“Followed by” should consult chronology relation

[1ec5]

The “Followed by” functionality is practically unused because it relies on undocumented followed_by:name and followed_by keys that are used four and 165 times worldwide, respectively. However, it requires followed_by to contain a URL, which is never the case as of writing. The inspector should instead link to the next feature in the chronology relation(s) of which the element is a member, assuming that the chronology relation is chronologically sorted.

https://github.com/OpenHistoricalMap/ohm-inspector/blob/901cd7e0f7063b3abb677e6a38cb66c9841fbc9d/openhistoricalmap-inspector.js#L262-L263

/ref OpenHistoricalMap/issues#189

[jeffreyameyer]

This type of tagging was very experimental; inspired by Who's On First's preceded_by, followed_by, and supersedes tags; and never followed up from earlier days of OHM, although it appears to have caught on slightly, recently. It's a fairly interesting concept, although it could probably be outboarded to Wikidata to some degree.

The automation of preceded_by and followed_by within a chronology makes sense and would be super cool, but that is not the full set of use cases for this type of tagging.

Sometimes there will need to be either automated connections through wikidata or some accomodation for manual tagging to support linkages between different chronologies.

Small examples in the US would include West Virginia preceded by Virginia or Tennessee preceded by the Southwest Territory preceded by UFT preceded by North Carolina, etc.

The need for supporting non-automated tagging is clearer in the case of commercial tenants of a building in neighborhood histories.

I'd suggest the lack of proper use is somewhat tied to a lack of documentation and education, but also not saying that if we had great docs and great training that it would still be more widely adopted.

[1ec5]

Small examples in the US would include West Virginia preceded by Virginia or Tennessee preceded by the Southwest Territory preceded by UFT preceded by North Carolina, etc.

To the extent that something needs to be explicitly modeled as a successor to something else, they can be part of a shared chronology relation. A feature can be part of multiple chronologies; a chronology represents whatever we’d describe as a definite historical progression, such as the concept of a successor state.

That said, I don’t see a strong need for explicit tagging in the case of West Virginia and Virginia. It’s already quite evident that West Virginia split from Virginia based on the spatial overlap. followed_by would be problematic in this case, because West Virginia only “followed” Virginia in some places but certainly not in others. In a spatial database, this is the job of geometries, not merely textual metadata. If necessary, perhaps the inspector can link to a spatial query showing which territories overlap in different time periods? An OHM-based Who’s On First could probably derive its followed_by properties from chronology relations and overlapping geometries rather than needing literal tagging.