How to Perform Evidence-Based Analysis with Flying Logic

How to Perform Evidence-Based Analysis with Flying Logic

This category of entity classes is suited to an environment when a more probabilistic mode of analysis is desired. One real-world scenario where Evidence-Based Analysis is useful is in Competitor Analysis. Usually an analysis is designed and then carried out over a period of time. During such time, Propositions may be discovered to hold, which may trigger further actions by the agency conducting the analysis.

Flying Logic Setup

There are two styles of Evidence-Based Analysis: belief-network and probabilistic. If the belief-network style is used, the Flying Logic document is set up with Proportional (::) for both the entity operator and default junctor operator. If the probabilistic style is chosen, the document is typically set up with Sum Probabilities (⊕) as the entity operator and Product (×) as the default junctor operator. This setup is analogous to the use of Fuzzy Or (OR) and Fuzzy And (AND) in Sufficient Cause Thinking.

Setup for Belief Network
Setup for Probabilistic


Propositions (also known as requirements) are questions for which the analysis is intended to discover the most likely answers. Propositions take the form of a statement that has some probability of being true. Determining whether the probability of the Propositions exceeds determined thresholds is a primary purpose of Evidence-Based Analysis. Propositions are analogous to goals in Effects-Based Planning, and thus are terminal, i.e., they are always successors and never predecessors.


Indicators are potential causes for Propositions or other Indicators, and can be considered analogous to Intermediate Effects in Effects-Based Planning. Another way of thinking of Indicators is as inferred evidence. Each Proposition typically has a set of Indicators that feed into it, each of which is considered to be a possible cause of the Proposition, and which together form a “template” for recognizing that the Proposition holds (i.e., that the requirement has been met.)

Each indicator in turn may have a set of more specific indicators which feed into it and form a “sub-template” for recognizing that the indicator in question probably holds. Indicators are usually both successors and predecessors.

In a complex analysis, individual analysts can be assigned responsibility for certain indicators, which places them in a supervisory role over all indicators that are predecessors of the indicators for which they are directly responsible.


Events represent direct evidence which becomes known throughout the life cycle of the analysis. In the intelligence community for example, Events may be derived from Signals Intelligence (SIGINT), Human Intelligence (HUMINT), or Open Source Intelligence (OSINT).

Events are always predecessors and are never successors. They are assigned a Confidence value based on their reliability (or probability.)


Knowledge represents pertinent facts known to be true about the situation under analysis. Knowledge can be built into the analysis before events are received, or can be added to the analysis in response to events as they occur. Knowledge entities are combined with Events to provide context and semantics either supporting or refuting the various indicators into which they feed.

Like Events, Knowledge entities are predecessors and not successors. They are assigned Confidence values based on their reliability (or probability.)

Edge Weights

Edge weights in the model are assigned based on the positive (or negative) correlation between each entity and its successors.