Proposal talk:Mapping disputed boundaries

Changelog

• Version 1.6
  • Defining terms for "controlled_by" tag to improve verifiability.
• Version 1.5.1
  • Adding role de_facto for boundary relations in Conflict Areas.
• Version 1.5
  • Eliminating Zones of Control as concept.
    • Permitting claimed_by and controlled_by tags to be placed directly on administrative boundary relations, eliminating those (now redundant) Zones of Control
    • Other Zones of Control become Boundary Claim relations.
• Version 1.4.2
  • Changing Crimea example to conform to current administrative boundary.
• Version 1.4.1
  • Changing "all" keyword to a list for the value of the "controlled_by" tag.
  • Adding "UN" as a special value for the "controlled_by" tag.
• Version 1.4
  • Using maritime boundaries instead of land boundaries
  • Eliminating redundant or unneeded relations:
    • De facto relation is eliminated; it is now the same as the existing administrative boundary
    • Minimal boundary is eliminated; it is now a Zone of Control with role "undisputed" in Master Claim
    • Master Claims and Zones of Control are eliminated when not needed, such as for countries with no disputes
    • Conflict Areas are explicitly made optional
  • Roles in Master Claim now differentiate how claimant and zone are related: undisputed, joint, de facto, claimed
  • Describing administered territories
  • Adding how to change the criteria for the List of Claiming Entities
• Version 1.3
  • Possible extensions page added
  • Flattening the hierarchy by removing Disputed and Undisputed Areas
  • Three Boundary Relations: de facto, master, minimal
  • All Zones of Control have the role zone in the three Boundary Relations
  • Eliminating Lines of Control
  • Country code tag introduced
• Version 1.2
  • Removing "according_to" tags
  • Adding Zones of Control and Lines of Control
  • Adding Disputed Areas and Undisputed Areas
  • Using type=land_area + land_area=administrative
  • Full country relations are no longer members of each other.
• Version 1.1
  • Adding "according_to" tag for relations
• Version 1.0
  • Initial proposal.
  • Land-based borders only; no maritime claims.
  • De facto and claimed borders and roles
  • List of Claiming Entities
  • OSM-designated borders
  • Claimed border relation becomes a member of the De Facto border relation, and vice versa

-- Johnparis (talk) 17:27, 9 December 2018 (UTC)

Criteria for the list of claiming entities

The nature of the map changes depending on which entities are permitted to make claims at the national level under the proposal for mapping disputed boundaries.

The following criteria for inclusion in the list of claiming entities are listed from loosest to strictest.

• Criterion 0: any entity that exercises control over territory.

This leads to maps that change daily in war zones. The map would also include, as a "country", entities like the Islamic State or armed rebel groups that control some territory. This criterion raises issues of verifiability.

• Criterion 1: any entity that controls territory and that is recognized by a member of the list.

Among the entities that qualify under this criterion are every country on the List of states with limited recognition except Somaliland. Four entities qualify under this criterion by virtue of recognition by South Ossetia: Artsakh (formerly Nagorno-Karabakh), Donetsk People's Republic, Luhansk People's Republic, and Transnistria. There may be others (more research required).

• Criterion 2: any entity that controls territory and that is recognized by a member of the U.N. General Assembly

This excludes the possibility of a "recognition ring", in which a group of entities recognize one another. This criterion produces the same Claiming Entities as Criterion 1, except that it excludes Transnistria, Artsakh, Donetsk People's Republic and Luhansk People's Republic, along with any others under Criterion 1 discovered by further research.

• Criterion 3: any entity that controls territory and that is recognized by at least two members of the U.N. General Assembly

This excludes Northern Cyprus, which is recognized only by Turkey.

• Criterion 4: any entity that controls territory and that is recognized by at least six members of the U.N. General Assembly

This excludes Abkhazia and South Ossetia. This is the criterion that matches the existing application by the Data Working Group of the Policy on Disputed Territories.

• Criterion 5: any entity that controls territory and that is recognized by at least 18 members of the U.N. General Assembly

This excludes Taiwan.

• Criterion 6: any entity that controls territory and that is recognized by at least 47 members of the U.N. General Assembly

This excludes the Sahrawi Arab Democratic Republic.

• Criterion 7: any entity that controls territory and that is recognized by at least 110 members of the U.N. General Assembly

This excludes Kosovo.

• Criterion 8: any entity that controls territory and that is recognized by at least 138 members of the U.N. General Assembly

This excludes Palestine and is equivalent to "any full member of the United Nations General Assembly."

Other possible criteria include:

• Criterion 0(a): any entity that meets the standard of the Declarative Theory of Statehood. "The state as a person of international law should possess the following qualifications: (a) a permanent population; (b) a defined territory; (c) government; and (d) capacity to enter into relations with the other states."

This is similar to Criterion 0, but does not have a requirement that an entity control territory.

• Criterion 1(a): any entity that controls territory and whose passport is accepted by a member of the U.N. General Assembly.

This is the same as Criterion 1, except that it includes Somaliland and excludes Transnistria and Artsakh.

• Criterion 1(b): any entity that controls territory and whose passport is accepted by a member of the list.

This is the same as Criterion 1, except that it also includes Somaliland, Liberland (a micronation), as well as any other entities whose passports are accepted by Liberland, forming a recognition ring. There may be other such anomalies; further research is needed.

• Criterion 4(a): any entity that controls territory and that has a two-letter code assigned under the ISO 3166 standard

This is the same as Criterion 4, except that it excludes Kosovo and Sahrawi Arab Democratic Republic. This is the stated criterion in the OSM wiki.

Still more criteria can be created by "mixing and matching", or by introducing new concepts. For example:

• Criterion A
  • Point 1: any entity that controls territory and that is recognized by a member of the U.N. General Assembly. (This adds Abkhazia and South Ossetia, by way of several countries, and Northern Cyprus, by way of Turkey.)
  • Point 2: Additionally, any entity recognized by an entity that qualifies under Point 1. (This adds Artsakh, Donetsk People's Republic, Luhansk People's Republic, and Transnistria, all by way of South Ossetia.)
  • Point 3: Additionally, any entity whose passport is accepted by an entity that qualifies under Point 1. (This adds Somaliland, by way of several countries, and Transnistria, by way of South Ossetia.)

By phrasing it this way, Criterion A would exclude a recognition ring. Sorry, Liberland. But that's a debate for another day.

-- Johnparis (talk) 16:43, 2 December 2018 (UTC)

While doing further research, I discovered several territories that have ISO 3166 codes but are linked in one way or another to U.N. General Assembly members. I count 51 of them. All of these have admin_level=2 boundaries, and would keep them under the proposal. However, none of them would be considered Claiming Entities. (I have listed them on the page of the List of Claiming Entities as examples of entities that do not qualify for the list.)

Greenland and the Faroe Islands, for instance, are countries within the Kingdom of Denmark (the other country being Denmark). The DK country code (and associated Master Claim) would represent the Kingdom of Denmark, which is a member of the U.N. General Assembly. For another example, Pitcairn Islands is a Non Self Governing Territory within the British Overseas Territories. It would not be a Claiming Entity and therefore would not have its own country code (despite have a two-letter ISO 3166 code). Again, the Netherlands, Aruba, Curaçao and Sint Maarten form the Kingdom of the Netherlands, a member of the UNGA that would have the country code NL and a Master Claim.

-- Johnparis (talk) 06:39, 16 December 2018 (UTC)

Backward compatibility and transition

Clearly, this proposal must fail if it breaks existing apps, including the OSM Carto map.

As a result, the proposal adopts a specific version of Criterion 4 for the List of Claiming Entities. I believe that no admin_level=2 tags on existing boundaries would change.

As for transition, we would need to tag the segments of the Boundary Claim relations and build those relations, the Master Claims and (optionally) Conflict Areas.

Depending on the sentiment of the community, we could start with a "test bed" area to see how this tagging works, before completing all the Master Claims.

After all the Boundary Claims and Master Claims are built, the new system would be fully in place. Individuals could produce "World According to Me" maps. Renderers would be able to choose between using the Administrative Boundary and the Master Claim, or some other combination of Boundary Claims.

At that point, taggers could create new Boundary Claim relations for existing boundaries that are doubtful or marked FIXME. Generally speaking, these are maritime areas. For instance, the waters off Ceuta have, in the past several months, been marked as under the de facto control of Morocco (current rendering) or the de facto control of Spain (previous rendering). In reality, neither exercises continuous and exclusive control, according to the news reports I have read. So that area would be tagged claimed_by=ES;MA + controlled_by=nobody.

At some point, the OSM community might decide to change the criteria used for the List of Claiming Entities. That should prove to be an interesting discussion.

One overall advantage of the proposal, to use Kosovo as an example, is that Serbia could use its claimed boundary in any maps, thus showing Kosovo as a province, while Kosovo could use its own claim, showing itself as a country.

Taggers would then be able to apply objective criteria, as opposed to the current system, whose criteria are vague and not objectively measurable. For example:

• "OpenStreetMap contributor opinion". This is impossible to assess, as there is no mechanism to put issues to a vote of all contributors, or even to conduct a scientific opinion survey of a randomly selected subset.
• "most widely internationally recognised" borders. This is false in the cases of Taiwan and Sahrawi Arab Democratic Republic, which are recognized only by a minority of the 193 members of the U.N. General Assembly.
• "best meets realities on the ground". This is false in all the cases that would be approved under Criteria 0-3.

-- Johnparis (talk) 17:56, 5 December 2018 (UTC) -- revised 17:23, 9 December 2018 (UTC)

Non-verifiability of territorial claims

Want to start by saying that my views on verifiability are probably well known. This comment is not about me pushing my opinion but to bring up the topic for discussion not to be ignored when preparing this proposal.

It is fairly clear that a territorial claim that does neither represent de facto control nor references a certain de facto administrative unit (like Ukraine claiming Crimera in the same boundaries with which it is currently de facto administered by Russia) is non-verifiable in the sense of the verifiability principle because by definition a territorial claim is the claim of a single authority and as such by definition cannot be verified independently.

It would not be the first no verifiable mapping concept in OSM obviously but if a proposal for mapping such claims was put up to a vote and approved it would to my knowledge be the first such case in OSM. --Imagico (talk) 12:28, 27 November 2018 (UTC)

To me, "The border of Ukraine according to the Ukrainian government" (and "The border of Russia according to the Ukrainian government") are unambiguous (one includes Crimea, the other doesn't). They are verifiable by looking at an official map from that government, or by reading a textual description from the Ukrainian government. Any other OSMer could come to the same conclusion. Is this not the heart of verifiability? --Rorym (talk) 15:59, 27 November 2018 (UTC)

Verifiability in OSM means other mappers will come to the same conclusion based on independent observations of the geographic reality. This is not the case with a claim that is defined through what a single authority says. This is the core of OSM verifiability in contrast to Wikipedia verifiability (which is based on reliable sources^tm).

I don't want to distract from the important discussion if and how OSM can represent territorial claims but i want everyone considering this to be aware that this is a departure from the idea of mapping a single, verifiable and inherently consistent geographic reality where statements can not be true and false at the same time.

It seems to me what John is trying to do here is trying to define a small part of the political domain tightly enough that it can be integrated into the observable geographic reality OSM is based on without major consistency issues. I am not sure this is possible.

--Imagico (talk) 16:46, 27 November 2018 (UTC)

But isn't that the same in many cases? The name of a shop can be changed by a single entity (the shop owner). The ref=* or official_name=* of a road is changed and defined by a single governmental entity. In Welcome to Wikipedia users, it says "But there's one related rule that is similar in OSM and Wikipedia: NPOV, aka Neutral Point Of View. In fact, we go further: No Point Of View. We only record facts, not opinions, even measured ones. OSM isn't the place for subjective quality judgments". To me, it's the absence of any subjectivity which makes something verifiable, rather than number of different sources (along with the on the ground rule) Rorym (talk) 08:22, 28 November 2018 (UTC)

And what is the absence of any subjectivity if not the independent, intersubjective verifiability? The name of a shop is verifiable if it is either physically manifested in the form of a sign or if is common knowledge among the locals (i.e. you can ask any local and they will give you consistently the name of the shop).

Again: I don't want to turn this into a discussion on verifiability in general, i just want the designers of this proposal to make a conscious choice about how it is supposed to relate to the verifiability principle.--Imagico (talk) 11:34, 28 November 2018 (UTC)

In my opinion, ground truth has little to no relevance to national borders. We cannot see administrative borders as a physical object (some national borders are not demarcated, those that are have non-zero thickness, and internal administrative borders within countries are predominantly defined in documents, not on the ground). What I mean is, demarcation and border itself are separate, though related, entities. Demarcation objects (border signs, barbed wire fences, walls) are physical objects that are covered with "Ground truth" rule and can be verified, border itself is more of a logical, virtual entity. Thus, we should separately verify physical objects and physical control (which are covered by "Ground truth") on one hand, and international recognition (which considers borders as logical entities and can be verified exclusively through documents, preferably top jurisdiction, like UN resolutions) on the other. I believe that since these two entities are different, they should have different sets of tags. We cannot compare international recognition with de-facto control (and surely cannot combine them in one tag as the current "DisputedTerritories" policy suggests) but can show them simultaneously as different properties of the same object, thus achieving at least first normal form and ensuring data integrity. Splitting a tag (property) into two tags (properties) and validating them independently is one of the typical approaches for such cases in theory of relational DBs. -- Kilkenni (talk) 23:13, 27 November 2018 (UTC)

I understand your point of view, so what tagging scheme would you suggest? The existing scheme (boundary=*) already indicates these notional, rather than physical, borders.

Or, by contrast, are you OK with the existing tagging scheme, but want to separate the ways representing, say, the centerline of a river, from the ways representing the national border? If that's the case, I wholeheartedly agree!

-- Johnparis (talk) 09:37, 28 November 2018 (UTC)

I believe we should use the "main" tagging scheme (currently used) for international recognition, and additional entities (relations, tags) for describing who currently controls the territory (if it is contested), and also actively encourage users of OSM (primarily companies) to support rendering of both tags to avoid confusion and lawsuits.

And yeah, I prefer to have borders as separate lines, not reusing lines of rivers or forest edges or such. It makes the database more transparent data-wise and doesn't lead to changing borders if all you need is to slightly shift the river bed or a piece of wood. Kilkenni (talk) 17:46, 28 November 2018 (UTC)

Can you give a good definition of "international recognition"? Often the phrase it used to mean "USA+allies". Rorym (talk) 18:01, 28 November 2018 (UTC)

When I say "international", I think United Nations and ISO. This approach has its issues, but I fear it is the best one we have. OSM already uses ISO 3166 to define what is a country, and ISO 3166 depends on UN recognition. If we already use this approach for countries, why can't we use it for national borders?

They way I see it, if we want to help popularize OSM and maintain independency, adhering to international laws and regulations is a pretty safe approach. Kilkenni (talk) 15:41, 1 December 2018 (UTC)

OSM claims to use ISO 3166 but in fact does not (I just edited the wiki to reflect this). Kosovo and Sahrawi Arab Democratic Republic are mapped at Level 2 but do not have codes assigned by ISO.

Also, I see a lot of assumptions here about "international recognition". For the most part, countries do not specifically endorse territorial claims by other countries. The best you can do are vague resolutions supporting either "territorial integrity" or "right of self-determination", which are usually code words for the opposite sides of the debate. You will almost never see someone saying "I support Country XX in its claim." Also, "recognition" refers to extending diplomatic relations, not acceptance of territorial claims. And forget about using UN maps. They always explicitly say they are NOT definitive when it comes to borders, and precisely for these reasons. -- Johnparis (talk) 16:52, 2 December 2018 (UTC)

Side note: ISO codes do not depend on UN recognition (think Taiwan, code TW, for example). If you want to restrict the meaning of "country" to members of the UN General Assembly, you're talking Criterion 8 above.

Clarification: ISO codes do depend in some sense on the UN (https://en.wikipedia.org/wiki/ISO_3166-1#Criteria_for_inclusion) but not on formal UN recognition

A country can be verified on the ground. Who has the ability to arrest you? If you don't pay taxes to a government and they can arrest you and put you in prison, then that's the country you're in. This is the 'de facto control' reasoning. Rorym (talk) 13:32, 28 November 2018 (UTC)

If we use your logic, ISIS and Taliban (the latter currently controls about 60% of Afghanistan territory) are countries and should be mapped. Some regions of the world are controlled by gangs (like Somalian pirates). Should these be mapped as countries, too? Kilkenni (talk) 17:46, 28 November 2018 (UTC)

Good point. Maybe we should map ISIS and/or Taliban areas. They are essentially engaged in horrible wars. We already map horrible regimes like Saudi Arabia. I don't know a lot about the current Afghanistan/Syria situation. Rorym (talk) 17:59, 28 November 2018 (UTC)

This raises huge issues of verifiability in my mind. The boundaries of Islamic State vary day to day. And micronations have already been mentioned. I will open a discussion of the criteria for the List of Claiming Entities if this proposal is approved. -- Johnparis (talk) 16:52, 2 December 2018 (UTC)

Examples

Just a quick note to say thanks for adding the examples on the dev server. I realise that they're still a work in progress, but they're a great help to visualise what you're proposing here. SomeoneElse (talk) 11:28, 1 December 2018 (UTC)

(for info for anyone else trying to use these on the development server) the "query" option in the OSM UI uses live data, but if you've got an account on the dev server (just sign up like you did the live server) drop into edit mode to query the data. See e.g. https://master.apis.dev.openstreetmap.org/way/4305291947 for a constituent way of the former Western Sahara and see what relations it belongs to). SomeoneElse (talk) 14:52, 3 December 2018 (UTC)

Including recognition in the main proposal

I think the most important aspect of any dispute is how to draw it, and this cannot be done unless we also account for the target audience. Thus, we must include which countries recognize whose claims. I think this must be part of this proposal instead of being relegated to a proposal extension. If we leave it for later, we wouldn't know if this proposal is a good fit for describing point-of-view, making it cumbersome for everyone. I do agree that other aspects like maritime borders and sub-national disputes could be left until later day. --Yurik (talk) 22:40, 3 December 2018 (UTC)

Thanks for that comment. On the extensions page, I explain why I removed this from the proposal. But to look at it from another angle, OSM records facts, not opinions. Land claims are facts. Points of view about those claims are opinions. A rendering of "the world according to Bhutan" (for example) would be translating its viewpoint into a map.

I suppose if the opinions were fixed and verifiable, I could reconsider. But in the cases I looked at, this wasn't the case.

The proposal does offer concrete relations for a country from its own point of view (the Master Claim), which is fixed and verifiable. So in the case of Western Sahara, for instance, there will be a Master Claim relation for SADR whose extent matches the Western Sahara borders. And there will be a Master Claim relation for Morocco that includes all of Western Sahara and an unbroken continuation into the rest of the country. Neither of these represents de facto borders.

I am interested in opinions. If the overwhelming opinion of others is that they should be included, I'll reconsider. And of course people are free to use any tags they like, on relations they create. -- Johnparis (talk) 22:55, 3 December 2018 (UTC)

Johnparis, recognition of country's territorial claims are also facts, just as country's own territorial claims - you can easily check if France recognized Crimea as part of Russia or as part of Ukraine. Going solely by the "controlled-by" fact is what has gotten OSM community into this mess in the first place -- that's why if we don't include this information from the beginning, the value of the proposal is greatly diminished. In some cases a country might not have published their opinion, in which case we should consider it siding with the majority/UN, until the country speaks up or we find some concrete evidence of their official position (e.g. a statement published by their foreign affairs office). --Yurik (talk) 23:07, 3 December 2018 (UTC)

Well, you certainly couldn't assign a POV to a country based on the "majority". Countries without a stated POV (even in my original proposal) would need to be omitted from the list. There are four possible values in each two-way dispute: support #1, support #2, neutral, and unknown. -- Johnparis (talk) 23:31, 3 December 2018 (UTC)

To return to your original point, I think this proposal is definitely a good fit for describing point-of-view, and provides the algorithm for doing just that. You provide me a database of your country's (or your butcher's) opinions on all disputes, and I will generate you a map for that. It's in the proposal. The problem with the present system is that EVEN IF you presented me with the database I describe, I could not generate a map of the world from the point of view of your country (or your butcher). -- Johnparis (talk) 23:40, 3 December 2018 (UTC)

Yuri and I had a discussion on the OSM US Slack channel #borders (which I have added to the resources in the proposal). While I am not persuaded to integrate this into the main proposal, I offer some ideas on how such an extension could be implemented on the possible extensions page. -- Johnparis (talk) 07:57, 4 December 2018 (UTC)

Special cases: Kashmir and Israel-Palestine

Upon reflection, I believe my original suggested tagging did not conform to the proposal. That is one of the nice things about it: because the standards are objective, a particular tagging is verifiable.

-- Johnparis (talk) 07:44, 4 December 2018 (UTC)

General comment

Hi, there are mappers who take no sides to either NATO or Russia in the Crimea dispute, among them being me (see https://www.openstreetmap.org/user/ika-chan!/diary/47095), and I am hopeful that there will be a technical solution to make it easier for data users to use the data as they see fit.

As for OSM-carto, the stylesheet should be truly neutral, showing all disputed borders as a dotted line version of the relevant administrative level. The sooner a this proposal is adopted, the sooner that can be reality. --Ika-chan! (talk) 16:28, 4 December 2018 (UTC)

I agree here on neutrality. While I do have a side in this issue (I wouldn't say "dispute" or "conflict" because we have an obvious aggressor here), I support the notion of OSM neutrality. Technical solution and a good and flexible tagging scheme is an important part of resolving the problem. Rendering disputed borders in a particular way sounds sensible enough.

(on a side note, not sure why you mentioned NATO - afaik it is a military alliance, not a political entity. This is more of a "Russia vs the World" case. But that's irrelevant to tagging) Kilkenni (talk) 21:01, 7 December 2018 (UTC)

Note that OSM-carto development is not obligated to implement/accept PR solely because change was voted on OSM Wiki Mateusz Konieczny (talk) 20:16, 9 December 2018 (UTC)

Software support?

If I'm reading this proposal correctly, no current software supports this scheme. Do you have any ideas of how to start with that, and how challenging that would be? Rorym (talk) 16:46, 4 December 2018 (UTC)

Thanks for that question, Rory. I created the examples using JOSM, so that software supports it, as does iD. The tags are fully backwards compatible (see above), so no existing software should break.

One problem I have run across, however, is differentiating between the different boundary types, which some software does using the "Name" field or the "Type" field, which for these boundaries will be the same. JOSM, for instance, has a special rule (apparently) when the boundary=administrative, to make those easier to spot. I could work around that fairly easily:

* type=boundary + boundary=master becomes type=master_boundary

Upon reading about osm2pgsql imports, though, I think type=boundary is probably necessary.

In terms of rendering, there will be no immediate change. All the renderers I see support any type of relation, so the new Master Claim boundary type should render right away. -- Johnparis (talk) 18:31, 5 December 2018 (UTC) -- revised 17:23, 9 December 2018 (UTC)

This is just a tagging scheme, of course JOSM supports arbitrary tags.

-- previous message posted unsigned by Rorym (talk)

In what way JOSM does "support it"? Does it allow you to merge relations and select based on them without actually changing the underlying data? SomeoneElse (talk) 10:25, 8 December 2018 (UTC)

Seeing as how I used JOSM to create all the examples, it "supports" these relations in the sense that you can perform any of the normal JOSM functions on them. It's pretty simple, for instance, to create a new relation with the boundary claim relations and their members, then remove any internal borders as needed. Removal of the borders can be done manually or based on tags (previously I used the "line_of_control=yes/no" tag for that purpose; it might be useful to restore it?). But other GIS software can also trivially handle these problems (GRASS GIS has the dissolve function, for instance). -- Johnparis (talk) 14:22, 8 December 2018 (UTC) updated Johnparis (talk) 07:10, 15 December 2018 (UTC)

Key-value for tags

Here's an explanation for why I chose certain key names and values. I'm open to altering any or all.

Ways

• boundary=claim -- on a segment of a Boundary Claim relation that does not qualify as boundary=administrative.

Some of these ways are already tagged, sometimes with boundary=disputed. I chose boundary=claim because all the ways involved are claimed by someone; not all are controlled and not all are disputed. The ones that already have a boundary=* tag with a value other than "administrative" would need to be modified one at a time.

Relations

• claimed_by=* (one or more country codes, or nobody) -- on a boundary relation.

Seems pretty straightforward. It was suggested to me that it might be better to avoid the semicolon-separated list, but from a data processing perspective I see little difference between the proposal vs. saying claimed_by:1=*, claimed_by:2=*. As far as I know, there are no areas claimed by three or more parties, but theoretically you would need claimed_by:3=* in that case. (Historically, the Trans-Karkoram Tract was claimed by three parties until 1963.)

• claim_level=* -- on a boundary relation where all claimants agree on the claim level.
  • claim_level:claimant_country_code=* -- on a boundary relation where the claimants assign different claim levels

I chose claim_level to differentiate it from admin_level. The subtag is necessary when the claimants disagree on the claim level.

• controlled_by=* (one or more country codes and/or UN, or nobody) -- on a boundary relation.

Straightforward.

• country_code=* -- on the Master Claim to indicate the country code

In a previous version of the proposal, I did not have this tag, instead putting both claimed_by and controlled_by on a master claim. So this tag is intended to simplify data processing and is available even for Claiming Entities that do not have ISO codes (currently two of them).

• boundary=master -- on the Master Claim.
• boundary=claim -- on a Boundary Claim relation
• boundary=conflict_area -- on an optional Conflict Area relation

As I mentioned in the thread above ("Software support?"), using this method (type=boundary) can lead to difficulties in using JOSM or iD, because the only boundary type they (currently) distinguish is boundary=administrative. But as I also note above, upon reading about osm2pgsql imports, I think type=boundary is probably necessary.

Note: If I count correctly (using Overpass Turbo), there are three relations and 165 ways tagged boundary=claim. All of these are covered under this proposal, so there is no need to create a new key-value pair for them.

-- Johnparis (talk) 09:54, 6 December 2018 (UTC) -- updated Johnparis (talk) 07:10, 15 December 2018 (UTC)

Comparison with other proposal

SomeoneElse asked for a brief comparison between RoryM's proposal and mine. I posted my commentary on Rory's proposal, and SomeoneElse replied with some comments on the two proposals, but also some specifically directed at this one, regarding the relations, so I am reposting those comments here.

[snip]

The biggest issue that we have with claims at the moment is not being able to show the claim of the state making the claim, not third parties. Mappers anywhere typically aren't creating a map of the world; they're creating a map of one area according to one world view. If a mapper from Bhutan wants to make a map of Europe according to Spain's view, then using this proposal they'd simply need to select certain relations over others (easy to do while in lua before rendering, to take one use case example). Under your proposal if they were to "simply to change the Bhutan tag on the France-Spain Conflict Area" it doesn't get them any closer to being able to do anything useful with the data - they'd need to somehow munge together the particular relations that they were interested in to create larger relations (perhaps using something a bit like the area/multipolygon processing in osm2pgsql, but as yet unwritten?) to get the data they want OR they could use the first party claims (which are also present in your proposal and just ignore the "conflict area" relations altogether).

Based upon what gets looked after in OSM and what doesn't, I suspect that your extra conflict area relations simply won't get maintained in OSM. I'm not saying that it's not a good idea to say what tags we should use for them, but I am saying that most people making maps according to a certain viewpoint won't use them, and they'll tend to get broken easily which will reinforce people not using them. SomeoneElse (talk) 10:17, 8 December 2018 (UTC)

Thanks for these comments, Andy, which I will respond to one at a time.

You say (and I agree): the biggest issue that we have with claims at the moment is not being able to show the claim of the state making the claim, not third parties.

This is one reason that I made the third-party claims an extension to my proposal. They are central to Rory's. My proposal does include a country's self-view.

You say: Mappers anywhere typically aren't creating a map of the world; they're creating a map of one area according to one world view. If a mapper from Bhutan wants to make a map of Europe according to Spain's view, then using this proposal they'd simply need to select certain relations over others (easy to do while in lua before rendering, to take one use case example).

But why would they choose Spain's view of Europe, rather than France's? And they're not the same. If they agree with France in the France-Spain dispute, sure, they'd pick France's, but what if they agree with France in their dispute with Spain, but Italy in their dispute with France? Nearly impossible with pre-built relations (as in Rory's proposal). For Southern Europe, you'd need to pick the "Spain minus France dispute" version of Spain, the "France plus Spain dispute minus Italy dispute" version of France, and the "Italy plus France dispute" version of Italy. With my proposal, you simply apply your own POV onto the existing tags for the boundary claim relations. If you're doing a map of Europe, do it with all the European boundary claim relations. Done and dusted.

You say: Under your proposal if they were to "simply to change the Bhutan tag on the France-Spain Conflict Area" it doesn't get them any closer to being able to do anything useful with the data - they'd need to somehow munge together the particular relations that they were interested in to create larger relations (perhaps using something a bit like the area/multipolygon processing in osm2pgsql, but as yet unwritten?) to get the data they want OR they could use the first party claims (which are also present in your proposal and just ignore the "conflict area" relations altogether).

I misspoke. That's part of the extension, which I was comparing with Rory's proposal. I am not proposing adding third-party views. But if someone has a database of their own views, then yes they would need to use something like osm2pgsql to generate their custom maps. And I provide pseudocode for the (very simple to implement) algorithms. As you note, the first-party claims are in my proposal. Spain from the POV of Spain is the Master Claim in my proposal. Also a lot of people will want to use the undisputed boundary; I frequently see Morocco, for one, and India, for another, mapped that way. But in fact under Rory's proposal you have more relations, possibly a lot more.

You say: Based upon what gets looked after in OSM and what doesn't, I suspect that your extra conflict area relations simply won't get maintained in OSM. I'm not saying that it's not a good idea to say what tags we should use for them, but I am saying that most people making maps according to a certain viewpoint won't use them, and they'll tend to get broken easily which will reinforce people not using them.

My gut feeling is that you're right. I almost wrote "Optional" for the Conflict Areas, because they are not essential to the proposal. They are useful for people with a neutral POV -- the United Nations, for instance, refers exclusively to Western Sahara, which "exists" only as a conflict area (that is, the sum of two particular boundary claim areas). Without the conflict area relation, you have the same situation we have today -- there is a Western Sahara relation, but it's not linked to anything. And under the proposal, without a conflict area, you'd have two adjacent boundary claims, but no name to display. I suspect that because of the contentious nature of Conflict Areas, any time one gets touched the Data Working Group will hear about it pretty fast. :) Not to mention that Wambacher (talk), or his tools, will probably lead to a revert of any changes even before that. There's even some guy named SomeoneElse_Revert who seems to have a habit of restoring such changes to the status quo ante. ;)

-- Johnparis (talk) 13:06, 8 December 2018 (UTC) updated Johnparis (talk) 07:10, 15 December 2018 (UTC)

Standard tile layer rendering

I wish to ask where the Standard tile layer currently gets its boundary data? Because if it renders boundaries from ways and not relations, we may be able to propose an interim measure that allows the Standard tile layer to show disputed international boundaries as dotted lines. --Ika-chan! (talk) 17:14, 9 December 2018 (UTC)

The osm-carto style, it seems to use Natural Earth Data (data source) for low numbered zooms, and boundary lines when you zoom in. However, I'm pretty sure I've seen the purple admin border lines in places where I'm sure the the admin area is tagged as a relation.

Regardless, this is where you suggest improvements to the osm-carto style, perhaps start there?

Rorym (talk) 17:50, 9 December 2018 (UTC)

Relations are, obviously, supported - see for example https://www.openstreetmap.org/way/339430371#map=19/50.08213/19.91569 Mateusz Konieczny (talk) 20:19, 9 December 2018 (UTC)

"all" and "UN" keywords for "controlled_by" tag

I have eliminated the special keyword "all" for the "controlled_by" tag and replaced it by a list. This simplifies the algorithms and also takes into account a special case where a territory is jointly controlled by two Claiming Entities but is claimed only by one. (That is, one Claiming Entity jointly controls the area but does not claim it.) I believe this might be useful in describing Area B of the West Bank.

I have also added "UN" as a special keyword for the "controlled_by" tag. While the UN does not claim territory, it can be considered as administering it. An example would be a buffer zone under the aegis of a UN peacekeeping force. -- Johnparis (talk) 15:02, 11 December 2018 (UTC)

Former Zones of Control

Private feedback led me to eliminate the Zones of Control. For purposes of the algorithms, I believe the claimed_by=* tag works as a proxy. If I'm wrong, we could introduce a zone_of_control=yes/no tag on the relations. -- Johnparis (talk) 08:07, 15 December 2018 (UTC)

Conflict areas

"The Zones of Control should be added as members with no role."

Why not using a role like zone_of_control? It makes clear what it is used for.

--Nospam2005 (talk) 21:15, 15 December 2018 (UTC)
(comment moved here from Possible Rendering page by Johnparis (talk) 06:09, 16 December 2018 (UTC))

I have no problem with that, though the Zones of Control are now referred to as boundary relations in version 1.5. To be consistent with other relations, it seems to me the role name would be de_facto. I have updated the proposal and examples using that. Thanks. -- Johnparis (talk) 06:13, 16 December 2018 (UTC)

The controlled_by tag

If the proposal is approved, no changes need to be made in any of the existing tags right away.

I'm suggesting specific time periods (the Resolution Period) in order to improve verifiability. There is little doubt in my mind that in times of war, the "fog of war" literally prevents verifiability. And even after the dust settles, it's possible that the next day -- or the next hour -- will bring a new geographic reality. So the proposal now offers specific waiting times before making a change in the tag.

This is appropriate, as OSM is a map, not a news service. At the same time, OSM is not an archive. Waiting more than four years to update a "frozen conflict" is, I think, too long. However, perhaps two years is too short. I'd like to hear discussion about the concept of the Resolution Period and the specific criteria for it.

Here are two examples of how I would have proposed tagging certain conflicts over time. My tagging might not be correct; the accuracy of the tags can be compared objectively to the criteria established in the proposal. These case studies are part of the discussion, not part of the proposal itself.

Case 1: Waters off Ceuta

Morocco and Spain dispute the Mediterranean waters off the coast of Ceuta. Moroccan fishermen in these waters are routinely arrested by the Spanish Coast Guard. Spanish fishermen in the same waters are routinely arrested by the Moroccan Coast Guard. Because the answer to the question, "Who will arrest you?" is "It depends," it is fair to say that nobody has exclusive control of this territory. The tags are claimed_by=ES;MA + controlled_by=nobody.

Case 2: Crimea

Here is a rough timeline as I understand it. Please feel free to correct any misimpressions, and I will update the timeline.

January 2014: Crimea is under Ukrainian control. Tags are claimed_by=UA + controlled_by=UA.

26 February 2014: Russia occupies main road to Sevastopol and establishes checkpoints. At this point, armed conflict has begun, but these are not considered "sustained hostilities" until 30 days have passed. (If Ukraine had successfully re-established control within 30 days, the tag would have remained controlled_by=UA continuously.)

2 March 2014: Unidentified forces take control of Crimea. Russia says these are "self-defense" Crimean forces, not Russian.

4 March 2014: Most armed conflict has stopped.

16 March 2014: Plebiscite held.

21 March 2014: Russia approves annexation of Crimea, retroactive to 18 March. This is Russia's first formal claim of the territory. Because fighting began less than 30 days earlier, territory is still considered under Ukrainian control. Tag is claimed_by=RU;UA + controlled_by=UA.

24 March 2014: Ukraine orders withdrawal of its forces from Crimea. Sustained hostilities have stopped. Resolution Period will begin in 30 days unless hostilities resume.

28 March 2014: Since conflict has been under way for 30 days, tags are claimed_by=RU;UA + controlled_by=nobody.

23 April 2014: Because sustained hostilities have been over for 30 days, the Resolution Period begins, retroactive to 24 March. If at any time, Ukraine formally accuses Russia of controlling Crimea, the Resolution Period ends (under the "mutual agreement" criterion). Otherwise, it ends on 24 March 2016 (under the "frozen conflict" criterion). Tags remain claimed_by=RU;UA + controlled_by=nobody during the Resolution Period.

Note that Ukraine sponsored two U.N. resolutions in March 2014, one before the Security Council (S/2014/189) and one before the General Assembly (A/RES/68/262). Neither resolution, however, formally named Russia. It is possible that Ukraine made other formal statements accusing Russia of controlling Crimea, and such a statement would bring the Resolution Period to an end, because both sides would agree that Russia controls the territory. Without such agreement, the Resolution Period would end on 24 March 2016 under the Frozen Conflict criterion. After the Resolution Period, tags are claimed_by=RU;UA + controlled_by=RU.

These tags trail the course of events somewhat, as intended, because of the Resolution Period.

-- Johnparis (talk) 07:59, 2 January 2019 (UTC)

Other cases

The following was posted by Phake Nick <c933103@gmail.com> on the tagging mailing list.

I think there are some cases that might not be sufficiently covered by the current proposal and it might be a good idea to explain how they can be tagged in example section of the proposal if they can be represented by it:

• Minamitorishima, where it is undoubtably a Japanese natural feature, however there are dispute on the nature of the island, which affect whether Japan is able to enjoy 200nm EEZ from the feature.
• Southern Sakhalin and Northern/Central Kuril Islands, where it is de facto controlled by Russia, and Japan have already renounced their right there, however Japanese government insist the ownership of these territories are not determined yet.
• Sub-national disputed boundaries, for instance the recent city-level dispute between Hong Kong and Shenzhen over the Sha Tau Kok River
• Different active level of claims for different parties, for instance Republic of China (Taiwan) still haven't renounced their claim on part of Russian and Myanmar territories, yet it doesn't seems right to list them as a party in territorial dispute between China (Mainland) and other surrounding countries on the same level as PRC itself
• Other different types of claim, for instance the 9-dotted lines which China claims "historical right" within the line
• The proposal supported by various governments around the world to turn Jerusalem into a corpus separatum
• Dispute between a national government and a sub-national entity, for example dispute between Somaliland and Puntland, where according to my understanding Somaliland is an unrecognized country while Puntland is an autonomous regional government that is intended to be part of Somalia.
• Dispute between regional government and their national government, for instance disputed in area for Kurdish autonomous region in Afghanistan
• Some special situation about United States - should Wake Island be controlled by US or UM (US Minority outlying islands)?
• Guantanamo Bay, where the controlling country (or force) doesn't claim the area but continues to control it anyway

It would be nice if the proposal can be extended to cover them.

Also, among the existing list of example, for Shebaa Farms, the claimed_by=* should also include Syria. For Israel-Palestine dispute, it should also separately list out Area A/B/C for West Bank as each of them have different status.

I forgot to mention a few other possible cases of disputed border, for one of them I would use the historical dispute of Sikkim's integration into India as an example, where most countries including India recognized the integration of Sikkin into India and there are also no independent government entity for Sikkim exists after such integration, however countries like China continues to claim there should be an independent country for Sikkim

Another case is that, how about an government in exile like historically for various European countries during WWII, especially when they have different claims on territory from existing government of the country, like Free France vs Vichy France? (in both the situation when such government in exile control some overseas territory and also in situation when such government does not control any territory)

-- Phake Nick <c933103@gmail.com> 13:24, 4 January 2019 (UTC)

Please note a couple of things:

1) This proposal is meant to cover only those things it states as its intention to cover. For instance, it does not cover EEZ boundaries. I have offered a couple of suggestions already on possible extensions, and if you think others are warranted, please feel free to propose any.

2) I am not an expert. The proposal is meant to provide objective standards for someone on the ground who can evaluate his or her observations and tag the reality correctly. If I offer an opinion on how to tag something, it is based only on my limited knowledge; there are not a lot of these places that I have personally visited.

Now, as best as I can, point by point ...

• Minamitorishima.

https://www.openstreetmap.org/relation/8004038 From what I can tell, it is clearly claimed by and controlled by Japan. Whether it is properly tagged as an island, an islet, or whatever, is not part of this proposal. Nor is this a "law of the sea" proposal. It extends only to territorial waters.

• Southern Sakhalin and Northern/Central Kuril Islands, where it is de facto controlled by Russia, and Japan have already renounced their right there, however Japanese government insist the ownership of these territories are not determined yet.

If Japan has formally renounced its claim, then you cannot tag it claimed_by=JP. It would have to be claimed and controlled by Russia. "Insisting ownership is not determined" is not the same thing as making a formal claim. It could, however, possibly considered as a third-party point of view under RoryM's proposal.

• Sub-national disputed boundaries, for instance the recent city-level dispute between Hong Kong and Shenzhen over the Sha Tau Kok River

Subnational claims are specifically not included in this proposal.

• Different active level of claims for different parties, for instance Republic of China (Taiwan) still haven't renounced their claim on part of Russian and Myanmar territories, yet it doesn't seems right to list them as a party in territorial dispute between China (Mainland) and other surrounding countries on the same level as PRC itself

If a claim has been formally made and has not been renounced, then it is "active". There are no "levels" of activity in this proposal.

• Other different types of claim, for instance the 9-dotted lines which China claims "historical right" within the line

I am not familiar with this. However, according to Wikipedia, "Despite having made the vague claim public in 1947, China has not (as of 2018) filed a formal and specifically defined claim to the area within the dashes." Therefore, no claim.

• The proposal supported by various governments around the world to turn Jerusalem into a corpus separatum

No Claiming Entity has made a formal claim along these lines. I think (I'm not sure) that this could be considered under Rory's proposal for third-party points of view.

• Dispute between a national government and a sub-national entity, for example dispute between Somaliland and Puntland, where according to my understanding Somaliland is an unrecognized country while Puntland is an autonomous regional government that is intended to be part of Somalia.

Somaliland is not a Claiming Entity under this proposal; neither is Puntland. Somalia is a Claiming Entity. If parts of Somalia are not under its control, these areas would be tagged controlled_by=nobody.

• Dispute between regional government and their national government, for instance disputed in area for Kurdish autonomous region in Afghanistan

Again, subnational disputes are not covered by this proposal.

• Some special situation about United States - should Wake Island be controlled by US or UM (US Minority outlying islands)?

UM is not a Claiming Entity. The United States is. Therefore, controlled_by=US.

• Guantanamo Bay, where the controlling country (or force) doesn't claim the area but continues to control it anyway

This is explained under Special Case: Administered Territory.

• Sikkim's integration into India as an example, where most countries including India recognized the integration of Sikkim into India and there are also no independent government entity for Sikkim exists after such integration, however countries like China continues to claim there should be an independent country for Sikkim

As with Jerusalem and Sakhalin, it is irrelevant what other countries think should happen. That is (possibly) covered under Rory's proposal. The only question is whether another Claiming Entity claims Sikkim. If that were the case, then Sikkim should be carved out separately and given its own set of tags. (I do not believe that is the case, however.)

• Another case is that, how about an government in exile like historically for various European countries during WWII, especially when they have different claims on territory from existing government of the country, like Free France vs Vichy France? (in both the situation when such government in exile control some overseas territory and also in situation when such government does not control any territory)

No governments in exile qualify as Claiming Entities (unless you count Taiwan). All claims by Claiming Entities would be tagged as needed. No other claims qualify for the claimed_by tag under this proposal.

Regarding Shebaa Farms, I have updated the example, thanks. Regarding Area A/B/C, I specifically note the tags needed to conform to the existing OSM structure. I agree that it would be a good idea to change those tags, assuming the proposal is approved, and the proposal provides clear guidance on those areas.

-- Johnparis (talk) 16:53, 4 January 2019 (UTC)

I think they sort of conform with existing OSM structure, see: https://www.openstreetmap.org/relation/3791783 https://www.openstreetmap.org/relation/3791784 https://www.openstreetmap.org/relation/3791785 . What would be the controlled_by for Area B, given the "Palestinian civil control and joint Israeli-Palestinian security control" description? Also, Is it correct to say that Kowloon Walled City would have been tagged as controlled_by=nobody according to this proposal?

Also, a correction: What I said as Minamitorishima in my previous message should actually be Okinotorishima. And then, if EEZ is not intended to be covered by the proposal, then how about some artificial islands on South China Sea? Artificial islands should not have their own territorial water per convention but there are some countries that claim otherwise and claim their own territorial water for those artificial islands while disputed by others. C933103 (talk) 17:51, 4 January 2019 (UTC)

https://www.openstreetmap.org/relation/3791784 (and the other two you cite) are not admin_level=2 relations. The correct current relations are 1473946 and 1703814.

I am not proposing specific future tags at this point, but Area B would, I presume, be controlled_by=IL;PS, because they have mutually agreed to jointly administer the area. (See "Special case: Cooperative jurisdictions".)

I am not familiar with Kowloon Walled City, but if it is not controlled exclusively and durably by a Claiming Entity, then yes, controlled_by=nobody is correct. By contrast, if it is controlled exclusively and durably by a Claiming Entity, then the controlled_by tag would get the country code of the entity. (I have just looked it up, and it appears to be a park in Hong Kong. Historically, if it was in fact not under durable and exclusive control by a Claiming Entity for more than 30 days, then it would indeed have been controlled_by=nobody.)

EEZ is not covered by this proposal. I do not intend to extend this proposal to other maritime boundaries. Others may make their own proposals. If any island (artificial or otherwise) currently has a territorial water boundary, it would retain it under the proposal. If it does not, it would not. In short, the proposal makes no changes from the current status. For disputed territorial waters (which I think are pretty commonplace), see my Ceuta example.

-- Johnparis (talk) 07:27, 5 January 2019 (UTC)

controlled_by=RU;UA

The page currently states ([1]) "Crimea: To conform to the currently approved de facto borders, the administrative boundary relation, which comprises all of Crimea, would be tagged type=boundary + boundary=administrative + admin_level=2 + claimed_by=RU;UA + controlled_by=RU;UA + claim_level=4." I do not understand why it is tagged as controlled_by both. The only other example of multiple control is controlled_by=DE;LU, but that doesn't seem comparable. The controlled_by guideline says "who will use force to stop you if you enter the area without permission". I can see that Ukraine would stop a person trying to enter Crimea _from Ukraine_ but that doesn't seem sufficient; this could also apply to entering Taiwan from China without proper permit. Can someone clarify? --Jarek Piórkowski (talk) 14:25, 27 January 2019 (UTC)

Change in meaning of boundary=administrative + admin_level=*?

The real world examples suggest that Crimea should be mapped with a boundary relation tagged type=boundary + boundary=administrative + admin_level=2. That would constitute a change in meaning of the admin_level tag which is not otherwise documented in the proposal. It would be important to clarify that - as well as the intended extent of the Ukraine and Russia boundary relations under the rules of the proposal. --Imagico (talk) 17:23, 27 January 2019 (UTC)

Control in the West Bank

The control over the West Bank was defined by the Oslo Accords, which define 3 areas:

• Area A, claimed_by=PS + controlled_by=PS.
• Area B, claimed_by=PS + controlled_by=PS;IL.
• Area C, claimed_by=PS + controlled_by=IL.

-- Zstadler (talk) 14:56, 3 February 2019 (UTC)