1 Introduction

As the working arrangements in today’s workplace become more flexible and distributed teams more ubiquitous, the challenges of managing and collaborating at a distance attract the interest of the mainstream management thought.

A manager deciding to work from home for one week to meet an important deadline suddenly places his team in a distributed environment. For a traditional organization with a large proportion of vertical communication, the manager’s absence from the office can lead to a noticeable slow-down in many of its collaborative processes. Prolonged absence from the office may also lead to anxiety and deterioration of trust as a result of the lack of “background” information on what is happening at the home base (Baskerville and Nandhakumar 2007). This kind of issue has important implications for a geographically distributed team because trust is said to be the crucial factor for its effective operation (Handy 1995).

In my earlier publication (Chadyuk 2007a), I introduced a concept of teamchat, a text-based single-thread persistent chat environment for the exchange of both work-related and social messages among all the members of the virtual team. This paper which, like the previous one, is intended to become a chapter of a book on teamchat, will focus on practical aspects of the teamchat application in a distributed work environment.

In the first part of this paper, I theorize on the correlation between the level of teamchat activity and the levels of trust and collaboration within the team.

Further, I consider how trust and collaboration are affected by the ability of the team members to meet each other face-to-face and how these effects are enhanced by the teamchat interaction between the meetings.

2 Trust and Collaboration

2.1 Measuring trust

In my earlier work (Chadyuk 2006), I argued that the existence of trust in the team was a critical factor for the success of a teamchat implementation. Obviously, there are much wider implications of trust for the virtual team operation: without trust efficient virtual team is impossible (Handy, 1995).

But how do you know whether trust is present in a team? And, for that matter, can anyone know whether a team works to its maximum ability - or its performance is just a simple sum of efforts of every team member pushing in his own direction with no regard for anyone else?

To begin with, a team should be set up for collaboration in the genuine sense of the word. However, there is often ambiguity in defining the ideal collaboration setup. Some distributed-work practitioners assume that the best approach to virtual team-work is divide-and-conquer: you break down the problem your team faces into smaller parts, give each part to a separate team member, then receive their outputs, arrange them into a complete solution - and you are done (Jones, Oyung, Pace 2005).

I believe this view of the challenges faced by an average virtual team today is too simplistic. Divide-and-conquer works well for the material-goods world where every process can be - and has to be - analyzed, optimized, and replicated, in order to achieve maximum production efficiency. In today’s knowledge economy, virtual teams face non-standard tasks; the elements of the problem are not known until after the solution is found and, increasingly, every serious challenge has to be addressed from a holistic point of view by the entire team working together, as opposed to each team member doing a separate bit of work.

Jones, Oyung and Pace (2005) propose the making of a coffee-table as an analogy for virtual team-work. I think the concept of software debugging is much more apt: you have a problem, you don’t know where it’s coming from, the symptoms don’t give you any clue to a possible solution; and even when you find the solution you were looking for, you discover that it has just allowed another bug to come to the surface. You get an impression that you are constantly chasing a moving target. This kind of challenge just doesn’t allow you to break down a problem into simpler parts, and it won’t let your team members succeed in isolation.

In real life, you get both “software-bug” and “coffee-table” kinds of problems. The trouble is, some “software bugs” look exactly like “coffee tables”, and when your team is built and managed like a “coffee-table” production line, some “tables” will clog your entire conveyor for no apparent reason, and leave you wondering: is it that the inputs that my team gets are faulty, or is it one of the team members taking a ride?

On the traditional production line, you always have an option to go down to the shop floor and see for yourself what is not working correctly, but a geographically distributed team cannot conduct this kind of close monitoring. Many more things need to be taken on trust by both managers and employees. Charles Handy in his seminal Harvard Business Review article ‘Trust and the Virtual Organization’ (Handy 1995) calls for a radical rethink of the way an organization goes about its critical control functions: a need to replace command and control with trust and accountability.

In this chapter, trust is a recurring theme in the context of team collaboration because team collaboration - not just a sum of efforts of isolated people - is the only efficient method of addressing a “software-bug” kind of problem whenever your team faces it. It has been observed that high levels of trust in a team are associated with its ability to manage the uncertainty and complexity of the task, as well as the challenges arising from physical isolation of the virtual team members (Jarvenpaa, Leidner 1999).

Still, coming back to the question that we raised at the beginning of this chapter, how do you measure trust? How do you know your virtual team members collaborate if you personally see them only so often?

Doing my research, I haven’t come across a good direct measure for either trust or collaboration to be used on a day-to-day basis: interviews and questionnaires are only good for complete strangers from a random sample - we all learn very quickly in our careers to tell the boss what he likes to hear. However, this has never stopped serious science: scientists learned to use indirect measures for things that are not readily measured. All you need is a good level of correlation between the two.

The research of Sirkka Jarvenpaa and Dorothy Leidner (1999), that I quoted above when I talked about the correlation between trust and collaboration, also notes that persistent communication between the team members increases the trust levels: in particular, social communication helps to build the links at the early stages of collaboration; predictable and timely interaction on the task-related issues helps maintain high levels of trust and commitment throughout the project.[i]

Thus I argue that when you analyze trust and collaboration, the most accessible measure is the level and quality of communication among the team members. Trust, collaboration and communication spring from each other and cultivate each other. By watching the levels of both social and business-related communication in teamchat, you can get a very good indicator of trust levels within your team.

Active interaction improves the team effectiveness: Luecke (2004) says that “communication lubricates the wheels of collaboration”. All three elements - trust, communication and collaboration - are intricately linked, and so a break-down in one will necessarily lead to deterioration of the two others.

Herbsleb et al (2002) noted (although never developed into a full-scale observation) how usage of their teamchat-like tool fell to zero after the team was affected by reorganization, the most reliable trust-buster known to contemporary man, a mere rumor of which will urge people to build fences in order to protect their status quo. Anxiety and turf-wars replace trust and collaboration. I observed very similar results in my team (Chadyuk 2006).

There may be other, less straightforward reasons for deterioration of trust but teamchat will always be a good indicator. Just like a canary used by an 18^th century coal miner to detect dangerous levels of carbon monoxide underground, when your teamchat stops chirping and singing, you know that trust in your team is being suffocated.

2.2 Impact of teamchat on trust development

At this point in our exposition, it probably makes sense to step back and consider what we mean by the word ‘trust’.

There has been a lot of confusion and ambiguity in the literature in the definition of trust, its roots and implications. For the sake of this analysis, I suggest considering the definition proposed by Roger Mayer, James Davis and David Schoorman (1995: 712):

Trust is the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party.

Note how this definition of trust touches on the issue of control, as we discussed earlier in the context of Charles Handy’s (1995) work.

Mayer, Davis and Schoorman (1995) propose a model that separates the factors that influence trust, from the trust per se, which influences the behavior of a trusting party. In particular, they identify the trustee’s abilities, benevolence, and integrity, as the key factors for the development of trust.

If, for example, you are a new team member, and I am your co-worker, the level of your trust in me, that is, your willingness to be personally vulnerable to my actions, will depend on your perception of my trustworthiness. This perception consists of three elements: your perception of my abilities, your perception of my benevolence towards you, and your perception of my integrity in my dealings with you[ii].

Having a certain level of trust in me (which is a function of the above three-fold perception of my trustworthiness), you estimate the risks associated with a proposed action, and only then make a decision to engage in it. The outcome of the action will influence your perceptions of my abilities, benevolence and integrity, which, in turn, will influence the level of your future trust in me (Mayer, Davis and Schoorman 1995: 720).

Roxanne Zolin and Pamela Hinds (2004) theorize that receiving preliminary information about me, before you and I start interacting, helps you form perceptions of my abilities, benevolence and integrity. But how exactly can that kind of information be delivered in a distributed environment? I argued that a teamchat is an ideal conduit for such information because it gives new team members an ability to “lurk”, i.e. to observe interaction between the colleagues without contributing to the actual interaction (Chadyuk 2007a).

This does not imply that if there is no preliminary personal information, there is no trust. Researchers (Jarvenpaa, Leidner 1999; Zolin, Hinds 2004) observed instances of blind or “swift” trust originally described by Debra Meyerson, Karl Weick and Roderick Kramer (1996) in temporary co-located project groups with no history of prior collaboration. “Swift” trust, however, is usually more task-oriented, rather than personal. It also requires that the team members come into the project with strictly defined roles within their temporary organizations, as can be seen in film or airline crews, emergency teams, and so on. When the team-member roles are blurred, swift trust, as a rule, is much weaker (Meyerson, Weick, Kramer 1996).

In permanent distributed teams that, in today’s volatile business environment, are equally permanently under-resourced, all team members are fairly versatile in their roles, and so for a newcomer it is usually quite difficult to understand who does what without a process manual (which never gets written due to, yes, under-resourcing). Under these circumstances, training and integration of a new team member may be addressed by appointing a coach to “hold the hand” of the newcomer and explain how things happen (which again exacerbates the resourcing issue since the experienced employee is taken away from doing actual work). Plan B is letting the newcomer “learn by doing” which may result in costly trial-and-error cycles.

As an alternative, “lurking” on the teamchat interaction, without replacing the need for formal training, provides the new employee with “learning by observation” that not only clarifies the multiple roles of the colleagues, thus supporting formation of the swift trust, but also provides relevant information on the colleagues’ competence in their roles, as well as benevolence and integrity in their dealings with co-workers, which assists in developing positive perceptions that form the basis of personal trust.

Charles Handy writes: “It’s hard to trust people whom you […] haven’t observed in action over time.” (Handy 1995) The teamchat environment gives new employees the ability to watch the way existing employees communicate, react to crises, and solve problems. This observation generates perceptions that lay the foundation of trust.

2.3 Importance of face-to-face communication for distributed teams

Almost everyone who writes on virtual teams contends that to build a trusting relationship within a team, it is necessary to meet in person from time to time. The biggest impact is achieved if the entire project team assembles at the very beginning of the project – ideally if the meeting involves a formally mediated team-building session (Zheng et al 2001: 293). This may include an independent analysis of the functions of every player within the team (e.g. Belbin, 1993), providing an important frame of reference for the future interactions with employees in a distributed environment where necessary social cues will be absent. For example, an employee who plays a role of Evaluator in the Belbin (1993) scheme has an important function of screening various options offered in brainstorming - and it is crucial that his distributed co-workers get to understand this through an independent observation. Without such preliminary understanding, they may end-up labeling him as “constantly criticizing” and avoid requesting his inputs in the teamchat environment because he “feels so negative all the time”.

Zolin and Hinds (2004) base their argument regarding the importance of an initial face-to-face interaction for the ensuing trust in a distributed team on the universally accepted notion that the first impressions are the strongest, and their effect is the longest-lasting one. But you can expect that your remote employees will trust you only if you give them a positive first impression. If you are late for your introductory session because your kid couldn’t go to school and the baby-sitter had been late, you have very little chance to convince your team that you are generally very punctual: Zolin and Hinds insist that perceptions of remote employees are much slower to change compared to collocated employees. So organizing a team-building session for your remote team can be a mixed blessing: you can expect positive effects only if the execution of the first meeting is flawless.

Bonnie Nardi and Steve Whittaker (2002) refer to the common wisdom that whenever you show up in person, you show commitment to building a relationship. This, I suppose, is critical in buyer-seller relationships, which are by far the most conservative with regards to face-to-face time. The buying decision depends heavily on the relationship and trust being built between the buyer and the seller because it is almost never possible to analyze and check all the selling points that are made by the sales person: many important elements of the decision should be based on trusting what the seller says.

However when we talk about the people who are part of the same distributed team working to achieve a common goal, even when risks are high and close personal relationships are not yet established, distrust almost never appears to be a starting point of the project. According to Zolin and Hinds (2004), trusting seems to be an “easier option”, and thus distributed teams do exhibit “swift” trust mentioned above (Meyerson, Weick, Kramer 1996; Jarvenpaa, Leidner 1999).

It is difficult to argue with the assertion that “you demonstrate an enormous amount of unconscious commitment when you actually take the time and the trouble to put yourself in the same place as the person you want to build a relationship with.” (Nardi and Whittaker, 2002: 93) But the policy of meeting everyone face-to-face may end up defeating the purpose of a distributed team. Majchrzak et al (2004) remind us that “much of the value of virtual teams is derived from members’ ability to be in two places at once.” Here I will not argue the benefits of distributed teams; there is enough literature for that purpose. What I do argue is that although constant face-to-face interaction is the best trust-building tool, it is not the only one, and the same results can be achieved when FTF, for some reason, is not an option.

Contrary to the common wisdom, Zheng and colleagues (2001) demonstrate that when people are able to communicate with each other via a text chat, even when they don’t meet each other face-to-face, they can develop a level of trust that is as good as if they had known each other personally. The social information processing theory proposed by Joseph Walther from Cornell University suggests that it simply takes longer for people communicating via CMC to form personal relationships, and lack of face-to-face interaction will not prevent, just slow down, that process (Walther 1992).

2.4 Limitations of Face-to-face communication

Nardi and Whittaker (2002) introduce a concept of “communication zones” that you can imagine as a shared playground that people create when they establish a social relationship. From that moment, all communication is acted out within the boundaries of this imaginary playground. Nardi and Whittaker postulate that the two main games played there are: building social bonds and managing so called “attentional contracts”: agreeing to pay attention to each other’s communication.

No doubt physical presence and face-to-face interaction gives both games a multitude of dimensions: as a species, we have had one or two millions of years learning to build social bonds with other people through body language, touch, sharing a meal. The eye contact as a proper behavior when somebody talks to us is the ultimate commitment to the word and spirit of the “deed of attention”.

But what can be assumed for a perfect scenario of an ideal face-to-face communication is rarely played out for longer than 5 minutes: our attention span for a social conversation is becoming shorter every decade. We come to business meetings with laptops to do our emails and barely look each other in the eye. And the word “complaisant” in the sense of “eager to be agreeable to someone else” completely disappeared from our language, with nothing left to do in it but to puzzle today’s reader of an 18^th century novel (Blair 2000).

I remember deciding once to go with my boss in a taxi to the airport, to see him off to Paris. Spending a couple of hours with him in a traffic jam in Moscow to mull over the meetings of the last two days, and get a bit of social chit-chat, seemed a pretty decent idea for a bleak winter afternoon. In the end, I spent three hours to the airport and two and a half back without uttering a word because he spent all the time to the airport speaking on the phone in French, of which I understood nothing.

I am not saying he spent these three hours inefficiently: he had a chance to catch up on the important projects happening elsewhere, discussed the outcome of the visit with his boss, spoke to his wife about his son’s grades, all of which was more important than sharing office gossip with me. Real-life priorities often play a trick on the best scientifically-grounded assumptions.

In this case, face-to-face interaction would happen in a lab setting, but in real-life it didn’t, even though the two people were in the same place, knew each other, were part of the same team, and worked towards the same overall goal.

2.5 Teamchat: supporting trust between face-to-face sessions

Let’s suppose, however, that whenever a distributed team meets in one place the members do interact with each other, and do build some level of trust as a result. But what happens when people return to their usual distributed mode of operation? Richard Baskerville and Joe Nandhakumar (2007) offer a hypothesis that trust dissipates over time without face-to-face interaction. The question is: Can computer-mediated communication (CMC) help maintain trust in the long periods between face-to-face meetings of the distributed team members?

Baskerville and Nandhakumar (2007) argue it can’t. However once you give their underlying field study a closer look, you see that the subjects of the research did not use CMC for socializing. They tried to setup teleconference coffee sessions where everyone felt too awkward to say something: the environment was too unnatural. This is exactly the issue that teamchat is intended to address. Being a primary communication environment, sometimes even replacing email for business-related discussion, teamchat becomes a natural medium for interaction; more than that: because of the participant accountability – ‘I know that you know that it’s me who is writing this’ – teamchat becomes a safe environment, a “discussion sanctuary” that feels conducive to personal communication (Bradner, Kellogg, Erickson 1999; Erickson et al.1999; Chadyuk 2007a).

Teamchat shares the capability of a near-synchronous exchange with instant messaging applications (Chadyuk 2007a). According Nardi, Whittaker, and Bradner (2000), informality of communication that we observe in IM is a result of its synchronous nature. Every interaction creates an immediate context that reduces the possibility of misunderstanding, in particular, misinterpreted emotions, so common with less synchronous media like email.

This emotion-compatibility of a synchronous environment is the reason why we see so much humor in teamchat posts. Obviously, there is no better trust-builder than humor.

It was observed in a number of sources (Iaconna and Weisband 1997; Jarvenpaa and Leidner 1999; Coppola, Hiltz, and Rotter 2004) that a continuous, frequent and predictable interaction was crucial in maintaining high levels of trust in distributed teams. Thus, if the team members continue interacting with each other via CMC in the periods between personal meetings, the trust levels can be maintained.

Coming back to the study conducted by Baskerville and Nandhakumar (2007), the people interviewed in it complained that their virtual workplaces did not provide access to “backstage activities”. They were concerned about the lack of information about what was going on in the central location. As could be expected, anxiety and mistrust resulted from information asymmetry, a common feature of many failed virtual teams (Chadyuk 2006). However, lack of trust did not come from the lack of social interaction in this case - it resulted from the lack of awareness of what was happening elsewhere.

The all-inclusive nature of communication in teamchat provides the background noise; however distracting it may sometimes be and however wasted the time reading though the nervous communication on a project that has nothing to do with your work feels, this noise creates a very positive, albeit not very obvious, effect. Just knowing what other people do makes you feel these people are much closer to you, even though you don’t see them. Your imagination supplies you with the necessary emotional stereotypes: when you see somebody’s project is delayed, you know you would be distressed if that happened to your project, you imagine how that person may feel, and then you see her showing a gracious reaction to the circumstance and you feel tacit respect for her. Now that you’ve seen her perform under pressure, as far as you are concerned you have personal history of a relationship with her that you can rely on for your future joint projects, even though you haven’t even communicated with her so far. This immediately links back into personal trust formation, discussed above.

Having to read through dozens of posts on a teamchat log that have no direct relevance to your current work, searching for the bits of the conversation that will give you the information you need, seems a very debilitating endeavor. However this chore gives you a peripheral view of the “backstage”, the lack of which gradually obliterates trust, smothers empathy, and obstructs collaboration.

3 Summary and Direction of Further Work

Social communication in the virtual environment is different because you do not get the bandwidth available when meeting in person. However Walther (1992) argues that computer-mediated interaction can deliver the same quality of social bond, it just needs much more time than face-to-face interaction to arrive there. The research of Zheng et al (2001) demonstrates that maintaining social communication in a virtual environment prevents loss of trust, and even makes it more robust.

According to Walther (1997), time is the only dimension that you need to consider when facing a dilemma of trust-building in computer-mediated communication as opposed to face-to-face interaction.

However not all tasks within the distributed environment can be addressed by teamchat with equal efficiency. For example, if you have a conflict between co-workers, you have to address it immediately, and you just don’t have the time required by the social information processing theory (Walther 1992) to address the trust issue before the crisis rages out of control. This seems to be a quite intuitive conclusion.

But there are less obvious situations where computer-mediated-communication seems to be doing OK, while face-to-face would do much better, and vice versa: where face-to-face ends up to be a complete waste of time (Majchrzak, et al 2004; also see example above). In the next paper, I will analyze which tasks are addressed efficiently by teamchat, and which require face-to-face interaction to achieve their goals.
List of Abbreviations

CMC Computer-mediated communication

FTF Face-to-face (interaction)

IM Instant messenger (chat)

Selected Bibliography

Baskerville, R.; Nandhakumar, J. 2007. Activating and Perpetuating Virtual Teams: Now That We're Mobile, Where Do We Go? IEEE Transactions on Professional Communication, Vol. 50, Iss.1, March 2007, Pp.17-34. DOI=10.1109/TPC.2006.890849

Belbin, R.M. 1993. Team Roles at Work. Butterworth Heinemann: Oxford.

Blair, D. 2000. Introduction and Notes to J. Austen. Northanger Abbey. Wordsworth: London. p. 169.

Bradner, E., Kellogg, W. A., and Erickson, T. 1999. The adoption and use of “BABBLE”: a field study of chat in the workplace. In Proceedings of the Sixth European Conference on Computer Supported Cooperative Work (Copenhagen, Denmark). Kluwer Academic Publishers, Norwell, MA, 139-158.

Chadyuk, A. 2006. Corporate blog: the critical communication tool for remote teams. http://www.chadyuk.com/Rushmore/

Chadyuk, A. 2007a. Teamchat: Requirements definition. http://www.chadyuk.com/Chadyuk2007/

Coppola, N.W., Hiltz, S.R., Rotter, N.G. 2004 "Building trust in virtual teams," IEEE Transactions on Professional Communication, vol.47, no.2, pp. 95-104, June 2004.

Erickson, T., Smith, D. N., Kellogg, W. A., Laff, M., Richards, J. T., and Bradner, E. 1999. Socially translucent systems: social proxies, persistent conversation, and the design of “babble”. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems: the CHI Is the Limit (Pittsburgh, Pennsylvania, United States, May 15 - 20, 1999). CHI '99. ACM Press, New York, NY, 72-79. DOI= http://doi.acm.org/10.1145/302979.302997

Handy, C. 1995. Trust and the Virtual Organization. Harvard Business Review, 73 (3), 40-50.

Herbsleb, J. D., Atkins, D. L., Boyer, D. G., Handel, M., and Finholt, T. A. 2002. Introducing instant messaging and chat in the workplace. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems: Changing Our World, Changing Ourselves (Minneapolis, Minnesota, USA, April 20 - 25, 2002). CHI '02. ACM Press, New York, NY, 171-178. DOI= http://doi.acm.org/10.1145/503376.503408

How Canaries Saved the Lives of Coal Miners http://www.petcaretips.net/canary-coal-mine.html Downloaded 4 June 2007.

Iaconna, C.S., Weisband, S. 1997 “Developing trust in virtual teams,” in Proceedings of Hawaii International Conference on System Sciences, vol. 2, 1997, pp. 412–420

Jarvenpaa, S.L, Leidner, D.E. 1999. Communication and Trust in Global Virtual Teams. Organizational Science. Volume: 10. November-December 1999, Number: 6. Pgs: 0791-0815.

Jones, R., Oyung, R., Pace, L. 2005. Working Virtually: Challenges of Virtual Teams. Hershey, PA, USA: IRM Press.

Luecke, R. 2004. Creating Teams With an Edge: The Complete Skill Set to Build Powerful and Influential Teams. Boston, MA: Harvard Business School Press.

Majchrzak, A., Malhotra, A., Stamps, J., Lipnack, J. 2004. Can absence make a team grow stronger? Harvard Business Review. May 2004; 82(5):131-137, 152. Reprint DOI: 10.1225/R0405J

Mayer, R.C., Davis, J.H., Schoorman, F.D. 1995 “An Integrative Model of Organizational Trust” The Academy of Management Review, Vol. 20, No. 3 (Jul., 1995), pp. 709-734 doi:10.2307/258792

Meyerson, D., Weick, K.E., and Kramer, R.M. 1996. Swift trust and temporary systems. In Kramer, R.M., and Tyler, T.R., Eds. Trust in Organizations. Thousand Oaks, CA: Sage. Pp. 166-195

Nardi, B.A. and Whittaker, S. 2002. The place of face-to-face communication in distributed work. in Hinds, P. and Kiesler, S. eds. Distributed Work, MIT Press, Cambridge, MA, 83—110.

Nardi, B. A., Whittaker, S., and Bradner, E. 2000. Interaction and outeraction: instant messaging in action. In Proceedings of the 2000 ACM Conference on Computer Supported Cooperative Work (Philadelphia, Pennsylvania, United States). CSCW '00. ACM Press, New York, NY, 79-88. DOI= http://doi.acm.org/10.1145/358916.358975

Walther, J.B. 1992. Interpersonal Effects in Computer-Mediated Interaction: A Relational Perspective. Communication Research, Vol. 19 No. 1, February 1992, p. 52-90.

Walther, J.B. 1997. Group and interpersonal effects in international computer-mediated collaboration. Human Communication Research, 23 (3), 342-369. Quoted in Jarvenpaa, Leidner (1999).

Zheng, J., Bos, N., Olson, J. S., and Olson, G. M. 2001. Trust without touch: jump-start trust with social chat. In CHI '01 Extended Abstracts on Human Factors in Computing Systems (Seattle, Washington, March 31 - April 05, 2001). CHI '01. ACM Press, New York, NY, 293-294. DOI= http://doi.acm.org/10.1145/634067.634241

Zolin, R., Hinds, P. J., (2004) Trust in Context: The development of interpersonal trust in geographically distributed teams, In Trust and Distrust in Organizations Eds. Cook, K. and Kramer, R. New York: Russell Sage Foundation.