The OSM Data Model determines how our data is structured in the OSM database. In general it is working well for us, but it has its problems. We have changed it in the past and we can change it again, if we find better ways of doing things.
This repository documents what we have and where problems are and here discussions can take place to see how we can evolve the data model into something better suited for the future of OSM.
At the State of the Map conference there was a talk "Modding the OSM Data Model" by Jochen Topf about this subject (slides, video). After that we had a workshop "Areas, Routing, and Diffs: Can we have Something Better than Relations?" by Roland Olbricht, which was also about this subject (sorry, no video).
At the FOSSGIS2018 we had a discussion on some of this already. A video is available (in German).
See also the wiki pages about Areas.
To find the best new data model we need statistics. And we need tools to convert from one format to another to analyze them. Some tools are available from https://github.com/osmcode/osm-data-model-tools.