After a lengthy hiatus, I’m pleased to say that the HED project is active once again. Work has been going on to a small extent since the last blog post in 2021, with a focus on developing a general strategy for creating multi-agent models with descriptions or specifications that are separate from the technologies used for their implementation.
As noted in previous posts, this independence from technology has traditionally been seen as one of the strengths of mathematical models, along with their deductive power and human readability. Gaining some of these benefits within the context of multi-agent simulations would be valuable for verification and validation of models, and also make them accessible to a broader audience.
Efforts to more directly translate the implemented version of the island game to something more conceptual, abstract and human readable using, for example, mathematical markup languages and tree structures (again see previous blog posts) eventually proved to be intractable for practical purposes.
Changing tactics, and taking inspiration from Conway’s Game of Life, which was originally carried out manually using paper and pencil, I realized that games are effectively a manual, mechanical, implementation of a multi-agent simulation and that, flipping this, the simulated environment of the island game could be written out in a way that made it into a ‘game’ that could be mechanically ‘played’ by people. Once the model was written out in this way, as long as any computer implementation of the abstracted, or conceptual, model was consistent with the game version, then the implementation could be considered to be a faithful representation of the conceptual model underpinning the simulation.
At this point, a manual game version of the island simulation has been created and is being tested for completeness and accuracy. The next step after this would be returning to the implemented model, confirming the synchronization of the two versions, and then proceeding.