On tags and places

Question

This is the second of a series of meta-questions reviewing our approach to tagging, in a follow up to Improving our tagging (redux). (The first was On tags and time periods)

PolyGeo has highlighted this related question: Applying geographic tags to questions?, to which I'll add How are we using geographic tags?

Very many of our questions need a 'place' tag -- events and the production of the associated historical records are frequently linked to one or a few specific places and can't be answered without knowing the locality. For example, Finding death records for family members? cannot be usefully answered, other than by producing a laundry-list of answers for places all over the world -- or preferably, by identifying which place is being asked about. (Of course, we also have questions that don't need a geography-tag but those aren't being discussed here.)

In what follows I'm assuming that geography tags meet the test of 'mapping what our community is about/not about' -- genealogy anywhere on this planet is in-scope (although it might be more difficult in some places than others); I reserve the right to open a discussion about Martian genealogy if and when the issue arises. It's also unlikely (but not impossible) that a geographic tag can work as the only tag on a question, so I'll not be considering that in any detail.

What I hope we're going to be thinking about:

• using tags 'to organise not summarise', helping people to find questions they can answer and answers that might address their own question
• optimising search engine results that might bring new users to our site

and importantly:

• providing clarity and simplicity so that an OP can easily determine what geographic tag to use for their question, or to filter by to reliably find relevant/related answers

So... What sort of geography-tags do we want to use?

continent? Probably too granular to be of much use, except where there are no countries at all. Antarctica is the only one that springs to mind, but I suppose we could get a question about whether the records for a particular Antarctic expedition survived and where they can be consulted, in case Great-great-uncle Freddie really met his end in the frozen wastes as the family legend says.

country? We definitely have questions where country is very relevant -- looking for Marriage records in England and Wales is a different process to looking for them in Scotland or France, and the information provided in such a record will differ as well (although there may be some things in common). So country tags seem sensible.

But... how to identify the countries? By modern country name germany that should be reasonably static (with some exceptions) ? By historic country name bavaria? united-states versus british-colonial-america? israel versus british-mandate-palestine?

Do we need to take into account if we can the sensitivities that exist around identifying places, particularly with regard to recent history and unresolved conflicts (Is the Crimea in Russia or the Ukraine? Does the answer depend on the question/time period concerned? Shall we just wait and see if a problem arises?).

We also need to consider which approach is simplest for those who may not be steeped in the intricacies of time-based geography. Many of our wikis direct users to use the modern country to include previous political entities, but this isn't consistent -- we don't have israel but we do have british-mandate-palestine. Should we encourage users to tag with both the current and historical name...? Or is that too complicated/too much to expect?

state-or-county? Locating records in individual US States or UK counties can be specific to that jurisdiction (other jurisdictions exist in other countries and should be assumed to be included in this discussion). So there are definitely cases where the answer to a question depend on defining a place more precisely than a country.

But... how do we decide which 'units' to tag for, and how to disambiguate them (e.g. new-york-city and new-york-state). If there could be ambiguity do we ensure that the tag names remove any doubt about which is being used? Or do we assume (for example) 'State' unless the tag says otherwise. And do we need any guidance up front or just handle things on a case-by-case basis?

Combining place tags

How do we/do we want to combine geographic tags (I'm using English examples here).

Do we want to tag only for the smallest possible unit that categorises the question accurately ... so in precedence terms county trumps country? If so, should we consider that, if the answers(s) turn out to be not specific to the County although the OP thought they would be, we should retag the question with the country to make it most widely useful (and edit the question to include the County name if it doesn't already do so)?

Or do we want to tag for both Country and County, in a belt-and-braces approach? Include all of wales glamorgan united-kingdom? Should we also have a england-and-wales where the question relates to those elements of the UK but not Scotland, Northen Ireland and all the other islands. Jersey isn't a country; it isn't part of the UK or Great Britain but is is part of the British Isles. Or do we treat it as a quasi-country for the purposes of tagging? How do we ensure consistentcy so that somebody looking for questions on Northamptonshire doesn't miss relevant questions tagged England?

Over to the assembled masses...

Answer 1

I favour a simple set of guidelines where questions are tagged by geography as follows:

• tag a country if the question and answer could be applied to anywhere in that country
• tag a county/state/province if the question pertains specifically to that county/state/province

What this means practically is that the vast majority of questions should only have one geography tag. A question should not be tagged with the county/state/province only because it mentions that county/state/province. A question about a very specific locality (e.g. county) should not be tagged with the country as well.

If you are thinking about adding multiple county/state/province tags to one question, then a country tag is probably more appropriate.

Many questions will be about ancestors who lived in more than one country, but this does not mean that every country or place mentioned in the question must be tagged. Choose the one, or possibly two, places that are pertinent to the question and tag accordingly. For example, a question about the death of an ancestor from Albania who immigrated to the USA does not need to be tagged albania because that has nothing to do with finding the death record.

The nature of country/state/province tags will vary by country, and the scope of such tags needs to be determined on a country-by-country basis.

Avoid tags with overlapping geographies where possible. I know this is easier said than done, but again I think has to be determined on a country-by-country basis to keep our set of tags manageable and useful.

Answer 2

My own thinking has moved on somewhat in the light of experience somewhat from when I posted this answer: https://genealogy.meta.stackexchange.com/a/3170/104.

I said then:

Does the geography matter to the answer? If not, don't tag for geography.

I still agree with this.

Do one or more geographic tags help questioners find similar/related questions that might let them answer their own question or at least expand their knowledge in area of interest; and steer them away from rabbit-holes.

This test is now leading me to prefer modern country names over historical ones -- non-experts asking questions are likely to tag for the modern country anyway (or their understanding of the modern country); and explaining why the modern country is not the best guide to where to look for records or a place name or whatever can materially contribute to making an answer widely useful.

We've had to explain this often enough that it's clearly not viable expecting 'old-name' tagging in the first place and correcting tags from modern to historical is a workload we don't need. The synonym system doesn't help as there's no simple mapping we can apply in many case. (This is a prime example of where I think our tag-wikis -- or the ones that don't currently explain what historical places lie within the modern boundary -- can be made more useful -- even if only as a place to point for the inevitable explanations that will be needed.)

This still doesn't address the sensitivities around some place names; we could address it in some cases by using a 'proxy' -- Crimea instead of calling it 'Ukraine' or 'Russia'. But how do we handle Britsh Mandate Palestine versus Palestine versus Israel? I don't know. Both modern places overlap the old one, as does Jordan.

I do think that it's important to be consistent -- tagging some questions Bavaria and some Germany isn't helpful when filtering, especially if you're not already aware that they are overlapping places. So I'd want to see a historical place name only ever in addition to the modern place name, not instead of it. Except... the Palestine question again :)

Do they help experts find questions to which they can contribute expert answers (I know a fair bit about the availability of records for Pembrokeshire, but nothing about Monmouthshire, for example).

This is leading me to believe that we should alwaystag from 'smallest' to 'largest' (City/County/State > Country). If there's only room for one geographic tag, tag it at the lowest level, but understand that it will be retagged if its more widely applicable.

Answer 3

A couple of thoughts about specific problems.

Questions involving migration, emigration and immigration may be problematic because the question might need to be tagged in two places. Questions about finding where someone died, like When did Abraham Decker (b. 1839) die? may be especially problematic because people can die anywhere, no matter where their usual residence is.

Historical vs. Modern names

One solution that exists for the modern vs. historical place name problem is to do what the Family History Library does -- their place hierarchy is often based on that described in a particular gazetteer. For instance, the German place names are based on Meyers Gazetteer of the German Empire for which there is now an online searchable version at http://www.meyersgaz.org/. We now have tags for all the jurisdictions that the online Meyersgaz calls "Level 1 -- Level I shows which Kingdom or State the place is located in." On the other hand, people could be doing research in the period before the unification of Germany, and not know which kingdom or state is appropriate, so they may want to use Germany for those questions anyway.

Answer 4

I think we would be remiss to not mention the discussion already had at Applying geographic tags to questions?

From that my preference when tagging geographically is in this order:

1. City tags only in exceptional circumstances e.g. new-york-city - see this answer to Applying geographic tags to questions?
2. State/Province/County in those countries where we have enough questions to warrant them.
3. Country when State/Province/County tags are not available or yet warranted

I also think historical county, country and (rarely) city names may be used if they come up frequently (e.g. prussia).

At the moment we have many questions with both a Country and State/Province/County tags and I think a loose guideline should be to phase Country out in favour of State/Province/County tags, particularly for the countries with lots of questions.

Ideally, I think a question should have only one geographic tag, but if other key aspects of the question have already been tagged and there are one or a few "to spare" then applying a second (or third more exceptionally) seems fine to me.