Paper #48. Supply Chain Networks in the Age of AI, Part I: Fixing Joint Doctrine

  • Thomas A. Drohan, Ph.D., Brig Gen USAF ret.
  • Cyber, Leadership, Strategy
  • No Comments

This paper is Part I of II that defines and applies information environment concepts to supply chain networks for an age of artificial intelligence (AI).  To avoid overtrained learning, I draw from supply chain examples and beyond.[i]

Part I fixes two fundamental flaws in joint military doctrine: (1) the failure to define the information environment in a testable way, and (2) the failure to define information at all.

Defining the Information Environment


To manage the pervasive influence of information including generative AI, we need a holistic definition of the information environment (IE). Holism is the idea that a whole is more significant than its interrelated parts—such as a critical supply chain with interactive partnerships. Ecosystems show us that their parts can interact such that “simple, local, decentralized processes can result in global patterns.”[ii] From an effects perspective, the whole doesn’t have to be greater than the sum of its parts. It has to have more significant effects. This feat is difficult to pull off. The massive, 19-year US-led combined arms and counterinsurgency campaign against Taliban armed groups failed to achieve one critical effect–breaking their will to fight. The Taliban’s long war of resistance and brutal intimidation involved fewer physical resources, but exhausted US will at the Presidential level of significance.


The opposite of holism is reductionism, which understands a whole only in terms of its parts. Such closed system supply chains are rare, but we still try to isolate or simulate them as if they were. “Not in my lane or job jar” attitudes and analysis perpetuate reductionism even as the IE blows up bureaucratic boundaries. So, it takes leadership to envision the bigger picture and motivate collaborative efforts.


There are so many interactions affecting supply chain networks that we need holistic synthesis to compete in the milieu. I’ll discuss dialectical synthesis in Part II. For now, let’s see what joint military doctrine gives us.


Information Environment

The joint military doctrinal definition of the information environment (IE) recognizes the IE as a holistic aggregate with various interactive parts:

Joint doctrine definition
“The aggregate of individuals, organizations, and systems that collect, process, disseminate, or act on information.”[iii]

The Joint Chief of Staff’s Joint Operating Concept’s definition treats the IE as a transcendent actor itself, an aggregator consisting of information other actors use to assign meaning and understand the world:

Joint Operating Concept Definition
“The IE is comprised of and aggregates activities that an observer interprets and uses to assign meaning. Informational aspects include but are not limited to timing, platforms, location, and duration. They are the sensory inputs used by individuals, systems, or groups to assign meaning and gain understanding about the world. The IE directly affects and transcends all OE.”[iv]

Notice that these definitions are descriptive, not explanatory according to the scientific standard of falsification. The problem is, they use “information” to define the information environment. This logical error is like defining the global environment of a supply chain solely by the supply chain’s parts. Plenty of people do it, but it yields thick description, as in cultural understanding, but not a disprovable explanation. As a result, we gain a partial understanding of the whole.

A fundamental reason for this flaw is another one–the lack of a definition of information in joint military doctrine. That may seem unnecessary, but its absence proliferates non-falsifiable definitions. How? There are plenty of information activities and functions that use “information” in their definitions. Examples include information operations, information management, information operations force, information operations intelligence integration, information-related capability, information report, information requirements, and information superiority:

  • Information environment: the aggregate that collects, processes, disseminates and acts on information
  • Information exchange: exchange of information essential to C2
  • Information management: managing information resources
  • Information operations: integration of information-related capabilities
  • Information operations force: a force that trains or integrates information-related capabilities against adversaries
  • Information operations intelligence integration: of disciplines and methods to characterize, forecast, identify vulnerabilities, determine effects and assess the information environment
  • Information-related capability: tool, technique, or activity employed within a dimension of the information environment for effects and conditions
  • Information report: of raw information
  • Information requirements: of the commander
  • Information superiority: an operational advantage

These definitions make sense to people who are used to them, spawning more circular definitions. The failure to define information in a testable, disprovable manner creates assessment problems. Anything can be information ops, information management, information force, and so forth. Commanders get frustrated because everything is information, so the concept lacks particular value. As a result, it’s harder to set priorities and easier to run irrelevant operations.

This dilemma of bad alternatives is fixable. We need to know when something is not information, just as we need to know what’s not part of our supply chain or not a target for influence. If everything is information, we can always find information to confirm what we think is true. We avoid disproving it. That happened in the run-up to the US war in Iraq when nobody looked for evidence of the absence of WMD. It wasn’t politically acceptable at the time.

Reducing the ambiguity of what is and what is not, is more obviously critical in supply chains. We can reduce them into parts, components, products, and services, including hardware and software. The need for holism comes in as competitors contest all aspects of a supply chain and anything else that affects them.

The following two illustrations represent each of the above definitions.

Depictions of the IE

Figure 1 is the JCOIE definition of the IE.


Figure 1

This depiction describes the information environment as presented in the joint concept.[v] The environment has three dimensions, one of which is information. This concept describes the information environment but does not help explain and test the causes of information. That’s important because to compete, we need to influence causes, not just effects after or as they occur. We want to fix what will break, not just what’s broken. In a dynamically competitive environment, “if it ain’t broke, don’t fix it” is loser’s advice.

MOP / MOE Example

Consider a physical network (physical dimension above) that influences the information environment. How do networks influence behavior? Our standard way to answer that is assessment, specifically by developing measures of performance (MOP) and effectiveness (MOE).

Suppose the MOP for the network is the number of marketing messages posted on social media. This measurement indicates the network is performing to a standard we chose. Suppose the MOE is the number of positive responses to our sales pitch. This measurement shows how effective the network is in achieving our objective—creating a positive sales and investment environment for our product or service.

increase sales and investment via social media

number of marketing messages we post on our social media website

number of positive responses to our message

The problem for assessment is that now we’re in the informational dimension that defines the whole we are trying to explain. We can’t disprove cause-and-effect propositions in the physical and cognitive dimensions without going into a dimension that’s indistinguishable from the information environment.

For instance, we can’t test whether changes to our physical network create better or worse sales and investment without considering data, which the joint concept places in the informational dimension. That doesn’t make sense because the causes of changes in sales and investment include data and other components in the physical and cognitive dimensions. The informational dimension of the information environment is redundant. What’s the problem with that?

We can apply MOP and MOE to most problem sets. To combat the scourge of ISIS recruitment, replace “sales and investment” with “religious tolerance and human rights” (Objective). This does not make execution to influence behavior easy, of course, but at least we can develop solutions for the problem at hand–the threat of ISIS recruitment.

Instead, our model of the IE diverts our attention to symptoms–we avoid the causes of information. As a result, we chase solutions and self-inflicted surprises. Some “black swans” emerge in the deluge of information because we have not anticipated them. Instead, we passively describe a complex IE and lose initiative. At best, we respond to problems and competitors. That’s reactive resilience, such as bidding for a new supplier in an already-disrupted market. Instead, we want proactive resilience, such as signing directed-buy contracts with multiple suppliers in advance. Managing risk is part of the solution, which requires leaders who want to understand potential causes.

Our basic failure was to conflate the part with the whole, so we had no other solutions to our problem. In the example above, we combined information = supplier (the part) with information environment = supplier (the whole). We needed to look beyond our supplier and adding more suppliers, such as investing in warehouses.

The second illustration solves this problem in two ways.

Figure 2

First, let’s analyze the information environment into two interactive, distinct dimensions. Keep in mind that these dimensions represent sources of information.

The physical dimension is the same as Figure 1 except for replacing “tangible, real-world” with “matter.” So, the physical dimension is matter-centric and includes energy. This analysis reflects our scientific understanding of physical reality—matter and energy that convert into many forms but aren’t destroyed.

The psychological dimension is broader than “cognitive” and includes behavior. The psychological dimension is data-centric and includes analog and digital forms of energy. The psychological dimension is no longer human-centric. Humans and other sentient beings, notably artificial intelligence (AI) for now, process data into a context, producing information. So, if we focus on one particular type of actor–humans, we will neglect AI-created information and aggregates created by non-human actors, including climactic forces.[vi] Cyborgs and technology-enhanced animals and insects, anyone?

These two dimensions isolate the causes of information as matter, data, energy, and how they interact. Now we need a specifiable definition of information itself.

Second, we define information as meaningful data in an operating system of inputs, changes, and outputs.

Defining Information 

Information: the values of characteristics in the input, change, and output of processes. 

[Note: this definition is an expansion of Robert Losee’s “discipline independent” definition: “the values of characteristics in the output of processes,”]

The power of defining information this way is to shape the information environment better than one’s competitor.





Figure 3

Note that the outputs become potential inputs for more information operating cycles. When we understand information as part of a dynamic operating system, information is not separate from operations. Information is an operational process, as in a supply chain. How does this work?

Information and intelligence are much more than understandings someone collects to support operations. We have to develop data and information to create the understanding to shape operations. Actors process data into information by assigning meaning as their capabilities and circumstances permit. They operationalize data, information, intelligence, and knowledge into actions. We’re used to this language that puts information in a “supporting” role for “supported” operations.

How do adversaries compete with that? They go holistic. They also informatize operations by creating data and processing it into meaningful information, contextual intelligence, and accepted knowledge that influence targets.

Most commanders, operators, planners, and strategists refer to all those as information effects. Supply chain networks, however, have all four types of information-related effects:

Comparing IREs & IRCs

Examples of Information-related Effects

  • Data effect: network-wide visibility (preferably end-to-end)
  • Information effect: improved delivery (what data visibility can mean)
  • Intelligence effect: increased customer satisfaction (a context of improved delivery)
  • Knowledge effect: accepted analytics (a reputational result of all the preceding)

Compare these information-related effects (IRE) with the joint doctrinal definition of information-related capability (IRC):

“A tool, technique, or activity employed within a dimension of the information environment that can be used to create effects and operationally desirable conditions.”[vii]

What’s the difference? IRCs are means and ways to influence rules, norms, and beliefs among influencers, vulnerable populations, and mass audiences.[viii] IREs help us clarify the ends; effects to establish desired conditions (also known as “end-states”). They inform our selection and development of competitive capabilities to realize our desired conditions. We also can use IREs to red-team any actor more thoroughly.

Competitive questions for the four IRE types include the following.

  • Data: is your network visibility as holistic as the competition?
  • Information: are delivery improvements continuously controllable?
  • Intelligence: is customer satisfaction suited to where the market is headed?
  • Knowledge: is your analytics reputation local, regional, and global?

Similarly, the information operating cycle depicted in Figure 3 can focus our efforts on what we can do in the IE to increase the value of our product or service and decrease that of adversaries. We can assess our efforts and those of adversaries when we define information in a disprovable manner. In other words, everything is not information or relevant information.

What is not Information is Critical to Success

The question, “what is not information?” is critical to assessing progress and lack of it toward our objectives and goals. According to our definition of information, if the data is not meaningful, it is not information. The data needs to be put into context, such as the temperature and location of a medicine shipment. Also, if we have meaningful data but do not consider how the data can be an input, a change, and an output, then the information is not relevant.  Temperature and location data are relevant as inputs to describing the medicine’s condition, prompting an alert if it’s outside parameters so we can make a change to ensure safety and prevent delays.

We make information relevant or fail to do so. How?

How to Make Understanding Relevant to Operating Cycles and Effects

How do we make our understanding of the IE relevant? Generally speaking, we can operationalize our understanding and use our operations to create new understanding. One way is to operationalize information and informatize the operations, as we explained below Figure 3 above.

Operationalize Info and Informatize Ops

That way is to operationalize information into actions and informatize operations. This process creates information effects.  However, there is more to understanding than information. This means we can do more than operationalize info and informatize ops.

First, we can operationalize data, info, intel, and knowledge, not just information. For instance, we use targeting data, give it geographic meaning and a political context to create operational intelligence. At any point in that process, we may stop or be stopped by a lack of understanding. Or not, unfortunately. An operation that relies on data others don’t have may still be a relative advantage. It’s best to understand what the data means in a particular cultural context, especially if one’s competitor does not. That’s an information and intelligence advantage.

Second, we can use operations to create data, information, intelligence, and knowledge effects, not just information effects (“informatizing”). At the “low” end of understanding, erroneous data that causes industrial machinery to malfunction is a significant data effect. At the “high” end, victimization narratives accepted as knowledge cause believers to comply with authoritarian regimes who claim legitimacy based on the narrative. That’s a huge knowledge effect. The overall point is to make our understanding (data-info-intel-knowledge) relevant to operating cycles (inputs, changes, outputs) and desired effects.

To clarify this integrated approach to information and operations, let’s return to our MOP / MOE example, where we want a positive environment for our product or service to increase sales and investment.

We create messages (an info input) that cause (the change) positive responses (an info, intel or knowledge output). Note that the outputs vary by category depending on whether they have meaning (“info”), context (“intel”), or broad acceptance (“knowledge”).  We have physical and psychological means, our inputs to improve product or service sales and investment. We don’t have a separate informational dimension because we break down the IE into value-added interactive physical and psychological components (“value-added” refers to our information definition–the values of characteristics).

What we do need are value-added parts we can control.

To be sure, the physical and psychological dimensions contain information. However, instead of separating information into its own dimension, these two dimensions integrate information as operational processes, just as a supply chain consists of distinct, interactive parts.

Now we can consider any physical component–such as network hardware, and any psychological component–such as a narrative, to enhance message quality. Examples include processor speed, visual appeal, and emotional traction. We can test our modifications as an end-to-end information operating system–data to knowledge. Our information definition is a cycle, so end-to-end reinvents itself through feedback.

The information operating cycle applies to any supply chain network. In fact, supply chain examples help us in other areas where parts of wholes are more difficult to discern, such as narrative warfare. In a supply chain, even hard, physical structures exist as information we choose to represent them. Companies develop new categories and arrangements all the time—from transportation, spare parts, manufacturers, distributors, and retailers to logistics, hub-and-spoke distribution centers, and sensor-based tracking.

Now that we have established an assessable definition of the IE and information, the next Paper will apply advanced analysis techniques adapted from JMark Services’ Information Environment Advanced Analysis course.



Author: Thomas A. Drohan, Ph.D., Brig Gen USAF ret.

Leave a Reply