9
1

An OSM contributor, Marc Mongenet, has recently turned several Italian province capitals from place=city to place=town because their population is <100k inhabitants. But, according to the relevant Italian wiki page, Italian province capitals with >50k inhabitants shall be regarded as cities as well; the international page is more or less on the same note, saying that the place=city tag is meant "to identify the largest settlement or settlements within a territory, including national, state and provincial capitals, and other major conurbations"; this other page indicates that a city should have "normally" a population of 100k inhabitants (but I personally consider this threshold as rather flexible as it's likely that it was originally meant for the US).

I have invited the user to revert his edits, but he refuses to do so because he does not agree with the Italian criteria, and claims that inserting all of those cities will make the map overcrowded at low zoom levels. So, who's right and who's wrong?

asked 03 May '14, 11:56

Decan's gravatar image

Decan
3517717
accept rate: 0%

edited 03 May '14, 11:57

1

@all the help site is not a general purpose discusion site, I would strongly recommend moving the discussion to the tagging list, or other suitable international list.

(09 May '14, 00:15) SimonPoole ♦

What follows is my personal view - there are other views and I'm sure that people will add them as alternative answers:

First - bluntly, if the whole Italian community is saying one thing and one user in Switzerland is saying another, then I'd say that the Italian community's view had more weight with regard to Italian data.

That's not to say that his argument doesn't have merit - the place=city wiki page has, since the earliest version, suggested that it should be used for large places, not necessarily places that are labelled cities.

However, in Ireland he made town/city edits that were reverted by the Irish community as soon as they found out. They have a very clear definition of what a city is there - the mistake that he made there was confusing distance between places with the difficulty of getting between them (for example Galway is every inch a city - I suspect no one around there would consider it in the orbit of Limerick, which he assumed that it was).

He did something similar in the UK. There the discussion wasn't as clear-cut (the majority in this and the previously linked discussion were in favour of "place=city == large" rather than "place=city == designated as a city"). Personally I'd still think there needs to be some way of recording "this place is, in the real world, a city" (perhaps "designation"?) in addition to "this is a large and important place" (and at some point I plan to investigate the suggestions in that list thread for usage clashes).

Unfortunately this users edits in the UK were comically poor for other reasons - promoting Telford to a city, adding a duplicate Glasgow, and tagging-for-the-renderer Wolverhampton as a town because it's near Birmingham. These got reverted.

I did try to explain to him how population tags of towns and cities can be very misleading when areas nearby that you might consider part of the larger area aren't counted in its population figures (Nottingham and Manchester in the UK are obvious examples). However, I do not believe that communication occurred.

So in short it does make sense to have a tag that says "this is a large and/or important place", which may or may not mean the same as "this place is, in the real world, a city" - but it's almost certain that one Swiss user's interpretation of "large and important place" is in many cases completely wrong and probably needs to be reviewed and if necessary reverted across Europe.

permanent link

answered 03 May '14, 12:33

SomeoneElse's gravatar image

SomeoneElse ♦
32.9k65341771
accept rate: 16%

edited 08 May '14, 09:37

3

As a minor aside, a rule that place=city|town etc., based on population produces unsatisfactory results (e.g., in Ireland, and would logically require all London Boroughs to be cities, as well as dormitory suburbs in certain US conurbations). We have a perfectly satisfactory tag for population, called population. It is entirely possible for data consumers to create rules which use both city and population for rendering and other purposes: anticipating rendering outputs with tagging does no favours to either the data or development of better rendering outputs. Wikidata should help in future.

(03 May '14, 12:45) SK53 ♦
3

I totally agree. Not only the matter has been debated nationally lots of times (without reaching a solid consense, even though there's loose agreement by most mappers), even recently there have been edits by contributors who imposed their own rule. This case in particular seems not ok to me.

(05 May '14, 10:23) Simone Saviolo

I also agree that the italian community has probably got a better view on this. Still, I've just read the points Marc Mongenet noted in a page of his [1], and I also believe that some international standardization would be necessary. Otherwise we might end up with a EU country with more cities than another, for instance. And not because one country does really have more cities, but simply because different criteria were used.

(05 May '14, 13:45) solitone
(05 May '14, 13:48) solitone
1

Yes, but this requires a serious discussion, whereas our user decided these edits by himself, not caring about either the Italian or the international guidelines, and not even in compliance with the rules he is making up now (Pisa is a striking example of this), possibly because he edited without really knowing the involved places. And I think that, until no new agreement is reached, these edits must be reverted. My personal idea to solve the matter is to discern between two levels of importance: maybe using two different tags, place=city and place=metropolis.

(05 May '14, 15:29) Decan
2

I also think that 50.000 is a bit too few for an Italian town to become city. IMHO population should not be mixed with the administrative functions. All province capitals in Italy have (or had in the past, maybe some have been removed in the meantime) the additional tag capital=6 (for admin_level 6 which is for a province). I was also opposing this idea of place=city for small towns, but appearently being in the minority didn't change them myself, still I believe it is a hack (mis-tagging for the current rendering rules).

(05 May '14, 15:42) dieterdreist
2

Yes, dieterdreist, I totally agree. One example is Aosta, which before Marc's edit was tagged as city, just because it is a regional capital, and not because it is a city indeed.

I also agree with Decan, when he points out that we'd better make changes like these when we deeply know the places we're editing.

I believe we need to clear up the international guidelines, and reach an agreement as to which places in Italy are cities and which are towns. We want to have a proper hierarchy of places to get a better map.

(06 May '14, 08:32) solitone
showing 5 of 7 show 2 more comments

My original question just meant to seek for opinions about how to consider the behaviour of an user that decides his own edits, which explicitly do not comply the guidelines of national communities, rather than proposing edits to these guidelines: this is not "sad bureaucracy", it's how a community works. But, since the discussion is now on a much deeper level, here is my point of view. There are lots of inhabited centres that are "minor cities" or "major towns". Forget about their mere administrative role, or their official status. Looking at the hierarchy of inhabited places, they attract traffics (whatever this means) from the surrounding towns, and to some extent are themselves attracted by bigger cities that are not too far away. There are diverging opinions about how to deal with these places, and the ideas that Marc gives in his discussion page, although substantially correct in my view, aren't too helpful with these cases (that are a lot!).

Looks like the Italian community (as well as other ones) has, roughly, decided to consider "minor cities" as proper cities when mapping on OSM. Their criterion that a province capital is a city when its population is >50k may sound inappropriate, but it works in practice if we keep in mind the intuitive idea stated above. And I like this, because when looking at the map on a regional or local scale, bigger cities may be missing at all, but the qualitative distinction between a town and a "minor city" is still evident.

Since the map is usually looked at from a local scale, rather than a continental one, I consider this distinction more important than the one between minor and main cities. But of course, these are again two different hierarchy levels. How to deal with those?

  • If the only problem is that the continental rendering by Mapnik is overcrowded, we just have to ask for a better algorithm of the rendering priority on low zoom levels; maybe one that is only population-based will work reasonably well in this case.
  • If the problem is more substantial than this then, as I was saying before, we may introduce a new place tag, that either represents big cities (so that only "minor city" will be tagged as "place=city"), or "minor cities" (so that only bigger cities will be given the tag "place=city").

Of course, we will still need to clarify the distinction between a big city and a minor city, or between a minor city and a town. But this distinction looks more natural to me than just flattening part of a natural hierarchy.

permanent link

answered 08 May '14, 11:57

Decan's gravatar image

Decan
3517717
accept rate: 0%

@Jack the Ripper There is consensus that it is inappropriate to tag place values within the US based purely on administrative status. If you disagree with that, the appropriate place for a discussion is the talk-us@ list, not the help site.

permanent link

answered 09 May '14, 06:37

pnorman's gravatar image

pnorman
2.4k52140
accept rate: 19%

You can certainly revert my edits. The edit you are interested in should all be commented with the same format. Of course, the map (taken as a whole) will not really get better with the revert, it's mainly a (sad) bureaucratic decision.

I am still refining the page http://wiki.openstreetmap.org/wiki/User_talk:Marc_Mongenet/town_and_city

permanent link

answered 06 May '14, 01:40

Marc%20Mongenet's gravatar image

Marc Mongenet
704
accept rate: 0%

1

Marc, you might want to be a little more careful about changing the place tag. For example, New York state has a specific legal definition of what constitutes a village, town, city, etc. that has no relationship to OSM's generic definition. Other U.S. states only legally recognize cities.

It might be a good idea to leave these changes to the local community.

(07 May '14, 23:57) Jack the Ripper
3

I agree with Jack. Each region will have its own definitions of what constitutes a city, and that's fine. The same type of regional judgment already occurs frequently with the highway=* tag, and I think that's generally accepted.

If too many cities are displayed on a particular map, that map's rendering rules should be changed, not the tag values.

Even if we do want to make a global definition change for the place=* tag, it's a major change and should be discussed and agreed-upon before going and making mass changes.

(08 May '14, 00:47) alester

I agree with Marc, in cases of doubt it doesn't matter for the place-value how your local jurisdiction calls something, it is important how OSM has defined the meaning of those and our tagging should be consistent to our own definitions. If you want to store the official classification (parallely I suggest), use another tag.

(08 May '14, 10:38) dieterdreist
2

@Jack the Ripper, the one thing I had hoped was clear is that the OSM place=* is not meant to reflect a country's legal meanings of such terms, and more than it being a proxy for population ranges. We have other tags suitable for describing legal status. Overloading place with such rules makes the tag less useful: e.g., a US incorporated city with 100 people should never be tagged place=city.

(08 May '14, 15:40) SK53 ♦

Hello,

Here are the main points of http://wiki.openstreetmap.org/wiki/User_talk:Marc_Mongenet/town_and_city

Orthogonality problem

The place tag is the generic tag to map human settlements. For capitals (national, state, provincial, any admin level), there is the admin_centre role.

Using place for both purposes (settlements and capitals) is an error, because many provincial, state, and even some national capitals, are not a large settlement, nor the largest settlement in the territory (examples: Privas, Appenzell, Vaduz). Settlement importance and capital status are orthogonal properties.

Besides, tagging capitals with place=city solves no problem, the admin_centre role being available to rendering software. But it creates problems, because tagging as city mere towns, or even villages, destroys all expectations (see below) the map user can have when seing a city on the map.

Locating is NOT mapping

This chapter could also be named: Map what's on the ground & Don't map your local legislation, if not bound to objects in reality

In many countries, there is one (or several) list(s) of cities, cornubations (or villes, or bourgs, or cités or agglomérations or whatever it is called in the local language). Quite often, these lists are based on statistics data, and give precious informations. But blindly tagging as city the list items on the map is NOT mapping. It is only using the map to show the location of items of some list. None of these lists was produced for OSM. So, all lists must be interpreted in the light of OSM criteria to decide if a place is a city (note that I write city in monospaced font, because it is only an OSM tag value).

Of course, some lists are of great help for the place tag, while other are of very limited use, and no sane mapper would consider that copying the list as is in the map makes sense (except, of course, with a custom tag).

International consistency

Places tagged as city are displayed at medium to low zoom levels. An low zoom levels, towns are generally not displayed.

At low zoom levels, several countries can generally displayed in the browser window without scrolling. On some renderers, Europe can almost be displayed west to east on a HD screen with city names.

This fact makes the place=city tagging consistency specially important to achieve a good, consistent, map, across the continent, and the world.

permanent link

answered 08 May '14, 01:11

Marc%20Mongenet's gravatar image

Marc Mongenet
704
accept rate: 0%

2

"blindly tagging as city the list items on the map is NOT mapping"

But it is if that's what it's called. People who live in village of Almond, New York call it the village of Almond, even though it has less than 500 people. They don't call it the hamlet of Almond. And someone travelling from out of state to visit them will have a much easier time of it if the map matches.

(08 May '14, 01:49) Jack the Ripper
1

I'd like to mention for the "Orthogonality problem" also the tag "capital" which can be used with the value "number of admin_level" to simply add the administrative importance to a place entity without using a relation or having to inherit from an administrative entity.

(08 May '14, 10:36) dieterdreist
3

Just one more anecdote about this - the idea of trying to keep any "wide-area" view of a map consistent is very difficult, and in some places, impossible. When the first widely available in the UK "Peters Projection" paper atlas was released in the late 1980s, the authors had tried to classify towns and cities by populaton only on each map page. Unfortunately this failed badly where dissimilar areas appeared on the same map page (north Africa and southern Europe was an example I think - it suggested no important places in northern Africa at all, which is just wrong).

(08 May '14, 20:46) SomeoneElse ♦
1

@Jack the Ripper

The place values "hamlet", "village", "town", "city" should not be used literally; they are only levels of settlement importance. Why not literally? Well, because they can only be applied literally in English speaking countries... So, should France be mapped with "place=hameau", "place=village", "place=bourg", "place=ville", and "place=cité"? Of course not.

An anecdote about the word "city": the most direct French translation is "cité". But nowadays, when you say "cité", you mean a poor suburb with big buildings built in the 60s. You will immediatly see what I mean by looking at https://www.google.fr/search?q=cit%C3%A9&tbm=isch. So, if France were mapped literally (after translation), the place=city would be reserved for poor suburbs built in the 60s. ;)

@SomeoneElse

A wide area consistency can be achieved without applying the same population threshold in all territories of the World. The population threshold can (and should, I think) depend on the population density. So, in territories with low population density, cities are generally much smaller than in high density territories. For instance, Nuuk Greenland is tagged as city with only 16,454 inhabitants. But the first larger town is hundreds of kilometers away. An Nuuk has an international airport, port, hospital, university, government buildings and institutions. So the city tag seems fairly reasonable.

(08 May '14, 22:14) Marc Mongenet

@Marc

I agree that the terms shouldn't be used literally, but they should be used to reflect their correct status within their local area. Going back to my New York state example, there are legal "towns" and "villages" that have the same name. In order to properly distinguish between them, there is no choice except to use their legal title.

And your example about direct translations into French, I think, makes my case. We can't take a generic template (cities are only capitals, etc.) and apply it worldwide. There are simply too many regional differences for that to work.

(09 May '14, 05:25) Jack the Ripper
Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "title")
  • image?![alt text](/path/img.jpg "title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Question tags:

×848
×89
×9

question asked: 03 May '14, 11:56

question was seen: 5,355 times

last updated: 09 May '14, 06:37

powered by OSQA