Printer Friendly

Information escrows.

1. Legal Issues

Moving beyond theory, the remainder of this Part grapples with difficult questions of implementation. We begin in this Section with the relationship between allegation escrows and the antidiscrimination mandates of Title IX and Title VII, and then turn to the legal relationship between the escrow "agent," the depositor, and the sponsoring institution.

a. Allegation Escrows in Post-Secondary Schools and Title IX

Any mechanism for reporting sexual harassment in colleges and universities receiving federal funds operates against the background of Title IX of the Education Amendments of 1972, which prohibits gender-based discrimination in federally supported educational programs. (79) In Franklin v. Gwinnett County Public Schools, the Supreme Court interpreted gender discrimination under Title IX to include sexual harassment. (80)

While Franklin declared that schools may be held liable when officials intentionally fail to end harassment, (81) the Department of Education ("DOE") has developed a complex regulatory scheme to ensure compliance with Title IX. (82) Specifically, school officials must (1) investigate and address harassment when they reasonably should know about the conduct, even prenotice, and (2) undertake some investigations even when victim confidentiality cannot be assured. (83) These requirements may make schools reluctant to implement escrowed reporting systems, but may also increase the potential benefits of such systems to victims. Risk-averse schools may fear that federal investigators would view any conduct related to an escrowed complaint as within those "reasonably should know" bounds, leading schools to incentivize public reporting as much as possible and to resist any "official" mechanism that fails to inform the school of alleged harassment. On the other hand, the requirement that all such complaints be investigated without guaranteeing confidentiality may heighten the concerns that keep victims silent in the first place. (84)

Nonetheless, an allegation escrow system might help schools fulfill their Title IX obligations. The Department of Education has suggested that one "reasonable" method of identifying harassment prenotice is seeking out cases that resemble previously submitted complaints. (85) Schools could utilize escrow systems to fulfill that obligation by submitting any claim received, and in the case of a triggered complaint, contacting the users whose escrowed allegations match. Administrators might further pursue their Title IX obligations by responding to aggregate data suggesting that a particular department, fraternity, or physical location on campus repeatedly engages in misconduct. While the specifics of unmatched deposits would not be released, the interim disclosure that a certain number of allegations of sexual misconduct had been deposited relating to a particular department might provide an impetus for administrators to take corrective action.

b. Allegation Escrows in the Workplace and Title VII

Closely related to harassment in colleges and universities is sexual harassment in the workplace, which is governed primarily by Title VII of the Civil Rights Act. (86) While Titles VII and IX share a common goal of preventing discrimination, differences in standards of liability and affirmative defenses may make employers even more reluctant to implement information escrow systems than schools. Under Title VII, employers are required to take a much more active role than schools in the prevention and investigation of sexual harassment in the workplace. (87) Indeed, the leading cases in Title VII suits for sexual harassment, Faragher v. City of Boca Raton (88) and Burlington Industries, Inc. v. Ellerth, (89) indicate that a primary aim of Title VII is encouraging employers to prevent and quickly address sexual harassment. Much of Title VII jurisprudence focuses on whether employers have taken reasonable steps to prevent and promptly remedy harassment in their workplaces. Reasonable steps are often interpreted to mean providing harassment awareness training and reasonably accessible mechanisms for lodging harassment complaints. (90) If an employee establishes evidence of a hostile work environment (which does not rise to the level of tangible employment action), the employer can avoid both compensatory and punitive damages by establishing the affirmative defense that the employer exercised reasonable care to address sexual harassment and that the subordinate complainant unreasonably failed to take advantage of remedies available in the workplace. (91) Such a focus on the employer's actions and the get-out-of-jail-free card that reasonable, employer-provided complaint mechanisms represent understandably make employers eager to uncover as quickly as possible any cases of harassment that could lead to litigation. (92)

Pennsylvania State Police v. Suders (93) gives large-scale employers even more reason to focus on demonstrable steps to prevent and correct harassment as a means of avoiding liability, and thus to fear unreported cases of harassment. Seemingly forging a middle path between the strict liability of tangible employment-action cases and the avoided liability of hostile environment cases, the Supreme Court held in Suders that Title VII allows constructive discharge claims, and that such claims can, in the most severe cases, rise to the level of tangible employment actions. (94) The Court also held, however, that in all but the most egregious constructive discharge cases, employers will have recourse to the affirmative defense that the employee unreasonably failed to take advantage of employer-sponsored steps to prevent and correct harassment. (95) This approach to constructive discharge claims broadens employee access to courts in Title VII suits, as it removes the affirmative defense as a basis for summary judgment where there is a reasonable question as to whether the employee quit in response to harassment that constituted both a hostile work environment and a tangible employment action. (96) At the same time, however, by limiting the strictest standards of liability to only the most grave constructive discharge cases, the Court also increased the importance and prevalence of the reasonable-employer-actions affirmative defense in Title VII cases. Suders thus serves to increase an employer's interest in having publicly demonstrated mechanisms for reporting harassment, but may also increase employer resistance to escrow-based reporting systems which do not automatically notify the employer of complaints.

As with schools, employers should be able to develop allegation escrow systems that help, rather than hinder, their effort to comply with the requirements of Title VII. The escrow service would merely represent an additional option for victims who are uncomfortable bringing forward direct allegations. But as long as victims could easily bring direct complaints and had the option of subsequently converting escrow deposits into go-it-alone direct complaints, an employer sponsoring an escrow service should be able to avoid Suders liability for failure to have adequate harassment reporting mechanisms. However, given the increasingly narrow focus on an employer's reasonable efforts to prevent and correct harassment, employers may nevertheless resist implementing escrow systems for fear that they will lead to a perception of employer indifference or even that employers are actively obstructing victims' claims from seeing the light of day. These concerns might be allayed by evidence that harassment investigations increase after the introduction of the escrow option. But employers may nonetheless be concerned that courts would focus on the large proportion of orphaned complaints that would eventuate in even successful implementations of the escrow. Whether or not their reluctance to implement escrowed reporting mechanisms is reasonable, employers may ultimately find the current Title VII framework too comfortable to risk changing.

c. The Escrow System's Relationship with Sponsoring Institutions

Though specific requirements vary by jurisdiction and job sector, nearly all employees enjoy a legal or contractual right to review documents used by an employer in making hiring or employment decisions. (97) As a result, any escrow system should be clearly and legally distinct from the companies that it serves. Before any escrowed allegations are accepted, the escrow service should establish that the employer has no claim to any of the complaints lodged with the service. Indeed, the escrow service might not even have a contractual relationship with the companies at all. The service might only enter into contracts with depositors to forward complaints under prespecified conditions. Regardless of whether the employer sponsors the escrow service, the employer should only be granted access to complaints when the escrow service chooses, in its judgment, to forward a match generated by the system.

The independent juridical status of the escrow service would further insulate the service from being held liable for the employer's lack of response to gender inequality in the workplace. In addition, the escrow service would benefit from making explicit its fight to use its judgment regarding matched complaints and whether they are suitable for forwarding to the employer or the authorities. Because there will inevitably be error in a process that operates in a realm of unproven allegations and vague descriptions, the escrow service should avoid any implication that it has an obligation to forward complaints that happen to meet certain criteria. Similarly, the legal independence of the escrow service can help insulate employers from potential liability for failure to respond to deposited allegations awaiting a match, since with an independent escrow service the employer will be unaware of such allegations.

A strong separation between the sponsoring institution and the escrow system is also important to protect the integrity of the allegation escrow system. Both schools and employers have an interest in identifying potentially damaging types of behavior as soon as feasible, and in the absence of proper separation, these institutions may put pressure on the allegation escrow agents to look for and share evidence of that behavior. Both schools and employers may seek information about illegal activity, for example. Schools might want to know about academically dishonest or risk-taking behavior, while employers would be eager to know about employees embezzling or mishandling company resources. Nonetheless, using submitted allegations to identify such information would be a significant subversion and abuse of the system, as well as of the trust users had placed in the allegation escrow mechanism. The best defense against such unacceptable institutional encroachment is a firm and clearly delineated distinction between the third-party escrow provider and the sponsoring school or employer.

d. The Escrow System's Relationship with Users

Tailoring the Service's Core Ministerial Duties. In any allegation escrow mechanism, managing the relationship between the escrow system and the user is of critical importance. At a minimum, the escrow agent would owe to a depositor the duty of making good-faith efforts (1) to preserve the deposit; (2) to keep the deposit confidential both prerelease and postrelease (with regard to all except the prespecified recipients); (3) to inquire whether the deposit matches with other deposits (and meets the requisite for release); and (4) if the release is triggered, to forward matching deposits to the pre-specified recipients. While game theorists tend to refer to the escrow service as an "agent" of the depositor, we tentatively think that the nonfiduciary contractual duties of good faith better balance the interest of the escrow agent and the multiple depositors rather than turning the service into an agent with multiple principals. (98) The escrow agreement between the depositor and the escrow service should establish the perception that the service is simply a ministerial channel through which allegations may pass. As with traditional escrows, these core duties are ministerial and largely nondiscretionary. Indeed, one could imagine an escrow contract that minimized escrow release discretion. Under such a system, the service would not substantively review (or even access) the content of the complaints. It would merely test whether two deposits were directed toward an accused with the same identifying information, such as an email address, and only then release the deposits to the prespecified investigative entity. We tentatively reject such a pure ministerial system.

To preserve the integrity of the escrow process, we recommend that the escrow contract specify that the escrow service have sole discretion to substantively review the content of deposits and have discretion not to forward deposits that the escrow service deems to be made in bad faith or to be insufficient in substance to warrant matching. The purpose of such discretion is to protect both bona fide depositors and the accused from a release triggered by complaints that the escrow agent deems to have been made in bad faith. Therefore, the contract ought to require complainants to acknowledge that the only way to ensure that someone will read the complaint is to pursue a stand-alone complaint. The escrow agent should develop, but keep confidential, processes to help ensure that the deposits are made by bona fide complainants. For instance, by requiring depositors to demonstrate that they have access to a valid student or alumni email, the escrow agent can begin to ensure that the depositors have not misrepresented their names and have at one time been in residence at the university. Finally, the escrow system should explicitly and clearly disclaim any responsibility for the content of the submissions and emphasize that depositors are warranting their representations to be true. Making clear that the escrow's obligations are dominantly ministerial will serve to minimize the risk that the escrow itself would be held liable for defamation. (99) Moreover, the independence of the escrow from the employer may allay some victim concerns of institutional bias. (100)

In a longer working paper version of this Article, we address in more detail the legal and programming issues involved in our attempt to bring a harassment allegation escrow into being as a nonprofit internet website. (101) But here we focus on a few game-theoretic choices about how to design the escrow agreement--choices concerning the escrow trigger and interim reporting--that might powerfully impact the escrow equilibrium. The triggering mechanisms determine the conditions under which the escrow deposits will be released, to whom the deposits will be released, and potentially for what purposes. Since the escrow mechanism is in essence a contract specifying the contractual duties of the escrow agent (who might or might not be a fiduciary of the depositor), escrow depositors might specify the contract in a variety of different ways--giving the depositors ex ante freedom and even various aspects of ex post freedom. One approach to choose among these particular design options would be to speculate about the types of contractual terms that are most likely to be favored by the victim class.

Tailoring the Disclosure Recipient. The mechanism might specify that upon receiving the prespecified number of matching deposits, the escrow agent would merely reveal the identity of the claimants to one another (or reveal their identity and their underlying claims) and let the claimants decide postmatch whether and how to proceed. Granting claimants this kind of ex post power might induce more deposits. On the other hand, it might also induce more fallacious deposits from individuals who merely make a deposit to learn the identity of other claimants. Because we want depositors to take the process seriously, and because we want to protect good-faith depositors from subversion of the system, we prefer a design where matched deposits are automatically forwarded to the proper authorities as actionable complaints. Potential depositors lose their own freedom not to proceed, but they gain the assurance that the claims of other depositors will be available for investigation. The relative inflexibility of automatic forwarding combined with the formal or social consequences visited upon fallacious complaints may also reduce the number of false deposits and resulting Type II errors.

Tailoring the Trigger. The mechanism must also specify the number of deposits lodged against a particular harasser that will trigger disclosure. We have thus far focused on an implementation where the escrow agent offers an across-the-board trigger of 2 (or possibly 3) deposits. But it would be possible to allow individual depositors to choose the trigger with which they are most comfortable. Under such a system, escrowed deposits would only be released if the set of deposits existed for which the trigger conditions were met for all the depositors in the set. For example, imagine a succession of deposits against a particular accused, where the claimants choose triggers of 4, 99, 2, 3, and 4. The third deposit would not trigger release, even though that depositor is comfortable with just two matching deposits, because there is not another depositor with a trigger of 2. The fourth deposit also would not trigger release because there are not three depositors with triggers of 3 or fewer (or subsets of two depositors with triggers of 2). (102) Like ex post control, we can imagine that an ex ante choice of trigger might increase deposits, and in particular might attract deposits from those complainants who are the most reluctant to go it alone and feel secure only with many co-complainants. But it would be difficult to communicate the workings of this escrow adequately to depositors--who might not realize that their escrow could be orphaned even though their personally chosen trigger requirement had been met. Moreover, our simulation leads us to think that depositor-chosen triggers are likely to lead in equilibrium to fewer harassment investigations on net (even if there are more escrow deposits). (103)

Tailoring Depositor Acceleration and Withdrawal Rights. The escrow contract must also specify whether a depositor can rescind or potentially accelerate the release of an as-of-yet unmatched claim deposit. Rescission and acceleration can be thought of as midstream alteration of the trigger number. Rescission effectively increases the trigger number to some unreachably high number to ensure that the deposit would never be released. Acceleration has the effect of decreasing the trigger number--potentially to 1, meaning that the claimant would be willing to "go it alone" and have her claim deposit forwarded immediately, unaccompanied by even a second claim. This is yet another dimension where theory and current data do not provide a clear, a priori solution. But we tentatively prefer an asymmetric system, in which a depositor can, anytime after making a deposit, accelerate her trigger to make a direct go-it-alone claim but cannot decelerate (or rescind) her deposit. We favor the acceleration option, because of the positive externalities of released claims. If a victim, after making a deposit, is willing to "go it alone," it furthers her private interest and the public interest in adjudication and deterrence to immediately lodge the complaint on her behalf. (104) Indeed, another advantage of the escrow system is that it can provide complainants with a simple method of creating a contemporaneous account of their allegations with the continuing possibility of turning the deposit into a go-it-alone complaint at any point in the future. Just the process of privately giving voice to their narratives as part of making a deposit might be sufficient to make some victims willing to move forward by themselves. The process of naming and blaming can itself be transformative and lead to claiming. (105)

The acceleration option might also be used by depositors who, over time, find themselves in a more empowered position. Indeed, because sexual harassment claims are subject to statutes of limitations, a depositor might eventually face a choice between submitting a direct complaint and foregoing the possibility of future litigation. By notifying depositors that the statute of limitations is about to run against the accused with regard to their claims, escrow agents might prompt depositors to transform an escrowed allegation into a go-it-alone complaint. Because we assume that most allegation escrow users would remain silent in the absence of an escrow system, that transformation also encourages the submission of official complaints regarding incidents that would otherwise likely have gone unreported. Creating the acceleration option can thus mitigate the problem of orphaned deposits, which stands as a chief contraindication of implementing an escrow regime.

The problem of orphaned escrows is also a chief reason why we (by a slight margin) prefer an asymmetric system where depositors cannot change their minds and rescind or cancel their deposit. Rescission or cancellation would effectively ensure that claims remain orphaned. Society would be deprived of the deterrence value of rescinded complaints. Moreover, we worry that rescindable deposits might be taken less seriously by depositors who retain the right to cancel if their deposit does not immediately match. We also worry that some depositors might too readily rescind if they find that their deposits did not trigger an escrow release. (106) These depositors would only make a short-term deposit to find out (by the immediate release) if another deposit was already outstanding against a particular professor. On the other hand, the requirement that escrows be nonrescindable might dampen the initial deposit rate more than the rescission option depresses the amount of deposits for potential match. In the absence of persuasive evidence on this issue, we have a slight preference for making deposits nonrescindable for a period of one or two years. Temporary nonrescindability makes clear to victims at the time of deposit that they cannot simply place a temporary deposit to test whether someone else has claimed against a particular professor. Depositors must be serious in making an escrow deposit because they will not be able to immediately change their minds about the fact of the matter at a later time. (107) Depositors should be exposed to some negative scrutiny if they make a deposit that investigators later judge to be fallacious.

Tailoring Interim Disclosures. Finally, the escrow contract should clearly delineate the uses, if any, that might be made of deposit information while the complaint remains unmatched. For example, it would be possible for the escrow agent to inform the accused that an escrowed allegation had been levied against him without revealing the name of the accuser making the deposit. The goal of such interim disclosure would be to potentially deter the accuser from harassing other students because the accused would be on notice of a potential future investigation. An additional benefit would be that such a warning might induce an innocent defendant to submit a defense escrow report, should he be concerned that a particular disgruntled student might have submitted a false claim. However, the downside behind such interim disclosure to the accused is that it might trigger accuser retaliation against harassment victims. (108) Accordingly, we propose a regime where interim anonymous reports to the accused are not made, and indeed in which deposits are treated as if they are in a black box until matching. However, this treatment might merely be a default that individual depositors could contract around if they were comfortable in interim notice being made, such as when a student depositor limits her complaint to inappropriate professorial conduct in a large, lecture-based class. In that case, the complainant may be unwilling to make her allegations public in a go-it-alone fashion, but she may also feel confident that the accused will be unable to identify her after learning that a student in the lecture has submitted an escrowed complaint.

As we discuss in greater detail below, interim reports of varying granularity might also be revealed to the public or the university to further other interests besides specific deterrence of the individual harasser. We favor disclosing to the public aggregate information on the number of allegation escrows that have been deposited with respect to employers with a sufficiently substantial number of employees (combined with information on the number of deposits released from escrow). By letting victims know that other people have been making use of the escrow mechanism, disclosure can raise the salience of the mechanism, and perhaps lead to limited general deterrence by maintaining a visible enforcement presence on campus. It might even be possible to disclose the number of deposits for subgroups of workers (for example, for different schools within a university). The factors limiting the degree of granularity should be whether a harasser is likely to infer that a complaint has been deposited against him, and whether any member of the university is likely to infer who is submitting complaints. The total number of Yale University deposits increasing by one deposit tells a particular professor very little. But the total number of Yale Law School deposits increasing by one deposit might tell a harassing law professor (shortly after an episode of harassment) that he has been accused. (109)

In thinking about interim reporting, we should distinguish between individual malfeasance and institutional malfunction. Besides deterring individual acts of harassment, the escrows might be designed with an eye toward alerting human resource administrators about a more systemic problem. Instead of designing an escrow system to solely respond to the problem of repeated harassment by particular professors, it might also be possible to design a system to respond to more pervasively hostile educational or employment environments. For example, imagine that the escrow agent learns that seven harassment deposits have been received accusing different professors in the math department. Even if the individual deposits are not sufficient to sanction any of the individual professors, good-faith administrators might, if informed of the separate allegations, have sufficient evidence to take other kinds of action to mitigate a hostile atmosphere in the department. (110) Escrow designers should contemplate whether there could be different release triggers and potentially different triggers for different types of proceedings or uses. (111) Thus, while the escrow agent might not publicly report department-specific escrow amounts, it might be useful to reveal these counts to administrators for departments that display an inordinate number of deposits. It would be possible for the escrow agent to go beyond these more granular department-specific counts and reveal to the administration the allegations themselves. Indeed, the escrow mechanism might even include a second trigger specifying release if a certain number of deposits were received relating to a department during a particular period of time. Thus, a victim depositing an allegation against a math professor might know that the allegation will be made public if either (1) another allegation deposit is received relating to the same professor, or (2) three other allegation deposits are received relating to harassment in the math department during any three-year period. This second trigger would be better tailored to investigate and root out more pervasive atmospheres of harassment or discrimination. And as before, depositors could rest assured that they were not alone in making their allegation in the sense that the second trigger would only be met if four relatively contemporaneous claims of department harassment were being made. We tentatively conclude against secondary triggers that complicate both agents' matching process and the explanation that must be made to potential depositors.

This brief discussion of triggers and interim reporting only scratches the surface of the manifold possibilities of escrow design. While allegation escrows initially seem as though they might be designed by a simple act of deposit and a subsequent release if prespecified conditions are met, we have shown that there are literally dozens of permutations on this basic design, as well as critical issues of judgment and context-specific tailoring. In the remainder of this Part, we explore in greater detail some of the more practical issues in designing and launching an information escrow mechanism.

2. Managing Salience

Allegation escrow systems depend on sufficient user participation in order to fulfill the objective of lowering first-actor barriers to action and assisting communication. Perhaps the worst outcome for an allegation escrow system would be having all or nearly all submissions orphaned, as any complaints that would otherwise have been submitted directly would represent a net loss in the number of official allegations submitted by sexual harassment victims. (112) Maximizing the number of users (113) is thus critical to minimizing the likelihood of an overall negative impact and maximizing the social and user utility of information escrows.

Attracting submissions from complainants who would otherwise have remained silent presents two primary obstacles: exposure and user-perceived inertia. The success of an allegation escrow depends on a sufficient number of harassment victims being aware of the system and how to use it. Similarly, it is critical to publicize effectively the confidentiality and other benefits of escrows to potential victims. However, escrow developers may also need to overcome a lack of faith in the system. Many complainants might consider the hassle and emotional strain of submitting an escrowed allegation worthwhile only if there is a reasonable probability that other victims will participate. Users who perceive the escrow to be untried or unpopular may thus neglect to submit an allegation for fear that the system will not generate matches when appropriate. This inertia problem is self-perpetuating. To combat this risk, allegation escrow developers should address the marketing aspect of introducing the new system, as well as the problem of user-confidence inertia.

In our view, the first iterations of allegation escrows should follow the model established by the popular social networking site Facebook. Facebook's successful rollout strategy began by limiting the site to individual college campuses. (114) The site was originally located at Harvard University where it was an immediate local sensation. (115) It then moved to other Ivy League schools before being introduced at all U.S. and some non-U.S, universities. (116) By the time Facebook was introduced to the general public, its popularity and reputation were well established. (117) Following that pattern of success, allegation escrow systems should begin by targeting well-known universities, such as Yale, that are already publicly addressing Title IX, harassment reporting, and other gender-based issues. Having (hopefully) gained some public attention as a result, the allegation escrow system should then be introduced to other universities and educational communities. Finally, if the escrow systems are well received and their utility demonstrated, they should be expanded to groups, communities, and sectors beyond education.

The purpose of a narrowly targeted rollout model is not to achieve or even aspire to the wild popularity of websites like Facebook and GMail, but rather is to take advantage of the built-in benefits that accompany locally tailored site introductions. Targeting universities for the first wave of allegation escrows reduces the likelihood and severity of institutional resistance. As we mention above, many universities are already revamping their harassment reporting processes or could benefit from doing so. (118) Large-scale employers, on the other hand, have several legal and practical reasons to resist allegation escrows for as long as possible. More importantly, targeting schools likely to respond positively to allegation escrows mitigates both marketing and inertia concerns. A closed universe of potential complainants and targets makes it easier to inform potential participants about the new escrow system and its benefits. Similarly, a limited population of participants and targets would reassure complainants that any potentially matching targets are almost certainly aware of the escrow system. Finally, beginning at schools that are already addressing Title IX, harassment reporting, and other gender-based issues fosters the recognition, enthusiasm, and energy that can lead to widespread use of allegation escrows.

3. The Matching Algorithm

Design of an algorithm to determine when two deposits match is an essential part of escrow design. For example, while the foregoing examples concern matching of escrow deposits based on the identity of the individual harassers, one might instead design the escrow to allow matching on the basis of group identity, such as conduct by an entire department. Allowing group-based matching complicates any allegation escrow mechanism, but may be necessary to establish patterns of conduct for a proper investigation. However, group-based matching may require an increased level of participation by an escrow system agent, thereby introducing labor costs and the possibility of human error. Escrow developers would also have to decide whether to infer a group complaint from an individual complaint. For example, six complaints each identifying a different faculty member in the math department could be forwarded as an anonymous group-based complaint. Finally, group-based matching would also probably require a separate and difficult-to-define triggering threshold for forwarding to officials and authorities. (119)

Perhaps the most difficult challenge for the escrow agent reviewing matched allegations is ensuring that the allegations are similar enough to warrant forwarding. In the university context, for example, an allegation of a faculty member requesting sexual favors on the threat of grade retaliation is fundamentally different from a complaint that describes the same faculty member's pattern of making inappropriate jokes in lecture. Both complaints are entirely valid and legitimate, but describe behavior of vastly different severity. Many escrow systems may treat the two claims as different enough that they fail to trigger a forwarded match, for fear that the complainant alleging the more serious conduct would essentially be placed in a position very similar to the one she avoided by refusing to submit a go-it-alone complaint. (120) The difficulty in making this judgment is determining where to draw the line between matched allegations that are similar enough to trigger official reporting and those that are not. Inevitably, many of the matched allegations will present close calls, significantly complicating the review process. (121)

The consequences of delaying matched allegations are potentially significant. At minimum, a screener's choice to treat two matched allegations as fundamentally different increases the risk that both allegations will be functionally orphaned. In addition, because this dilemma implies that one of the complaints alleges serious harassment, even a short delay in forwarding matched allegations exposes the complainants and others to the risk of continued harm. On the other hand, the screener must assume that but for the option of waiting for a sufficient match, the complainant would have chosen to remain silent. To the extent that officially submitting two vastly different claims effectively forces the users to go it alone, forwarding poorly matched allegations subverts the aims of the allegation escrow system and directly harms complainants. Ultimately, the screener should decide whether to err on the side of over- or underreporting matched allegations, and then accept that some controversial or regrettable decisions are inevitable. (122)


Information escrows have a broad range of potential applications in addition to our central example of sexual harassment complaints. In this Part, we will rely on the theoretical foundation developed in Part II to explore whether the prerequisites for useful application of the escrow tool exist. The goal of this Part then is to identify new potential applications and to assess whether an escrow is likely to be on net socially beneficial.

A. Applications for Allegation Escrows

1. Sexual Harassment, Date Rape, and Other Sexual Assaults

It is common in high-profile incidents of sexual harassment to have other victims step forward with similar accusations. Perhaps the most currently salient and disturbing is the story of Jerry Sandusky, who was a popular defensive coordinator for the Pennsylvania State University football team. After Sandusky's arrest in early November 2011 on charges of child molestation, ten additional victims came forward and alleged similar abuse. (123) The tremendous damage that the university suffered as a result of the sheer number of accusations, to say nothing of the harm to the children, highlights both why schools might feel pressure to uncover as many claims of abuse as possible, and why they might benefit from the use of allegation escrow systems in that effort.

Stories of high-profile follow-on complaints are also common in national and global politics. Shortly after Nafissatou Diallo's accusation led to Dominique Strauss-Kahn being charged with sexual assault, (124) Tristane Banon, a French journalist, publicly accused Strauss-Kahn of attempted rape. (125) We saw this same pattern when Bill Clinton was accused of multiple instances of sexual misconduct throughout the 1990s in the wake of the Paula Jones sexual harassment lawsuit. (126)

Beyond these high-profile cases, however, empirical studies on sexual harassment support the idea that information escrow systems can add value to current reporting systems. Escrow systems are of course useful only inasmuch as a perpetrator is likely to harass multiple victims. An early study on sexual harassment in the federal government found that "many women and men reported that their harasser had also bothered others at work." (127) A more recent statistical study based on labor arbitration decisions summarized the number of victims per perpetrator: (128)

Variables       M      SD     N      1      2      3      4      5

1. Number of   2.32   2.12    92     --

8. Number of   4.77   3.58    92     .32    .09    .18    .30    .42

Variables       6      7      8

1. Number of

8. Number of    .38    .55    --

Data from Margaret A. Lucero et al., An Empirical Investigation of
Sexual Harassers: Toward a Perpetrator Typology, 56 Hum. Rel. 1461,
1470 (2003).

This data indicates that perpetrators of sexual harassment typically harass multiple victims, comporting with research that has found a recidivism rate for general sex offenders of 61.1 percent. (129)

Nonetheless, there are significant barriers to reporting sexual harassment, thus making it difficult to assess the extent of the problem. Studies seeking to measure the incidence and prevalence of sexual harassment tend to rely on surveys that, while useful, are subject to selective response, lack of response, and other methodological issues. (130) A leading direct survey suggests that the prevalence of sexual harassment on postsecondary campuses remains shockingly high. (131) Of those students surveyed, 62% indicated that they had been sexually harassed in some way. (132) Even more alarming, in a phone survey, 2.8% of the college women respondents indicated that they had experienced either an attempted or a completed rape (133) in the previous 6.91 months. (134) A follow-up study further found that just 4% of rape victims inform a college official, while only 2.1% of victims report incidents to police. (135) Finally, a 2007 study found that college rape victims are more likely than victims of other types of crime to be repeat victims. 136 Such repeat victimization may contribute to a high rate of underreporting, particularly when measured on an incident-by-incident basis.

There is reason to be skeptical of the conclusions in both of these studies, since the very factors that make underreporting a problem may also lead to unreliable responses to a direct survey. Studies focusing on existing reports are clearly insufficient, and surveys are both highly sensitive to design issues and notoriously inaccurate. (137) School administrators and policymakers do not need to know the full and precise extent of sexual harassment on campus, however, to recognize that both harassment and underreporting are serious problems. Because the likelihood that matched reporting will augment total reports increases as the prevalence and underreporting (138) of harassment rise, the data above suggest that colleges and universities are good candidates for allegation escrow systems.

Studies addressing sexual harassment among working adults indicate that the prevalence of sexual harassment in the workplace is equally alarming. Some studies suggest that perhaps as many as 50 percent of women in the workplace have experienced some type of sexual harassment and that harassment targets tend to be repeat victims. (133) Female employees also often decline to apply the "harassment" label to incidents that otherwise meet all definitions of sexual harassment, perhaps in an effort to improve their working environments by ignoring inappropriate conduct, or because they are reluctant to acknowledge that the incidents have upset them. (140) As with educational settings, data indicating a high prevalence of sexual harassment in the workplace suggest that workplace harassment reporting may be a particularly promising application for information escrow systems.

2. Whistle-blowing and Allegations of Nonsexual Wrongdoing

Beyond our principal examples of sexual harassment reporting in universities and workplaces, properly tailored allegation escrow systems have the potential to significantly and positively impact information sharing not only with regard to other types of harassment and discrimination, but also with regard to a wide variety of other types of misconduct. The allegation escrow tool could be deployed to ameliorate potential first-mover claiming disadvantages regarding whistle-blowing. The antiretaliation protections afforded to whistle-blowers (141) are not always sufficient to allay a potential whistle-blower's concerns that she will be subjected to serious social and economic consequences if she reports. For example, the Los Angeles Police Department ("LAPD") formerly had a practice of assigning reporting officers to "freeway therapy," whereby the offending reporters would encounter punitive transfer far from their homes and colleagues, thereby also significantly extending their commutes. (142) Unofficial retaliation need not be so blatant to effectively deter reports of wrongdoing. Whistle-blowers often worry that they will be labeled "troublemakers," passed over for promotion when more than one equally qualified candidate exists, and suffer social consequences in the workplace. (143) Matched reports of wrongdoing in the workplace would benefit from the credibility enhancing and group-safety functions of allegation escrows. While an escrow would probably not eliminate unofficial retribution, it would make egregious patterns easier to prevent, identity, and prove.

Moreover, directed allegation escrow systems could benefit users anytime there is reason to establish a credible record of repeated conduct before making a public complaint. For example, individual citizen oversight of complaints of police or other government misconduct may fail to attract attention in the absence of strong community support. What's more, state authorities have been criticized for making little more than perfunctory investigations of citizen complaints unless and until a pattern of misconduct has been established. (144) Establishing such a pattern presents a problem to complainants, however, since allegations lodged after an initial go-it-alone complaint can be accused of copycatting. Escrow systems that allow communities or groups to collect a set of complaints before making them public might thus help to both establish a pattern of wrongdoing and support the credibility of the individual complaints.

Allegation escrows could also ameliorate first-mover disadvantages when bringing private qui tam actions. Under the False Claims Act, a private party may bring a civil action in the name of the United States against a defendant for defrauding the government. (145) To encourage such private suits, the Act establishes an award of up to 30 percent of the proceeds of the action, depending on whether the Department of Justice subsequently intervenes in the case. (146) Moreover, the Act expressly limits this right to the first person to bring the claim. (147) Similarly, the Dodd-Frank Act established a whistle-blower program awarding 10 to 30 percent of the proceeds obtained from original information leading to the successful enforcement of securities laws. (148) Thus at first glance, these Acts seem to establish a first-mover advantage. However, a complainant might worry that she lacks sufficient proof to prevail alone, and may wish to proceed only with other plaintiffs. This is a particularly poignant concern because despite statutory antiretaliation protections, (149) case law has recognized the right of qui tam defendants to bring counterclaims against whistle-blowers. (150) A potential qui tam plaintiff might justifiably worry that she could be exposed to substantial liability were she not to prevail. Allegation escrows thus provide "safety in numbers."

While whistle-blowers are often insiders employed by an entity engaged in wrongful conduct, the example of adverse drug events ("ADE"s) shows that escrows might encourage reports of different kinds of information by corporate outsiders. Pharmaceutical manufacturers are required to report adverse reactions reasonably related to the use of a drug. (151) But patients and physicians are often reluctant to submit voluntary ADE reports because they lack incentive to report and fear negative repercussions or embarrassment if prescription error or patient noncompliance is blamed. (152) An allegation escrow could lead to greater reporting of ADEs by reducing the risk of embarrassment or negative consequences from direct communication. Moreover, an allegation escrow system could also take into account individual reports of the severity and types of reactions, addressing the concern that current reporting leads to little continuity of care when physicians assess the severity of reactions. (153) Nonetheless, ADE escrows would require a significantly higher threshold to trigger the release of allegations, perhaps a level similar to the 10,000 ADEs submitted to the FDA when physicians began to publicize concerns with the controversial diabetes medicine Avandia. (154) Finally, because ADE submissions are controlled by the Health Insurance Portability and Accountability Act of 1996 ("HIPAA"), (155) patients would benefit from legal protections against the use of their medical information outside of the safety-review context, but the potentially serious consequences of an accidental or bad-faith breach of confidentiality significantly increases the importance of information security measures. (156) Nonetheless, given the current underreporting of ADEs, an ADE escrow may ultimately prove worthwhile.

In some settings, insider or outsider whistle-blowers might prefer to create informant escrows, which prespecify conditions of release of allegations to newspapers or other media outlets to ensure further protection against retaliation. (157) When government or corporate actors are unwilling or are perceived to be unwilling to take appropriate action, whistle-blowers may prefer to spark a public debate by prompting news coverage of some controversial set of facts. Informant escrows can analogously allay the retaliation fears of potential sources by adding to a reporter's anonymity the assurance that the reporter has multiple sources.

3. Suspicion Escrows

Allegation escrows might also be useful where individuals suspect misconduct on the part of another. For example, mutual friends of a married couple might be aware of one spouse's unfaithfulness, but hesitate to tell the other spouse for fear of having mistakenly construed the situation. This type of situation demonstrates the problem of wrongdoing uncertainty discussed above. (158) An allegation escrow would permit friends to report a suspicion of adultery that would only be forwarded if a triggering number of other suspicion reports were received. Setting the appropriate threshold for forwarding such suspicion reports would seem to present a challenge, but an average of users' estimates of the number of other individuals aware of the misconduct might be the best approach. Moreover, suspicion escrows might benefit from optional reporting anonymity to encourage friends to share their suspicions without the risk of endangering their relationships with either spouse.

In designing suspicion escrows, it is important to keep in mind that the intended beneficiary may not wish to receive the information. Some spouses may prefer ignorance to unproven allegations of adultery. As discussed above, suspicion escrows could be designed with either a presumption of victim interest or disinterest--allowing the victim spouses to "opt-in" or "opt-out" of receiving suspicion reports. To preserve the victim spouse's interest in not knowing, we favor suspicion escrows that presume disinterest and accordingly require victims to affirmatively opt in to receiving escrow disclosures. As such, suspicion escrows combine aspects of both allegation and shared-interest approaches. A related approach might notify users that the threshold number of reports has been reached, thereby enabling any user to notify the beneficiary that she may wish to opt in and learn of the suspicion. Finally, as the example of Iago teaches, (159) the harm resulting from collusive or bad-faith suspicions could be substantial: a spouse may be unable to repair the damage to the relationship. As with shared-interest escrows, an opt-in approach can minimize this harm by ensuring that spouses voluntarily expose themselves to suspicion information with full knowledge of the risk of bad-faith or collusive reports.

4. Insecurity Escrows

A variation on suspicion escrows could ease communication in sensitive group settings. Often delicate questions arise for which one might desire honest feedback from trusted friends and colleagues, but social customs, insecurities, and concerns about awkwardness prevent a forthright conversation. For example, a professor might wonder if his colleagues think it is time for him to retire, or if his lectures are boring. On a more personal level, a person might want to ask friends whether his recent weight gain is noticeable, or whether he has bad breath. Often, his friends, for fear of provoking hurt feelings, anger, or awkwardness in the relationship, will offer nothing by means of direct communication but politically correct platitudes or tempered opinions.

One way to gather information ambiguously and indirectly is through anonymous surveys such as Survey Monkey and Google Surveys. While perhaps better than a direct conversation, such web-based and purportedly anonymous surveys also include several drawbacks. A colleague might fear that his responses would be identifiable. For example, if everyone who receives the survey gives the same answer, then the professor would be able to infer that all of his colleagues think it is time for him to retire. The colleague might also worry that if nobody responds, the professor would be able to infer that each colleague failed to respond to the survey request.

An insecurity escrow could minimize these concerns. For example, an insecurity escrow might (1) only report a random subset of anonymous responses and (2) only report the random subset if a minimum number of responses is received. Thus, if a professor asked ten colleagues for an opinion, the escrow would only report the anonymous results if at least five colleagues responded and only send on five responses--choosing five at random from the submitted responses. Unlike other shared-interest escrows, an insecurity escrow seeks to fulfill a shared interest while preserving a lack of common knowledge as to respondents' identities and opinions. The triggering threshold ensures that the requesting party could never conclude that no one had responded to his request for information. Similarly, to preserve this ambiguity, respondents would never know whether their feedback was actually forwarded. Finally, the random sampling process would ensure that only a subset of actual responses would be reported to the requesting party. This random subsampling would preserve ambiguity as to whether the results reflected the entire group invited to respond.


This Article has tried to do three things. First, in providing a metatheory for information escrows, we have tried to reveal relationships between a wide array of existing practices. Seen through the lens of information escrows, one can see connections among the disparate practices of everything from Cybersettle and criminal expungements, to adoption consent registries and even GoodCrush. In each of these contexts, private information is deposited with an escrow agent who is only to pass on the information under prespecified conditions.

Second, we have tried to suggest other contexts where information escrows might provide value. In addition to explaining existing practice, we have tried to show that a better understanding of information escrows can help generate new areas where they might be beneficially deployed. We have suggested a dizzying array of possibilities--including insecurity escrows, shared-interest escrows, and even suspicion escrows--as well as highlighting a number of crosscutting design choices--including, for example, presumed consent and interim reporting--which give greater flexibility in managing the potentially disruptive impacts of common knowledge.

Third, we have gone beyond a cataloging of mere possibilities to provide sustained arguments for deploying sexual-harassment-complaint escrows and workplace-dating escrows. Our theory provides no a priori arguments in favor of escrows, but it does suggest conditions when intermediated communication by escrow agents can produce socially enhanced equilibria. Our core application of allegation escrows concerning sexual-harassment-complaint escrows might alleviate the currently significant underreporting problem and trigger more investigations with more credible evidence. Extending the information escrow idea to this new context might play a role in improving the quality of life in our places of work.

(1.) Susan Estrich, Teaching Rape Law, 102 YALE L.J. 509, 518 (1992). In a study conducted by the American Association of University Women ("AAUW"), the AAUW found based on student survey responses that only 7 percent of harassment victims on college campuses report incidents to a school employee, and that 35 percent of harassment victims do not discuss their experience with anyone. CATHERINE HILL & ELENA SILVA, AM. ASS'N OF UNIV. WOMEN, DRAWING THE LINE: SEXUAL HARASSMENT ON CAMPUS 32 (2005) [hereinafter AAUW REPORT].

(2.) See, e.g., RANA SAMPSON, U.S. DEP'T OF JUSTICE, ACQUAINTANCE RAPE OF COLLEGE STUDENTS 9-10 (2002) (listing several reasons for victim underreporting, including "[f]ear that the prosecutor will not believe them or will not bring charges"); Louise E Fitzgerald, Suzanne Swan & Karla Fischer, Why Didn't She Just Report Him? The Psychological and Legal Implications of Women's Responses to Sexual Harassment, 51 J. Soc. ISSUES 117, 122-23 (1995) (listing concerns that "nothing can or will be done" as a reason why some victims do not report sexual harassment and documenting the low success rate of victims who ultimately litigate).

(3.) The stakes in this challenge are high, as the costs associated with continued harassment can be significant. As Chelsea R. Wiliness and others argue in their 2007 meta-analysis of sexual harassment survey data, sexual harassment in the workplace appears to be negatively correlated with job satisfaction, employee productivity, and organization commitment, and positively correlated with task and job withdrawal. Chelsea R. Wiliness, Piers Steel & Kibeom Lee, A Meta-Analysis of the Antecedents and Consequences of Workplace Sexual Harassment, 60 PERSONNEL PSYCHOL. 127 (2007).

(4.) For the sake of simplicity, we refer to a victim of sexual harassment as "she," and to the wrongdoer as "he." Of course, men are also victims of harassment, women are also aggressors, and harassment can occur to both genders. AAUW REPORT, supra note 1, at 3. Similarly, within the university context, we discuss cases in which a professor is accused of harassing a student, when the reverse is also fairly common. See Eric L. Dey, Jessica S. Korn & Linda J. Sax, Betrayed by the Academy: The Sexual Harassment of Women College Faculty, 67 J. HIGHER EDUC. 149, 157-61 (1996); Elizabeth Grauerholz, Sexual Harassment of Women Professors by Students: Exploring the Dynamics of Power, Authority and Gender in a University Setting, 21 SEX ROLES 789 (1989).

(5.) We discuss the possibility of collusion below. See infra note 69 and accompanying text.

(6.) As we discuss below, although follow-on allegations based on triggered reinterpretation of the past are a good thing, escrows can also help when individuals are uncertain about whether they are victims. See infra text accompanying notes 54-56.

(7.) See, e.g., Brian Martin, Whistleblowing and Nanviolence, 24 PEACE & CHANGE 15, 19-20, 23 (1999) (noting bow, in the whistle-blowing context, "isolated resistance is ruthlessly crushed" and suggesting that whistle-blowers would gain much from mobilizing and working together with other whistle-blowers); see also William L.F. Felstiner, Richard L. Abel & Austin Sarat, The Emergence and Transformation of Disputes: Naming, Blaming, Claiming ..., 15 LAW & Soc'v REV. 631 (1980) (identifying other barriers to victims bringing claims).

(8.) See, e.g., SAMPSON, supra note 2, at 4; fan Ayres & Katharine K. Baker, A Separate Crime of Reckless Sex, 72 U. CHI. L. REV. 599, 637 (2005).

(9.) Escrow mechanisms, however, at times, give escrow agents discretion on whether to accept information deposits. For example, bills have been proposed where convicts may have to petition a court conditionally to expunge a criminal conviction. See infra note 16 and accompanying text.

(10.) The nonintuitive possibility that giving people an option of impeded communication would produce a more informed equilibrium parallels an argument made by Jennifer Gerarda Brown and Ian Ayres in discussing the information-filtering role of caucus mediation mechanisms. See Jennifer Gerarda Brown & Ian Ayres, Economic Rationales for Mediation, 80 VA. L. REV. 323, 357 (1994).

(11.) As this Article was being edited for publication, Colin Sullivan of Yale University's Office of Cooperative Development discovered a patent application that relates to some of the escrow implementations discussed in this article. In 1998, John Ogilvie filed a World Intellectual Property Organization ("WIPO") patent application under the Patent Cooperation Treaty, No. PCT/US98/23841 (filed Nov. 11, 1998), for a business-method invention to use "computer networks to both protect message contents by keeping them secret until a specified condition occurs." Id. at 1. The patent application describes an "information escrow" that might be used, inter alia, by "a potential whistle-blower or other witness to some wrongful act or plot." Id. at 1. The patent application describes the implementation of an information escrow with disclosure triggers and a network of dispersed escrow agents that would operate as "a 'hidden choir' which will sing when desired and otherwise remain silently ready in the background." Id. at 4. In contradistinction to the allegation escrow described below, the Ogilvie patent does not contemplate escrow disclosure that is triggered by the deposit of a "matching" escrow deposit by another user. We discuss the existence of other information escrow patents infra in notes 3 l and 40.

(12.) THOMAS C. SCHELLING, STRATEGIES OF COMMITMENT AND OTHER ESSAYS 78-79 (2006); see also IAN AYRES, CARROTS AND STICKS 86-89 (2010) (discussing the Denver addiction clinic and other examples of self-blackmail).

(13.) It would be possible to create a commitment escrow with information that the depositor wants to be communicated but that will only be released by the escrow agent if the depositor achieves some prespecified goal. For example, a commitment escrow might specify that a desired proposal of marriage will only be forwarded if the depositor loses ten pounds. We have not encountered these "carrot commitment escrows" in practice, though.

(14.) One male worried that he wouldn't be sufficiently sell-conscious about having his overweight picture shown on TV, and so he also deposited into escrow an unflattering photograph of his spouse as additional commitment motivation. To this same end, one could easily imagine the commitment website, www.stickKcom (which fan Ayres cofounded), giving users the option of depositing into escrow embarrassing information that will only be released to the public or prespecified email addresses if the users fail to achieve their committed goals. See STICKK, (last visited Aug. 16, 2012). The potential disclosure of some wrongdoing would be an additional layer of accountability that might powerfully motivate users to lose weight or complete a dissertation or business plan. While this Article was being edited for publication, a version of Nalebuff's commitment escrow was offered by a new website. See AHERK!, (last visited Aug. 7, 2012).

(15.) Michael D. Mayfield, Revisiting Expungement: Concealing Information in the 117-formation Age, 1997 UTAH L. REV. 1057, 1057 (1997); Fruqan Mouzon, Forgive Us Our Trespasses: The Need for Federal Expungement Legislation, 39 U. MEM. L. REV. 1, 5 n.15 (2008).

(16.) See, e.g., CAL. PENAL CODE [section] 1203.4 (2004) ("However, in any subsequent prosecution of the defendant for any other offense, the prior conviction may be pleaded and proved and shall have the same effect as if probation had not been granted or the accusation or information dismissed."); MICH. COMP. LAWS [section] 712A.18e (2002) (providing for expungement of juvenile criminal records for one-time offenders lacking felony convictions); H.R. 5393, 92 Leg. Reg. Sess. (Mich. 2003) (bill proposing that expungement be conditional on receiving no subsequent convictions for four years).

(17.) See, e.g., SCHELLING, supra note 12, at 11. This type of dynamic may also be seen in the kidnapping context, where "[b]oth the kidnapper who would like to release his prisoner, and the prisoner, may search desperately for a way to commit the latter against informing on his captor." THOMAS C. SCHELLING, THE STRATEGY OF CONFLICT 43 (1981). A downside of being without sin is that you may not have the means to resort to this protective strategy. People who have ignoble secrets have something to exchange.

(18.) Adam Clymer & Don Van Natta Jr., Family of Robert E Kennedy Rethinks His Place at Libra,, N.Y. TIMES, July 12, 2011, at Al.

(19.) Time capsules are another example of posthumous escrows--they allow the depositors to preserve the deposited information and communicate with future generations.

(20.) A future testator may revoke a prior will and leave the information undisclosed.

(21.) JUSTINCASEIDIE.COM, (last visited Nov. 22, 2011).

(22.) Id.; see also FREDERICK FORSYTH, THE DAY OF THE JACKAL 64 (1971) (describing how a person who made a gun for an assassin used an escrow to make sure he wasn't killed by the assassin by giving a letter to others to be opened if he didn't come home alive).

(23.) This is also known as "source code escrow." For a discussion of source code escrow and a summary of arguments for and against its use, see Walter D. Denson, The Source Code Escrow: A Worthwhile or Worthless Investment?, l RUTGERS BANKR. L.J. 1 (2002). There are several companies that provide such services, such as EscrowTech International, ESCROWTECH, (last visited May 4, 2012), one of the leading source code escrow companies.

(24.) Kalyan Chatterjee & William Samuelson, Bargaining Under Incomplete Information, 31 OPERATIONS RES. 835 (1983).

(25.) See J.L. AUSTIN, HOW TO DO THINGS WITH WORDS (2d ed. 1975) (introducing the concept of "performative utterances").

(26.) Chatterjee and Samuelson also posited that the price might be set so as to give a bargainer with greater bargaining power a larger traction of the overlapping gains from trade (however, with an attendant loss in allocative efficiency). Chatterjee & Samuelson, supra note 24, at 841.

(27.) Roger B. Myerson & Mark A. Satterthwaite, Efficient Mechanisms for Bilateral Trading, 29 J. ECON. THEORY 265 (1983).

(28.) Robert H. Gertner & Geoffrey E Miller, Settlement Escrows, 24 J. LEGAL STUD. 87, 93 (1995).

(29.) Linda Babcock & Claudia M. Landeo, Settlement Escrows: An Experimental Study of a Bilateral Bargaining Game, 53 J. ECON. BEHAV. & ORG. 401 (2004).

(30.) NEB. REV. STAT. [section] 25-2924 (2008) (repealed 2009). The statutory provisions governing settlement escrows can be found in NEB. REV. STAT. [section][section] 25-2922 to -2929 (2008) (repealed 2009). The state's Office of Dispute Resolution made available a useful guide to the escrow process. See NEB. OFFICE OF DISPUTE RESOLUTION, A GUIDE FOR SETTLEMENT ESCROW DISTRICT COURT ONLY! (2003), available at for-party-wishing-to-participate-in-24567.html (last visited May 4, 2012). The program was discontinued in 2004 and the statutory provisions were repealed in 2009. See L.B. 1, 101st Leg., 1st Spec. Sess. (Neb. 2009); Telephone Interview with Rachel Lempka, Dir., Neb. Office of Dispute Resolution (Aug. 5, 2011). While the escrow option was in place, litigants deposited 252 offers, but only one case ever settled using the escrow mechanism (apparently because most of the escrowed settlement offers were unreciprocated). Interview with Rachel Lempka, supra.

(31.) See U.S. Patent No. 6,330,551 col.4 11.51-56 (filed Aug. 6 1998) (issued Dec. 11, 2001) ("The computer matches the settlement offer against the claimant's demand and performs its programmed calculations in order to determine whether or not a settlement has been achieved. Where the demand and offer intersect in accordance with preestablished conditions, settlement is reached."). The patent application was filed three years after the publication of Gertner & Miller, supra note 28, but fails to mention Gertner and Miller as prior art. See U.S. Patent No. 6,850,918 (filed Nov. 29, 1999) (issued Feb. 1, 2005); U.S. Patent No. 6,954,741 (filed Aug. 6, 1999) (issued Oct. 11, 2005); U.S. Patent No. 7,249,114 (filed Oct. 10, 2003) (issued July 24, 2007) (failing to reference Gertner & Miller, supra note 28). In 2001, the game theorist Barry Nalebuff with a host of coinventors unsuccessfully filed a patent application expressly invoking the concept of "information escrows." See A Negotiation Protocol Using a Third-Party Information Escrow, PCT/US01/12081 (filed Apr. 12, 2001).

(32.) How Cybersettle Works, CYBERSETTLE, (last visited Oct. 6, 2011). The Cybersettle mechanism allows each disputant to deposit bids for up to three different rounds of bidding--testing each successive round for overlapping deposits. Russell Weiss, Some Economic Musings on Cybersettle, 38 U. TOL. L. REV. 89, 91 (2006). The Federal Circuit in construing the '511 claims concluded that to practice the independent claims would "require the receipt of at least two demands and at least two settlement offers." Cybersettle, Inc. v. Nat'l Arbitration Forum, Inc., 243 F. App'x 603, 609 (Fed. Cir. 2007). The Cybersettle mechanism also modifies the Chatterjee and Samuelson split-the-overlap price with a "20% exception": "[T]he final settlement can never exceed the demand by more than 20%. For example, if the offer is $100,000 and the demand is $5,000, the case will not settle for the median ($52,500); it will instead settle for $6,000 (20% above $5,000)." Weiss, supra, at 91:see also Bruno Deffains & Yannick Gabuthy, Efficiency of Online Dispute Resolution: A Case Study, COMM. & STRATEGIES, Oct. 1, 2005, at 201, 201 (arguing that the Cybersettle mechanism "creates some crucial inefficiencies").

(33.) See Melissa Conley Tyler & Di Bretherton, Seventy-Six and Counting: An Analysis of ODR Sites, in PROCEEDINGS OF THE ODRWORKSHOP.ORG, EDINBURGH, June 28, 2003, at 13 (A.R. Lodder et al. eds., 2003) (providing as additional examples ClickNsettle, Intersettle,, MARS, Settlement Online, WeCanSettle, The Claim Room, WebMediate, and Dispute Manager).

(34.) See, e.g., The Last Chance Match, HARVARD COMPUTER SOC'Y, (last visited Oct. 6, 2011).

(35.) Nicholas Zamiska, Before Graduation at Yale, A Last Chance for Romance, N.Y. TIMES, May 30, 2004, [section] 14CN, at 6. The average depositor submitted more than six crushes-leading to more matches than people submitting. See id. This suggests that at least some submitters were not looking to connect with that special someone but rather those special someones.

(36.) Crushfinder, GOODCRUSH, (last visited May 4, 2012). See generally Anonymous Matching, WIKIPEDIA, (last visited Oct. 6, 2011) ("Anonymous matching is a matchmaking method facilitated by computer databases, in which each user confidentially selects people they are interested in dating and the computer identifies and reports matches to pairs of users who share a mutual attraction."). A half-dozen other websites have provided crush escrows-including (targeted to the teen market and claiming "more than 1.6 million users and ... more than 600,000" matches),, LiveJournal Secret Crush meme, (claiming 100,000 successful matches),,, and Crush Notifier (Facebook application). Id.

(37.) Crushfinder, supra note 36; see also Hannah Seligson, Will My Love Say Yes? College Kids Ask This Website, WALL ST. J., July 16, 2010, at W9.

(38.) See THE SOCIAL NETWORK (Columbia Pictures 2010).

(39.) Patricio Martinez, Online Dating Site Aims to Aid Lovestruck Cornellians, CORNELL DAILY SUN, Feb. 12, 2010,

(40.) U.S. Patent No. 5,950,200, at [54] (filed Jan 24, 1997) (issued Sept. 7, 1999) (titled "[m]ethod and apparatus for detection of reciprocal interests or feelings and subsequent notification").

(41.) See generally U.S. DEP'T OF HEALTH & HUM. SERVS., ACCESS TO ADOPTION RECORDS: SUMMARY OF STATE LAWS (2009), available at (surveying the approaches of various states with regard to access to adoption records and showing that some states allow access for adult adoptees while others are more restrictive).

(42.) See id. at 5 & n.13 (listing "Alabama, Alaska, Maine, [and] Oregon" as allowing adult adoptee access to original birth certificate upon request). Kansas and New Hampshire allow access to birth records as well. KAN. STAT. ANY. [section] 65-2423(b) (2002) (providing that an adoptee's original birth certificate "may be opened by the state registrar only upon the demand of the adopted person if of legal age or by an order of court"); N.H. REV. STAT. ANN. [section] 5-C:9 (Supp. 2011) ("Upon written application by an adult adoptee, who was born in this state and who has had an original birth certificate removed from vital statistics records due to an adoption, the registrar shall issue to such applicant a non-certified copy of the unaltered, original certificate of birth of the adoptee....").

(43.) See, e.g., ALA. CODE [section] 22-9A-12(c) (LexisNexis 2006) (granting adoptees nineteen years or older access to original birth certificate).

(44.) Two of those forty-one states, Alabama and Michigan, are open adoption states. In these states the shared-interest escrows may serve as a middle ground between no contact and direct contact.

(45.) See U.S. DEP'T OF HEALTH & HUM. SERVS., supra note 41, at 4 ("Approximately 30 States have established some form of a mutual consent registry.... However, eight States will release information from the registry upon request unless the affected party has filed an affidavit requesting nondisclosure.").

(46.) Id.

(47.) See id. at 5 & n.9. Private (or public) information escrows might provide additional options to adoptees and biological parents contemplating contact. For example, in open states, a nonexclusive escrow might facilitate contact between adoptees and biological parents in a more discrete and privacy-respecting manner than direct, unsolicited contact by the adoptee. Indeed, an escrow could allow the biological parent to indicate in advance that he or she would prefer not to be contacted, thereby preventing the invasion of privacy that might result from even receiving notice that the adoptee would like to establish contact. Similarly, while existing websites often allow adoptees to contact their biological parents, an independent escrow could facilitate contact in the reverse direction as well. Through the escrow, biological parents could notify adoptees of the parents' willingness to exchange information. Finally, though difficult, shared-interest escrows could support connections between adoptees and biological parents in the three pure "closed" states. Matching would need to occur among adult adoptees only, and would require sufficient information to match on bases other than name, such as date and place of birth. These practical realities might limit the number of successful matches in closed-adoption states. The high emotional, medical, and other stakes might nevertheless justify the effort of developing and maintaining the adoption escrow mechanism.

(48.) Joseph Farrell & Garth Saloner, Installed Base and Compatibility: Innovation, Product Preannouncements, and Predation, 76 AM. ECON. REV. 940, 940 (1986).

(49.) In a "stag hunt" game, two hunters must individually decide whether to hunt for a stag or a hare. Unlike a prisoner's dilemma, it is a pure game of coordination, where each player would prefer to match the other player's strategy. See John B. Van Huyck, Raymond C. Battalio & Richard O. Bell, Tacit Coordination Games, Strategic Uncertainty, and Coordination Failure, 80 AM. ECON. REV. 234, 235 (1990).

(50.) THOMAS R. EISENMANN, PLATFORM-MEDIATED NETWORKS: DEFINITIONS AND CORE CONCEPTS 7 (2006) (discussing Farrell & Saloner, supra note 48).

(51.) Vince Kuraitis, Overcoming the Penguin Problem: Setting Expectations for EHR Adoption, E-CAREMANAGEMENT BLOG (Aug. 2, 2009), overcoming-the-penguin-problem-setting-expectations-for-ehr-adoption.

(52.) As the disturbing example of former Pennsylvania State University coach Jerry Sandusky (which we describe in detail below) indicates, the need to mitigate these first-mover disadvantages is real and urgent. Many boys endured years of sexual abuse at the hands of an oft-repeat offender. It wasn't until a first mover finally made some of the charges sufficiently public--years after the alleged abuse began--that the full scope of the allegations became known. See infra note 123 and accompanying text.

(53.) And even when follow-on claimants do come through, the first accuser might experience relatively greater overt and reputational harm.

(54.) See AAUW REPORT, supra note l, at 38.

(55.) See Bonnie S. Fisher et al., Reporting Sexual Victimization to the Police and Others: Results from a National-Level Study of College Women, 30 CRIM. JUST. & BEHAV. 6, 8 (2003).

(56.) Similarly, in less egregious cases of harassment, complainants may be more willing to excuse a harasser if the harassment appears to be a one-time incident or mistake. Knowledge of repeated episodes might thus make complainants both more certain of and more offended by improper behavior.

(57.) See AAUW REPORT, supra note 1, at 32-33 (discussing underreporting by students).

(58.) In the real world, harassment takes place over time and victims choose over time whether to file complaints. However, to allow for more tractable estimation, we've assumed that the harassment takes place in an initial stage that might span five years, while victims (uninformed about whether other students were harassed by the same professor) must simultaneously and independently choose whether to file an initial complaint. While the assumption of simultaneity abstracts from reality, many universities keep the identity of harassment defendants nonpublic, see discussion infra accompanying notes 60-61, so that students are often uninformed about whether a prior complaint has in fact been filed--which creates an analogous strategic setting for potential complainants.

(59.) Ian Ayres & Cait Unkovic, (last updated 2012).

(60.) See, e.g., REPORT OF THE COMMITTEE ON SEXUAL HARASSMENT AND ASSAULT PREVENTION EDUCATION IN YALE COLLEGE app. (2008) (providing a survey of Yale University reporting policies).

(61.) For one school policy that deals explicitly with the privacy rights of the accused and the general interest in confidentiality, see BROWN UNIVERSITY, BROWN SEXUAL HARASSMENT POLICY (2012), available at

(62.) After stage 3, the escrow advantage declines because the larger proportion of silent victims in the "no-escrow" regime catch up by bringing more follow-on claims.

(63.) For example, the simulation does not allow for different propensities of silent victims to file stage 3 follow-on claims in escrow and no-escrow regimes. We also hold the distribution of harassing professors constant in the two regimes, even though the higher likelihoods of investigation and higher number of expected claims might deter some recidivist harassers. We also assume perfect or nearly perfect information with respect to reporting options. Relatedly the necessary publicity might inform some professors that their behavior is reportable, and potentially deter their harassing behavior. However, insofar as one ultimate goal is to reduce the prevalence of sexual harassment, such deterrence might nevertheless justify the use of an escrow system.

(64.) It is of course possible that a higher trigger will induce more silent victims to file escrowed allegations. However, our investigation suggests that the increased probability of converting "silents" to "escrows" would have to be unreasonably large to offset the increased orphaning effect of a larger trigger. For example, we estimate, using the parameters of Table 1, that [[alpha].sub.s] would need to approach 36 percent (more than double our current assumption) in order to offset an increase in the trigger from 2 to 3.

(65.) Available survey data suggest that l0 percent is closer to reality, but the sensitive context and vast secrecy make accurate estimates difficult. See AAUW REPORT, supra note 1, at 33 fig.10. The purpose of the table is to make clear that for certain parameter values the escrow option would degrade the information equilibrium.

(66.) See supra notes 60-61 and accompanying text.

(67.) Indeed, we can imagine making participation in the escrow system a first-stage requirement for even those victims who choose to lodge direct complaints. Such primacy might help to publicize the system, and generate trust and interest for those victims who are reluctant to report harassment directly.

(68.) See Felstiner et al., supra note 7 (introducing the distinct requirements of naming and blaming as prerequisites to claiming): see generally Judith Berman Brandenburg, Sexual Harassment in the University: Guidelines for Establishing a Grievance Procedure, 8 SIGNS 320 (1982) (providing a guideline for establishing a grievance procedure and an evaluation of the Yale College Grievance Procedure).

(69.) Of course, the risk of collusion may be higher in escrowed complaints than in direct, public complaints. This risk of collusion might be treated with other types of evidence--for example, evidence that the two escrow complainants attended school at different times and live in different cities would make it less likely that they colluded in escrowing allegations against the same professor. In contrast, the risk of copycat allegations can occur without collusion, and collusion can occur in the absence of allegation escrow systems.

(70.) See generally David S. Davenport, The Confrontation Clause and the Co-Conspirator Exception in Criminal Prosecutions: A Functional Analysis, 85 HARV. L REV. 1378 (1972).

(71.) The most concerning context is one in which the accuser lodges a complaint based on entirely fictional events. In such cases, the professor would have no reason to fear charges or make a defense escrow submission.

(72.) In the extreme biblical example from Genesis, Joseph might have had an inkling that Potiphar's wife would make a false accusation of harassment after Joseph spurned her advances. See Genesis 39:7-20. Similarly, in To Kill a Mockingbird, Tom Robinson might have reasonably worried that Magellan Ewell would falsely accuse him after be spurned her advances. HARPER LEE, TO KILL A MOCKINGBIRD (1960).

(73.) Although if an escrowed world radically increased the salience and publicity of harassment allegations, one might imagine that even false claimants would come to see direct, initial complaints as more "available." See generally Russell Korobkin & Chris Guthrie, Heuristics and Biases at the Bargaining Table, 87 MARQ. L REV. 795, 800 (2004) (discussing the availability heuristic).

(74.) As we'll discuss in more detail below, allegation escrow systems can and should employ techniques to prevent such fishing expeditions, in large part because they can result in premature identification of the complainant and accused. No system is entirely impervious to bad-faith actors, however.

(75.) However, the social disutility of false escrow claims in a world with first-mover claim disadvantages is less clear. The maid, Nafissatou Diallo, who falsely accused Dominique Strauss-Kahn in some way improved social welfare because her false accusation was a but-for cause of Tristane Banon's subsequent accusation which had lain dormant. See generally infra notes 123-124 and accompanying text. A world in which a false escrow deposit triggers release of a truthful deposit has, at a minimum, more complicated welfare effects.

(76.) See Ayres & Baker, supra note 8, at 637-40.

(77.) See supra Part II.

(78.) Indeed, many institutions resist the idea of any seemingly anonymous complaint mechanisms. At Yale, for example, a 2010 report issued by the specially convened Sexual Misconduct Committee said the following:
   We do not recommend the creation of a specific website, hotline, or
   other venues through which anonymous reports are encouraged.
   Instead, we want to encourage confidential reporting: and, indeed,
   we worry that the explicit provision for anonymous reporting might
   discourage non-anonymous reporting. Non-anonymous (but still
   confidential) reporting is more useful both to the complainant and
   to the University.

Report from the Sexual Misconduct Committee, YALE UNIV. OFFICE OF THE PROVOST (July 9, 2010),

(79.) 20 U.S.C. [section][section] 1681-1688 (2006).

(80.) 503 U.S. 60, 75 (1992). Franklin addresses the claim of a female student that she had been repeatedly harassed by a male teacher, and that the school administration had both failed to remedy the situation and had pressured her to forego litigation. Id. at 75. Franklin builds on an earlier case, Cannon v. University of Chicago, in which the Court held that Title IX creates an implied cause of action for victims of sex discrimination, and that students need not exhaust administrative remedies before pursuing private litigation. 441 U.S. 677 (1979). In Cannon, the Court addressed Geraldine Cannon's allegation that she had been denied admission to medical school on the basis of her gender. Id. at 690.

(81.) See Franklin, 503 U.S. at 74-75. The intentional-failure-to-act standard imposed in private suits is higher than the standard imposed in administrative enforcement cases. In actions seeking administrative enforcement, injunctive relief can be granted on the basis that school officials knew or reasonably should have known that sexual harassment was occurring. See RUSSLYNN ALI, U.S. DEP'T OF EDUC., DEAR COLLEAGUE LETTER: SEXUAL VIOLENCE 4 & n.12 (2011), available at The letter is identified as "a 'significant guidance document' under the Office of Management and Budget's Final Bulletin for Agency Good Guidance Practices, 72 Fed. Reg. 3432." Id. at 1 n.1; see also OFFICE OF CIVIL RIGHTS, U.S. DEP'T OF EDUC., REVISED SEXUAL HARASSMENT GUIDANCE: HARASSMENT OF STUDENTS BY SCHOOL EMPLOYEES, OTHER STUDENTS, OR THIRD PARTIES 12--13 (2001) [hereinafter REVISED SEXUAL HARASSMENT GUIDANCE], available at

The requirement that a school or its responsible administrators demonstrate some intentional failure to remedy issues of harassment distinguishes private suits alleging Title IX violations from private suits in the employment context (under Title VII of the Civil Rights Act of 1964, 42 U.S.C. [section][section] 2000e to 2000e-17 (2006)). As we discuss below, the regulatory implementation and judicial interpretation of Title VII establish a slightly lower standard of liability for employers facing private litigation for monetary damages.


(83.) ALI, supra note 81, at 5-6.

(84.) See AAUW REPORT, supra note 1, at 4 (noting that fear of embarrassment is a reason for underreporting); Nicole Allan, Confusion and Silence, YALE ALUMNI MAG., July-Aug. 2011, at 38, available at titleix.html. The 2001 DOE Revised Sexual Harassment Guidance and other official documents direct schools to consult complainants' wishes with respect to confidentiality and honor them where possible. See REVISED SEXUAL HARASSMENT GUIDANCE, supra note 81, at 17-18. However, the documents acknowledge that schools may not be able to maintain complainant confidentiality. Id. Furthermore, the DOE acknowledges that the accused's due process rights may force a school to choose between breaching the victim's confidentiality and being unable to pursue the investigation at all, which itself may carry administrative penalties. See id.

(85.) See REVISED SEXUAL HARASSMENT GUIDANCE, supra note 81, at 13 & n.77.

(86.) 42 U.S.C. [section][section] 2000e to 2000e-17.

(87.) That is not to say that employers operate under significant liability burdens, however. As we discuss here, the existing Title VII framework, while less friendly to employers than Title IX is to schools, is still nevertheless very employer friendly.

(88.) 524 U.S. 775, 807 (1998) (noting "Title VII's ... basic policies of encouraging forethought by employers and saving action by objecting employees").

(89.) 524 U.S. 742, 764 (1998) ("Title VII is designed to encourage the creation of anti-harassment policies and effective grievance mechanisms.").

(90.) Indeed, Lauren Edelman and colleagues argue that employers played an active role in shaping this area of Title VII law, and that the focus on employee sensitivity training and other internal mechanisms became entrenched in the law after first beginning in industry. For one of her first pieces on this issue, see Lauren B. Edelman, Christopher Uggen & Howard S. Erlanger, The Endogeneity of Legal Regulation: Grievance Procedures as Rational Myth, 105 AM. J. Soc. 406 (1999).

(91.) Ellerth, 524 U.S. at 765 ("[A] defending employer may raise an affirmative defense ... (a) that the employer exercised reasonable care to prevent and correct promptly any sexually harassing behavior, and (b) that the plaintiff employee unreasonably failed to take advantage of any preventive or corrective opportunities provided by the employer or to avoid harm otherwise.").

(92.) What's more, employers also operate under some pressure to identify cases of harassment as soon as possible, as the amount of monetary and punitive damages are dependent in part on the duration of harassment of which the employer should have been aware. See, e.g., Blackmon v. Pinkerton Sec. & Investigative Servs., 182 F.3d 629, 636 (8th Cir. 1999) ("[D]uration of harassment [is a] relevant factor[] when determining the appropriateness of punitive damages....").

(93.) 542 U.S. 129 (2004).

(94.) See Suders, 542 U.S. at 148.

(95.) Id. at 148 ("To be sure, a constructive discharge is functionally the same as an actual termination in damages-enhancing respects.").

(96.) The Court also makes clear in Suders that even when the affirmative defense is available to the employer, the burden of showing that the employee unreasonably failed to take advantage of existing employer-provided systems falls on the employer. Id. at 146 ("Ellerth [and] Faragher ... place the burden squarely on the defendant to prove that the plaintiff unreasonably failed to avoid or reduce harm."). This burden allocation further limits the utility of the affirmative defense in summary judgment and other pretrial motions.

(97.) E.g., WISC. STAT. ANN. [section] 103.13(2) (West 2012) (granting employees broad right to access personnel records subject to narrow exceptions).

(98.) See Mkt. St. Assocs. Ltd. P'ship v. Frey, 941 F.2d 588, 595 (7th Cir. 1991). The Frey court notes that,
   [t]his duty [of good faith] is, as it were, halfway between a
   fiduciary duty (the duty of utmost good faith) and the duty merely
   to refrain from active fraud.... The concept of the duty of good
   faith like the concept of fiduciary duty is a stab at approximating
   the terms the parties would have negotiated had they foreseen the
   circumstances that have given rise to their dispute.

Id. Fiduciary relationships are ill advised in part because different depositors might have conflicting views about whether two deposits are sufficiently matching to trigger an escrow release. As one author explains,

[t]he flat commission custom can thus be understood as a kind of compromise which offers the agent more of a return on marginal effort than does a flat-fee arrangement because it gives the agent a financial incentive to produce a higher rather than lower sale price, while greatly--but not entirely--reducing the conflict among principals, each of whom must still wonder whether the agent is maximizing joint revenue by preferring another principal's (more expensive) property.

Saul Levmore, Commissions and Conflicts in Agency Arrangements: Lawyers, Real Estate Brokers, Underwriters, and Other Agents' Rewards, 36 J.L. & ECON. 503, 508 (1993).

(99.) An escrow system would likely be immune from liability for defamation under section 230 the Communications Decency Act as an "interactive computer service provider" as long as it was clear that the submissions were provided by third parties and the website was not requiring or inducing users to post defamatory statements. See 47 U.S.C. [section] 230(c)(1) (2006) ("No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider."); Zeran v. Am. Online, Inc., 129 F.3d 327, 330 (4th Cir. 1997) (online bulletin board provider not liable for users' defamatory messages). Even if section 230 immunity did not apply, it is likely that the qualified-privilege defense would shield the escrow service from defamation liability. This common-law defense to defamation protects speech made "to someone who may reasonably be expected to take official action to protect a public interest." John Jay Fossett, Defamation in the Workplace: "The New Workhorse in Termination Litigation", 15 N. KY. L. REV. 93, 105 (1988) (citing W. PAGE KEETON ET AL., PROSSER AND KEETON ON THE LAW OF TORTS [section] 115 (W. Page Keeton ed., 5th ed. 1984)).

(100.) As Jennie Kihnley points out in her 2000 article, schools in particular often express two semicontradictory aims of their sexual harassment-reporting processes: (1) to empower victims and reduce sexual harassment, and (2) to limit their own liability in any relevant litigation. Jennie Kihnley, Unraveling the Ivory Fabric: Institutional Obstacles to the Handling of Sexual Harassment Complaints, 25 LAW & SOC. INQUIRY 69, 70 (2000). The potential tension between these two goals can force complainants to navigate a confusing maze of administrators and procedural rules. A relatively distant relationship between the complainant and the third-party escrow agent should help to clarify at least one part of the harassment grievance process. See id. at 72-73.

(101.) Ian Ayres & Cait Unkovic, You Are Not Alone (2012) (working paper) (on file with authors).

(102.) It is possible that the first depositor (with a trigger of 4) would be happy to have her complaint released so long as four deposits had been placed in deposit, even though only three are releasable at the time of the match.

(103.) See supra Section II.A (discussing simulation with escrow trigger of 3 instead of 2). In addition, as with irretrievable complaints, the relative formality of a system-determined trigger might reduce the number of bad faith "fishing" and frivolous complaints deposited.

(104.) Owen M. Fiss famously wrote about the social values of public adjudication in Owen M. Fiss, Against Settlement, 93 YALE L.J. 1073 (1984).

(105.) See Felstiner et al., supra note 7, at 635-37.

(106.) There may be a first-mover disadvantage to making deposits to an escrow. The rescission option makes it potentially too easy for depositors who come to learn that they were the first to make an allegation deposit against a particular person to withdraw that deposit.

(107.) It would also be possible to construct a mechanism where deposits would become void if unmatched after a certain number of years--possibly tied to the relevant statute of limitations.

(108.) If interim disclosures to the harasser were made, it would be useful to only disclose with some randomized lag time so that the accused might have a harder time identifying who was making the allegation. But even with randomized lag times, harassers might be able to infer that a complaint was made by the student who was most recently harassed.

(109.) For this reason, we also support regular and scheduled periods of aggregate reporting, such as twice per semester, so harassers are less likely to know specifically when complained-of conduct occurred. Similarly, we also support centralized reporting of such aggregate data, so the accused cannot be identified on the basis of their supervisor reporting incidents of harassment.

(110.) For example, administrators may employ sensitivity training to prevent sexual harassment. Kenneth M. York, Lizabeth A. Barclay & Amy B. Zajack, Preventing Sexual Harassment: The Effect of Multiple Training Methods, 10 EMP. RESPS. & RTS. J. 277 (1997).

(111.) The Internal Revenue Service, for example, has successfully completed detailed audits of taxpayers with the understanding that the audits would only be used to assess the system levels of tax underpayment and not used to sanction those audited for any discovered underpayment of taxes. See Ian Ayres & Barry Nalebuff, Why Not? Winning the Audit Lottery, FORBES, NOV. 30, 2009, at 116. While we would argue strenuously against removing sanctions of the accused as a way of extracting aggregate or nontraditional data from escrowed complaints, we can imagine that many schools could find analogous ways of utilizing the information in escrowed complaints even before they were matched. Assuming sufficient complainant protections could be assured, such early uses of the complaints should serve current and future harassment victims, as well as the sponsoring institutions.

(112.) We consider such a negative net effect to be unlikely, even if few submissions are successfully matched. As with all areas of research on the underreporting of sexual harassment, however, there is no way to be entirely certain in such predictions, and as a result we cannot discount the possibility entirely.

(113.) To be more precise, developers should seek to maximize the number of escrowed submissions from those harassment victims who would otherwise not submit go-it-alone complaints.

(114.) Sarah Phillips, A Brief History of Facebook, GUARDIAN, Jul. 24, 2007,

(115.) See id.

(116.) Id.

(117.) See id.

(118.) Yale University is currently revamping its sexual harassment-reporting processes in response to recent complaints from students and Title IX investigators. As a result, Yale and similarly situated schools might be prime candidates for early rollout locations.

(119.) This difficulty results from differing group characteristics: should the large history department be subject to the same threshold as the tiny classics department? Should sports teams and student organizations be subject to a threshold applied to faculty? While selecting a blanket group-threshold would be difficult and perhaps ultimately unfair, so too would predefining an appropriate threshold for all possible groups. Group matching might, however, be a realistic goal for subsequent iterations of allegation escrow systems that can benefit from early experiences at a given institution.

(120.) Of course, some escrow system developers might forward any matched, good-faith allegations. Such a decision would simplify the review process significantly and lower the risk of unnecessarily orphaned claims. At worst, however, it could also lead to go-it-alone style harms for some users, thus negating the purpose of the system as a whole. A pattern of such harms might ultimately also have a chilling effect on submissions.

(121.) Of particular relevance in this decisionmaking process would be the screener's guess as to the preferences of the complainant submitting the more serious allegation.

(122.) As with information security here too, independence from the sponsoring institution and its liability-generating responsibilities becomes important.

(123.) See Mark Viera & Jo Becker, Ex-Coach Denies Charges Amid New Accusations, N.Y. TIMES, Nov. 15, 2011, at B13, second-mile-resigns-amid-penn-state-scandal.html. If true, the charges against Sandusky suggest that he repeatedly and severely sexually abused young boys in his care. Beyond the indescribable tragedy of the harm done to the children, the scandal surrounding Sandusky's arrest has already had a profound impact on Penn State as an institution. See id.

(124.) See Al Baker & Steven Erlanger, I.M.F. Chief Apprehended at Airport, Is Accused of Sexual Attack, N.Y. TIMES, May 14, 2011, at A1,

(125.) See Maia de la Baume, A Writer Frees Herself by Speaking Out, N.Y. TIMES, July 23, 2011, at A6, Neither case was successful, as U.S. prosecutors eventually dropped all charges against Strauss-Kahn and French prosecutors declined to pursue criminal charges as well. See Steven Erlanger & Maia de la Baume, Strauss-Kahn Is Not Charged in French Case. N.Y. TIMES, Oct. 14, 2011, at A1, banon.html. John Eligon, Strauss-Kahn Drama Ends with Short Final Scene, N.Y. TIMES, Aug. 24, 2011, at A1,

(126.) See Francis X. Clines, Testing of a President: The Accuser: Jones Lawyers Issue Files Alleging Clinton Pattern of Harassment of Women, N.Y. TIMES, Mar. 14, 1998, at Al, (reporting attempt "to portray Mr. Clinton as repeatedly engaging in sexual harassment of female underlings from his executive positions in government," including testimony from Kathleen Willey); Francis X. Clines & Jeff Gerth, The President Under Fire: The Overview; Subpoenas Sent as Clinton Denies Reports of an Affair with Aide at White House, N.Y. TIMES, Jan. 22, 1998, at A1, with.html (reporting the Monica Lewinsky scandal); Woman Says Clinton Made Advance in '91, N.Y. TIMES, Feb. 12, 1994, at A8, (reporting the Paula Jones sexual harassment lawsuit).


(128.) Margaret A. Lucero et al., An Empirical Investigation of Sexual Harassers: Toward a Perpetrator Typology, 56 HUM. REL. 1461, 1470 (2003).

(129.) Ron Langevin et al., Lifetime Sex Offender Recidivism: A 25-Year Follow-Up Study, 46 CANADIAN J. CRIMINOLOGY & CRIM. JUST. 531, 545 (2004).

(130.) For a discussion of the concerns associated with direct surveys asking questions about sensitive topics, see ARIJIT CHAUDHURI & RAHUL MUKERJEE, RANDOMIZED RESPONSE: THEORY AND TECHNIQUES 2-24 (1988). We also imagine that the data available from surveys seeking to identify perpetrators of sexual harassment on college campuses is particularly vulnerable to such concerns.

(131.) See AAUW REPORT, supra note 1.

(132.) Id. at 15 fig.2.

(133.) Under Title IX, rape and sexual assault are considered serious, violent forms of sexual harassment.


(135.) Fisher et al., supra note 55, at 24.


(137.) For a discussion of surveys and studies measuring the prevalence of rape, see BONNIE S. FISHER, MEASURING RAPE AGAINST WOMEN: THE SIGNIFICANCE OF SURVEY QUESTIONS (2004), available at

(138.) Or, more accurately, the likelihood that information escrow systems will support reporting increases as the rate of underreporting that would be reported via allegation escrow systems rises.

(139.) Kimberly T. Schneider, Suzanne Swan & Louise E Fitzgerald, Job-Related and Psychological Effects of Sexual Harassment in the Workplace: Empirical Evidence from Two Organizations, 82 J. APPLIED PSYCHOL. 401, 402 (1997).

(140.) E.g., Beth A. Quinn, The Paradox of Complaining: Law, Humor, and Harassment in the Everyday Work World, 25 LAW & Soc. INQUIRY 1151, 1151, 1167, 1181 (2000). However, because the experience of sexually harassing behavior is formed in large part by the reactions of the recipient, it is exceedingly difficult, if not impossible, to identify which of those women who resist the term "harassment" do so for which reasons.

(141.) E.g., 15 U.S.C. [section] 2087 (2006) (protecting employees from retaliation for reporting violations of safety rules under jurisdiction of Consumer Product Safety Commission); 18 U.S.C. [section] 1514A (2006) (establishing broad protections for whistle-blowers complaining of fraud or violations of securities laws); 29 U.S.C. [section] 660(c) (2006) (prohibiting retaliation against employees who complain or testify about violations of the Occupational Safety and Health Act). See generally Employment Law Guide: Whistleblower and Retaliation Protections, U.S. DEP'T OF LABOR, (last visited Oct. 2, 2011 ).

(142.) One veteran LAPD officer claimed in 2008 that he'd been demoted and subjected to freeway therapy for defending a female colleague after she was harassed and discriminated against on the basis of her gender. In a lawsuit about the retaliation, he described being removed from his K-9 unit and assigned to a division that was a four-hour commute from his home. A Los Angeles jury awarded him $3.6 million in damages, and he agreed to collect $2.5 million from the city in return for the city's agreement to forgo appeals. Victoria Kim, Jury. Awards Damages to Officer, L.A. TIMES, Nov. 13, 2008, at B1, http:/

(143.) See generally Janet E Near & Marcia E Miceli, Retaliation Against Whistle Blowers: Predictors and Effects, 71 J. APPLIED PSYCHOL. 137-45 (1986) (discussing relationship between whistle-blowing and retaliation); Marcia A. Parmerlee, Janet E Near & Tamila C. Jensen, Correlates of Whistle-Blowers' Perceptions of Organizational Retaliation, 27 ADMIN. SCI. Q. 17 (1982) (exploring organizational responses to whistle-blowing and identifying factors that increase likelihood of retaliation).

(144.) See. e.g., Christine Hauser, Embattled Chief Leaves Police Review Board, N.Y. TIMES, Apr. 2, 2009, at A24. The New York City Civilian Complaint Review Board has a mandate to investigate every complaint received and refers the complaint to the New York Police Department if found to be substantiated. See N.Y.C. CIVILIAN COMPLAINT REVIEW BD., Frequently Asked Questions, (last visited Oct. 2, 2011).

(145.) 31 U.S.C. [section] 3730(b) (2006).

(146.) Id. [section] 3730(d).

(147.) Id. [section] 3730(b)(5).

(148.) 15 U.S.C.A. [section] 78u-6 (West Supp. 2012).

(149.) Id. [section] 78u-6(h)(1); 31 U.S.C. [section] 3730(h).

(150.) E.g., United States ex rel. Mikes v. Straus, 931 F. Supp. 248, 263 (S.D.N.Y. 1996) ("[T]he modern trend does not support a ban on compulsory counterclaims which are based on damages which are "independent' of the qui tam claim." (citing United States ex rel. Madden v. Gen. Dynamics Corp., 4 F.3d 827 (9th Cir. 1993))); Burch ex rel. United States v. Piqua Eng'g, Inc., 145 F.R.D. 452, 455-57 (S.D. Ohio 1992) (compulsory counterclaims must be permitted for due process reasons despite the seemingly contradictory intent of the False Claims Act).

(151.) See MedWatch: The FDA Safety Information and Adverse Event Reporting Program, U.S. FOOD & DRUG ADMIN., (last visited Oct. 3, 2011); EudraVigiliance, EUR. MEDS. AGENCY, ema/index.jsp?curl=pages/regulation/document_listing/document_listing_000239.jsp&murl= menus/regulations/regulations.jsp&mid=WC0b01ac05800250b5 (last visited Oct. 3, 2011). In December 2010, the European Parliament and European Council adopted a regulation and directive updating the adverse drug reaction reporting system in the European Union. The new legislation went into effect in July 2012 and will change the way that both pharmaceutical manufacturers and physicians report adverse reactions. The basic system will remain voluntary for physicians, and thus should still be considered "passive oversight." Both Regulation (EU) No. 1235/2010 and Directive 2010/84/EU are available at 2010 Pharmacovigilance Legislation, EUR. MEDS. AGENCY, regulation/general/general_content_000492.jsp&mid=WC0b01ac058033e8ad (last visited May 6, 2012).

(152.) Patient noncompliance, both intentional and unintentional, is exceedingly common and is also responsible for a nontrivial proportion of ADEs. Prescription error is much less common but by no means unheard of. Drugs with similar sounding names, for example, can lead to prescription errors. See U.S. GEN. ACCOUNTING OFFICE, ADVERSE DRUG EFFECTS, GAO/HEHS-00-21 at 6-8 (2000), available at

(153.) For more on the disparity of judgments among physicians, see Jeffrey A. Linder et al., Secondary Use of Electronic Health Record Data: Spontaneous Triggered Adverse Drug Event Reporting, 19 PHARMACOEPIDEMIOLOGY & DRUG SAFETY 1211 (2010).

(154.) Based on a search of, a private, for-profit website that makes the FDA's ADE database publicly available. See Search MedWatch Drug Adverse Events (AERS), FDABLE, (last searched Aug. 28, 2011).

(155.) For more information on HIPAA and its scope, see The Health Insurance Portability and Accountability Act of 1996 (HIPAA) Privacy and Security Rules, U.S. DEP'T OF HEALTH & HUM. SERVS., visited Oct. 3, 2011).

(156.) See generally MedWatch Online Voluntary Reporting Form, U.S. FOOD & DRUG ADMIN., (last visited Oct. 3, 2011). Despite the fairly strict HIPAA requirements governing the release of patient information, drug manufacturers are still required to submit some types of ADE reports, and encouraged to submit as many as possible. See id.

(157.) However, releasing reports to media outlets might expose the escrow service to defamation liability for abuse of the qualified privilege. See supra note 99 (discussing defamation and privilege).

(158.) See supra text accompanying notes 54-56.

(159.) See WILLIAM SHAKESPEARE, OTHELLO, THE MOORE OF VENICE act 3, sc. 3, where Iago engineers a scheme to convince Othello that his wife Desdemona is having an affair with Cassio by planting Desdemona's handkerchief in Cassio's room.

Ian Ayres, William K. Townsend Professor of Law and Anne Urowsky Professorial Fellow in Law, Yale Law School.

Cait Unkovic, Ph.D. Student, Jurisprudence and Social Policy, University of California-Berkeley. We are grateful to Bruce Ackerman, Mark Branch, Jeremy Bulow, Barry Nalebuff, and William Samuelson for useful comments. Joshua Mitts provided excellent research assistance. In parallel with the writing of this Article, we are developing a nonprofit website,, which initially will be dedicated to escrows concerning sexual harassment at a small set of universities.
COPYRIGHT 2012 Michigan Law Review Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2012 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:II. Using Allegation Escrows to Mitigate Initial Claim Aversion B. Additional Costs and Benefits of an Allegation Escrow Regime 1. Legal Issues through Conclusion, with footnotes, p. 172-196
Author:Ayres, Ian; Unkovic, Cait
Publication:Michigan Law Review
Date:Nov 1, 2012
Previous Article:Information escrows.
Next Article:Outsourcing regulation: how insurance reduces moral hazard.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters