Cutbacks in defense spending have forced all branches of the Armed Services to do more with less. Secretary of Defense Donald Rumsfeld has advocated making the right changes to transform the military into a "leaner and meaner" force that's capable of winning "every single battle that this military is faced with." The Office of Naval Research's Science and Technology Manning Affordability Initiative has set as its goal a "50-percent personnel reduction while demonstrating operational utility for all functions." 

To design new systems and re-engineer existing systems in order to maintain optimal performance with fewer people.

Below are the key findings of an extensive analysis of successful manpower reduction efforts in the corporate and military sectors conducted by Klein Associates for the Manning and Affordability Initiative sponsored by the Office of Naval Research (Militello, Klein, Crandall, and Knight, 1998). Along with each finding is a discussion as to how it can be addressed with CE methods.

1. Identify leverage points. Leverage points are places where modest changes can have high payoffs. To identify leverage points, Militello et. al. recommend combining an understanding of current operations with a top-down analysis of function that is not constrained by the current state of operations. To gain an understanding of current operations, a variety of CE methods are particularly relevant. These include Interviewing and Observing techniques, such as Contextual Inquiry, as well as a variety of other Knowledge Elicitation techniques, including Process Tracing and Conceptual Methods. Cognitive Task Analysis methods such as the Critical Decision Method and Applied Cognitive Task Analysis (ACTA) can be used to identify and analyze the critical decisions of the domain. The Cognitive Function Model technique can also be used to identify and analyze tasks with a high degree of cognitive complexity. For the top-down analysis of function, techniques such as Cognitive Work Analysis (CWA), Applied Cognitive Work Analysis (ACWA), Goal-Directed Task Analysis (GDTA), and Cognitive Function Analysis (CFA) are particularly relevant. In addition to mapping the functions of the domain, the ACWA method can also be used to identify the key decisions and the information required to make those decisions.
   
   
2. Iterate. Not all changes resulting from a reduction in manpower can be predicted a priori. Therefore, a period of time is needed to correct mistakes and iron out unintended consequences, particularly those resulting from a top-down approach involving many alterations. Modeling and simulation techniques, such as Computation Cognitive Modeling and Computational Task Simulation, can help designers envision the implementation of a new system and iron out critical deficiencies. Interface inspection techniques, such as Heuristic Evaluation, can eliminate serious flaws in the interface design. However, there is no substitute for testing a system with actual people, utilizing techniques such as Rapid Prototyping, Storyboarding, and Formal Usability Studies, as well as constructing high fidelity simulations and realistic testbeds.
   
   
3. Determine a clearly-stated goal. It is important to have a clearly-stated goal to be used as a guiding principle and motivator for people at all levels of the organization. With the goal of reducing crew size, all new technologies must be evaluated with respect to this goal.
   
   
4. Address training requirements. Training is a key issue in systems that operate with fewer people. It is likely that only the most qualified people will operate systems where most of the mundane tasks have been taken over by technology. Thus, there will be less opportunity for on-the-job training. To address training development, a host of CE methodologies are relevant. Among the most relevant are CTA techniques such as the Skill-Based framework, ACTA, PARI, DNA, and COTA, which can uncover differences between expert and novice performance so that novices can quickly be brought to the level of experts. These techniques can also help guide the development of intelligent tutoring systems.
   
   
5. Determine implications of manpower reduction across multiple modes of operation. While a certain level of staffing may be adequate for routine situations, it may not be adequate when operating in a crisis mode. Thus, it is necessary to conduct analysis for multiple possible scenarios and conditions, not just the most likely scenarios. The Critical Decision Method and the Critical Incident Technique are particularly relevant in the identification and construction of non-routine scenarios. Computational Cognitive Modeling and Computational Task Simulation techniques can also be used to assess the impact of reduced staffing in multiple scenarios. These techniques can also address the impacts on cognitive and physical workload with various levels of staffing under a variety of situations.
   
   
6. Take the larger world context into account. Manpower reduction efforts must be based on an accurate forecast of future missions. The conditions that the military operates in are constantly changing. Designers must question the relevance of previous scenarios and world conditions for the future. One way to address this issue is to have people articulate the means by which a legacy system supports the current goals of an organization. Techniques such as CWA, ACWA, and Goal-Directed Task Analysis can help people articulate the current goals and functions of the organization. In addition, there may be cultural barriers and established traditions that people have become accustomed to, but don’t have a good rationale for existing in the present. The functional decomposition of the domain generated by CWA and ACWA can be used evaluate the rationality of such traditions and determine whether or not they have a place in future systems.
   
   
7. Manage the implementation of changes. Effective management of the re-engineering effort involved in manpower reduction is critical. People at all levels of the organization must be kept apprised of changes that will impact them. The installation of new technology, training, and workload peaks should be coordinated so that personnel are not overloaded during the implementation.
   
   
8. Address technologies that best support manpower reductions. Militello et. al. identified three categories of technologies that best support manpower reduction: a.) tools to build and maintain situation awareness, b.) technology to support the use of remote specialists, and c.) central monitoring systems that allow fewer, centralized humans to monitor a larger area.
   

1. Tools to build and maintain situation awareness. These are tools for maintaining, updating, and communicating the big picture. They allow for more effective teamwork and fewer team members. CE methodologies are particularly relevant in the construction of such tools. Most relevant is Goal-Directed Task Analysis, which focuses on situation awareness information requirements. CWA and ACWA are also particularly relevant, since they produce functional decompositions of a domain and facilitate the creation of displays that integrate system variables into a meaningful picture, allowing for rapid assessment of information.
   
   
2. Technology to support the use of remote specialists. The use of centralized specialists whose expertise is available to several remote locations is another solution to reduce manpower.
   
   
3. Central monitoring systems. The introduction of computerized systems for monitoring and control tasks can allow for reductions in manpower.
   

9. Develop measures to provide design criteria for optimized manning. Measures devised by Militello et. al. that were deemed relevant to manpower reduction include: a.) the ratio of decision makers to overall staff, b.) the percent of time spent on overhead and information management tasks, c.) the ratio of handoffs/transactions, and d.) the ratio of information received versus information sought.
   

1. The ratio of decision makers to overall staff. Militello et. al. surmise a ratio of 1:5 or better should be the goal. To compute this ratio, the actual decision makers must first be identified. A variety of CE methodologies are useful for determining who the decision makers are, including CWA, ACWA, ACTA, GDTA, and the Critical Decision Method. Decision-centered design approaches such as these are particularly relevant to manpower reduction.
   
   
2. The percent of time spent on overhead and information management tasks. This relates to a u-shaped function that describes the relationship of staff size to workload. As staff size increases to a certain point, workload decreases. However, once staff size passes a certain point, workload actually begins to increase. This is because the information management activities needed to coordinate staff members continue to grow. Cognitive Engineering methods that can aid in the construction of effective teams, such as CWA and GDTA, are particularly relevant. Metrics to assess shared situation awareness and team cognition are also useful. For example, if two people need to coordinate frequently, perhaps their position should be merged. Anything preventing the merging of a position, such as increased workload or the need for new expertise, should be evaluated. Identifying areas where the flow of interaction gets stopped would also be useful. Positions that hold up others and ways to clear such gridlocks should be identified.
   
   
3. The ratio of handoffs/transactions. This is the number of times a data element gets rerouted before a meaningful transaction occurs. One way to trace such transactions is by constructing an "audit trail" of the decisions that were made to generate a product, such as a plan. CE methods can help by identifying decision points and by identifying artifacts (e.g. plans) and their purpose in supporting decisions. Methods that are particularly relevant include Decision Analysis, CWA, ACWA, and the Critical Decision Method. Field Observations and Ethnographic techniques may also be applied.
   
   
4. The ratio of information received versus information sought. This is the degree to which people spend time passively scanning information as opposed to actively seeking it. Sometimes as new technology is introduced, people become more passive in their information gathering because they lack of a good mental model of how the system works or a lack of understanding in how to search for what they need. Another issue is the degree to which skilled decision-makers are separated from raw data. Skilled decision makers need selective access to raw data, and a rule of thumb proposed by Militello et. al. is that if raw data is any more than three steps away, expertise is being compromised. CE methods can aid in these assessments with two different approaches: 1.) By determining how work is currently performed using methods such as Field Observations, Interviews, ACTA, and the Critical Decision Method to predict the impacts of new technology, and 2.) By mapping out the characteristics and constraints of the problem space, including the decisions that are made and the information that is required to support those decisions, using methods such as Decision Analysis, CWA, ACWA, and GDTA. These methods can also help identify the correct model of how the work domain operates independent of any system design. Such a model is an invaluable asset in the development of training programs to cultivate a workforce of active information seekers.
   

Advances in technology have made possible new levels of automation that have had many effects on operational settings. In settings such as airplane cockpits and power plants, automation has changed the role of the operator from one of manual control to one of supervisory control. In the command and control domain, systems have been developed to automate certain lower level decision making tasks, such as pairing weapons with targets. A variety of unforeseen issues and setbacks have arisen, including: 1.) Systems often don’t provide operators with justifications for their actions or conclusions, leading operators to ask of the automated system what is it doing? why is it doing that? what is it going to do next?, 2.) People aren’t aware of the uncertainty and unreliability of information propagated by the automated system, leading to issues such as overtrust or undertrust in automation, and 3.) People may lose situation awareness and have difficulty taking control when an automated system fails.

To build automated systems that amplify and augment human cognition.

Traditionally, the human factors community has regarded automation as a problem of function allocation. That is, the problem is fundamentally a matter of deciding which system functions should be performed by machines and which by people. With this approach, either the human does all the work or the machine does all the work. The current approach advocated by the Cognitive Engineering community is one that views the humans and machines as forming a joint cognitive system where people and intelligent agents work together to perform the work of the system. So the goal is to join people and automation together into a successful collaborative team. People need to be able to coordinate, trust, supervise, and cooperate with automated systems.

Based on studies of successes and failures in human-human and human-machine collaboration, David Woods and his colleagues have identified a series of issues that must be addressed in the design of automated systems (Christoffersen and Woods, 1999; Woods, 2001). A Function Allocation Roundtable held by the Office of Naval Research’s Manpower and Affordability Initiative highlighted other challenges in the design of automated systems (Militello, 1998). The list of design challenges and recommendations below is a combination of those addressed in these efforts, along with recommendations as to how they can be addressed with the methods of Cognitive Engineering.

1. Design for observability of automation activities (Woods, 2001). People must be aware of what the automated system has done, what the automated system is currently doing and for how long, why the automated system is doing what it does, and what the automated system is going to do next. These four situation awareness issues correspond to the most common questions asked of automation in airplane cockpits. Being aware of these aspects of an automated system helps people to build and maintain situation awareness and step in if the automated system fails. Further, by understanding the reasoning behind an automated system’s decisions, people can rectify differences between the system’s model of the state of the world and their own mental model of the state of the world. Cognitive Engineering  methods can help articulate aspects of the world that make decision making difficult. Goal-Directed Task Analysis (GDTA) can be used to get at the situation awareness information requirements. CWA and ACWA can also address the information required to make decisions as well as define the problem space that both the people and the machines operate in. CFA can also be use to elicit and construct cognitive functions that are involved in specific tasks.
   
   
2. Design for re-directability of automation resources (Woods, 2001). Means to re-direct or constrain the activities of automated systems as situations and workload change are necessary. The high potential for unexpected events in military command and control makes this challenge particularly relevant. What is necessary is for the automated system to be able to convey its intent to the operators so they can understand what the system is trying to do and decide whether or not what it is trying to do is congruent with what it should be doing.
   
   
3. Design for resilience in adapting to surprise (Woods, 2001). The high potential for surprises in military operations means that automated systems must be resilient in adapting to them. In command and control, understanding the commander’s intent can be used to develop adaptations to surprises. Failures to adapt to surprise often occur when people fail to adapt plans and procedures to local conditions, or when people adapt plans and procedures without considering the higher-level goals and constraints in the situation.
   
   
4. Design for control of attention in multi-threaded situations (Woods, 2001). When operators are performing more than one task or interacting with multiple automated systems, there is a need to shift their attention. Switching among these multiple threads of reasoning and activity requires balancing coherence with sensitivity to new events and information. Automated cueing of the operator’s attention to new information and events can be incorrect and errors can have high costs. Thus, there is a necessity for designers to have a thorough understanding of the operator’s goals and the information that is important to the fulfillment of those goals. The GDTA method breaks down the operator’s goals and analyzes the situation awareness information requirements for each of those goals. Computational Cognitive Modeling methods can also be used to assess the effectiveness of operators in monitoring multiple channels of information.
   
   
5. Design for building common ground across multiple agents (Woods, 2001). To be effective, people and automated systems must have a consistent model of the world and a common ground. Common ground breaks down when people think they are talking about the same things or believe they are making the same inferences, but do not realize that confusions have entered. Common ground between people and automated agents can only be established if people understand the region of interest, activities, and goals of the agent. This goes back to the principle of designing for observability of automation activities. CE methods that can be used to design displays that make activities of agents visible to human monitors are particularly relevant. Methods such as ACWA, COADE, and ACTA, which are particularly suited to develop visualizations for decision support systems, may also be relevant in the design of representations of automated activity.
   
   
6. Convey the uncertainty of information propagated by the automated system (Militello, 1998). Users of automated systems often have little or no information regarding the uncertainty and unreliability present in the information underlying the system’s operation. There is a need to somehow convey this information so that people can calibrate an appropriate level of trust in different contexts.
   
   
7. First understand the problem space (Militello, 1998). Before decisions can be made as to how system functions should be coordinated between people and machines, the problem space must be analyzed and understood. CE methods that analyze and represent the entire problem space include CWA, ACWA, CFA, GDTA, and Hierarchical Task Analysis (HTA). CWA, ACWA and HTA concern themselves with the functions of the domain, GDTA concerns itself with aspects of situation awareness, and CFA concerns itself with cognitive functions that can be performed by people or agents. Cognitive Task Analysis methods such as ACTA and the Critical Decision Method can also be used to address key decision-making activities and what makes those decision-making activities difficult. Generally, a thorough understanding of the problem space comes from combining analytical CE methods that define the entire problem space and the constraints it imposes on agents, such as CWA and ACWA, with methods that get at the strategies that experts have developed to cope with those demands, such as Structured Interviewing and Process Tracing techniques.
   
   
8. Iterate (Militello 1998). Simply changing the configuration of people and machines changes the functions themselves. It is often not easy to predict the consequences such changes will have. Prototypes, mock-ups, and field studies are needed to conduct empirical investigations of the system and make refinements.

In some situations, minutes or even seconds can make the difference between success and failure. For the Air Force, this situation occurs in the prosecution of Time Critical Targets (TCTs). TCTs are those for which there is a limited amount of time available to work through the kill chain cycle (i.e. find, fix, track, target, engage, and assess, or F2T2EA). In a speech at the 2002 C2ISR Summit, Air Force Chief of Staff John P. Jumper lamented "If you want the Air Force to hit it, by god you’d better have it three days in advance."

To design new systems and re-engineer existing systems that allow optimal decisions to be made in less time.

Many of the Cognitive Engineering applications discussed in Reducing Manpower and Improving Coordination Between People and Automation apply equally to Reducing the Timeline, since automation will most likely play a large role in any timeline reduction effort. Human-human collaboration also plays a critical role in the successful prosecution of time critical targets. Thus, optimization of both human-machine and human-human collaboration will most likely be the driving factors behind a timeline reduction effort. Below are some of the findings from the above sections on Reducing Manpower and Improving Coordination Between People and Automation that may have the most impact in a timeline reduction effort.

1. Identify leverage points. In timeline reduction, leverage points may be places where there are too many degrees of separation between decision makers and raw data, there are gridlocks in information flow, or there is a large percent of time spent on management tasks. Traditional task analysis techniques, such as Timeline Analysis and Operational Sequence Diagrams, may be used to develop optimized task execution procedures. Process tracing techniques, such as Exploratory Sequential Data Analysis, may be used to uncover common sequences of interactions and identify bottlenecks. The Information Flow Analysis technique may be used to find bottlenecks in information flow and suggest how they may be alleviated. Also, techniques that identify the critical functions, such as ACTA and the Critical Decision Method, are useful in identifying leverage points relevant to timeline reduction. These techniques can be used to find the critical cues that experts rely on to make decisions in time-pressured situations that novices may easily miss. A technique known as the Cogwheel Experiment (Militello et. al., 1998) may also be applicable. With this technique, a scenario is executed with command post functions speeded up to two or three or four times the normal pace. This results in a period of confusion, followed by a rapid development of workarounds. The experiment exposes the critical functions and identifies potential strategies for improving effectiveness.
   
   
2. Iterate. Again, the hallmark of a successful re-engineering effort is iteration. Changes made to a system with the goal of timeline reduction may inadvertently eliminate seemingly redundant functions that actually served to provide a measure of double-checking to reduce errors. These unintended consequences can only be exposed through a program of iteration and testing. Modeling and simulation techniques, such as Computation Cognitive Modeling and Computational Task Simulation, can help designers predict the results on the timeline with a new system design. High fidelity simulations and realistic testbeds will also be necessary. Human Reliability Analysis techniques can also be used to assess the effects timeline reduction may have on error rates.
   
   
3. Address training requirements. Training is key in systems that operate with a reduced timeline. And since optimization of human-human collaboration plays a key role, the training of teams is of particular relevance. Team members must know what to do and when, they must know when and how to compensate for their teammates, they must know which materials and information to provide teammates, and they must know how to fulfill responsibilities and manage resources without prompting by other members (Blickensderfer, Cannon-Bowers, Salas, & Baker, 2000). To develop effective training, Cognitive Engineering methods must analyze team knowledge of tasks procedures, sequences and timing, team member roles and responsibilities, teamwork behaviors, and how teams build a dynamic shared understanding of an unfolding situation. A variety of CE methods are appropriate for the analysis of teams and design of team training, including the Goal Directed Task Analysis method, which focuses on uncovering the information elements that are important for developing a shared situation awareness. Conceptual techniques, such as rating and sorting tasks, can be used to elicit knowledge regarding the domain elements and relationships among those domain elements. Interviewing and observing techniques, including group interviews and simulations, may also be used to examine team cognitive processes. Cognitive Task Analysis techniques such as ACTA and the Critical Decision method can be used to determine the critical decisions teams make and how they go about making those decisions.
   
   
4. Address technologies that best support timeline reduction. Optimizing the manner in which a team builds a shared picture of an unfolding, dynamic situation is one of the keys to timeline reduction. Thus, technologies that allow people to build and maintain situation awareness are of paramount importance. These tools for maintaining, updating, and communicating the big picture foster enhanced team communication and reduce bottlenecks that result from different team members having different interpretations of the current situation. Cognitive Engineering methods are particularly relevant in the construction of such tools. Most relevant is Goal-Directed Task Analysis, which focuses on situation awareness information requirements. CWA and ACWA are also particularly relevant, since they produce functional decompositions of a domain and facilitate the creation of displays that integrate system variables into a meaningful picture, allowing for rapid assessment of information.
   
   
5. Design for building common ground across multiple agents (Woods, 2001). Not only is it important for team members to have a consistent model of the world and a common ground, it is also necessary for people and automated systems to share that same common ground. In human-human collaboration, common ground depends on the accuracy of assumptions between team members regarding goals, knowledge states, workload, and priorities (Klein & Pierce, 2001). Common ground is something that must be maintained and calibrated as situations go along. When common ground is lost in a time-critical situation, there may simply not be enough time to re-establish it. Klein, Armstrong, Woods, Gokulachandra, and Klein (2000) found that technology that allowed team members to monitor the stance of others, including things such as their workload, fatigue level, and focus of attention, were most effective in sustaining common ground. Cognitive Engineering methods such as Goal Directed Task Analysis, ACTA, CWA, and ACWA may be particularly useful in identifying the information elements that underlie key decisions. Understanding the key information elements is critical to designing shared displays that allow team members to maintain common ground with other team members as well as with automated systems.
   
   
6. Create integrated rather than system-oriented displays. System-oriented displays are those in which the states of variables within the system are displayed. Integrated displays are those in which states of system variables are combined and presented in a manner that is congruent with the decision making goals of the operator. Integrated displays allow for rapid situation assessment. Techniques particularly relevant to the creation of integrated displays include Goal-Directed Task Analysis, CWA, ACWA, and Cognitive Function Analysis.