Measuring the premium on common knowledge in computer-mediated coordination problems Oded Nov a, * , Sheizaf Rafaeli b a New York University, Polytechnic Institute, 6 MetroTech Center, Brooklyn, 11201 New York, NY, USA b University of Haifa, Graduate School of Management, Mt. Carmel, Haifa 31905, Israel article info Article history Available online 21 September 2008 Keywords: Common knowledge Coordination Communication Knowledge management Email abstract Common knowledge, whereby everybody knows something, and everybody knows that everybody knows it, and so on ad infinitum, is claimed to be central to coordination in organizations. However, this claim has so far not received empirical support, as a method to empirically compare common knowledge with other forms of knowledge has not been available. In this article, we present a novel method through which we empirically estimate the common knowledge premium—the level of users’ preference of com- mon knowledge over ‘‘knowledge by all” (where everybody knows something, but not necessarily every- body knows that everybody knows it). Using the method we show that a ‘‘premium” of common knowledge over ‘‘knowledge by all” does exist consistently, across populations and measuring variations. The findings provide empirical support for the centrality of common knowledge. Implications of the study are discussed. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Common knowledge (CK) is the subject of research in various disciplines (e.g. Binmore and Brandenburger, 1990; Fagin, Halpern, Moses, & Vardi, 2003). Something is CK if everybody knows it, and everybody knows that everybody knows it, and everybody knows that everybody knows that everybody knows it... and so on ad infinitum (Lewis, 1969). This is different from something that everybody knows, but not necessarily everybody knows that everybody knows it, etc. We shall refer to the latter case as ‘‘knowl- edge by all” (KA). CK is preferable to, or can said to have a premium over KA in coordination situations (Chwe, 2001). Fagin et al. (2003) provide an illustration of this: a society wants all drivers to know that a red light means ‘‘stop” and a green light means ‘‘go” (this is an example of knowledge by all). However, drivers would feel much safer if they also knew that everybody else knows these rules, and everybody knew that everybody knew them, and so on (this is an example of common knowledge). CK is seen as key to coordination in organizations: Grant’s (1996) Knowledge Based Theory of the Firm, according to which firms exist in order to coordinate the specialized knowledge of their employees, involves CK as underlying such coordination. A similar concept to CK (Carlile, 2004), is that of ‘‘mutual knowl- edge”, where parties to a communication share certain knowledge in common and know they share it (Cramton, 2001). In distributed and virtual organizations such mutual knowledge is seen as key to effective communication and collaboration (Alavi & Tiwana, 2002; Cramton, 2001). However, while the claim of CK importance and its premium in a social or organizational context is used extensively (e.g. Carlile, 2004; Chwe, 2001; Cramton, 2001; Cramton, 2002; McInerney, 2002; Ramasubbu, Krishnan, & Kompalli, 2005), it has not been supported so far by empirical evidence. The CK premium is said to be empirically estimable, at least in principle (Chwe, 2001), but so far no method to do so was presented. In this article, we address this issue. Being able to support or refute the CK argument has implica- tions for knowledge and coordination research and practice. With the growing trend toward collaborative approach to knowledge management (Awazu & Desouza, 2004; Lutters, 2004; Nov & Rao, 2008; Ong & Lai, 2007), and as new knowledge management sys- tems (KMS) offer different communication channels with varying degrees of reliance on CK (Sabherwal & Becerra-Fernandez, 2005), it is important to know if CK is indeed desirable for coordi- nation, and to what extent. Thus, in this work we present a method through which we dem- onstrate the existence of a CK premium in a coordination problem, and estimate its magnitude empirically. 1.1. CK premium estimation The difference between CK and KA arises in email exchange. Sending an email to an undisclosed list of recipients (using the 0747-5632/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.chb.2008.08.005 * Corresponding author. Tel.: +1 718 260 3562. E-mail address: onov@poly.edu (O. Nov). Computers in Human Behavior 25 (2009) 171–174 Contents lists available at ScienceDirect Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh