PA as a Stochastic Process...A stochastic process might characterize either an unknown system that is acting on persons and conditions, or a system with random effects that interacts

Anti-Essentialism in Public Administration Conference A decentering tendency has undermined the foundations of public administration theory Fort Lauderdale, FL March 2-3, 2007 Theorizing Public Administration as a Stochastic Process Catherine Horiuchi University of San Francisco Working Paper: Not for attribution or citation with author’s permission Abstract

The capacity of information networks to capture and manipulate ever-larger

streams of globally acquired, real-time data accelerates fragmentation of traditional

public administration protocols, away from managing stable states toward temporary and

permeable framing of governmental and corporate interests. Technologies simultaneously

provide historic opportunities for dissent, individualism, and small-d democratic

movements. Intermittent, overlapping governance – characterized by private

government, small wars, state failures, and opportunistic shifts of power and capital from

public stewardship to private parties – results in ideological or pragmatic retreats from

and progressions of institutional boundaries.

Ephemeral balances rather than negotiated long-term settlements demark the

edges of public and private. This fluidity of realms increasingly affects the duration and

allocation of administrative responsibilities between formerly firmly edged divisions of

local, state, and national governments. The assumption of a static state in public

administration theory does not hold. Government becomes a metaphorical fluid, an eddy

that retains its shape more or less, long enough to become an object of analysis and

action. The new assumption of administrative fluidity invokes a world of measurement

estimating impacts of partially ordered and partially stochastic events. Sensemaking

derives from sophisticated evaluative and probabilistic analyses. Traditional construction

of the field with its assumption of durable governmental operations may no longer be a

best-fit theory for multi-layered, ephemeral states.

Theorizing Public Administration as a Stochastic Process Introduction

A casual observer might be forgiven for thinking academic preparation for

practitioners in the field called public administration involves traversing rudimentary

courses in a loose collection of social sciences and administrative training, grounded in

scholarly studies and practitioner reports of these practices in collective action. These

concepts and practices have been gathered together over the past century in an

occasionally haphazard but generally straightforward fashion, responding to the ebb and

flow of public judgment about the roles and responsibilities of governing institutions.

This constructionist approach to public administration extends to the training of its

professoriate and the development of the field’s curricula.

Public Administration finds its basis in a series of postulates and assumptions of

public governance, first established in the earliest written documents, periodically tested

and revised. Bedrock of these assumptions in modern public administration is that of a

stable state. In its absence, we assume humanity exists in anarchy or anomie. This

stability is assumed to last, at minimum, long enough for any of the field’s trained

practitioners to act and be recognized as a successful manager of public interests through

application of this study of collected concepts. In the one-class-per-field model of public

administration, a typical curriculum skims organizational theory, economics, law, policy

formulation, program implementation, ethics, and management, with a smattering of

statistics and analytic techniques.

The closing decade of the 20th century and the opening years of the 21st century

offer considerable evidence suggesting this assumption of stability results in practitioners

Horiuchi: Stochastic Process 1

who despite this collective study may not respond appropriately in the likely range of

situations a typical administrator will encounter. Professional education and research

based on this implicit postulate does not offer practitioners the best possible preparation

for public participation and service.

Long before the terrorist attacks on September 11, 2001, and accelerated

thereafter in concert with the increasing power and capacity of electronic data systems,

the US government’s law enforcement and national security interests sought the capacity

to capture, catalog, and analyze data related to subjects of investigation. With ubiquitous

systems and networks came the possibility to capture and store for an undetermined

period of time every imaginable bit of information on all possible distinct persons.

The exponential increase in storage capacity and new software programs to

capture and assign meaning to streaming data is reframing the role of government and its

relationship to corporate interests whose cooperation is essential, as much of the data

management infrastructure is privately owned and managed. These and other

technologies of mass data transfer simultaneously provide historic opportunities for

dissent, individualism, and democratic actions. No event is too mundane to be captured

and posted online, potential affecting overlapping government structures, from

neighborhoods to nations-states through the immediacy offered in network

communications.

There would be little impetus to commit so much government planning and

assessment to these technological advances if existing administrative practices worked

superbly in all instances. To the degree these adoptions reflect explicit or implicit

problems, they should be examined to consider whether they improve or worsen


outcomes and democratic processes. One frequent observation is the speed and

constancy of changes which governments must address, that is, a perceived need for

administrative fluidity and responsiveness.

An assumption of administrative fluidity invokes a world of measurement

estimating impacts of partially ordered and partially stochastic events. The use of

computers to assist decision makers and governance operatives in their sensemaking

under variable conditions and time-sensitivity, demonstrates an intention to include as

many known and measurable variables as possible to better model the entire universe of

possible causes and effects.

A sense of fluidity and impermanence is nothing new, particularly in politics,

where officials can rise and fall with bad practice or bad luck. As Lord Palmerston, 19th

century British Prime Minister noted, a state enjoys “… no permanent friends and no

permanent enemies, only permanent interests.” (Lord Palmerston, 1848) These concerns

were less troublesome in the past, however. Modern life has given us tightly-coupled

systems that are not capable of being completely free of unexpected effects. (Perrow,

1984; Taleb 2004)

Public administration theorists and practitioners might consider radical

restructuring, beginning with revision of fundamental assumptions and concepts. This

can lead to the development of a new set of competencies to manage stochastic and short-

lived administrative constructs. I begin by reviewing the concept of randomness, and

continue with a discussion of the problems in existing assumptions. The technical

response to the problem is described, governmental efforts to reduce randomness and

error through complete information. Three cases are considered for the role of


randomness. Lastly, ideas are offered toward restructuring assumptions that might result

in greater explanatory power regarding public responsiveness.

Regarding Randomness, or Stochastic Events and Processes

A stochastic event or instance can be consider a random, uncontrolled input that

affects a process of otherwise fully specified character so that, despite following known

rules and acting within acknowledged constraints, the outcome cannot be accurately

anticipated. A stochastic process might characterize either an unknown system that is

acting on persons and conditions, or a system with random effects that interacts with a

known process in a non-periodic, unforeseeable manner. Central to these descriptions is

the concept of randomness, at least from the point of reference of those acted upon by the

stochastic process. Inherent is a tantalizing potential for reducing seemingly random and

stochastic effects through more complete knowledge.

If a substantial number of events are random, can we argue that public leadership

simply cannot manage to minimize damage to populations? Perhaps, but such an

argument is pointless and fails the test of political legitimacy. Rather, similar to the

Heisenberg uncertainty principle and the Hawthorne studies, let us assume we affect any

system merely through observation and study. If so, the field of public administration

may develop useful knowledge and modify praxis through acknowledgement of and

investigation around these random effects.

Eagle (2005) addresses what he considers an unjust neglect in the study of the

concept of randomness, which he considers a special case of a process’s unpredictability.

This randomness is not mere indeterminism, and a better understanding of the concept as

unpredictability makes it useful in discriminating between various theories. Randomness


is a key concept in chaos theory, and for any dynamic process that is modeled with a

probability component. Human behavior, and by extension human organizational

systems, are also usefully modeled as random processes (Eagle, 752). Limitations on

predictability can be categorized as epistemic, computational, or pragmatic. These

limitations prevent the correct recognition of the state or situation under consideration

and may result in inappropriate response. Reducing these limitations, therefore, should

increase the correct identification of conditions and diminish unpredictable, random,

chaotic, and stochastic events with the concomitant risks associated with operating under

mistaken assumptions.

A random effect does not map uniformly across all units of analysis. Consider

marine accidents. These rare events were investigated in Perrow (1984) and are of

interest in any question of when and why insurance is purchased in case of loss. Any

particular ship captain will most probably experience zero accidents. The sea as a whole,

will most likely experience some loss each year. In these cases, external requirements for

insurance are the most likely method to ensure no uninsured losses. Goulielmos (2004)

tested smaller areas, and found that a non-random number of ships were lost per area

(using areas much smaller than the ocean as a whole) while the number of ships lost per

month remained random, using a new statistical measure for nonlinear dependence

In its data-driven effort to increase surety of security through complete

knowledge, the US government’s main interest is reduce unpredictability. But the act of

measuring creates system changes, and the measurements themselves introduce process

errors. And in developing a computational model of a open system, only a subset and not

the full set of measurable inputs can be included, thereby introducing more uncertainty in


the process outputs of the models, including whether the unpredictable model states

reasonably resemble in nature or number unpredictable or stochastic states in experiential

conditions. Since many governance functions revolve around response to uncertain and

unfavorable events, any irreducibility in randomness indicates bona fide risks to

successful operation or indeed viability. To what degree do existing theory and praxis

offer a protection against randomness?

Existing Theory Offers Incomplete Response to Risks of Random Events

Current public administration theory streams propose governments operate on

specific organizational principles, and determine courses of action using specific

decision-making models. In this section several principles and models are considered for

their effects under randomness.

Public managers must facilitate equitable collective social behavior despite a

rising number of unsteady, unstable states. Public administrators are expected to operate

successfully regardless of exposure to a range of outside mega-forces: geopolitical (wars,

annexations), religious (fundamentalism), economic (the ascendancy of capitalism and

globalization), and physical (climate variation, manmade or not). These external forces

result in internal financial, institutional, and political processes that disrupt the “normal”

flows of government performance. Public administration is not alone in facing these

forces, but may be later than other fields in addressing them beyond a peripheral

recognition. These immensity of these forces is illustrated by concerns raised in other

fields. The question of sustainability underlying research initiatives across a diverse

range of physical and social sciences serves as acknowledgment of the serious risks

facing the general population if these forces are ignored. Indeed, administrative theory


on the question of an unstable and random world, and the unsteady circumstances in

which administrators may operate in a non-deterministic time frame, are of pre-eminent

public interest.

Existing theory and practice in the field of public administration “resolves”

questions of random introduction of new issues by accruing additional specializations

such as emergency management, homeland security, leadership, and diversity, and by

developing models of new stabilities, most recently networks of governance. As an

example, these stable networks supplement existing models of traditional local and

national units, which become placeholders in the study of how managers deal with

contracts for services in a hollowed state. A displacement strategy fits well with an

accretive or conglomerate model of the field, as faculty can be added to teach courses in

the new core public administration concepts, and phasing out less significant faculty and

practitioner skills. Occasionally the field’s programs or entire schools have been

renamed or split off to reflect shifting interests and knowledge gaps, from public

administration to public affairs to public policy and public management. Perhaps

unsurprisingly, even the terms “public” and “administration” vanished in the case of the

University of Southern California’s School of Public Administration merger with its

School of Urban Planning and Development to become the School of Policy, Planning

and Development.

Policy schools provide an interesting case of the consequences associated with a

determination that an essential knowledge cannot be gained through a cursory review of

material, the type of review that is ubiquitous in a public management curriculum

designed to offer students a limited introductions to the deep bodies of knowledge that


underpins collective action. A rationalist perspective infers that more complete

knowledge of a policies design and potential effects could result in better decisions and

improved results. In order to develop the technological/analytical capacities needed for

this model of the policy process, students take additional quantitative courses in

quantitative and evaluation techniques; rather than graduating generalists targeting

positions as public managers, these programs create policy analysts who might become

technical specialists or consultants to governments.

Governance derives most fundamentally from bureaucratic organization, allowing

for predictable collective action. Merton (1957) describes bureaucracy as clearly defined

patterns of activity in formal, rationally ordered social structures overseen by

administrative, positional authority, defined as “the power of control which derives from

an acknowledged status.” (Merton in Shafritz et al., pg. 103.) The negative effects of

bureaucracy develop from three personality dislocations experienced by the trained,

salaried experts who operate under strict categorization: Veblen’s “trained incapacity,”

Dewey’s “occupational psychosis,” and Warnotte’s “professional deformation.” These

human personality responses to the demands of working in bureaucracy result in

“inappropriate responses under changed conditions.” (his italics, pg. 105.) Under the

limitations of human personality, then the unsurprising response of government

bureaucracies to any unexpected and randomly timed situation would itself be random

and unclear, and success in no measure certain.

In a society where administrative controls are few and far away, say, a rural office

hundreds of miles from the state capital, a bureaucratic workforce has long enjoyed a

natural barrier to tightly coupled control systems and should have experienced fewer


incidents of serious personality or behaviors disturbances that might be related to the

bureaucratic model. However, under a new technological regime where every phone call

can be pre-scripted and recorded, and every keystroke cataloged, these remote workers

may fall prey to the same bureaucratic disorders. To the general public, this situation

may be recognized or characterized as “unresponsive” behavior. It is in fact highly

responsive, however it is responding to the demands of the permanent bureaucratic

control records rather than the ephemeral citizen who perhaps will hang up the phone or

exit the office in short order.

The most prevalent models of decision-making do not fare much better when

considering the frequency and inevitability of random events on a systemic reference

scale. The most commonly agreed-upon and applied models, Herb Simon's satisficing

bounded rationality and Charles Lindblom's incrementalism operate under different

assumptions, but each produces sub-optimal results. Using these two models, the

preferred outcome of an inappropriate (“wrong”) decision or no decision would be a

small adverse impact. Indeed, smaller decisions with smaller impacts lie at the heart of

the logic supporting incrementalism. But for certain decisions, in a environment of tight

coupling, adverse effects may well be unacceptable, ranging across a continuum from

small failures to catastrophic. A small failure might be the 2001 California energy

market collapse with an estimated cost of $100 billion paid by taxpayers or ratepayers

over an extended period, 10 to 20 years. The largest involve catastrophic losses, such as

nuclear technology failure, unanticipated escalation of small wars into global events, or

biological annihilation though accident or deliberate acts.

Comprehensive Data Collection as a Method to Address Stochastic Risk


In this setting, seeking complete knowledge as a guard against missteps seems an

obvious choice. If analyzing more data leads to more recognition of risks, perhaps it can

prevent catastrophic outcomes. This section of the paper describes how this data is

captured and catalogued. A conundrum regarding unique identifiers provides a logistical

hurdle. Popular antipathy toward some law enforcement use of data, and concerns about

data security and provides, result in a political hurdle.

Data capture of all information on all persons follows a straightforward method.

The essence of the idea is quite simple: as electronic data streams through a worldwide

network of wires, each bit can be copied as it goes by a collection point. As the data is

collected, or at any later point, computer programs analyze the data and present summary

or detailed information for analyst attention to determine whether machine-targeted

patterned data is suited for specific attention or intervention. A model of this collection is

provided in Diagram 1.

ŅtotalÓ information

multiple-sourced data feeds

Diagram 1: Sensemaking through Comprehensive Information Assessment

This strategy involves data mining, a technological capacity criticized for ethical

concerns by privacy advocacy groups such as the Electronic Frontier Foundation and the


American Civil Liberties Union. More pragmatic concerns also exist; tests of data will

certainly result in failure to reject the null hypothesis in instances where a number of

hypotheses are tested by a number of analysts, regardless of coordination. Large data sets

contain patterns, so the method also invites participant bias (Denton, 1985; Frank, 1998)

Statisticians have developed strategies to address data cleansing (Hernandez and Stolfo,

1998) and use of random sampling within large data sets (Owen, 2003); the former

method requires several passes through the data and the latter method does not

interrogate the full data set. Philosophical questions have been shelved in the main, and

both the commercial world and governments alike develop ever larger and more

sophisticated data mining applications to manipulate these astronomically large data

structures, often using visualization techniques that improve comprehension of large data

sets. (Tufte, 2001; Rogowitz and Treinish, ND)

After passage of the USA PATRIOT Act in 2001, sustained funding and

administrative support flowed into data collection and mining. The “Total Information

Assurance” proposal of John Poindexter was closed under public pressure in 2003,

shortly after it was announced. Its initiatives have in the main continued, and new

databases have been developed and are growing steadily. (Electronic Privacy Information

Center (2005; American Civil Liberties Union, 2004) As one example, the National

Security Administration requested access to AT&T telecom switches so traffic could be

subjected to electronic capture and analysis. According to published reports, some

portion of internationals call traffic was diverted and routed through two large switches

situated in the US to increase access to objects of surveillance.


In the case of Homeland Security analysis of large data sets, any real-time

processes preclude data cleansing. Security concerns also likely limit random selection

of reduced data for patterns, out of concerns that seemingly irrelevant data contains

signals of impending terrorist activity (that is, we must first find the dots, then connect

them.)

Matching specific individuals to each message or data record is an ongoing

challenge. This requires fixed and unchanging identifiers for each person to facilitate

cross-indexing and searches. Most US citizens can be associated with a specific 9-digit

number issued by the Social Security Administration initially designed for use in

collecting payroll taxes and determining retirement benefits. Others with certain

employment rights may apply for a unique tax identification number. Yet the same US

government acknowledges and tacitly facilitates the widespread use of and dependence

upon a substantial underclass of foreign-born labor who have entered the country with no

official authorization to live or work in the US. Regarding this community, the California

journalist and author Peter Schrag has asked, “Is the border still a line, or is it now a

region?” (Schrag, 2007) Once this population moves through the border region and starts

working, two strategies predominate. The worker and employer can jointly agree to an

unreported labor agreement, where the employer pays no payroll or other taxes, or the

worker and employer can use (and reuse) a valid Social Security Administration number

(SSN) that uniquely belongs to another person. Reports on reuse suggests many numbers

are used dozens or hundreds of persons; retired workers who have returned to their home

countries participate in a market where they trade use of their identifier for a cash rent.


These behaviors increase false matches and complicate the search through data for

patterns of interest to law enforcement and security agencies.

The undocumented or mis-documented workforce is desirable in no small

measure due to their status: without legitimacy, these individuals have few bargaining

rights, are unlikely to report workplace violations related to pay, safety, and are unlikely

to receive the full range of workers’ benefits that otherwise make the US workforce one

of the most highly compensated in the world. For millions operating on falsified

documents or no documents at all, the type of entries generated in the databases create

persistent problems in data mining applications.

With more constraints on using the SSN as an identifier in non-official

applications, a collection of secondary identifiers are commonly used as part of the

process for establishing connections between databases that store individuals. These

include employer- or school-assigned identifiers, and the unique number on a person’s

state-issued driver’s licenses. To reduce forgery, and increase security in the issuance of

driver’s licenses, the Real ID Act in 2005 (H.R. 418) was enacted. It requires states to

change processes around the granting and manufacturing of official drivers’ license and

state identification cards. The states are balking at the requirements and a number of

legislatures have proposed refusal.

Evaluating the Explanatory Power of Different Assumptions: Reflecting on Cases

The quality of a theory rests, in part, on its ability to accurately predict outcomes

and completely explain why certain outcomes are more likely, especially in an applied

field such as public administration. The application of theory in administrative praxis

results in acts of power that can help or harm large numbers of persons, indeed, can result


in the prosperity or deaths of millions. Given the serious outcomes of flawed theoretical

constructs, there should be no limit to the number of ideas considered and evaluated.

Yet, how can ideas be tested, especially when the field encompasses several research

paradigms that do not have equal creadence among all? One method is to consider a

collection of cases. Three are considered here as to whether possible changes to

assumptions based on recognition of stochastic events and ephemeral governance provide

better explanation of outcomes. This consideration results in the development of more

equitable, favorable options that might improve outcomes. Many cases, in the US and

internationally, exemplify the limits of the traditional constructs and suggest the value of

revising the field to establish new emphases. Three well-documented US cases are

selected here as illustrations: Hurricane Katrina in 2005, the NASA loss of Challenger in

1986 (followed by the loss of Columbia in 2003); and the 1889 Johnstown Flood. These

cases span the 19th, 20th and 21st centuries of American governance, and showcase the

strength of ephemeral governance theory against situations and traits both stable and

unstable. New tools and technology may offer the possibility to minimize government

missteps and reduce the intensity and duration on the public, but this may require changes

in the way administrations are structured and managed. Because the NASA and Katrina

cases are more recent and have been widely discussed, the bulk of the description here is

of the earliest event.

The Johnstown Flood of 1889 was the worst civil disaster the United States had

yet suffered with over 2200 lives lost. (Frank, 1988, 63) Publicly commissioned and

constructed, the dam was only briefly operational as originally intended before being

turned over to private interests. Built to support 19th-century river canal traffic in an area


with intermittent waterways, the earth and rock dam’s construction was delayed several

times as other federal projects took priority in funding. By the time it was finished in

1852, railroad tracks had been laid to support commercial interests in the area, ending the

public utility for the dam, which had been drained in 1855 to repair leaks before being

sold to the railroad in 1857. Subject to benign neglect by its new owners, the dam

partially washed was five years later in 1862, with little damage downstream, emptying

most of the man-made lake. The railroad sold the non-operational dam in turn in 1879 to

private interests, industrialists from nearby Pittsburgh, who rebuilt the dam to better serve

their purposes in developing the South Fork Hunting and Fishing Club.

On Nov 15, 1879, the charter was approved and signed in the Court of Common Pleas in Allegheny County by Judge Edwin H. Stowe, who for some unknown reason ignored the provision in the law which called for the registration of a charter in the ‘office for recording in and for the county where the chief operations are to be carried on.’ Nor did the sportsmen make any effort to conform to the law. Perhaps it seemed a minor point and was overlooked by mistake. In any case, the charter was secured with out the knowledge of the authorities in Cambria County, and there would be speculation for years to come as to what might have happened right then and there had they and Judge Stowe gone about their business in strict accordance with the rules.

McCullough (1968, 49)

Capitalizing on an enlarged lake, members built a lodge, and private residences

for their recreational use. Dam design modifications included abandoning the original

discharge system, a stone culvert at the base with five sets of cast iron pipes and

discharge valves. Instead, water filled the dam and exited near the dam via a spillway

originally designed only to handle periodic extraordinary flow. A set of iron screens

were placed along the support for a bridge built to cross the formerly dry, now flowing

spillway, to prevent fish from exiting the lake via the spillway. The original dam was

further modified, reduced in height by two feet, in order to widen the road across its top


to support two-way traffic. Fundamental to understanding the contribution of governance

failure in these private party modifications, is the absence of communication between

Alleghany county, where the industrialists chartered their club, and Cambria county,

where the club was sited. The dam reconstruction was overseen by a club employee with

experience building railroad embankments. Because the discharge valves had been

removed, there was no means to release water and perform preventative maintenance or

repair small leaks that appeared. Nor was preventative maintenance performed on the top

of the dam, where settling of earth at the center requires rebuilding to the original level.

Limited public oversight of this private project resulted in only a few persons with both

engineering knowledge and knowledge of the dam’s reconstruction and operation

understanding the potential for catastrophic failure. Eight years after the dam was rebuilt,

it collapsed following a major rainstorm on May, 31, 1889.

Failure of an O-ring sealing segments of a booster rocket caused the explosion

that destroyed the space shuttle Challenger on January 28, 1986. However,

organizational culture also contributed. (Presidential Commission on the Space Shuttle

Challenger Accident, 1986; Vaughan, 1990) Vaughan used the term “normalization of

deviance” to describe a culture where management and regulators compromised their

own processes and controls, creating unreasonable expectations of success and

minimizing open discussion of problems between the government and its contractor,

Morton Thiokol.

Katrina, one of the strongest storms impact a US coastal zone in the past 100

years, made landfall as a Category 3 hurricane on August 29, 2005. Government services

failed at all levels. Infrastructure weaknesses included telecommunications equipment


located below sea level, flooded when the levees failed, worsening official

communication problems. The problems of governance during Katrina were neither rare

nor specific. The public sector response to Hurricane Katrina’s landfall illustrated some

of the same problems in bureaucratic governance as seen in the Indian Ocean earthquake

and tsunami event less than a year earlier. (Takeda and Helms, 2006)

Presenting a similar if not quite so likely risk, major levee failure in California’s

central valley could damage one of the world’s most productive agricultural regions and

stop the transfer of water to the populous south. One third of southern California’s fresh

water is transported from the north via the Central Valley Project, and a substantial

number of these levees are under private ownership, managed for agricultural support. A

2005 report estimated the economic damage in the range of 30 to 40 billion dollars.

In each instance existing governance failed, according to generally accepted

measures of expected performance based upon the degree of public investment into the

organizational systems. Responses often involved new interlocking structures, new data

measurements and collections. If the problem is we are not connecting the dots, then we

need more dots, leading us back to the concept of total information.

Johnstown Flood

Hurricane Katrina

Challenger/ Columbia

Weather X X X Rule “bending” X X X Infrequency X X X Groupthink X X X Dependence on technologies

X X X

Table 1: Shared explanations for state and institutional failures

Note a pre-eminent role of weather as a factor in three of these cases, which then

becomes a favored explanation for those in political authority. However, the weather per


se did not create the failures. In Johnstown, the dam broke in large measure because it

had been re-engineered without discharge valves at the base and with iron grates at the

top so the private reservoir operated in closer alignment with the interests of the sport

fishermen. In Hurricane Katrina, the levees were rated for a Category 3 storm, but two

failed, and there were no secondary water-retention structures. In the Challenger disaster,

the Thiokol engineers were unsuccessful in convincing NASA managers that the morning

temperature was dangerously cold, based on the effects of blow-by in earlier flights.

Restructuring Assumptions to Address Randomness

This section covers two points of discussion related to revisions of public

administration theory and praxis. First, I reflect on small-d democracy, then alternative

assumptions that address randomness are suggested. The impact of these reflections on

democratic action closes the section.

Implicit in democratic activity is the concept of individual, self-directed activity.

In a sense, one might hope that activities appear inherently random through self-direction,

and thus this concern ties closely to discussion regards any right to privacy in the US. If

the government (or banks, or any commercial firm, for that matter) seeks to capture

transactions at an individual level and to apply pattern seeking in order to successfully

target its interventions, it suggests that individual actions are probabilistically determinist.

US history, for instance the Populist movement of the 19th century, indicates that

unexpected activity by individuals and groups can be unfavorably received by the

politically powerful institutions.

At the high of the populist movement, the party experienced a wave of political

successes. (Schattschneider,1975) In the election of 1890, eight state legislatures went


Populist. In response, conservatives of both parties in the 1896 election acted to

destabilize the Populist base (pg. 76) forming blocks of Northern business Republicans

and Southern conservative Democrats (the “solid South”) This split the agrarian radicals

in the south and the west, and crushed the Populist movement. In short, “one party

politics tends strongly to vest political power in the hands of people who already have

economic power,” (pg. 78) reducing the impact of democratic movements. Even when

some democratic processes are repaired, signals of continued dispowerment may persist.

For example, after the Voting Rights Act (similarly the subsequent granting of the vote to

citizens between the ages of 18 and 21) a voluntary reluctance to vote (pg. 95) appeared

and persists. When many do not vote on a voluntary basis, political platforms can pay lip

service to the role of the public and the power of democracy, with little fear that the

people will actually show up. Schattschneider suggests this derives from the thin veneer

of democracy on institutional structures that are in reality anti-democratic (pg. 100) Thus

the struggle for democracy is still going on, merely shifted from the right to vote to the

ability to form political organizations of real power. “Nowadays the fight for democracy

takes the form of a struggle over theories of organization, over the right to organize and

the rights of political organizations, i.e., about the kinds of things that make the vote

valuable.” (pg. 100) Similarly, following the disputed 2000 presidential race the Help

America Vote Act of 2002 (HAVA) is enacted. The effect of HAVA has in the first few

elections appears to result not in improved exercise of voting rights and political power;

rather the nation has seen hundreds of millions of dollars awarded to a small handful of

producers of voting technologies. This misstep is made obvious as states begin to roll

back from fully electronic machinery to more comprehensible paper systems. Similarly,


health care reform does not seek to remove the entire machinery of insurance companies

and billings (clearly not a zero-cost operation) even if the nation goes to universal care.

The electronic patient record, another massive data collection system that will be linked

to existing structures, becomes the new marker for cost savings and mistake reduction.

Governance structures at random intervals experience instabilities of random

duration. The Populist movement can be characterized as an instability of the entrenched

system; changes in the demographics of voters can be viewed similarly. Events can be

minor, for instance storms, protests, parades, or major: war, disease, or climate changes.

They can be highly localized to a single site, or involve numerous governance units in a

multi-state/international power outage. The degree to which any government continues

to operate as expected varies. Some, such as changes in the party in power, might have

either minor or major effect. Table 2 offers assumptions that might be revised to better

address the questions raised in desiring to maintain good governance under randomness.

Alternative assumptions are offered, and some ideas about how these alternative

assumptions might change public administration research and professional training.

Classic Assumption Revision Implications for Praxis Governing structures are stable

Instabilities are frequent if not predictable

Bargain with non-states, unauthorized parties

Institutions define rights Conditions of the moment create rights Technologies constrain rights

Redesign institutions for loose coupling option Alternative educational strategies

Incrementalism works Incrementalism unnecessarily risks catastrophe

Insurance? Alternative decision models

Government equalizes Government stratifies, earmarks, redistributes

Incentives for ethical administration

Data are comprehensible Data are incomplete, obscured, or suggest contradictory explanations

Educational and coping strategies

Table 2: Revising basic assumptions for public administration


Diagram 2 offers a revision to Diagram 1 that suggests an alternative to the “total”

information collection and retention model. It involves more probabilistic analysis,

which might be either quantitative, or qualitative in nature. By considering an option for

non-quantitative, reflective, or discourse-oriented considerations of probable outcomes,

an open is created for more democratic action. It allows variation in approaches to the

study of randomness across the continuum of approaches to the study of public

administration Raadschelders (2005) from “scientific knowledge” to interpretive

postmodern expressions.

perceptually relevant, knowable information

multiple potential outcome scenarios, each with a calculable probability

ŅtotalÓ information

Diagram 2: Sensemaking through Scenario and Probability Development

Concluding Remarks

Were public administration schools to acquiesce to a data-driven model of

governance, they could mirror a strategy adopted by some business schools: more

courses in technical data management (a specialization in confidential data mining,

anyone?) supplemented by an ethics course to remind students that governance includes a


moral dimension. It is unclear, however, whether this approach has improved the quality

and success of business enterprises. Indeed, developing operational generalists who can

recognize and communicate across specializations is offered in Raadschelders (2005) as

one useful bridge to the field’s divergent approaches to characterizing core knowledge

and functions.


References American Civil Liberties Union. (2004). “Total Information Awareness.” Accessed February 20, 2007 at http://www.aclu.org/privacy/spying/14956res20040116.html Denton, Frank T. “Data Mining as an Industry.” Review of Economics and Statistics, 67:1 (February), pp. 124-127. Eagle, Antony (2005). Randomness Is Unpredictability.” British Journal for the Philosophy of Science. 56:4 (December), 749-790. Electronic Privacy Information Center (2005) “Total ‘Terrorism’ Information Awareness.” Accessed February 20, 2007 at http://www.epic.org/privacy/profiling/tia/default.html Frank, Walter (1988). “The Cause of the Johnstown Flood.” Civil Engineering, 58:5 (May), 63-66. Goulielmos, Alexandros M. (2004) “A treatise of randomness tested also in marine accidents.” Disaster Prevention and Management. 13:3; 208-217. Hand, David J. (1998). “Data Mining: Statistics and More?” The American Statistician, 52:2 (May), 112-118. Hernandez, Mauricio A. and Stolfo, Salvatore J. (1998) “Real-world Data is Dirty: Data Cleansing and The Merge/Purge Problem.” Data Mining and Knowledge Discovery. 2:1, 9-37. Kiefer, John J. and Robert S Montjoy, Robert S. (2006). “Incrementalism before the Storm: Network Performance for the Evacuation of New Orleans.” Public Administration Review. Washington: Dec 2006.Vol.66 pg. 122, 9 pgs McCullough, David (1968). The Johnstown Flood. New York: Simon & Schuster Paperbacks. Merton, Robert K. (1957). “Bureaucratic Structure and Personality” in Shafritz, Jay M., Ott, Steven J. and Yong Suk Jang, Classics of Organization Theory. 6th edition. Belmont, CA: Thomson Wadsworth, 2005. Norman, Geoffrey R. and David L. Streiner (1999). PDQ Statistics, 2nd Edition. Lewiston, NY: BC Decker. Owen, Art (2003). “Data Squashing by Empirical Likelihood.” Data Mining and Knowledge Discovery. 7:1, 101-113.


Perrow, C. (1984). Normal accidents: Living with high-risk technologies. New York: Basic Books. Palmerston, Lord Henry Temple. 1848. Cited in “Morality: Does national interest always come first in Foreign Affairs?” Accessed January 25, 2007 at http://news.bbc.co.uk/hi/english/static/in_depth/uk_politics/2001/open_politics/ foreign_policy/morality.stm Presidential Commission on the Space Shuttle Challenger Accident. (1986). Report of the Presidential Commission on the Space Shuttle Challenger Accident (Rogers Commission Report). Accessed February 20, 2007 at http://science.ksc.nasa.gov/shuttle/missions/ 51-l/docs/rogers-commission/table-of-contents.html Raadschelders, Jos C.N. (2005). “Government and Public Administration: Challenges to and need for Connecting Knowledge.” Administrative Theory & Praxis, 27: 4 (December); 602-627. Reeves, Jay (2007) “VA Notifying 1.8M of Missing Data,” February 12. Accessed February 12, 2007 at http://apnews.myway.com/article/20070213/D8N8GIU00.html Rogowitz, Bernice E. and Treinish, Lloyd A. (ND). “How NOT to Lie with Visualization.” IBM Thomas J. Watson Research Center. Accessed February 20, 2007 at http://www.research.ibm.com/dx/proceedings/pravda/truevis.htm Schattschneider, E.E. (1975, 1960). The Semisovereign People: A Realist’s View of Democracy in America. Fort Worth: Harcourt Brace Jovanovich College Publishers. Schrag, P. (2007). Lecture at California Association of State Counties. January 17. Takeda, Margaret B. and Helms, Marilyn M. (2006) “ ‘Bureaucracy, meet catastrophe’; analysis of Hurricane Katrina relief efforts and their implications for emergency response governance.” International Journal of Public Sector Management, 9:4, 397-411. Taleb, N. N. (2004). Fooled by randomness: The role of chance in life and the markets, 2nd ed. New York: Texere. Tufte, Edward (2001). The Visual Display of Quantitative Information, 2nd Edition. Cheshire, CT: Graphics Press. Vaughan, Diane (1990). “Autonomy, Interdependence, and Social Control: NASA and the Space Shuttle Challenger.” Administrative Science Quarterly, 35:2; 225-257.


PA as a Stochastic Process...A stochastic process might characterize either an unknown system that is acting on persons and conditions, or a system with random effects that interacts

Documents