in providing a reward for
receiving attention is especially important in the case of attentional networks.
This particular case has a number of strengths and weaknesses. The distinctive strength
of this study is that Congress is a well understood institutional microcosm. There is significant
homogeneity in the resources that offices have—eliminating many alternative explanations-- and
the sources of heterogeneity are generally well understood. It is therefore a useful petri dish in
which to study innovation in a decentralized system. However, there are certainly ways in which
the institution and the particular innovation that we study (use of the Internet) are unusual that
limit the generalizability of our findings—particularly to a federal system of decentralized
government. As a point of comparison we compare the results found here to two other studies
that have similarly structured data. The first involves the diffusion of practices among state and
local forensic DNA laboratories (Ref suppressed; Ref suppressed); the second knowledge
sharing among State Health Officials (SHOs—Ref suppressed). Both cases involved a similar
combination of quantitative and qualitative data. A comparison of the findings in this paper
11 Attentional networks likely have distinctive topographies, with, for example, constraints on the number of outgoing
ties and no constraints the number of incoming ties. Thus, for example, while some academic papers have many
thousands of references to them (and most have just a handful), no academic papers contain thousands of
references. By contrast, there are likely significant upper limits in terms of the number of reciprocal friendships an
individual can sustain.
highlights the contextual variables that interplay with the factors that we focused on for this
The most striking difference among the cases is the role that interpersonal networks play.
Our theoretical prior was that the apparent homogeneity and physical proximity of Congressional
offices would lead to networks being more important in this case than in the other two. The
reality was the opposite: interpersonal communication was least important in this case, and most
important for DNA laboratories. A comparison across cases suggests two factors as important.
The first is longevity in position. As discussed above, most individuals in charge of their
websites in Congressional offices had those responsibilities quite briefly. Personnel in DNA
laboratories are typically in a particular position, and in a specific laboratory for many years
(SHOs are in between). The second is longevity in career. The Hill is a starting point for many
careers and end point for few. Except for a small slice of individuals, the bulk of staff visit for a
few years and move on to positions elsewhere. The resources embedded in relationships are
highly perishable, and thus less likely worth the investment. In contrast, personnel in DNA
laboratories are self consciously building task related relationships across the country because of
the long run resources embedded in those relationships. Even if people shift organizations, they
rarely shift careers.
However, attentional networks were far more important in the case studied here than the
other two. This clearly reflects the nature of the innovation, where merely observing what other
Congressional offices are doing with their websites is quite cheap. By comparison, much of what
other DNA laboratories or SHOs are doing is impossible to observe without interpersonal
In all three systems, there are important hierarchical mechanisms that play the functions
of constraining and subsidizing experimentation, as well as disseminating information. In the
case of the DNA laboratories, the FBI has a key authority role, both because it guards access to
the national database system—if local agencies do not adhere to national standards, local profiles
will not be uploaded to the national database. The FBI sets those standards, and also pays for
attendance at an annual conference on the DNA database system. For SHOs, because of the
many policy domains in which they are involved there is a multiplexity of mandates, in which the
SHOs are less directly involved in (likely because subordinates are directly overseeing more
Markets play a key role in the DNA domain, because the tools (e.g., equipment and DNA
kits) used by DNA analysts are complex and expensive. As a result, market actors play an
important role in the DNA informational ecosystem, through support of those tools. This closely
parallels the role that market actors play in the Congressional ecosystem (although on a much
larger scale). In contrast, the SHOs, as managers of large organizations, are somewhat insulated
from information provided by market actors.
While our main focus is on illuminating the presence of multiple institutional logics of
innovation, a natural question is what the joint failings or success of the overall system is. The
research highlights a number of troubling patterns with respect to this particular informational
system. First, there is relatively little feedback into the system regarding success and failure.
From what we observed, there is rather little evidence with respect to what actually works. This,
we would argue, reflects a serious failing of the system, where it is in no single actor‟s interest to
conduct costly experiments from which others might benefit.12 Second, there is relatively little
interpersonal knowledge sharing regarding experiences within the institution. Instead, there is a
reflective mimicry process, with subjective evaluation of others‟ visible practices. The net effect,
perhaps, is a consolidation of practices, but without a strong foundation of information on what
actually works. In fact, these are exactly the conditions under which one might expect
information cascades, because the sharing of public information is high and private information
These observations also point to places for follow up research. Our data collection here
focused on the points of sovereignty in the system—the offices that make decisions as to how to
create their websites. What was not included, generally, was research on other key actors in the
system—the vendors, leadership, and HIR. It is clear that these actors played a key role in
driving the homogenization of the system; what role did they play in addressing the three
governance challenges we outlined initially? Did these actors, for example, aggregate
12 A potential exception might be if the parties or the firms invested significantly in evaluating potential effective
practices. However, no one suggested this was the case, and there certainly would have been an incentive for firms
information, by analyzing the relative effectiveness of different websites? Did they invest
resources, more generally, to evaluate what features of Congressional websites were and were not
effective? The indirect evidence that we gathered, from offices, suggests relatively meager
investments in this regard. Follow-up research can also focus on the second generation of
Internet tools, such as the inclusion of social networking services as innovative parts of
Congressional websites and how these new technologies are integrated into the existing Internet
and parties to reveal such investments in order to persuade offices to purchase new services (for the firms) or adopt
effective, new, practices (recommended by the parties).
Comparison of Sample and House
MC term (average)
Median district income
District % urban
District % Bachelors
Region of country
Bikhchandani, S., D. Hirshleifer, and I. Welch. 1992. A Theory of Fads, Fashion, Custom, and
Cultural Change as Informational Cascades. The Journal of Political Economy 100
Bourdieu, P. 1986. Forms of social capital. In Handbook of Theory and Research for the
Sociology of Education, edited by J. G. Richardson. Westport: Greenwood Press.
Brown, Trevor L., Matthew Potoski, and David M. Van Slyke. forthcoming. Contracting for
Complex Products. Journal of Public Administration Research and Theory.
Carpenter, D., K. Esterling, and D. Lazer. 1998. Strength of Weak Ties in Lobbying Networks:
Evidence from Health-Care Politics in the United States. Journal of Theoretical Politics
Coleman, J., E. Katz, and H. Menzel. 1957. The Diffusion of an Innovation among Physicians.
Sociometry 20 (4):253-270.
Coleman, J. S., E. Katz, and H. Menzel. 1966. Medical Innovation: A Diffusion Study.
Congressional Management Foundation. 2004. 2004 House Staff Employment Study. edited by
U. S. H. o. R. Chief Administrative Officer. Washington D.C.
DiMaggio, P. J., and W. W. Powell. 1983. The Iron Cage Revisited: Institutional Isomorphism
and Collective Rationality in Organizational Fields. American Sociological Review 48
Esterling, K. M. 2009. Does the Federal Government Learn from the States? Medicaid and the
Limits of Expertise in the Intergovernmental Lobby. Publius: The Journal of Federalism
39 (January): 1-21.
Esterling, K. , D. Lazer, and M. Neblo. 2005. Home (Page) Style: Determinates of the Quality of
House Members‟ Websites. International Journal of Electronic Government Research.
Granovetter, M. 1973. The Strength of Weak Ties. American Journal of Sociology 78 (6):1360-
———.  1995. Getting a Job. Chicago, IL: University of Chicago Press.
Hagerstrand, T. 1967. Innovation Diffusion as a Spatial Process. Chicago: University of Chicago
Hamel, Michael, and Charles M. Schweik. 2009. Open-Source Collaboration: Two Cases in the
U.S. Public Sector. First Monday 14 (1).
Isett, K., Mergel, I., LeRoux, K., Mischen, P., Rethemeyer, K. (2011): Networks in Public
Administration Scholarship: Understanding where we are and where we need to go, in:
Journal of Public Administration Research and Theory (J-PART), Minnowbrook III: A
Special Issue, Special Issue Editors: Beth Gazley and David M. Van Slyke, Vol. 21,
Supplement 1, January 2011, pp.i157-i173
Katz, E., M. L. Levin, and H. Hamilton. 1958. Traditions of research on the diffusion of
innovation. American Sociological Review:237-252.
Katz, R., and T. J. Allen. 1982. Investigating the Not Invented Here (NIH) syndrome: a look at
the performance, tenure, and communication patterns of 50 R & D Project Groups. R & D
Management 12 (1):7-19.
36 Download full-text
Mergel, I., D. Lazer, and M. Binz-Scharf. 2008. Lending a helping hand: Voluntary engagement
in a network of professionals. International Journal of Learning and Change (IJLC) 3
Ostrom, E. 2009. A General Framework for Analyzing Sustainability of Social-Ecological
Systems. Science 325 (5939):419 - 422.
Ostrom, Elinor. 1990. Governing the Commons: The Evolution of Institutions for Collective
Action. Cambridge: Cambridge University Press.
Powell, W. W. 1990. Neither Market nor Hierarchy - Network Forms of Organization. Research
in Organizational Behavior 12:295-336.
QSR NVivo 2.0.163.
Rogers, E. M. 1995. Diffusion of innovations. 4 ed. New York.
Rose, Richard. 1991. What is Lesson-Drawing? Journal of Public Policy 11 (1):3-30.
Ryan, B, and N. C. Gross. 1943. The diffusion of hybrid seed corn in two Iowa communities.
Rural Sociology 8:15-24.
Salisbury, R., and K. Shepsle. 1981. U.S. Congressman as Enterprise. Legislative Studies
Strang, D., and S. A. Soule. 1998. Diffusion in Organizations and Social Movements: From
Hybrid Corn to Poison Pills. Annual Review of Sociology 24:265-290.
Strumpf, Koleman S. 1999. Does Government Decentralization Increase Policy Innovation? .
http://ssrn.com/abstract=168210 or doi:10.2139/ssrn.168210.
Van den Bulte, Christophe, and Gary L. Lilien. 2001. Medical Innovation Revisited: Social
Contagion versus Marketing Effort. American Journal of Sociology 106 (5):1409-35.
Walker, J. L. 1969. The Diffusion of Innovations among the American States. The American
Political Science Review 63 (3):880-899.
Williamson, Oliver E. 1971. The Vertical Integration of Production: Market Failure
Considerations. The American Economic Review 61 (2):112-123.