We might have a ruins problem

Based on the tag description on the wiki, building=* should be used to represent the initial intended destination for the building we are mapping. However, we have nearly 300k use of building=ruins which seems a bit odd that there are so may buildings that were made to resemble ruins. I suspect that the intended tagging scheme for those buildings should be either building=* + ruins=* or ruins:building=*. I know that a mechanical edit cannot be made (since there can be some cases where building=ruins is the correct tagging schema), but maybe we can start by editing those that we know are clearly wrong.

Moreover, I think that this error was made easier to make thanks to an iD preset (which I’ve just asked to change since it seems to be adding an incorrect tagging schema).

Personally I am not fan how we tag ruins (see User:Mateusz Konieczny/ruined building - OpenStreetMap Wiki for my attempt to find a better schema - though I have no idea how much it is liked by others)

But I would not describe building=ruins as altogether wrong.

Personally I would prefer

  1. building= ruins=yes for “Ruined building”

  2. ruins:building= for “Remnants of Ruined Building”

but I am not sure would it be considered as valid distinction, I guess it can be argued that these are the same thing.

destroyed:building has problem of being widely used for cases where all remnants are gone.

Invent a new prefix for “there are remaining ruins but object is thoroughly destroyed”?

Whether good or not, building=ruins has been (mis-)used for so long and so often to mean a building that has fallen into ruin, that it reasonably has to be interpreted like that and not as ‘a building that was intentionally constructed to look like ruins’.

I would not recommend using it for any new objects and try to phase it out in favor of building=* + ruins=* or ruins:building=* (depending on how much of the building is left and if if can still really be considered a ‘building’).

It’s a problem, but it’s not a real problem. There are only about 400,000-600,000 of them worldwide. :slight_smile:

Normally, I dislike mapping buildings, because where I live people add lots of extensions, sheds, garages and other outbuildings and things become complicated. It takes a lot of effort to map these buildings accurately. This wouldn’t be so bad, except that the overhead imagery in half the country is old, blurry, has too much shadow, satellite-only, poorly aligned or just generally awful. Then there are tall trees. :frowning:

However, some areas have awesome :heart_eyes: aerial photos and I’ve been mapping buildings in one of those areas over the last month. A proportion of those buildings are ruins of various building types, construction types, ages, and states of dilapidation. There are some that were only ever built to the ground floor slab with/without some walls and abandoned, while others were ruined and restored, or ruined and found an alternative use. At times, I have questioned myself on what tags to use.

So, [EDIT: I was lead to believe] building=ruins was originally meant for fake Greek temples, etc. in rich peoples’ gardens. I think they (there can’t be that many of them) need to be changed to man_made=folly and/or building=folly to go with Tag:historic=folly - OpenStreetMap Wiki Only having them as ‘historic’ fails to appreciate that some of them are modern.

We then need to agree (a) rules what what is/was/will be a building and (b) a comprehensive lifecycle scheme for the scale of never-built, under construction, never-completed, vacant, existing, disused, abandoned, dilapidated, ruined, demolished …

I agree with @Mannivu. building=ruins means a building constructed as a fake ruin acc. to the building=* tagging scheme. As the actual use is apparently wrong in most cases we should try to correct it instead of retroactively legitimize it with the justification that it has been used in the wrong way so often.

not convinced by this

I expect that it would be logical extrapolation, but have you confirmed that it ever was dominant use?

maybe building=sham_ruins for actual fake ruins? As it is subset of them.

Following is a historic=folly but not a sham ruin:


This file is licensed under the Creative Commons Attribution 2.5 Generic license. Attribution: Newton2

Over the years, I have mapped dozens, perhaps hundreds of buildings using this tag, simply because iD offered it as a preset. As time went on, I looked into it on the Wiki and Taginfo. The documentation seemed rather unclear, while the usage numbers showed that the tag was widely used across the community, so I continued using it. Today, I believe those numbers make this tag a standard-de-facto , much like the generic building=yes.

However, @Mannivu ’s point is valid, and I think it is right to address the issue.

I agree with @Mateusz_Konieczny ’s proposal, which is essentially what I had in mind: introduce a new preset using a properly structured combination of tags, with equally accurate translations across iD’s various localizations (and perhaps some wording in JOSM could also be clearer). However, I would keep it simple, without too many variants.

At the same time, the wiki should also be updated with clear, precise guidance on how to map ruined buildings correctly, without ambiguity and with photos that are not misleading.

We will also need to decide what to do with the roughly 300,000 incorrectly tagged buildings. Should we encourage local efforts to correct them through dedicated tasks on MapRoulette or similar platforms? Or should we accept them and leave them as they are, leaving renderers to deal with the issue?
I do not think a mass edit would be feasible, as we would have no reliable way to identify false positives.

Didn’t things happen the other way round? Up to 2018 the wiki page for building=ruins simply said that it was for mapping ruined buildings. There was no hint that it should be reserved for “sham” ruins. Meanwhile the understanding and documentation of the building key (also far from clear up to that point) has evolved, so that the ruins tag now appears inconsistent with the overall approach. So I think the suggestion is more about retroactively delegitimizing a mapping practice that was entirely reasonable at the time.

Do you have grounds for thinking that? Based on old wiki pages and buildings I have found still in OSM data that were mapped a long time ago, it seems far more likely to me that early mappers simply used it to indicate ruined buildings.

I don’t mind which way round it happend. It does not fit the actual meaning and causes confusion so we should change it.

It seems a good idea to have a specific tag for this, to remove any further doubt about the mapper’s intentions for at least the small number of buildings tagged this way.

That would probably need to be well coordinated with renderers to be successful. Many ruins are prominent landmarks and mappers may think they have done something wrong if they disappear when a lifecycle tag is added. I seem to remember that happening to me many years ago which encouraged me to stick with building=ruins.

Likewise. A building was destroyed in a fire recently; it still has all 4 3-story walls standing proud, but the roof is long gone and there is nothing left of the upper floors. I spent a week waffling about what the correct tag was, I think now it’s destroyed:building=* and doesn’t show up on the map anymore. The fenced-off construction area around it does though, so the message still gets across.

Have you mapped the walls? :slightly_smiling_face:

I don’t think mapping it as walls makes sense as in this case the ground floor is still (by and large) intact. I believe the best representation for this is still a building that has been partially destroyed, rather than just 4 walls. Additionally tagging the walls would make it show up on https://www.openstreetmap.org, but I think that constitutes Tagging for the renderer - OpenStreetMap Wiki .

What worse exactly this tagging is also used for utterly destroyed buildings gone without any trace left :/

This poll is intended to only check is it worth it to make a RFC asking for more feedback.

What would be an acceptable solution here?

  • Consider building=* ruins=yes and ruins:building=* as expressing different things (ruined building that is still a building vs these remains are not a building anymore)
  • Introduce new tagging to cover this distinction (building=* heavily_damaged=yes + remnant:building=)
  • Introduce new tagging to cover this distinction but a different one (please, make post and propose own)
  • Do nothing, current situation is fine
  • Something else should be done (please, make a post with your proposal)
0 voters

Ruined, still a building:

commons file page


Author: Zbyszko Siemaszko. Public domain image

Ruined, not a building:

commons file page


This file is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license. Attribution: Lsowada

Currently we have no established way to distinguish between such cases.

At the moment, there is a lot of confusion. The iD preset is translated differently in different languages: for example, in German we have “artificial ruins”, while in English, French and Spanish we have “ruined buildings”. In Italian, we have also had “artificial ruins” for the past few days. And these are just some of the main languages.

I think we could start by adding the new preset, building=* + ruins=yes, which would be correct anyway and would not conflict with the current situation. Then, once things are clearer for everyone, we could take further steps.

Personally, I would prefer to “keep it simple”, so:

Situation Tagging
The building is there and is in use or usable building=*
The building is there, but is ruined, damaged or collapsed and is not usable building=* + ruins=yes
A building used to be there; now only ruins remain ruins=* (or ruins:building=*)
A building used to be there; the remaining ruins have historical importance historic=ruins
The building was deliberately built to look like ruins man_made=ruins?

To me, the difference between “ruined” and “destroyed” is not about appearance, but about what happened to the building. A ruined building generally falls into ruin over time through neglect. A destroyed building has suffered an external event: a fire, a flood, an impact or an explosion. Personally, I would not separate the two cases through the main tag: I would use ruins=yes for both, then add another tag for the condition or a description if that information is known and considered useful.

Lifecycle tags are certainly very useful, although they make things more complicated for renderers. To prevent buildings that no longer exist from being added again, I was familiar with razed:building=*. As for ruins:building=*, the distinction makes sense to me, as long as it is clearly explained with examples, so that mappers can easily choose the right tagging.

I would also be fine with the “Do nothing, current situation is fine” option, but the wiki and the preset translations would still need to be improved and aligned.

Street side imagery of structures similar to the first example may show evidence of residence. Clotheslines and so on. So not a ruin?

I personally don’t like building=* + ruins=yes and definitely don’t like it having a different meaning to the ruins:building lifecycle tagging.

Having the same word (ruins) meaning different things is poor from a tagging perspective, confusing linguistically, and bound to lead to error.

And, if instead there were to be no distinction, then I’d prefer we use lifecycle prefixes. True it is a bit of a mess, e.g. disused, but having both ways is redundant.

I think what you’re trying to convey with ruins=yes can already be conveyed with abandoned:building:

Still visible but fallen into serious disrepair and which could only be put back into operation with considerable effort