The Defense Intelligence Agency’s (DIA) Knowledge Lab faced the challenge of how to move accurate intelligence up the chain of command. Too frequently the intelligence analysis, painstakingly generated by front line analysts, was delayed and often severely modified by a chain of superiors before it reached policy makers who could act upon it.
Zeke Wolfberg, the Director of the Knowledge Lab and I, as the knowledge management consultant to the Knowledge Lab, recently published an article about how we addressed that problem at DIA. The article appeared in Reflections, the SOL Journal on Knowledge, Learning and Change (Vol. 10, 4). I provide a summary of the article in this post. The full article can be downloaded under My Publications on the left.
DIA provides military intelligence to prevent strategic surprise and deliver a decision advantage to warfighters, defense planners, and policymakers. The analysts who develop intelligence assessments are divided into teams of 30 plus with each team focused on a different region (e.g. South East Asia) or a different functional area (e.g. missiles), as well as task forces that are periodically assembled to address crises (e.g. the Mumbai terrorist attack).
The intervention we undertook to enable telling truth to power, was called “Critical Discourse.” The format was to bring together a team of analysts along with their supervisor, to jointly analyze the actual conversations that occurred between members of the team and between team members and their supervisor. The conversation analysis was based on the work Chris Argyris, using his left and right hand case format to create scripts of difficult past conversations. Each analyst and supervisor selected and wrote out the dialogue that occurred in three difficult conversations they had engaged in. Each team met over a period of three months to analyze each other’s cases and to practice Argyris’ Model II skill set with the goal of reducing the misunderstandings occurring in their conversations. Each team member also received individual coaching between group meetings.
Model II is a way of interacting that results in knowledge and learning for everyone who is engaged in the conversation. The goal of Model II interaction is not to win an argument, but to create a space for joint learning. A person using the Model II skill set holds in their mind the underlying assumption that, “I have part of the information and others have other parts that I may not be aware of.” The skills are used to get all the information that is available and pertinent into the conversation. The Model II skills that produce full and accurate knowledge are:
Advocate your own position and share all relevant information related to the issue, not just the data that supports your view
Encourage others to question your position so if your reasoning is faulty or if there is information you have missed you will discover it
Ask questions about the position of others, not to find where you might “catch them” in a mistake, but to fully understand their reasoning and data.
The Argyris Model II skills are simple to state, but take a great deal of practice to put into place in a real conversation. The difficulty occurs because the default way of interacting in difficult situations in nearly every organization is Model I, which defeats gaining learning through conversation. Model I skills are:
Advocate my position in a way that discourages inquiry into it
Keep my reasoning private so others cannot challenge it
Don’t ask others about their reasoning
Assume I understand the situation and anyone who sees it differently does not
A team using Model I skills in a contentious or potentially embarrassing situation would exhibit interactions that are familiar to all of us. For example, 1) a team leader states his opinion on an issue in such a way that it is clear he doesn’t want to be challenged; or 2) a team member offers an idea that no one understands, but other team members don’t ask him the questions that would make his thinking available to them - his idea just dies on the table, or 3) a team member offers a solution, providing convincing evidence about why it would work, but doesn’t reveal the very real risks he knows are involved.
Using Model I interaction teams make poor decisions because all of the available data about the issue does not get into the discussion. Team members are more interested in winning their point than in learning. Yet Model I interaction is so ubiquitous that team members are largely unaware they are employing it in a conversation. To use Model II interaction skills, team members first have to become aware of their Model I interactions. Fortunately, others can see what we cannot see in ourselves. So a team’s joint analysis of their cases allows members to help each other become aware of and then unlearn Model I skills. And provides them a safe place to practice using Model II skills.
Over a two year period, I worked with 15 DIA teams using the Critical Discourse intervention, including teams at the highest executive level. During that time team members wrote hundreds of conversational cases. The analysis of those cases resulted in two outcomes that shifted the culture of DIA; 1) both seniors and analysts developed the skills to challenge each other effectively – to tell truth to power, 2) dysfunctional patterns that were preventing accurate information from being exchanged across the whole of DIA were identified. These patterns, that Argyris calls defensive routines, had become embedded in the fabric of the culture in such a way that made it unlikely any that any individual, on his or her own initiative, would be able to act in a manner other than Model I.
To illustrate the process through which the culture changed I will focus on one of the teams that was set up to look at an emerging intelligence issue. The code name for this team was “Fresh Look.” Fresh Look is discussed in the article but in less detail than I write about it here.
Fresh Look was given the task of finding a solution to a long-standing problem that many other teams had been unable to resolve. The team was to report their findings to a high level executive team – tell truth to power.
Previous DIA teams that had struggled to find a solution to this issue had worked virtually, dividing the work up so that people could function as individual contributors with minimal interaction. Fresh Look, however, was co-located to give them the advantage of bringing together their collective knowledge to solve this difficult problem. But co-location meant that to be successful in using all the team’s knowledge, the team would have to interact in a Model II manner to jointly address content issues. To assist the team, early in their time together, the Knowledge Lab provided the Critical Discourse intervention described above.
Using the Critical Discourse format, each team member wrote three cases about their interactions with other Fresh Look team members. By reading and discussing each other’s cases, team members were able to see, laid out in black and white, the negative impact of their actions. With practice they began to recognize the conversational tactics that were in the cases being exhibited in their real time conversations and begin to call each other on it when they recognized Model I tactics. With further practice of the Model II skills, they began to be able to share their knowledge more fully and to speak openly about their concerns related to the problem they were addressing.
A turning point in terms of new thinking for the Fresh Look team came about six weeks into the project. For some time the team had been struggling with how to organize themselves to deal with the central issue of their analysis. This issue was not only at the heart of the Fresh Look topic, but was an issue that touched on the very nature of how DIA does analysis – so how the team structured themselves and their work was a very relevant topic for the team.
At a Friday meeting, with two of the team members absent, the team finally made a decision to organize into several smaller teams, each devoted to a piece of the larger question. The norm within the intelligence community is that when a meeting is convened and decisions are made, those who were absent are expected to accept the decision and support it. The rationale being that there is not enough time to rehash decisions. There is also an implied perception that if the team member had really wanted to be present at the decision meeting, they would have attended.
On the following Monday the absent team members were back in place for the morning meeting. The absent team members respectfully challenged the decision-making norm by asking the rest of the team to go over their logic for reaching the decision of dividing into several smaller teams. Instead of rebuking the two, the team not only discussed their logic of the previous Friday, but the entire topic of methodological pursuit was reopened for debate: should the team employ the status quo method of structuring itself or should they behave differently? During that discussion, the team members who were present on Friday acknowledged that their decision was likely driven by being tired at the end of the week and wanting to end the day to start the weekend. The willingness to challenge and be challenged that had begun to develop within the team during the previous weeks of practice using Model II was for the first time being fully exercised.
Encouraged by the group’s response to this initial challenge, a few days later another team member was willing to raise a much more critical issue that had been troubling her - how the team had framed the analysis question. Her position was that the team was restricting the analysis by unthinkingly accepting as their starting point the assumption currently held among subject matter experts. The team took her concern seriously and was able to hold an honest and open interaction that resulted in significantly broadening the scope of their analysis. The reframing of the question eventually led the team to discover a remarkable finding - that the intelligence community had ignored a plausible counter-assumption without ever collecting data to prove or disprove it! Essentially leaving the intelligence community blind on a critical issue.
The on-going practice the Fresh Look team had in using their inquiry skills to elicit knowledge from each other and to respectfully challenging each other prepared them to speak truth to power when their scheduled briefing with high level DIA seniors occurred at the end of the project. By the time of the executive briefing, team members had grown confident in using the Model II skills in increasingly challenging situations.
The following are two descriptions of that meeting, first from the perspective of a senior executive who was in attendance. The second from the perspective of one of the Fresh Look team members who engaged the seniors in conversation.
Senior Executive Perspective
"I was one of five senior executives attending the debrief of Fresh Look. When I walked into the meeting room I could see that the physical arrangement of the seating area was very different than what I was used to. Normally, seats were arranged in rows facing a podium. On this day the seats were arranged around a set of tables so that everyone sitting at the table could see each other. During the team’s introductory remarks they mentioned how they intentionally designed the physical space to promote mutual learning between the seniors and themselves rather than the typical transmit then respond mode of communication.
The Fresh Look team explained that their briefing content would be different from the usual format as well. The team said that the goal they had in mind was that everyone around the table would learn with and from each other. I felt okay with this approach but wondered if they could pull it off. They started with brief presentations of their findings during which several members of team offered their own perspective, providing a variety of views, again not a typical debrief. Instead of just saying, “Here are our findings.” the team spoke about the challenges they had faced and how they had navigated those challenges, providing context for how their conclusions had been reached. Having laid out their findings, the team indicated they were ready to open the floor for dialogue.
I noticed that during the give and take of the conversation the team members were able to strongly advocate the findings they had developed as well as demonstrate their openness to our (the executive team) views. However, one of the team members, Carrie (not her real name), said something that really bothered me. She made a claim that a certain piece of information was non-existent. Now, I was aware that she was wrong, that piece of information was available. I said just that, perhaps a bit more forcefully than I intended. And I said that if the team had been thorough enough in their search they would have found it. However, rather than becoming defensive, Carrie said she found my comment interesting, because even the senior member of their team who had full clearance to look at all the available data did not come across it. She began a discussion among us all about the availability of information across silos. Out of that discussion I gained a new understanding about the issue of information availability from the perspective of analysts doing a search. It was a very stimulating conversation."
The Fresh Look Team
"The meeting with the executives was even more successful than we had thought it would be. The executives were impressed with the quality of our findings. And we were pleased with the skills we were able to exhibit in being able to respectfully detect and overcome the “old guard” approach.
As a team we were particularly proud of our interaction with Sam, the senior executive who reacted critically when he realized that the team did not know about the existence of a critical piece of information, saying, “Everyone knows it exists.” We were able to recognize the role-based behavioral pattern in Sam’s remark and were able to avoid the emotional trap that could have deflated and sidetracked us away from the issues. We were able to detail the process we went through to find the data, what was made available to us, and what had been hidden from even the most senior analyst member of our team who specialized in the issue. This detailed explanation led to a more general discussion of how information is shared and accessed which brought new insights both to us and to the high level executives present. Through using the critical discourse skills, our goal of using review meetings to jointly develop knowledge was realized.
The Director of DIA attended the debrief and told us later that what he witnessed was the embodiment of team behavior - the kind of team behavior DIA needs for detecting new threats. He said the discussion with the team was cathartic for him."
Fresh Look is one of many examples of changes in the ability to speak truth to power at DIA. To do so analysts and seniors first needed to recognize their own contribution to the dysfunctional patterns that existed inside DIA. Secondly, they needed to learn a new way of interacting and practice those new skills in a relatively safe environment. Finally, they needed to use the skills to interact in real world situations to gain confidence in their own ability to interact more openly and honestly. Through this slow developmental progression they were able to bring knowledge into a conversation that otherwise would have been hidden and they were able to speak truth to power.