How should we map uncertainty... indeterminate boundaries, disputed territories, etc?

jeffmeyer · January 18, 2024, 7:19pm

Here’s a compilation of topics around how OHM should depict uncertainty and boundaries.

My take is that if we come up with a basic consensus around a tagging and rendering treatment, I think we could come up with a much more interesting map, especially with the number of historical boundaries I’m anticipating we’ll import in the coming months.

Indeterminate Boundary Treatments for OHM

see also:

OHM Issue #204: Need a smart, sensitive approach for representing indigenous peoples’ land relationships
OHM Issue #553: Need a way of highlighting disputed/contested areas

Reasons for trying to depict uncertain boundaries

Irrelevance to culture and people being mapped
Specific boundaries can have no basis in historically verifiable claims
Differences of historian opinion about location & desire to show multiple views at once
Land disputes of any sort

Solution process

The primary goal of this forum topic is to identify a path for a solution, both in order to work on that solution and to have an approach we can coherently discuss with partners.

As this question has been percolating for almost 3 years without resolution, I recommend we consider anything incremental that moves us forward even a little and doesn’t block us from a “better” long-term solution. Any number of maxims apply here: “Don’t let the perfect be the enemy of the good,” “Shipping is a feature,” “Move fast and break things,” etc.

How to tag uncertainty with certainty

Reasons for uncertainty

Whether the source of the uncertainty matters needs further discussion. For now, this is a non-exhaustive list of reasons for uncertainty for consideration and thought:

Boundary itself is unknown
Boundary is disputed by on the ground parties, such as those during wartime or even in less contentious dispute resolution processes
Boundary is disputed by historians with multiple opinions or conflicting evidentiary claims
Boundary is irrelevant to those being depicted

Indeterminate geometry considerations

Points: indeterminate points can be grouped together in relations. The options for rendering these in a way that reflects uncertainty needs more thought.
Lines: Boundary edge segments can be of differing levels of confidence. E.g., the 48th parallel is a well-defined boundary (even if the US-Canada border does not follow that line exactly), while older boundaries, such as “the top of a mountain ridge between river X and valley Y” might have been enforced a little more loosely in the medieval period. Or, a boundary can be disputed along a particular subsegment of the entire boundary.
Areas: areas seem to be the most straightforward for identifying disputes.

How to tag is currently unclear

From @1ec5 at #553:

As far as tagging, OpenStreetMap has more or less settled on disputed_by, claimed_by, and to some extent recognized_by, but the values are still up in the air. The proposal was limited to countries with ISO 3166-1 country codes, so it didn’t consider subnational disputes and obviously didn’t consider disputes between historical entities. I’ve been quietly bucking that consensus around country codes by mapping disputes between a city’s neighborhood councils.

OSM data consumers such as OpenMapTiles use these tags to filter boundaries according to an American worldview, Chinese worldview, etc. Is there a view towards doing something similar with OpenHistoricalMap, or is the intention more to display the parties’ names to the user? If the latter, then maybe integration with Wikidata would be less messy in the long run. We should ensure that every disputed territory is linked to a Wikidata item specifically about that dispute and not about tangential topics, as in some of the examples above. Then we could rely on Wikidata statements like territory claimed by (P1336), item disputed by (P7378), statement supported by (P3680), and statement disputed by (P1310) to come up with the list of parties pretty straightforwardly in the inspector.

No practices from OpenStreetMap seem heavily used

OSM Taginfo for:

claimed_by: <700 uses
disputed_by: <1600 uses

Rendering obstacles and solutions

Areas vs. lines

As described above and suggested below, both areas and segments or boundary lines may be disputed and best be rendered differently from undisputed areas and lines.

Depiction at various zoom levels

As discussed by @jeffmeyer in #553, whatever treatments are used do not necessarily need to be the same at all zoom levels:

I’m open to whatever suggestions we have for approaching this, but I’m thinking that it doesn’t need to be apparent for too deep of a zoom, or that maybe there are different treatments at different zooms - e.g.:

browser zoom 4-8 = transparent area highlight

browser zoom 9-12 = some sort of transparent pattern (diagonal lines?)

browser zoom 2-3, 13-20 = nothing

Examples on OHM today

Possible alternative treatments

Considerations:

Not all treatments have to show up at all zooms – e.g., there could be a zoom level dedicated for large-area territorial disputes that emphasizes / highlights the disputes, while the disputed areas are more muted at other zooms.
Not all disputes have to be treated the same way. Maybe there’s a special type to call out disputes like large-area wars.
High contrast to the basemap color is used in mockups, but may be less helpful in production.

Text following a curve implies some area of coverage, even without lines.

Changing boundary styles by segment could reflect varying degrees of confidence - note the dashed segment on the northern side of this boundary.

Boundaries can have fading effects to show indeterminate nature.

Or a varying offset to show min and max estimates of reach

Or a pattern instead of a solid.

Or, a combination of treatments.

Comparables

Related background

jeffmeyer · January 18, 2024, 7:48pm

My biggest question about the Wikidata system is I don’t really see the difference between territory claimed by (P1336) & item disputed by (P7378).

For example, when would they ever be different? And, if n>1 entities claim something, isn’t it by default disputed? I guess both Californians and Americans might claim California, but so might Mexico or Oregon, so how to mark which claimants are disputees? The Venn diagrams here might get interesting.

Minh_Nguyen · April 10, 2024, 5:29pm

There’s a lot in this thread; I’m not sure how much of it is notes versus something you’d like the community to find a solution for. For what it’s worth, I think there’s a limit to how faithfully our renderer map can depict a situation based on OHM data alone. Long-term, we could consider Wikidata as a companion source for non-geographic attributes, as some OSM data consumers do to a limited extent. Or we could decide that our rendered map comes with caveats but nonetheless serves as a starting point for further research.

I’ve encountered indefinite boundaries in a variety of contexts. Some cities are divided into poorly or subjectively defined neighborhoods. Some boundaries are well-defined except for extremely remote, unpopulated areas where the precise boundary is a moot point (except to cartographers). Some boundaries were intended to be well-defined, but the description has inconsistencies or ambiguity that defy a precise representation on a map, for example by referring to obscure metes and bounds. Some boundaries are disputed in areas where it only matters for specific purposes like enforcing fishing licenses, so mapmakers split the difference with an indefinite boundary for a general audience.

Eventually, we may need a different approach for representing each of these situations in the database or the tile schema, so that rendered maps can distinguish them. For now, simple annotations such as indefinite=yes, note=*, description=*, and fixme=* at least allow mappers to communicate these nuances to each other.

For one thing, there can be more than one party to a conflict, each party having conflicting claims and disputing some or all of the other parties’ claims. Some countries take a position on disputes half a world away that don’t directly affect them. There are lots of examples of this complexity in Palestine and the South China Sea. Additionally, the item disputed by (P7378) property applies to more than just claims about ownership of a territory; it can also apply to a place name, an establishment date, the legal status of a jurisdiction, a jurisdiction’s flag, etc.

jeffmeyer · April 10, 2024, 7:06pm

Indeed!

I think my intent was to both summarize thoughts on this and to also invite suggestions for how to handle these issues cartographically, particularly in the use cases identified - primarily official territorial disputes or indeterminate boundaries. Perhaps those should be split into different topics, as those each have their own issues? Regardless, we should have some attempt at guidance for mappers, either here or on the wiki.

My main concern, too, is that we outline some practices for mapping that:

Identify key use cases, some of which are included above
Remove barriers to people mapping - e.g., people not mapping an indeterminate area because it isn’t clear how to do so. Where to tag? Boundary lines? Relations?
Lay the groundwork for eventual cartographic treatment

This is exactly the type of guidance I’m hoping we can turn into recommendations. E.g., for neighborhoods (which I haven’t thought a lot about), do this; or, for territories, do this; for countries invaded, do this; etc.