Responsibility for Failures of Government: The Problem of Many Hands

The problem of many hands—the difficulty of assigning responsibility in organizations in which many different individuals contribute to decisions and policies—stands in the way of investigating and correcting the failures of government. The problem can be mitigated by giving greater attention to the design of processes of organizational responsibility. An independent investigation can identify both the individual actions and the structural defects that contributed to an organizational failure. Then, specific individuals can be designated as overseers, who are held responsible for monitoring the structure and making changes as necessary. Three cases—the official responses to terrorist attacks on the World Trade Center in 2001, the Deepwater Horizon oil spill in 2010, and the financial crisis that began in 2007—illustrate how this prospective approach of designing responsibility could work in practice.

When terrorists attack, oil spills, banks fail, and other "stuff happens," 1 we naturally look for individuals to blame.But in modern society, the most serious damage is usually done by large organizations including governments.Because many different individuals in an organization contribute in many ways to the decisions and policies, it is difficult even in principle to identify who is responsible for the results.This difficulty is known as the problem of many hands. 2 The problem poses a challenge to the task of assigning responsibility for organizational failures in government.To investigate and correct organizational failures, we need to be able to locate specific individuals to hold responsible.Yet when many hands are involved, individuals who may bear some responsibility for the harm are less likely to see that they do and less likely to be held responsible by others.The profusion of agents obscures the location of agency.If an individual leader "takes responsibility," as in the familiar ritual of command responsibility, he or she typically suffers few consequences (at least in political organizations), and the search for responsible agents is often cut short. 3 Some try to avoid the problem by turning from individual to collective responsibility.This collectivist approach has been prominent in the philosophical literature. 4It also finds favor with some writers on corporate law (Friedman, 1999(Friedman, -2000)).It is claimed to have two principal advantages: If we target only the organization, we have identified an agent that we can hold responsible without unfairly blaming individuals; and targeted the agent that has the greatest capacity to provide compensation and undertake reforms.But this is not really an escape.Whether collectivities are considered moral agents or only legal entities, it is still the individual members of the organizations who suffer many of the consequences that follow from the ascription of responsibility, and it is still individual officials who will have to respond.Furthermore, if we hold only organizations responsible, then either all members are equally blameworthy or all are excused, regardless of the degree of their responsibility.
We do not have to reject the possibility of holding organizations and other collectivities themselves responsible in some way, and certainly should not rule out subjecting them to legal or financial sanctions.Some such responsibility, with or without moral agency, is necessary and appropriate in modern society. 5It is the "only" that is objectionable.We still need to find a role for individual responsibility.
Without an adequate account of individual responsibility in organizations, the standard ideas of moral and legal responsibility will be violated either by targeting individuals who are not responsible, or by exonerating individuals who may be at least partly responsible.The result is not only unfair to the individuals (and sometimes the organizations as well) but also undermines the set of incentives on which the organization and society depend to encourage responsible action.If the incentives are not directed toward actions or omissions for which individuals and organizations could reasonably believe they are responsible, the incentives and therefore the means of accountability are less likely to be effective. 6That is, responsibility (being a contributor to an outcome) needs to track accountability (being required to answer, or subjected to sanctions, for it). 7 We can create a greater role for individual responsibility in organizations if we give more attention to responsibility for their design.Investigation of the failures of government should seek to identify responsible individuals as well as structural defects that led to the outcomes, but with the primary aim of making changes in organizational design.The changes should provide greater incentives for taking individual responsibility in the future and should designate specific individuals to continue to monitor the organizations with the objective of fostering a culture of responsibility.The members of the body that carries out the investigations should themselves be held accountable according to criteria of individual responsibility.

Individual Responsibility and Its Limits
The first step is to recognize that assigning individual responsibility even for organizational outcomes is not as hopeless as a simple statement of the many hands problem might suggest.We can strengthen the role of individual responsibility for failures if we are more careful in interpreting the criteria for ascribing it.An individual is generally said to be responsible for an outcome insofar as the individual's actions or omissions are a cause of the outcome, and the actions or omissions are not done under compulsion or in ignorance. 8An individualist approach that adopts these criteria is more robust in organizational settings than usually assumed, provided the criteria are properly specified (Bovens, 1998(Bovens, , 2007(Bovens, , 2010;;Luban, Strudler, & Wasserman, 1992;Thompson, 1980Thompson, , 2005)).How they are interpreted and applied matters.
If we characterize an outcome in very general terms-say, the environmental harm from a massive oil spill-we may not be able to identify any individual as a cause on a strict criterion.But if we describe the outcome more specifically-the failure to take usual safety precautions in drilling the well-we may have more success in locating responsible individuals.Or take the excuse of ignorance: We should not have to show that an executive should have foreseen the specific act of particular subordinates (e.g., that the drilling team would neglect to follow certain procedures).In organizations, certain patterns of fault are common enough that we should expect any competent official to anticipate them and to take reasonable precautions to avoid them or at least to minimize their harmful consequences.
Nevertheless, in many cases the problem of many hands persists, because even if the individuals who made mistakes and contributed to the outcome can be identified, the consequences of their acts (as well as their omissions) are often disproportionate to the harm brought about by the organization.The harmful effects of the oil spills, the financial crisis, or terrorist attacks go beyond anything that any individual may have done or could be reasonably held responsible for.Once we apportion blame and apply sanctions to everyone as far as possible, we would still find that the enormity of the outcome exceeds even the collective sum of the actions of individuals.
The individualist approach thus reaches a limit in trying to attribute responsibility for harm brought about by organizational failure.But so does the collectivist approach because it cannot escape relying to some extent on individual responsibility in imposing sanctions and carrying out reforms.Once we have reached this limit and exhausted the supply of individual responsibility for an organizational failure, is there nothing more to be said or done about responsibility?If no one (or alternatively if everyone) is responsible, no one is likely to do anything about it, whatever it is.That is not a promising approach for making either individuals or organizations more responsible and accountable.

The Shift to Design Responsibility
We can still find a role for individual responsibility if we shift our perspective from the responsibility for outcomes to the responsibility for the design of organizations.The source of many of the organizational disasters in recent years can be found in the structure of the organization and its relationship to other organizations.But as we go looking for the individual designers, we will confront the problem of many hands again.The organizational design is often the product of many decisions and many nondecisions by many different people over a long period of time.Potential designers who knew about the defects may not have had the power to fix them, and those who had the power may not have known (though often they should have known).If we look only for design faults in the past, we are likely again to find too many hands with too little responsibility.
Therefore in shifting to design responsibility, we also need to adopt a forwardlooking conception of responsibility-what may be called prospective design responsibility.We examine past failures-but chiefly for the purpose of preventing future ones.In carrying out such an examination, we first need to locate, as far as possible, not only the structural defects in the organization but also the individual actions that may have contributed to the failure.That is necessary so that we can separate the structural defects from the individual errors.Then prospectively, we designate specific individuals or groups of individuals as overseers, who can be held accountable for monitoring the structure of the organization and making or recommending changes in it and the organizational culture as necessary.In the future, if they fail to fulfill that responsibility, we will know whom to blame, even if the organizational failure itself is the result of the actions of many hands.
The shift to prospective design responsibility can preserve a role for individual responsibility in many hands circumstances that would otherwise frustrate attempts to assign it.However, overcoming the many hands problem in this way requires more than merely adding design responsibility to the job descriptions of existing officers and members of the organization.It requires establishing new offices or institutions with individuals specifically charged with overseeing organizational changes to correct structural deficiencies that could result in disastrous failures.Ironically, it requires creating more hands-but with more precisely defined responsibilities.
How design responsibility could work can be shown by examining its application in three different cases of major governmental failure: the official responses to terrorist attacks on the World Trade Center in 2001, the Deepwater Horizon oil spill in 2010, and the financial crisis that began in 2007. 9In each case, an official body was appointed to investigate the failure, and some of these bodies came closer than others to following the approach proposed here.The aim here, then, is dual: to examine from the perspective of design responsibility both the failures of responsibility and the bodies that carried out the investigations of the failures.

Unprevented Terror
After the terrorist attacks on September 9, 2001, that killed nearly 3,000 innocent people and destroyed New York's World Trade Center and part of the Pentagon, Congress created a 10-member bipartisan commission to investigate the failures of government and to recommend ways to avoid them in the future.The commission's 567-page final report, issued in July 2004, presented a riveting narrative of the policies and events leading up to the attacks, detailed descriptions of the government's response, and a set of recommendations for changes in the practices and organization of many agencies of government (National Commission on Terrorist Attacks Upon the United States, 2004).The initial reaction to the report was favorable.Remarkably well written and replete with new information, the document was turned into a book and immediately became a best seller.
But on closer reading, many found the report deficient.The critics included not only the families of the 9/11 victims but also a former aide in the Bush White House and a senior adviser to the Commission (Falkenrath, 2004; E. R. May, 2007; E. R. May, Zelikow, & Falkenrath, 2005;Shenon, 2008).The most salient criticism they raised is directly relevant to the problem of responsibility.The report failed to hold any individual accountable; it declined to pass judgment on individuals who made the key decisions.The Commission adopted a "no fault" theory of government-"an imprecise, anodyne and impersonal assignment of responsibility for the U.S. government's failure to prevent the 9/11 attacks" (E.R. May et al., 2005, Falkenrath, p. 211).
The Commission's decision to avoid singling out individuals was deliberate: "Our aim has not been to assign individual blame."Later, one of the co-chairs explained more fully: ". . .if we had come up with a list of bad actors, it would have blown the commission apart and it would have blown any credibility we had . . ." (Shenon, 2008, p. 405).Although the Commissioners were commendably conscientious and public spirited, the Commission was, in origin, composition and foreseeable reception, a political body, and its scope for action was inevitably shaped by political considerations.The understandable political calculation the co-chairs made had serious costs.First, an analysis that neglects individual responsibility is an inadequate guide for decision makers in the future.Refusing to "name names" is "exactly the wrong message to send to future government officials and the people who train them" (E.R. May et al., 2005, Falkenrath, p. 211).More generally, for the purposes of redesigning incentives for responsibility in the future, we need to know which were more or less effective and under what circumstances in the past.
Second, avoiding discussion of individual responsibility weakened the recommendations for institutional reform.The Commission's proposals for change were only loosely connected to its analysis of the failures.That analysis relied on very general, impersonal concepts such as a "failure of imagination," the most prominent of the four major "failures" it cited (National Commission, 2004, pp. 339-347).Presumably, everyone should be more imaginative.Analyzing the "management failure," the report retreats to the passive voice: "information was not shared . . .analysis was not pooled.Effective operations were not launched . .." (National Commission, 2004, p. 353).Responsibility designers would look to the report in vain for the agents who failed to share information, pool the analysis, or launch the operations.They would be at a loss to know which roles or offices should be redesigned to prevent these failures in the future.While the diagnosis generally avoided singling out individuals, the prescriptions did not.Many of the recommendations called for changes that would assign definite responsibilities to specific individuals (National Commission, 2004, p. 339).The connection between the diagnosis and the prescription remained obscure.
Bush's counterterrorism adviser Richard Clarke charged that National Security Adviser Condi Rice (and others) failed to take the terrorist threat seriously (Shenon, 2008).In the Clinton administration, Clarke's office gave him more regular and direct access to the President and other principals than it did in the Bush administration, which (in Clarke's view) downgraded the office.Was this dereliction on the part of Rice and others, or was it the result in part of organizational defect that should be corrected?Similarly, for many of the other specific failures described in the report, it would be essential to know why those who could have acted (presumably a small and identifiable number of individuals) failed to do so.
Even if we do not want to discipline individuals, and indeed if we believe that some should be excused because of structural defects over which they had no control, we cannot redesign the roles or offices in which they acted without knowing what decisions which individuals took or failed to take, and what realistic alternatives they might have had.We need to know who did what, not to ascribe blame or mete out punishment, but to guide the design of the roles and organizational culture to prevent future failures-including failures of responsibility.The former Bush aide's critical summary indicates why the report could not be such a guide: Even if authority is widely and confusingly spread around the executive branch . . . the starting point in any after-the-fact governmental analysis should always be the concept of personal responsibility.The 9/11 commission instead focused on a handful of amorphous, impersonal causal factors, none of which is nearly as compelling as the notion that an identifiable set of government officials made bad decisions about where to apply their energies and, as a result, failed to do the job that the American people had the right to expect them to do.(E.R. May et al., 2005, Falkenrath, p. 211) The experience of the Commission suggests several lessons for design responsibility and the bodies charged with promoting it.First, if a postmortem on a major organizational disaster such as 9/11 is expected to produce recommendations for organizational change, a bipartisan commission or other political body is probably not the best instrument.In this case, the supposed advantage of influencing Congress did not materialize.Many of the recommendations were ignored, and others watered down.A commission may be appropriate, but it should not be so closely connected to the political parties, and should include experts and citizens rather than partisan political figures.It should in its own design be more independent than the 9/11 commission.
Second, an essential aim of any attempt to redesign responsibility in an organization after a disaster should be to pursue individual responsibility as far as possible.The critiques of the 9/11 commission make clear that the possibilities for locating individuals who failed as well as those who succeeded in fulfilling their responsibilities are greater than is usually assumed, even in a massive, large-scale disaster of this kind.We do not have to hold individuals responsible for the whole disaster to hold them responsible for specific and substantial mistakes.Distinguishing the structural from the individual sources of these mistakes is necessary for the task of designing responsibility.The assignment of individual responsibility is the handmaiden to the prevention of collective disaster.
Finally, part of the redesign of any organization should involve specifying more clearly which individuals should be responsible for which decisions, and giving them the independence and information to fulfill their responsibilities.Some of the Commission's recommendations moved in that direction, but their more comprehensive efforts were weakened by their own analysis, which more readily suggested impersonal remedies such as the implication that everyone should be more imaginative or that Congress should be better organized.The Commission suggested some specific reforms, but was often silent about who should be responsible for carrying them out (e.g., National Commission ,2004, pp. 419-21).

Spilt Oil
In April 2010, in the Gulf of Mexico, a deep water oil drilling rig operated by British Petroleum (BP) exploded, killing 11 workers and causing a massive gusher that eventually released nearly 5 million barrels of crude oil into the Gulf.The spill caused extensive and continuing damage to marine and wildlife habits and to the fishing and tourist industries on which most of the residents of the Gulf depend.The government held BP responsible for managing the clean-up and compensating victims, but the various investigations and legal proceedings have still not entirely settled which of the several companies and individuals involved are responsible for the spill and its effects.
In May 2010, President Obama appointed a seven-member National Commission to investigate the spill with the aim of "providing recommendations on how we can prevent-and mitigate the impact of-any future spills that result from offshore drilling" (White House Press Office, 2010).Although the Commission was called "bipartisan" because the co-chairs were identified as a Democrat and a Republican, the composition and the mission were not political in the way that the 9/11 Commission was.Both the co-chairs had held positions and had experience directly relevant to the environmental disasters of this kind.The other members were recognized independent experts, mostly researchers or leaders of apolitical institutions.
Even though its explicit charge was forward-looking, the Commission realized that it had to examine the causes, and that such an examination would require giving some attention to decisions that individuals had made.The commission did not try to assign specific blame for a catalog of mistakes and shortcuts taken by the companies and their employees, but it is clear from the report that the major agents engaged in highly risky behavior that neither senior management nor government regulators properly oversaw.(Broder, 2011, p. A14) The Commission concluded that the immediate causes of the Macondo well blowout can be traced to a series of identifiable mistakes made by BP, Halliburton, and Transocean that reveal such systematic failures in risk management that they place in doubt the safety culture of the entire industry.(National Commission on BP Deepwater Horizon Oil Spill and Offshore Drilling, 2011, p. vii) The list included many mistakes that could be attributed to specific individuals or small groups of individuals, such as the supervisors who ignored early warnings that key pieces of equipment such as the blow out preventer might fail.Many individuals had both the authority and knowledge to change the practices that led to many of these errors, but the pressure to save time and money evidently drove the decision making more than concern for safety (National Commission, 2011, pp. 125-26).
Eventually the numerous legal proceedings began to identify some individuals who were more responsible than others, and allocate some responsibility to the several corporations involved.But undoubtedly, even the cumulative total of legal liability will not be commensurate with the damage caused by the spill.The more constructive effort focuses on assigning responsibility for preventing or reducing the risk of similar disasters in the future and the first line of response here is governmental oversight.The report of the Commission, which because of its relative independence and expertise, carried out an investigation more constructively in this way than did the 9/11 Commission.The report and the analyses of other observers point to significant failures of design responsibility.
There were several different agencies responsible for oversight, and no one had overall authority.The responsibility design was diffuse, which probably contributed to the disaster.Furthermore, the principal agency for regulating the drilling, the Mines Minerals Service (MMS), granted so many exceptions and overlooked so many violations that its officials may be as much responsible for the disaster as many of those at BP (Urbina, 2010).The Commission made clear that agency officials had missed many opportunities for redesigning the regulatory system (National Commission on BP Deepwater Horizon Oil Spill and Offshore Drilling, 2011, p. 71).
Although the MMS failures were partly the result of manifest corruption that had long plagued the agency, they were made worse by a design problem.The agency was charged with two different and conflicting tasks: promoting the industry (encouraging drilling) and regulating it (ensuring that safety was the highest priority).Even without the corruption, the promotional efforts would have been likely to overshadow the regulatory responsibilities.One lesson is that organizational designers should divide conflicting responsibilities by assigning them to different agencies.That is in fact one of the steps that the government has now taken in this case, splitting the previous agency into three different parts to avoid the conflicts (U.S. Secretary of the Interior, 2010).
But the problem persists because the individuals who know the most about the industry and how to regulate it effectively come from the same backgrounds, and often move in the same circles as the people they are regulating.Even if they are not looking for opportunities in the industry, they are more likely to see the world from the perspective of those whom they are regulating than from the perspective of the citizens who may be harmed by mistakes that the industry (or the regulators) make.To the extent that government seeks the most competent experts to conduct oversight, this design problem cannot be avoided at this level.
Another lesson, then, is that to address this kind of problem, we need an additional body to ensure that some oversight responsibility is assigned to people who have a different perspective.We need a body composed of members who would give more weight to the effects on citizens, and who are more willing to challenge expert opinion.One method that has been tried in similar circumstances is a citizens' advisory council (Applegate, 1998).Such a body was set up after the Alaskan Valdez disaster, but legislation to require that it be established in other regions failed to pass in Congress.The National Commission briefly reintroduced the proposal: any new "structure should therefore include a citizens' advisory council to provide formal advice and a direct line to citizens' concerns" (National Commission, 2011, p. 212).The Commission did not specify the form that such a council should take, but an earlier report by another federal panel set out some of the requirements a council should satisfy.It should be an independent public body charged with providing policy and technical advice for specific projects, sites, or regions.It would consist of 10 to 20 members, including directly affected parties, and also unorganized "individual residents that live in the communities or regions in which [the] site is located" (Federal Facilities Environmental Restoration Dialogue Committee,1996, pp. 56-57).Governmental officials would serve as nonvoting members, and governments would provide professional staff.
Thus, two of the lessons of this episode-divide conflicting tasks, and add checking authorities-point to responsibility reforms that would address the problem of many hands by multiplying the hands.That solution might seem to recreate the problem it is supposed to solve.But the multiplication is not the same.The difference is that the hands would be specifically charged with oversight and nothing else, and they would be independent in the sense that neither their mission nor their interest would conflict with their responsibility for oversight.
This proposed multiplication of oversight responsibility for oil drilling would occur on a single level of authority; each of the authorities would have a somewhat different function but would be equal in the sense that neither would oversee the other.The type of structure is what may be called horizontal responsibility for oversight.Such a structure could of course create a problem of coordination and potentially give rise to conflicts.Those problems could be mitigated by rules requiring regular consultation and joint meetings, and specifying which body takes priority in cases of conflict.
Rules of this kind could also obviate the need to establish a higher authority to oversee both bodies, which would create a further problem.It would in effect introduce a form of vertical responsibility to the structure.The problem with vertical oversight responsibility is that it tends to duplicate functions at each level, recreating the many hands problem.It also invites a reiteration of the question as to who will oversee the overseers, generating a regress of oversight that has no logical termination (Thompson, 2005, pp. 261-62).However, the vertical model may be necessary in some cases, and with the appropriate modifications can avoid these problems, as consideration of the failures of responsibility in the financial crisis beginning in 2007 illustrate.

Failed Banks
The financial crisis that plunged the world economy into the worst depression since the 1930s was set off when the housing bubble burst and a liquidity shortage developed in the United States in 2007 (Blinder, 2013).Such an immense and complex calamity had many causes, and not surprisingly, the list of individuals and organizations that could be plausibly blamed is distressingly long. 10The crisis manifests the problem of many hands in its most florid form.
The most prominent of the many investigations was conducted by the Financial Crisis Inquiry Commission, created in July 2009 by Congress, which appointed all 10 of its members (Financial Crisis Inquiry Commission, 2010a).The members had considerable expertise in financial matters, but less independence from their political supporters.The Commission's analysis and recommendations were as a result less helpful in advancing the aims of design responsibility.Unlike even the 9/11 Commission, it split along partisan lines and did not issue a unanimous final report (Financial Crisis Inquiry Commission, 2011).The majority report made some effort to identify individuals and institutions that were responsible but included so many culprits that the minority report was provoked to object: "When everything is important, nothing is" (Financial Crisis Inquiry Commission, 2011, p. 414).Yet the "ten essential causes" summarized in the minority report itself emphasized broad impersonal forces, such as the credit bubble in the world economy, giving less attention to the role of individual decision makers (Financial Crisis Inquiry Commission, 2011, pp. 417-419).Nevertheless, the Commission's report contains information and analyses that are helpful in examining the failure of design responsibility in this case and potential changes to prevent such failures in the future.
Although a full assessment of responsibility for the crisis would examine many hands, one set of institutions-the rating agencies-merit special attention because they illustrate how the vertical model for designing oversight responsibility might be applied.The agencies, which include once respected organizations such as Moody's and Poor's, were not the best known villains in the popular exposes of the crisis, but their failures contributed significantly to the crisis (Financial Crisis Inquiry Commission, 2010b;2011, pp. 43-44, 212, 426, 418).The agencies were the "reputational intermediaries" who enabled the banks to persuade investors that the securities were safe (Walter, 2010).The Commission majority concluded, "The three credit rating agencies were key enablers of the financial meltdown" (Financial Crisis Inquiry Commission, 2011, p. xxv).
Could agency executives have recognized the risk sooner?Even if they were unaware (and some surely were not), their ignorance does not seem excusable.Some of their own analysts knew that they were giving high ratings to nearly worthless securities (Lowenstein, 2008;U.S. Senate, 2010).If they knew or should have known, could they have downgraded the securities sooner?The pressure from the investment banks to give high ratings was relentless.The business of rating these securities accounted for nearly half of Moody's revenue in the year before the collapse.But the agencies could have revised their ratings-as they eventually did anyhow, and with worse consequences for everyone than if they had acted sooner.
It would be possible to identify the individuals who were responsible, but for purposes of organizational design that effort should be in the service of locating the structural problems that contributed to the failure.The most salient problem is another institutional conflict of interest-not the functional conflict we saw in the case of the MMS in oil drilling, but a classic financial one.The rating agencies' interest in providing accurate assessments conflicted with their interest in financial gain.Any agency that declined to give good ratings to the securities that their client eagerly wanted to sell would soon lose that client and probably others. 11 Although this structural problem was recognized well before the financial collapse and some steps have now been taken to address it, the conflict remains (Financial Crisis Inquiry Commission, 2011. pp. 211-12;Lynch, 2009; Office of the Special Inspector General for the Troubled Asset Relief Program, 2009, pp. 136-37;U.S. Congress, 2010).The banks pay the agencies who rate the products that the banks want to sell to investors.Investors have no incentive to pay for their own ratings (which are not confidential and could be easily used by other, free-riding investors).The government is not in a position to take over the rating process itself, or even oversee it at any depth.Under these challenging circumstances, how could responsibility for monitoring the ratings be designed into the system?
A proposal described in an amendment to the Dodd-Frank bill, adopted by the Senate but dropped in the final legislation, would go some way toward addressing the problem (Congressional Record, 2010;Herszenhorn, 2010).It would require every new asset-based bond issue to be rated not by a rating agency chosen by the investment bank offering the security but by an agency assigned by a new independent board, on the model of a public utility, appointed and overseen by the Security and Exchange Commission.The board would choose the agency based on its competence and performance.The agencies would still be paid by the banks, but the banks could not shop around for their preferred ratings.They would pay the agency regardless of whether they liked the ratings.This structure would thus eliminate or at least drastically mitigate the institutional conflict of interest.
For our purposes the significance of the proposal is to be found in the way the oversight responsibility is designed.Responsibility is assigned to the independent board, not for reviewing the ratings themselves, not even for overseeing the practices of the rating agencies in their day-to-day business, but for maintaining standards intended to encourage agencies to produce accurate ratings.It is a version of the vertical model of responsibility mentioned above, but without the duplication of function and potential regress that less differentiated versions of the model can generate.It also preserves individual responsibility.The hands that choose the agencies are identifiable, and they produce a record that could be used to identify the individuals in any agency who are not following best practices.Having this kind of responsibility regime in place could reduce the likelihood that blame would have to be assigned in the future because it increases the incentives for blameless behavior in the present.

Conclusion
The problem of many hands is inherent in any complex organization.The failures of governments are usually the result of decisions and nondecisions by many different individuals, many of whose contributions may be minimal and unintended.Yet to assign responsibility and maintain accountability for an outcome fairly and effectively, citizens have to identify individuals who knowingly and freely contribute to it.This individualist approach is necessary even if the purpose is not to punish or discipline individuals but to make changes in the organization to reduce the chances of adverse outcomes in the future.
We can use the results of investigations into responsibility for past outcomes as a guide for making changes to clarify individual responsibility for future outcomes and future oversight.Specifying the responsibility for monitoring the reforms in the structures and in the culture of responsibility-design responsibility-is often neglected but is no less important than assigning responsibility for outcomes.
The commissions that examined the 9/11 response, the Deepwater Horizon oil spill, and the financial crisis were in effect engaged in this kind of investigationidentifying past failures of responsibility to prevent future ones.They were successful insofar as they were independent and knowledgeable, and unsuccessful insofar as they were not.Notice that the requirements of independence and knowledgeability parallel the criteria for individual responsibility.In effect, we assess the responsibility of the commissions with criteria analogous to those by which they should assess the responsibility of officials.
The commissions not only examined failures in the past but also proposed institutions to strengthen individual responsibility in the future.The most constructive reports at least implicitly recognized that changes in organizational design are not self-executing, confront many obstacles (such as the "traps" described by Argyris, 2012), and require ongoing attention and action.For that reason, some of the most significant recommendations are the proposals to establish oversight bodies, such as the citizens' advisory council for oil drilling or the independent board for choosing rating agencies.The council, an example of a horizontal oversight, would bring a different perspective to balance those of government experts and industry executives.The board illustrates how vertical oversight could avoid the duplication and regress in designs that address the problem of many hands by multiplying the hands.
Whatever their other functions, oversight bodies such as these could be charged with holding individuals in the organization responsible on a continuing basis, and most importantly with exposing organizational defects that obstruct individual responsibility.They would in effect take responsibility for designing responsibility.
Commissions and oversight bodies, properly constituted, can be important devices for mitigating the problem of many hands, but no less important is the rationale for establishing such institutions.We would be better able to identify the individuals who contribute to the failures of government and thereby reduce the chances of future failures if we refine the criteria of responsibility and extend their scope to encompass the design of institutions.This modified individualist approach to the problem of many hands can serve as a guide in the continuing effort to find ways to strengthen individual responsibility in government and hold its officials accountable to democratic citizens.

Notes
1.The responsibility-denying phrase comes from Secretary of Defense, Donald Rumsfeld, defending the military response to looting and disorder in Baghdad after the U.S. invasion: "Think what's happened in our cities when we've had riots, and problems, and looting.Stuff happens!" (U.S.Department of Defense, 2003).2. The problem was first analyzed in these terms in Thompson (1980Thompson ( , revised 1987Thompson ( , reprinted 2005)).The "many hands" articles have generated a large literature that continues to grow.The most extended discussions are by Bovens (1998Bovens ( , 2005Bovens ( , 2007Bovens ( , 2010)), who carries the analysis further and proposes responses to the problem that are consistent with Thompson;and Smiley (1992), who criticizes Thompson's conceptual criteria for responsibility.A formal analysis of the problem is given by Mastop (2010).The Thompson formulation of the problem has proved relevant in a wide range of subjects and disciplines: public administration (Burke, 1986;Christensen & Laegreid, 2002;Dobel, 1990;Pesch, 2008;Petter, 2005;Tong, 1985;van der Heijden, 2010), blame attribution (McGraw, 1990(McGraw, , 1991)), city planning forecasting (Wachs, 1990), computing accountability (Nissenbaum, 1996), corporate management (Kaptein, 1998;Wempe, 1998), crisis management (Boin, Hart, McConnell, & Preston, 2010), engineering (Cohen & Grace, 1994;Doorn, 2012;Doorn & Fahlquist, 2010;van Gorp, 2005), criminal justice (DiIulio, 1988;Steiker & Steiker, 1995), European law (Bovens, 2007), financial analysis (Hildreth, 1983), garbage collection ethics (Rich, 1996), government regulation (Black, 2008), health care (Leichter, 1992;West, 2000), European local government (Becquart-Leclercq, 1982;Denton & Flinders, 2006;Sullivan, 2003), law of war (Bluhm & Heineman, 2007), nanoethics (van de Poel, 2008), organization theory (Dubnick, 2003;Jos, 1988), political campaigning (Schedler, 1998), professional ethics (DeMartino, 2011), research and development networks (van de Poel & Zwart, 2010), resignation in protest (Dobel, 1999;Mulgan, 2002), school governance (Allen & Mintrom, 2010), and technology assessment (Zwart, Van de Poel, Van Mil, & Brumsen, 2006).3. "With regular incantations of 'I accept full responsibility,' an official strengthens his or her own political standing-by reassuring the public that someone is in charge and by projecting an image of a courageous leader who does not pass the buck . . . the ritual often quells public debate about a controversial decision or policy, effectively blocking further inquiry into the genuine moral responsibility of all of the officials involved . .." (Thompson, 1980, p. 907).4. For example, May and Hoffman (1992).An exception is Miller (2006) who has developed a cogent individualist account of collective responsibility.For a valuable overview, see Smiley (2011).5.For a discussion of how under certain conditions collectivities can be blamed even if they are not properly regarded as human agents, see Scanlon (2008, pp. 160-66).6.An organizational structure characterized by "many hands," usually required by the technical rationality of organizations, may be regarded as another "mask" for what Adams and Balfour (2009) call administrative evil (acts of "pain and suffering and death" inflicted by officials, who typically do not see that they are acting wrongly, pp.4-5, 11-13).To the extent that we can show that the presence of many hands does not eliminate individual responsibility, we in effect "unmask" another significant source of administrative wrongdoing.7. Responsibility and accountability are often used interchangeably, and the concepts overlap in many contexts.But if they are distinguished roughly in the way suggested in the text, the challenge of the problem of many hands can be seen as finding a way to bring responsibility more in line with accountability.The aim is to try to make sure that the individuals who through their actions or omissions contributed to a failure are those who are held accountable.Design responsibility is one way of bringing about this alignment: it would create structures in which in the future the agents to be held accountable will actually be responsible.For a wide ranging critique of the various concepts of responsibility and accountability in public administration, see Harmon (1995).8.These criteria, which parallel the classic definition in Aristotle (1963, Bk. III.1-5), raise notoriously difficult philosophical issues.For a survey, see Eshleman (2009).For a legal analysis that informs the application of the criteria here, see Hart and Honoré (1959).9.The accounts presented here are necessarily selective, and should be regarded more as illustrative sketches than conclusive assessments.Also, there are many other cases that invite a similar analysis and suggest similar lessons, notably the government response to Hurricane Katrina (Boin et al., 2010), the Enron scandal (Coffee, 2002;Gordon, 2002), and the Challenger disaster (Adams & Balfour, 2009, Chapter 5 ;Vaughan, 1997).10.The book that identifies the most hands is appropriately titled All the Devils Are Here (McLean & Nocera, 2010).As one reviewer wrote, "Some of us want to tie the financial crisis to Wall Street and Washington.Others want to blame greedy and ill-informed consumers, rouge traders and brokers, out-of-control lenders and people with a Pollyanna view of the world.McLean and Nocera make a convincing argument that it's all of the above.And more" (McNay, 2010).For a discussion that links the financial collapse to broader historical and cultural trends in American society, see Adams and Balfour (2012).11.A similar institutional conflict of interest contributed to the Enron scandal: Enron's auditors (Arthur Andersen) did not ask hard questions partly because they had an especially close relationship with Enron executives (Thompson, 2005, pp. 246-57, 261-62).Like many accounting firms, Arthur Andersen had a large consulting contract with Enron, which was more lucrative than the auditing arrangement.Many of the financial executives at Enron had earlier worked at Arthur Andersen.