OMF Members Council meeting 2026 - Breakout discussion on standards

[cmbarton]

Follow up discussion on proposed standards

[alee]

Issues raised during the discussion

1. the ODD document "breaks up the narrative" and is not an ideal way to share the model for different audiences.

This might be obviated as a semi-"generated" artifact, where the underlying data forming the ODD is the primary source of truth, e.g., the intermediate reviewable artifacts that capture the following:

Research Question
        │
        ▼
Conceptual Model
        │
        ▼
Assumptions
        │
        ▼
Implementation Plan
        │
        ▼
Code
        │
        ▼
Verification
        │
        ▼
Results

[alee]

Who is the ODD for? People? Machines? Both?

how do we version our standards? calver seems to be the right fit.

should there be an OMF metadata standard for models. CodeMeta may be too general but we should look into what it might take to make it sufficient instead of building our own. DataCite may be good enough as well

metadata examples

• CSDMS model info: https://csdms.colorado.edu/wiki/Model:DVMDOSTEM
• ML Model Cards: https://deepmind.google/models/model-cards/ https://skops.readthedocs.io/en/stable/model_card.html
• CoMSES.Net model info: https://www.comses.net/codebases/?ordering=-first_published_at&peerReviewStatus=reviewed
• Make Models FAIR: https://tobefair.org

suggested names for the new ODD

Let's come up with some names for the first OMF Documentation Standard based on ODD. It shouldn't have ODD in the name, but should convey its intent, which is to efficiently transfer information from one modeler to another and serve as a lingua franca for all computational models

Model Description Standard (MDS)
Model Understanding Standard (MUS)
Model Interpretation Standard (MIS)

[cmbarton]

A Standard for Documentation and Description of Computational Models
A Protocol for Documentation and Description of Computational Models

[cmbarton]

Actually I'm leaning toward Model Description Standard (MODS) or OMF Standard for Model Description (OSMD). Short and understandable.

Since this is designed to be expandable to other models, we probably don't need to qualify it as applying to computational process models.

ODD is the stricture of the standard.

[alee]

also need standards to describe data inputs / outputs - the Model Interface

[cmbarton]

Like BMI but more general?