OODA Loop 2.0 = "OODAR" - The Art of Fighting Fires with Water-less Buckets in a Flammable Suit

Adapting Boyd's OODA Loop for the Modern Enterprise Environment... A systematic approach to increasing the velocity and accuracy of critical decision making from crisis operations all the way to strategic planning.

Typically when unexpected really bad things happen most organizations lack the capability to effectively identify much less make critical decisions in a timely manner. In our new existence a.k.a. COVID19 reality - organizations and industries are having to adapt and pivot quickly to remain in business much less remain relevant. I would like to bring a military concept into the professional world or make it more front and center - that being Colonel Boyd's O.O.D.A loop.

In order to adapt the loop to our new reality I propose Version 2.0 or "OODAR" to incorporate "After Action Reviews" or Lessons Learned as the appending "R". This improved version of feedback intelligence will better support Organizational Critical Decision Support both in crisis operations as well as long term strategic planning. Lastly, there is a "Decision Classification Model" and OLAP lesson to help identify the critical from mundane as well as how to decision efficiently (last bolded section at the bottom).

Points covered in the article to follow:

• Brief history Colonel Boyd & the OODA Loop
• Gartner version of OODA Loop for Enterprises

What's New:

• Proposed Adaptation of OODA to OODAR & Identified Goals
• OODAR Chronologically applied (for evolving situations)
• OODAR Hierarchically applied (for Organizational understanding and improvement)
• OODAR applied for individual process improvement
• Decision Classification Model (an effort to help weed out the critical decisions as opposed to the less critical and a way for Enterprise Organizations to quickly identify candidates for potential workflow or decision automation)
• Online Analytical Processing (OLAP) Cubes - basic definition and some practical examples
• How OLAP Cubes actually work and how they can help identify trends in extremely complex data sets quickly (Roll-up, Drill-down, Slice & Dice and then Pivot / Rotate)

Brief History on Boyd & the OODA Loop to set context:

"Forty-Second Boyd", also known as Colonel John Boyd (1927-1997) had a standing offer to all pilots - if they could defeat him in simulated air-to-air combat in under 40 seconds then he would pay them $40. As an instructor at the Fighter Weapons School (FWS) at Nellis AFB, he fought students, cadre pilots, Marine and Navy pilots, and pilots from a dozen countries, who were attending the FWS as part of the Mutual Defense Assistance Pact. He never lost. He was responsible for revolutionizing the way fighter aircraft are designed, tested and evaluated (E-M Theory) as well as making other impacts like the Fighter Mafia. Most notably he was the creator of the OODA Loop as depicted below:

Colonel John Boyd was a critical decisioner ahead of his time but all formulas can potentially be improved over time. In Boyd’s era the Air Force didn’t perform after action reviews/reports (AARs) – in his defense – had the service performed them I am sure he would have incorporated them in the first OODA / Feedback Intelligence model. The basic steps of the OODA loop are described below:

O - Observe - Observe is centered around our own hypothesis' and assumptions. The goal being to discern any misalignment between our perceived reality and the true reality around us. This protects critical decisions from any cognitive or cultural biases.

O - Orient - The Goal of Orienting is to position your organization into a position of advantage based on observations / outmaneuver the opponent. Typically this is accomplished by orienting to an enemy’s or situation's identified strengths & weaknesses / vulnerabilities – five main considerations):

• Cultural Traditions
• Genetic Heritage
• Previous Experience (this will be enhanced by future AAR takeaways)
• New Information
• Analysis & Synthesis of the prior four categories)

D – Decide (most efficient way to act with least risk of exposure). This is typically unique to each organization but can be improved by the "Decision Classification Model" described at the bottom of this article.

A – Act (as quickly or violently as possible until the objective is achieved) There is an old military adage here that goes something like - a bad plan executed violently often works.

Gartner Version: the O.O.D.A. Loop for applications in commercial Enterprises -

What's New:

OODAR Proposed Adaptation:

This article is proposing OODA Version 2.0 as "OODAR" now to incorporate "After Action Review" or Lessons Learned as the appending "R". This improved version of feedback intelligence will better support Organizational Critical Decision Support both in crisis operations as well as long term strategic planning.

OODAR Goals - Why is adding the "R" so important:

(hopefully the German translations are not offensive as the idea is better alignment on common critical objectives across the organization)

The object being to speed your loop up faster than an enemy allowing you to survive/succeed and them ultimately to succumb. This typically takes different forms at organizational levels however it continues to lack one prime element which all Service Branches have adopted post Boyd’s era – that is some form of the After Action Review / Report etc. Typically, an AAR takes shapes by defining the following four (4) questions:

What was supposed to happen?

What actually happened?

What was good / bad?

What do we plan to do next time?

This portion of the loop would not take place in the immediate action phase but should be incorporated in unit level tactics in between engagements if faced with a new enemy or opponent. With new enemies come new challenges and unknown vulnerabilities – sadly the surviving members of any initial skirmishes must map these and their initial experiences might provide a decisive advantage for future force interactions.

We can enhance the existing OODA loop by continuing the feedback loop post-action to enhance tribal knowledge or “data-on-previous-engagements” or as some would call “d.o.p.e.”. In the private sector we call this “Continual Service Improvement” or “CSI”. This principle combined with the existing decision model improves long-term outcomes at the higher organizational level.

If the “R” or “Review” component could be incorporated into the loop at the Organizational level we should be able to improve outcomes over longer term engagement windows by capturing the results both positive and negative and adapting tactics, techniques and procedures (TTPs) / theater or adversary-based strategy to incorporate the feedback intelligence gained by prior engagements.

OODAR Chronologically applied:

The concept can be applied chronologically at an organizational level in order to identify the most optimal approach in order to achieve a desired or "intended" future as seen below:

OODAR Hierarchically applied for Organizational Understanding and Internal Improvement:

This type of concept (if it were to be combined with Artificial Intelligence) could yield better Organizational understanding, better resource utilization models, greater efficiency etc. This type of generational growth could be viewed similarly to Maslow's Hierarchy of needs - almost like an Organizational Maturity Model - concept visually described below:

OODAR applied for individual Process Improvement:

In addition to potentially helping improve the chances of Organizations integrating a concept like OODAR with Artificial Intelligence and a Big Data set of information (goal being to create a recommendation engine or eventually an automated decisioner / actioner) this concept can also be applied at the individual process or workflow level. Below is a diagram regarding how OODA was applied to improve the efficiency of a killchain:

Although this has a particularly military-focused application - imagine the primary processing changing from "Monitor", "Target" and "Prosecute" to "Customer Request Intake", "Customer Request Processing" and "Customer Request Delivery/Execution"... etc.

We can show a specific commercial Enterprise application of OODA in the associated diagram (use case Enterprise Event & Incident Management):

Decision Classification Model

How do we identify and classify Critical Decisions in terms of which desperately need automated actions/alerting related to them? All of this is great - but how can we leverage OODAR and AI to make better decisions faster - thanks to Chris Rafael (only reference I could find) please see the proposed "Decision Classification Model" below. This matrix could also be used by Enterprises to identify decisions/workflows as automation candidates:

Once an Enterprise has identified which decisions desperately need to be decisioned in a more automated nature - how do we identify the trends in the data quick enough to make actionable decisions?

Enter the Online Analytical Processing (OLAP) Cube.

An OLAP cube is defined by Wikipedia as a multi-dimensional array of data and is typically a computer-based technique of analyzing data to look for insights. The term cube refers to a multi-dimensional dataset, which is also sometimes called a hypercube if the number of dimensions is greater than 3. In the graphics pulled from Google below we can see multiple applications of 3D OLAP cubes to better understand Enterprise issues at a Macro scale:

The image to the right tracks sales-related-information for various fuel types month-over-month. In short, OLAP cubes are great are doing multivariate comparisons over time-based intervals within a given environment.

The same type of analysis can also be instantly applied for healthcare related trend analysis as seen in the OLAP cube image below:

How does the OLAP cube actually work?

Credit to the following site/author for laying it out concisely: https://www.guru99.com/online-analytical-processing.html

There are four (4) primary types of analytical operations in OLAP and they are:

1. Roll-up
2. Drill-down
3. Slice and dice
4. Pivot or Rotate

1) The Roll-up: this is typically referred to as the aggregation phase - and this can normally be accomplished by doing one or both of the following activities.

A. Reducing Dimensions

B. Climbing up the concept hierarchy. The concept hierarchy is a system of grouping things based on their order or level.

These can be explained by the graphic below:

In the graphic above the domestic cities get "rolled-up" or aggregated into national metrics. In order for the "roll-up" technique to be considered effective - one or more dimensions need to be removed. In the example above the data set is further simplified for rapid decisioning by removing the city variable altogether. Depending on what decision needed to be made - this might not be a relevant variable - i.e. if we wanted to know the contribution that PCs, Books, Shoes and Clothes made on US & Australia's respective GDPs - the city would not be as relevant. However if we wanted to compare east and west coast US to different areas of Australia then perhaps the regional trend would be a much more critical data point. As the decision support team are going through variables these are the types of questions that need to be asked to see if rolling up is appropriate given their specific situation.

2) The Drill-down: In the simplest form this is the opposite of the Roll-up. In the process of attempting to make a decision there come times where key variables are still not understood at a level appropriate enough to make an competent decision. This is done on the OLAP cube by moving further down the concept hierarchy or increasing/sub-dividing one dimension. This is demonstrated in the picture below when the Time dimension is further divided from Quarters to individual Months:

3) Slicing and Dicing: Once the data set is appropriate for the issue at hand (all Rolling-up or Drilling-down has been completed and the decision is appropriately understood given its criticality - then comes - how do we accelerate making the best decision.

The first way to do this is to slice the problem into consumable chunks that can be actioned or analyzed individually. In the example to the left apparently the critical decision involves having a better understanding of what occurred during Q1 specifically. If this was supporting some type of quarterly-earnings-related decision - perhaps where to spend Q1 marketing dollars based on prior data...etc.

Sometimes looking at the variables across only one slice is not enough and there are specific conditions or situations that need to be identified quickly within a data set. Dicing is when we look across multiple dimensions - this can be in any relational direction. The image below details a fairly rudimentary dice in the form of looking at a semi-annual data segmentation (Q1 & Q2) as opposed to just the Q1 slice above. The main point here is slicing and dicing can also be diagonally through the OLAP cube but for a practical visualization - please see the image below regarding dicing:

4) Pivot or Rotate: Often times while attempting to develop solutions to complex problems we often end up feeling like we have been staring at the same problem day after day and we just cannot seem to find a new angle. In these scenarios or ones where completely unconventional or unexpected solutions are necessary sometimes it is best to Pivot or Rotate. This is where the magic happens - and this is where Artificial Intelligence can take human decision support into the next generation in the form of an AI Advisor / Recommendation Engine. Rotating / Pivoting is rapidly reshaping the cube to put different variables/dimensions on a common face. In the Rolling Up example from the beginning we would look at Countries consumption rates of various clothing items with the time variable being the depth metric to the cube. Or put simply - pretend you are an externally viewer of the OLAP cube and literally rotate the cube so that it is viewed from a different face or perspective. Sometimes these different perspectives can inform the viewer of trends that were not otherwise easily visible or identifiable.

The reason OLAP cube style decision processing is so efficient is that it allows a user to rapidly sort a massive amount of information into a useable format from which the majority of the data (if unneeded for decision support) can be discarded instantly making the critical variables that much more apparent (as well as their related trends). OLAP cubes are great at immediately identifying complex correlations between seemingly unrelated variables otherwise and can provide key decision makers the advantage when time is of the essence.

Colonel Boyd's OODA becomes OODAR; Artificial Intelligence and Big Data bring an opportunity to make more timely and more accurate Enterprise Decisions through the real-time application of OODAR and we can identify which workflows to focus on as automation candidates with the Decision Classification Model. Lastly once the automation candidates are identified we can leverage the OLAP Cube to process the data and action in a near-real-time manner.

In summary - OODA V2 for larger Enterprise Organizations becomes OODAR :)

Hope it was worth the read; please leave your comments.

#One2Watch #40SecBoyd #OODA #OODAV2 #OODAR #CriticalDecisionMaking #OrganizationalDecisionModels #DecisionClassification #RapidDecisioning #RapidPrioritization #CriticalIncidentResponse #CriticalEventResponse #CyberSecurity #WhatAreTheCriticalDecisions #ExecutiveDecisionSupport #ArtificialIntelligence #AI #BigData