Archives of Persona Papers ex libris Ludwig Benner, Jr.
   - - - - - -Last updated 5/29/17

     [ Investigation Research Roundtable Site ] [Investigation Catalyst ]  
[ Contact "me" at ]

MES Investigation Task Rules

This is an abbreviated set of rules for implemening the investigation system more fully described and accessible at the locations noted at the end of this rule set.

Selecting and Acquiring Data
  1. Everyone and everything has to be someplace, doing something during accident process: track their actions
  2. Force yourself to think "who or what did what when" during process
  3. Look for the actor + action in observations and data
  4. Give each actor a name and always use only that name
  5. Give priority to tracking the change makers
  6. Be alert for physical, mental, sensory or "programmer" actions
  7. Break down (decompose) actors or actions to clarify what happened
  8. Let data, not experience, drive search for data.
  9. Observe the "Do no harm" rule with witnesses and objects
  10. Get permissions before recording witness interviews
  11. Actions change conditions: read conditions to identify actions
  12. Quantify actions where possible
  13. Build yourself a mental movie frame by frame
  14. Get "dids" instead of "did nots"
Building EBs
  1. An "event" = 1 actor +1 action, with related dimensional attributes
  2. Transform observed data into actor + action formatted Event Blocks (EBs)
  3. Always record Actor first, then action
  4. Use ? as placeholders until you get needed data
  5. Use one specific and unique name for each actor
  6. NEVER ever put two actors or actions in an EB
  7. Use consistent grammatical tense
    • a. Past tense for past occurrence
    • b. Present tense for planned or operating systems.
  8. Use glossaries for consistent actor and source entries if possible
    • Avoid poison words like <ul
    • pronouns - she, he, they
    • plural nouns - crew, group, squad
    • passive voice- was, were,
    • conjunctions - and, or, but,
    • ambiguous terms, jargon, acronyms, categories, factors
  9. Avoid judgmental or opinion words
    • inadequate, poorly, faulty, unsafe (subjective conclusions)
    • did not (implies error, masks process)
    • failed to or violated (accusatory,)
    • the cause (subjective attribution)
    • root cause (subjective attrtbution)
  10. Record at least one source for every EB
  11. Keep your experience, other external data sources out of EBs
  12. Confirm only observed times as factual; indicate estimtes
  1. Concentrate on data generated by the occurrence
  2. Cite only incident sources that can be accessed,retrieved or reproduced
  3. Cross-reference any codes used to identify sources
  4. Show more than one source for EB when available
  5. Try to maintain a chain of custody for all source objects, documents and media used
Integrating EBs
  1. Crete a blank time/actor matrix for integrating EBs into a flow chart describing what happened
  2. Add each EB to the matrix as it is created
  3. Align each new EB along actor row in temporal and spatial sequence relative to all existing EBs
  4. Actor row should contain EBs describing everything the actor did during the process. Gaps help identify and define unknown unknowns and data to pursue
Linking EBs
  1. Use input/output logic flows to define links: do not assume or guess or rely on intuition
  2. Use solid arrows to show validated input/output links in interactions
  3. Use dashed arrows to show tentative links and then confirm tentative links with data if possible
  4. Strive for solid black arrow links
  5. Key logic test: will linked input behaviors ALWAYS produce the resulting action(s) 6. Use logic trees to bridge gaps to define additional data needed to confirm hypotheses
Matrix validation (quality assurance)
  1. Review logic flow of EBs on Matrixes when completed
  2. Have disinterested third party review completed matrix for logic flow
  3. Verify accessibility of all sources cited
  4. Explain reason for all remaining ? and gaps in linkages (uncertainties)
  5. Remove extraneous EBs and comments to finish Matrix
  6. Title and Sign Matrix if required.
Problem Definition
  1. This is an analysis task.
  2. Examine sequentially every linked EB pair and set, and their links on the finished matrix
  3. Determine whether interaction is a problem that should be addressed
  4. Prepare list of identified problems or lessons Learned
Recommendation Development (if required)
  1. Work through problem list in sequence
  2. State pros and cons for making changes
  3. Identify candidates for changes
  4. Define how success of change will be identified
  5. Tie all recommendations to an EB, link or EB set.
  6. Don't ask anyone to do something you would not want to do yourself
  7. Check carefully for hidden assumptions
  8. Check for unjustified conclusions or arguments
Investigation report
  1. Use the matrix content to define the reported description
  2. if permitted, separate the description of what happened on the matrix from analyses of that description, cause statements and other less reproducible assertions.
Test Plans
  1. Whoever Owns the Ball Calls the Game. (If its your money, you are in charge!)
  2. NO PLAN, NO TESTS! (one of the basic commandments for Investigation, especially if you own what is to be tested!)
  3. Don't destroy it before you get the data it holds now_
  4. Keep Test(s) Relevant to purpose. (i.e., get event blocks)
  5. Scale the Plan to the value of the data it will produce. (Are EBs worth cost?)
  6. Progressive destruction demands priorities
v1.03 Posted 2004-2017 by Ludwig Benner Jr

Additional resources

  • - numereous papers, presentations and a book are available on line. Peruse the list of on-line publications, starting with 1975 papers "Accident Theory and Accident Inveestigation" and "Accident Investigaiton: Mutilinear Events Sequencing methods."
  • for MES guides, models and tutorials, especially the Human Behavior model for individual interviews
  • - variousauthors' papers about investigation problems
  • - for technical notes about what the MES investigation system is and is not.