e-Reputation: Building Trust in Electronic Commerce

• Position: Professor of Law, Fordham Law School

Professor of Law, Fordham Law School. Many thanks to the participants in the LSU Law School Symposium on Unifying Commercial Law and the Fordham Law School Legal Theory Workshop for their coments and encouragement.

Who would place a bid in an online auction in which a stranger, identified only by pseudonym, offers to sell an object based on a brief written description and a grainy digital photograph? Why would any set of strangers engage in commerce over the Internet, whether retail, wholesale or tag sale? If something goes wrong in the electronic transaction, legal redress may be problematic. Where e-commerce involves small dollar amounts or transaction partners in distant or foreign jurisdictions, breach is unlikely to be followed by litigation since litigation may be too time-consuming, expensive, or unpredictable.

The academic community has long understood that, while clearly articulated legal rules and predictable legal enforcement of those rules enhance commercial transactions, commerce can occur, and may even flourish, in their absence. Commercial norms can supplement commercial law,¹ while verification institutions² and non-legal sanctions³ supplement judicial enforcement of that law. Moreover, traders can manage the risks that underlie commercial transactions by acquiring information about others' reputation in the market,⁴ but "relying only on direct personal experience is both inefficient and perilous: inefficient, because any one individual will be limited in the number of exchange partners she or he has, and perilous, because one will discover untrustworthy partners only through hard experience."⁵As a result, markets have developed mechanisms⁶ for sharing and selling reputation information.⁷

In Reputation and Intermediaries in Electronic Commerce,⁸Clayton P. Gillette considers whether reputation intermediaries and reputation sanctions offer plausible substitutes for contractual enforcement in e-commerce,⁹ and focuses his attention on the performance of feedback forums like that provided by eBay, the largest online auction.¹⁰ In the end, he is skeptical that the eBay feedback forum can provide customers with a reliable means for transmitting accurate information about traders' commercial reputation and sanctioning those who do not engage in reasonable commercial conduct in online auctions.¹¹ He argues that the usefulness of a reputation system is undermined by participants' limited capacities to verify the accuracy or inaccuracy of reputation information available over the Internet.¹²

Gillette's ambivalence about feedback forums' success stems from the fact that only a small fraction of online traders post negative or even neutral commentary with eBay. He suspects, not just that too little feedback is posted, but, more damning, that participants under- report their dissatisfaction with online auctions. He raises both theoretical and practical reasons why dissatisfied buyers and sellers would refrain from posting negative remarks more frequently than they would refrain from posting positive ones. Gillette is quick to note the existence of empirical studies, such as that conducted by Paul Resnick and Richard Zeckhauser, which show a significant correlation between a feedback score and the likelihood that an auction will be concluded successfully.¹³ Nonetheless, Gillette remains unconvinced by this data. He disagrees that these studies demonstrate, as Resnick and his co-authors claim, "that reputation systems appear to perform reasonably well" online.¹⁴ Gillette fears that the data show only that "successful bidders might have believed that Feedback Forum ratings conveyed valuable information, even though that was not the case, and even though unsuccessful visitors more appropriately discounted the quality of the signal."¹⁵ According to Gillette, traders' perception that eBay feedback forums accurately convey reputation information, their naive faith in this reputation mechanism, is not enough to establish eBay's success as a reputation intermediary.¹⁶ Resnick and Zeckhauser contest this view and instead claim that "it is the perception of how the system operates, not the facts, that matters."¹⁷

In this paper, I consider the circumstances under which an online reputation system enables participants to deter fraud, share information about commercial quality, and otherwise encourage market participants to engage in electronic commerce. Part I considers whether there exists an inherent selection bias against negative commentary on eBay's feedback forum, as Gillette suggests. Part II asks whether imperfect feedback forums can still "perform reasonably well," as Resnick, Zeckhauser, and others argue, because they employ informal mechanisms for punishing online transgressions, at least some of the time. Finally, Part III explores the importance of perception to electronic reputation systems and the development of trust in e-commerce.

I

Commerce-including electronic commerce over the Internet- creates a sort of Prisoner's Dilemma. Both of us would be better off if our bilateral performance obligations were completed, but the temptation to shirk may preclude us from maximizing our collective welfare. I am tempted by the thought that I will be best off if I receive the goods or services you agreed to provide, but fail to pay for them; you are tempted by the thought that you would be best off if you were to pocket my payment, but deliver boxes of sawdust rather than valuable widgets. The common law sets contract damages at levels intended to discourage the temptation to shirk and resolves this Prisoner's Dilemma. However, for all of the reasons set out by Gillette-small transaction size; imprecise legal rules; traders in distant, perhaps, foreign locations-legal rules may not sufficiently deter shirking on the Internet.

In the end, e-commerce requires commercial actors to trust one another just as trust binds commercial actors in "bricks and mortar" marketplaces. Scholars explain that traders trust one another, in part, based upon their reputations in the marketplace. Economic actors who conduct themselves consistent with commercial norms create and preserve a reputation for good conduct in the marketplace.¹⁸ A reputable commercial actor is unlikely to shirk, not only because there is a good chance that the trader will act as reputably in future transactions as she has in the past, but also because the trader knows that breach will be sanctioned in the commercial community by the loss of her valuable reputation. Scholars predict that reputation sanctions work best against large-sized businesses,¹⁹ especially those that frequently engage in similar transactions and must return to the market with their reputations intact.²⁰ This literature also ties the effectiveness of reputation sanctions to the market's ability to assess a potential trader's reputation. Markets limited to a small, homogeneous group of participants easily share reputation information,²¹ whereas diffuse markets may only be able to share information about a trader's reputation using technology.²²Reputation systems that utilize advanced information technology are said to work best when they evaluate simple and objective information²³ regarding present or past occurrences.²⁴ Where these costs of assessment are affordable, traders can avoid "bad deals" ex ante by evaluating the reputation of their counterparty before entering into the transaction.

At first blush, the Internet seems especially well suited to assist traders or their intermediaries in acquiring and broadcasting reputation information to interested parties.²⁵ Advanced information storage and retrieval functions enhance potential traders' abilities to access and manipulate the reputation information at a low cost. Moreover, the effectiveness of reputation sanctions need not be limited to geographically or culturally tight-knit commercial communities because the information can be disseminated with ease and little cost over the Internet.

Despite the obvious technological advantages of conducting a reputation system over the Internet, commentators have uncovered a number of obstacles, some theoretical and some practical. In their paper Facilitating Trust in Internet Interactions, Paul Resnick, Richard Zeckhauser, Eric Friedman, and Ko Kuwabara describe the significant challenges that reputation systems face in eliciting reputation information from participants.²⁶

First, Resnick, Zechhauser, Friedman, and Kuwabara contend that the system must overcome collective action problems inhibiting people from taking the time to provide any feedback.²⁷ I am uncertain that eBay participants face a collective action problem in deciding whether to post feedback. Feedback imposes minimal costs and creates a significant benefit for users, at least for those who anticipate returning to the auction. A participant is unlikely to build an online reputation by failing to comment on a counterparty's conduct, and, thus, has every incentive to post. In any event, actual practices contradict what the collective-action theory predicts. As Gillette notes,²⁸ empirical studies of eBay show that buyers and sellers post feedback far more frequently than one would have thought. Resnick and Zeckhauser examined a large set of eBay transactions from February 1 to June 30, 1999,²⁹ and found that buyers commented on sellers for 52.1% of the items and sellers on buyers 60.6% of the time.³⁰ Thus, surprisingly, a higher percentage of users post feedback on eBay than voted in the last Presidential election.³¹

Resnick, Zeckhauser, Friedman, and Kurawara also note that it is difficult for a reputation system to ensure that it is eliciting honest commentary, and particularly difficult to obtain traders' negative feedback.³² Gillette concurs, and goes on to argue that the "information that eBay provides appears to be heavily skewed in a manner that renders its utility...