From OpenStreetMap Wiki
Jump to navigation Jump to search


Squares are often of interest for tourists and navigation to them is sought after. Examples:

I disagree with the need for an addr: tag. Streets and places do not have addresses, they have names that become part of the address of adjacent objects. So the fifth house adjacent to Красная площадь gets the appropriate address, not the square itself. An address is something I can post a letter to, I cannot send one to a place. The soviet government did not sit on the Red Square, they sat in the Кремль, which is adjacent to the square, and uses the square in its address.
As for navigation, the questions from tourists you cite demonstrate that they have the same misconception that an address is the only thing you can programme into your satnav. You can equally navigate to a named point of interest, or a place. Having the square in the place=* namespace provides the advantage, that routing engines most likely have the key already in the database, so navigating to place=square + name=Красная площадь is as easily implemented as navigating to place=city + name=Москва.

--Polarbear w (talk) 23:41, 30 October 2015 (UTC)

I also disagree with the need for an addr:* tag by the same reason. Also, saying that USSR government had an address on the Red Square is completely incorrect (Wikipedia doesn't have reliable source reference for that statement) - first, "USSR Government" is a blanket term, which refers to several administrative offices, and all of them had own buildings (and addresses) somewhere else. --BushmanK (talk) 17:25, 8 February 2016 (UTC)
I also disagree with adding address tags at squares, squares are part of addresses, but they don't have an address on their own. You can route to these place=square objects like you can route to any other POI. --Dieterdreist (talk) 09:18, 28 June 2016 (UTC)


Only one thing is disturbing me: the place=* tag is normally being used to indicate something administrative - I'd rather have preferred to use landuse=* tag instead which is for describing areas... --katpatuka (talk) 05:53, 6 July 2016 (UTC)

actually the tag place is used to describe something geographical, it's the tag boundary=administrative that is used for indicating administrative entities / their boundaries. Landuse on the other hand is an attribute and not suitable to define features (e.g. it can be arbitrarily split without changing meaning) --Dieterdreist (talk) 06:39, 6 July 2016 (UTC)

Difference between place=square and highway=pedestrian

What's the difference between tagging an area with place=square and with highway=pedestrian (aside from rendering, which is non-existent with place=square)? Or are both to be used in combination? --Absay (talk) 00:41, 10 March 2017 (UTC)

place=square is "orthogonal" to other tags, since the square can comprise several things, like a pedestrian area, streets, ways, a park, green areas, etc. It can also completely consist of one of them, in this case the same object would have the physical tag and the place tag. Rendering place=square is applied for and will come once there are enough in the database, most likely initially with label only. --Polarbear w (talk) 08:32, 10 March 2017 (UTC)


It's a pity that the place=square is still not rendered in the standard Mapnik layer (at least not for single nodes), although there are multiple thousand instances mapped. You search, you find, but you see nothing on the map. This will make people revert to "map for the renderer" using other tags.--Johsin18 (talk) 17:31, 30 October 2017 (UTC)

There are two open tickets in Carto, #2673 and #2203, you can contribute your discussion there.--Polarbear w (talk) 20:52, 30 October 2017 (UTC)
  • the following statement moved from the feature page to this discussion:

Note that simply using "place=square" on an area, without combining it with other tags, will not result in any rendering in any layer. Talk:Tag:place=square#Rendering. Jidanni (talk) 23:15, 22 May 2020 (UTC)

As far as I understand , place=square gets its name rendered in Carto both on nodes and areas. --Polarbear w (talk) 00:28, 23 May 2020 (UTC)

Changed definition and consequences

With the changed definition of the tag i wonder how the mapper is supposed to decide if an area that qualifies as highway=pedestrian or as leisure=park or landuse=village_green is supposed to be tagged place=square. Is any open space in an urban centre a place=square and if not what are the criteria? --Imagico (talk) 23:13, 30 October 2017 (UTC)

I see setting the place node/area for the square independent from what is on that square, so you would still map the parks, streets, ped areas present. In analogy, setting place=locality or place=suburb would not prevent you from mapping what is there physically. Also, there is a focus on having a name, which was in the definition already: "used to name squares", which of course has to be verifiable.--Polarbear w
I repeat and reformulate my question: How should mappers decide if they should map an open space in an urban centre as place=square - either as a node or as a polygon (and if as a polygon how to delineate it). The previous definition said: It depends on if it is named Square or one of the listed equivalents in a number of languages. Is that still the case (i.e. is Terminology part of the definition)? Note in German speaking countries there are things named Platz that do not match the new definition (i.e. are not open space or not in an urban environment). --Imagico (talk) 09:22, 31 October 2017 (UTC)
Maybe you could give examples/counterexamples to illustrate your concern. Alexanderplatz, Berlin, does match my imagination of place=square, while de:Sportplatz does not. But, do we need to defend definitions against multiple meanings of the translated word in other languages? de:Kiefer/en:pine does match natural=tree; de:Kiefer/en:jawbone does not. --Polarbear w (talk) 19:21, 31 October 2017 (UTC)
I am not questioning the definition at this time, i am just trying to understand it. So again: Is Terminology part of the definition? Does a place need to be named square or equivalent to be tagged as place=square and if not, how can i - as a mapper - determine if something is to be tagged place=square or not?
A few examples i find unclear (not necessarily exactly the feature linked to but the area around): --Imagico (talk) 10:15, 1 November 2017 (UTC)
Thanks for the examples. I think the multilingual terms were a vehicle for finding a proper definition. Thus it was meant to include objects that would fit the concept in all those languages, and exclude the geometrical figure en:square or any location ending with de:-platz. I wonder (@Dieterdreist) if we should add the requirement of architectural signifcance and a name, to exclude those small patches of grass that were shown in the list above.--Polarbear w (talk) 11:16, 1 November 2017 (UTC)

@Dieterdreist - thanks for today's improvements on the definition. What about names, should it be required that a square has a name, or should we keep that in useful combinations only?--Polarbear w (talk) 19:27, 31 October 2017 (UTC)

Alternative Tagging

It looks like most place=square can also be tagged in one of these 4 ways:

1) leisure=park - For squares/plazas which are a small urban park in the center of a town, village or city neighborhood, usually rectangular and surrounded by streets (or pedestrian streets). In this case a majority of the are is vegetated with grass, trees, shrubs, flowers etc, though a significant minority may be paths or hard surface.

2) highway=pedestrian + area=yes - For squares which are a paved open area in the middle of a town, village or quarter, traditionally used for parades and markets, also good for demonstrations, protests, etc. - in this case, a majority of the area of the square is a hard surfaced area where pedestrians may walk freely.

3) place=neighbourhood for a named neighborhood surrounding a square, plaza or traffic circle, which is named after the central square (or a historic square which is no longer there). Eg: Harvard Square (Cambridge, Boston), Trafalgar Square (London), Times Square (New York).

4) junction=yes for a road junction or street intersection.

So I wonder when `place=square` is really needed.--Jeisenbe (talk) 00:26, 27 April 2019 (UTC)

I don't have much of a position on the merits of `place=square`, but the wiki page talks about "typically crossed by streets". So a situation roughly comparable to 2), but with a part of the area dedicated to motorized traffic, might not have a direct equivalent in other tags. --Tordanik 20:28, 6 May 2019 (UTC)

@Polarbear w: - See addition of junction=yes for named junctions. It seems there is an alternative, more popular, way to map almost any type of "square". -- Jeisenbe (talk) 01:55, 20 March 2020 (UTC)

The four taggings you mentioned above, as well as named junctions, are not the same as a square. The pedestrian areas or parks can be part of a square, or fill it completely. Neighbourhood is a different concept. Junction=yes is used for named junctions that do not form squares. This is a concept used in some countries, e.g. Japan. --Polarbear w (talk) 12:58, 20 March 2020 (UTC)
Re: "The pedestrian areas or parks can be part of a square, or fill it completely." If a leisure=park fills a square completely, how is it then a place=square as well? What makes it a place=square instead of just a park?
Re: "Neighbourhood is a different concept." Yes, because the page says to exclude the buildings that are "at the square" but include things "on the square". But why is this the case? Why aren't the buildings part of the square? Can you have a square with no buildings around it? Note that neighborhoods are mapped as nodes, which avoids the problem of non-verifiable or fuzzy borders.
Re: "Junction=yes is used for named junctions that do not form squares". It is also used for unnamed junctions (Only 31% have a name: How is a mapper to decide when a junction is a square and when it isn't, based on the the current definition of this tag? I read it that basically any junction could be considered a square, if it is in an urban area.
The problem here is that this definition of place=square is so broad that it can include almost any urban park (most urban parks are surrounded by buildings and they are an "open public space), it can include any urban junction, and any urban highway=pedestrian area. If it were just mapped as a node, this would not cause any more harm than using place=locality (though that has problems too). But by suggesting that mappers should use this on areas in a particular way, we are just replacing a verifable geometry like a leisure=park area or highway=pedestrian area (or junction=yes node) for a non-verifiable, imaginary boundary around a place name. Why including the streets around the square, but not the buildings around the square? Such a definition only makes sense in a particular culture: European urban culture, and if you look at the data from the rest of the world, this definition doesn't work. --Jeisenbe (talk) 13:48, 20 March 2020 (UTC)
As I have started the feature page Square over the old redirect, it would be helpful if you could illustrate your thoughts there, e.g. by adding pictures with non-European cases. --Polarbear w (talk) 16:29, 20 March 2020 (UTC)

Inconsistent use of this tag

See discussion and analysis of how this tag is used, with images, on github:

Summary: An analysis of the use of place=square in 8 places (regions or cities) in North America, South America, Europe and Asia showed that many cities and countries are using place=square in ways that are not consistent with the definition on this page.

Many of the "squares" were also mapped as highway=pedestrian, or were a similar hard-scaped open area which could have been mapped in this way. Some were road junctions, where the area included roads and surrounding open space. Others were parks, or were inside of park. Some were small hardscaped open areas between a building and a street. Only a minority were a "town square" as suggested by the photos and mapping advice on this page.

Brussels was the only city checked where more than half of place=square features were used for a town square, but even there the mapping is inconsistent, and many of the proper squares are mapped as highway=pedestrian. --Jeisenbe (talk) 07:17, 19 March 2020 (UTC)

Have you analysed when the incorrect mapping was introduced, and if that could have been influenced by a misleading editor preset? --Polarbear w (talk) 12:51, 19 March 2020 (UTC)
The date the features were mapped does not seem to have any bearing. In JOSM the preset offers an icon for a node and a rendering for an area. I have not checked iD since I no longer use that editor, but please investigate. However, I suspect that the problem is due to the lack of definition in the tag: if the tag is not defined here on the wiki, how should the editors be expected to provide a better definition? --Jeisenbe (talk) 13:51, 20 March 2020 (UTC)
What lack of definition do you mean? Today I fixed over a dozen place=square w/o a name. All, 100% of them, were introduced by iD after 2018 (thus after the 2nd refinement of the tag definition), and all but one were just mapping tiny footway or servie areas. Thus iD must be doing something wrong, again. --Polarbear w (talk) 20:08, 20 March 2020 (UTC)

History of the tag

Originally this tag was poorly defined:

"The place=square tag can be used to name [W] squares (English: square, plaza; French: place; German: Platz; Italian: piazza, largo, piazzale, campo; Spanish: plaza; Russian: площадь, Chinese (simplified): 广场. ).

Over 2 years later in October 2017, the definition was changed to "a town or village square ... an open public space common in urban centres, such as cities, towns or villages". However, there are many instances in the database of this tag being used for any feature that includes one of the words above in it's name, such as any "plaza" in a Spanish-speaking country (though not all plazas are a "town square").

There are also many cases where this tag is used for any flat, open area, including those in parks or military bases, which are not surrounded by buildings. These are more commonly tagged highway=pedestrian + area=yes, as are most hard-surfaced open areas in towns, where pedestrians may walk. While the definition has always been limited to named features, there are a significant number of uses for unnamed flat, open areas.

The previous definition said "They are typically crossed by streets, but can also be pedestrian areas or more rarely green areas," and some features with this tag are vegetated areas might be tagged with leisure=park instead.

See also wikipedia about [W] squares.
(above analysis of the page history originally written 08:09, 19 March 2020‎ by Jeisenbe) moved over from main page to this discussion --Polarbear w (talk) 12:46, 19 March 2020 (UTC)

Both, the original definition in 2015 and the refinements in 2017 were the result of discussions in the community, the latter in particular triggered by a response in Carto that the 2015 definition was insufficient. --Polarbear w (talk) 13:45, 19 March 2020 (UTC)
I'm aware that the changes in definition were meant to make this tag more verifiable and consistent. However, most features do not match this definition. Why isn't this information relevant to the wiki page? The current definition is brand-new, and does not match the usage of the tag for the majority of features in the database. Tag: pages should correctly document how a tag is used in practice, so that database users will know what to expect when they find these tags in the database, in addition to suggesting the optimal mapping method. --Jeisenbe (talk) 01:40, 20 March 2020 (UTC)
I appreciate your analysis, but as you said elsewhere it confuses a mapper when it is on the front page of a tag. As for the definitions, they are not brand-new, they are as old as the usage of the tag which started in 2015 when the usage was around 10 or 15. Up to this point, people abused individual elements of a square to tag the name. If now people are mis-using the tag, the reasons have to be found, and not the definition changed. If, fictionally, people would tag blue roofs as natural=water, would you widen the definition of water? --Polarbear w (talk) 12:54, 20 March 2020 (UTC)
The definition of natural=water has been widened to include landuse=reservoir, if you want a real example. If a majority of features with a tag are mismapped this should be noted on the page. Imagine the opposite: water tagged as "roof:material=stainless_steel" due to faulty AI or armchair mapping, and now a majority of ""roof:material=stainless_steel" features are actually reflections off of ponds. In that case it would be good to mention.
Re: "Up to this point, people abused individual elements of a square to tag the name." Ah, I see: you disagree with adding name=* to the highway=pedestrian + area=yes way, since you think the square also including the surrounding streets and patches of grass and fountains etc.... But who is to determine where the square ends? In Latin America, I always assumed that the Plaza was the part in the middle that is like a little park with trees and walkways, not the streets around it. And if you include the streets, why not the surrounding buildings: the Cathedral, Town Hall, shops? I see your point that putting the name=* tag on the highway=pedestrian area is not precise, but the solution would be to map the named square as a node at it's center. Everyone can agreee that the center of the "square" is part of the square. But telling mappers that the square needs to "share nodes with surrounding buildings" but not include them is just trying to enforce a particular cultural view of what is a "Square/Place/Platz/Plaza", and clearly it has not worked: even in Europe most place=squares are not being mapped this way. --Jeisenbe (talk) 14:01, 20 March 2020 (UTC)

monitoring older synonyms

If you know places with this tag, verify if it could be tagged with another tag.
Automated edits are strongly discouraged unless you really know what you are doing!
If you know places with this tag, verify if it could be tagged with another tag.
Automated edits are strongly discouraged unless you really know what you are doing!
If you know places with this tag, verify if it could be tagged with another tag.
Automated edits are strongly discouraged unless you really know what you are doing!

Status "in use" or "de facto"

I believe the status "de facto" should be reserved for tags which are the most common way to tag a feature (and which were not approved). If another tagging method is more common, the status should be "in use". In this case, most "town square" features in Openstreetmap are mapped as one of these 4 alternatives:

  1. leisure=park
  2. highway=pedestrian + area=yes
  3. place=neighbourhood
  4. junction=yes

So place=square is not the established tagging, and it should be "in use", not "de facto". --Jeisenbe (talk) 02:06, 20 March 2020 (UTC)

Personally I don't fight hard for distinguishing "in use" vs. "de facto". However I find the same mistake repetitively in your argumentation, that you compare the tagging numbers of different tags without relating them to how many objects of this type are there in reality. Yes leisure=park is tagged more often, but there are many more parks than squares, and not every park is a square. --Polarbear w (talk) 11:20, 20 March 2020 (UTC)
My claim is not that leisure=park is more common (that is obvious), but that more features which fit the definition on this page or the previous definitions are currently mapped as leisure=park, highway=pedestrian + area=yes or junction=yes than as place=square. And I additionally I suspect that each of those 3 tagging options is individually more popular for mapping named "squares" than the tag place=square.
See for example (warning, really big queries. Might be faster to download all the features and search in JOSM): (leisure=park ways and nodes with name including the string "Square" - relations and other languages excluded because the query is already really long. (highway=pedestrian ways with area=yes and "Square" or common translation - really long query) (junction=yes with "Square" or a common translation).
This situation could be improved by excluding named junctions from the defintion of place=square, or excluding things that could be considered parks, though that would still leave highway=pedestrian + area=yes as a very common synonym. As you've mentioned, it has the disadvantage that it can only be added to the actual area of hard surface, open to pedestrians, but this also gives the advantage of the geometry being verifiable, unlike the suggested geometry for mapping a place=square according to the current definition and usage. --Jeisenbe (talk) 13:13, 20 March 2020 (UTC)

The situation is comparable to drawing landuse=residential and adding the name of a village to the landuse, vs. adding place=village either as a node or to an outline comprising the residential and non-residential features of that village. These are orthogonal concepts. The OSM-historic problem is that place=square was not in use until 2015 and mappers resorted to improper solutions to tag the name somewhere. If your pedestrian area fills 100% of the square, I have no problem with adding place=square to the same outline. But I have several situations here in Berlin where the square contains all sorts of things, crossing roads (some named as the square, some not), patches of greenery, fountains, ped' areas. I agree that just a named junction is not per se a square, but there are squares where roads cross. --Polarbear w (talk) 13:30, 20 March 2020 (UTC)

Re: "comparable to drawing landuse=residential and adding the name of a village to the landuse, vs. adding place=village either as a node" - Yes, very good example. In that case adding the name to the residential area is an unnecessary duplicate of the important node place=village, and incorrect since a village is not purely a residential area but usually includes schools, farmland, places of worship, shops, etc. If you instead try to add place=village to an area, that is a mistake because a village does not usually have verifiable boundaries, so a node at the center is better.
So I can see the benefit of using place=square for the name=* for hardscaped highway=pedestrian areas.
But this suggests that place=square should not be used for a "named road junction", which can be tagged as junction=yes. And it shouldn't be added to a leisure=park, since parks should be named and they have a clear definition. It should only be used for a named, hardscaped open pedestrian area in an urban area which is surrounded by buildings. And I would recommend mapping it as a node, like other urban places, at the center of the square, to avoid debates about what to include. --Jeisenbe (talk) 14:09, 20 March 2020 (UTC)