To map the thing or to map the purpose and their illegitimate offspring

General talk
,
1

This is a somewhat philosophical pondering.

There seems to be to different tagging principles people believe in strongly. One is mapping the thing, and the other is mapping the purpose. Or you might see that as tagging the specifics or tagging the generics.

Say there is a big rock on the road. Some people see that and want to map that there is a big rock on the road. Other people see that and what to map a motor vehicle barrier. They both get mostly what they want in the current tag schema, with barrier=boulder.

Both principles have their merits, and we can make them work together.

This is where it gets complicated.

There was recently a discussion about a steel swing/swivel gate across a road. The product has different names and many applications. Some insisted that the name of this product is a cycle barrier, and insisted on tagging it as a cycle barrier even if the purpose was to be a motor vehicle barrier on a designated cycleway.

When people start naming things after one of the thing’s purposes, and then insist on tagging it as such regardless of the actual purpose, it feels like the bastard offspring of the two principles.

People were saying “The name of this arrangement of steel bars is a cycle barrier”. The wiki has at some point been updated with a text stating that a cycle barrier – counter-intuitively – may be used as a barrier to other vehicles (citation turns out not to say that, but that’s another story).

In another discussion, we’re discussing if a cycleway can be a sidewalk. In this instance, a sidewalk – perhaps counter-intuitively – is not a place for walking, it’s that thing over there, that looks something like this, regardless of purpose. Yet another bastard offspring, but only a half-sibling to the other bastard. Now we take the sole purpose and make that the name, and the thing can be used for any purpose.

OK, so obviously I’m a disciple of one the principles above, but it still feels like the two principles can co-exist. These bastards just make that so much harder.

It feels like we’re in a deep, murky hole here. How do we get out?

2

natural=rock + barrier=boulder of course. Sorry, I couldn’t resist :wink:

You’re right, and I completely agree with what you’re saying, but I don’t see how these 2 things can be fixed, especially not in a world-wide playfield, where something means A in one language, and B in another, or where the man-on-the-street calls it X, but the laws call it Y. The only “proper” way to fix this would be to come up with completely independent definitions of even the most basic things, using our own (OSM’s I mean) words for these things, so people aren’t tempted to read something into a word, because in their language it means “Z”. But this sounds so unrealistic, that I would vote against it immediately :wink:

1 Like
3

If you are referring to my statements on the topic then I feel I’ve been quite misunderstood. What I’ve said is that a path on the side of the road can sometimes be both a cycleway (because it is for cycling) and a sidewalk (because it also for walking). I would never call a dedicated cycleway (one that is not for walking) a sidewalk.

4

No, I wasn’t thinking about your statements in particular. There were many people in the discussion, and the general idea has been voiced many times, in several discussions.

1 Like
5

Functions are volatile. A typical church can be a community centre now, and an artists’ retreat tomorrow, and it still looks a church to passers-by. A former orchard can be a camping site now and still look like an orchard, probably still called “the orchard”. Come to think of it, I camped in the “orchard” several times, no harvesting going on there, but I still would not hesitate to tag it as an orchard.

6

Church is literally a function.

7

it is also literally a building, and literally an organization

2 Likes
8

Literally an organization, yes. I agree. Church of England is an organization, Den norske kirke (the Norwegian church) is an organization.

Literally a building? Sure, you might look at a building and say “This looks like a church based on my limited set of objects that are churches”, so there’s a false equivalence, but how would you expect people to agree on what looks like a church?

This is an office building (the name on the front may have been removed lately, I don’t know):

image
image1219Ă—1135 163 KB

This is a church:

P1050306_semilarge
P1050306_semilarge1920Ă—1281 364 KB

This is a restaurant (now it’s Pincho Nation):

McDonalds-Kristiansand
McDonalds-Kristiansand800Ă—533 77.1 KB

Let’s say you were to tag the first building as building=church, because it was built to be a church, and you don’t tag it with amenity=place_of_worship because if’s an office. How would you tag the other church if it ceased to be a church? Still building=church because it was built to be a church?

The restaurant was built as a bank, so does that mean the correct tagging for this building is building=bank + amenity=restaurant? If not, why is it different from building=church?

And what happens if I build a new office building looking exactly like the one in the first photo, and then later it is repurposed into a church? Is it now a building=office_building + amenity=place_of_worship? Or is it a building=church?

Btw, this is not meant to be about churches specifically, but about the principle.

The reasoning that makes building=church “correct” is exactly the same reasoning that makes barrier=cycle_barrier “correct”, regardless of the actual purpose of the structure.

9

We don’t just map by function. We don’t just map by physical appearance. It’s a mix. Sometimes function has the upper hand, sometimes physical appearance wins. And sometimes it’s mixed with other things, e.g. number of inhabitants or importance. And mappers and communities have styles and preferences.
That’s the way it is. Still, OSM produces nice maps, can you believe it?

10


|
|

  • | - |

How would you tag the other church if it ceased to be a church? Still building=church because it was built to be a church?

to decide this, you would have to be familiar with churches in this context, to me it would not have looked like a church at first sight, but if it does to someone living there, (or if it is clear from the inner structure, decorations etc.), it might be tagged as a building=church as well.

The restaurant was built as a bank, so does that mean the correct tagging for this building is building=bank + amenity=restaurant?

sure, it is also clearly recognizable as a bank

And what happens if I build a new office building looking exactly like the one in the first photo, and then later it is repurposed into a church? Is it now a building=office_building + amenity=place_of_worship?

the idea of “building” is to look at the architectural type, now there are different kind of types, there is the outer appearance (volumes, facades, etc.) and the inner structure (floor height, loadbearing system, rooms, specific facilities, etc.).
A church is typically a very big room (and optionally some minor rooms), while an office is usually not (floor height according to room size, trying to maximize volume utilization).
Traditionally, architects (usually) seeked for “truth” (outside and inner working consistent), but there can also be a “lobotomy” (Rem Koolhaas, Delirious New York) where the outer aspect does not allow to infer what is inside, or iconic buildings where the outside is like a sculpture and anything can be inside (or similarly outside just a big box) (Robert Venturi, Learning from Las Vegas).

So an office building “looking” like a church, will probably not be church type of building, just have the outer aspect like a sculpture, but not the inner structure which would be useful for a church. Therefor I would not necessarily tag this hypothetical kind of office building as church and would not believe there can be a conversion that keeps the inner office structure but just uses it as a church. If indeed the office was converted into a church (likely by removing the floors and rooms in the inside), then it would probably become a building=church.

Cheers,

Martin

11

I don’t disagree. Like I said, we can mostly make it work. That doesn’t mean there’s no problem.

To me, insisting on building=church for non-churches is quite similar to insisting on barrier=cycle_barrier for non-cycle barriers. There’s a differentiator in that one can argue that the building was originally built to be a church, whereas I’ve never seen anyone argue that the motor vehicle barrier was originally built to be a cycle barrier. I do find the “built to be a church” argument quite weak, though. Why does it not apply to other buildings? Why churches? I’m genuinely curious to know, if anyone feels like taking the time to reflect on that.

And I need to emphasize that the problem I’m raising is insisting on a particular tagging, with appearence – blended with one’s own limited view of the world – as the most important criterion, causing the OSM object to represent sometimes the opposite of the actual purpose.

12

Well, there are options how to tag, and I tolerate a lot more options than I myself would pick. And, over time, I am not even consistent myself. I have found that most other mappers have a similar attitude.

13

In which context is this clearly recognizable as a bank? I’m really curious to know.

14

I agree. We learn more, the world changes, so we adapt.

So what to do when people insist that a motor vehicle barrier is called a cycle barrier – or that the purpose is irrelevant, unknown or not easily determined, but it’s still named a cycle barrier – therefore they are correct in continuing to tag those as cycle barriers?

15

I think there is a difference, in that there is a fairly consistent conceptual separation in OSM between physical buildings and the amenities or facilities they contain. Even for the smallest type of building, there are examples near me of building=kiosk, shop=lottery. If the lottery seller closed the shop tag could be removed or changed to disused, while leaving the building so long as it exists. So using building= for the form of the building doesn’t lose any information about its current function.

In contrast, for a barrier there seems to be no similar way to separate form and function.

1 Like
16

Well, one possible approach is to consider the practical implications of the choice. Does one or other option lead to renderers or routers giving misleading information? (This is not definitive, of course - renderers and routers can be wrong too. But it is one factor to consider).

17

It makes sense for kiosk, because the dictionary “definition” of kiosk is a small, detached building with an open front.

It doesn’t make the same amount of sense for church (except in a specific cultural context) or bank (see above).

The current schema for building is a mix of forms and functions. About half and half, from what I can see.

18

Whaaaaat??? My world is shaking!

1 Like
19

Sure, and as it happens there are lots of similar ongoing discussions about highway tagging where top level lags are also a mix of functions, physical characteristics, and legal restrictions, which can make progress difficult.

My point was simply that it’s usually not so important for buildings because there is typically some kind of independent tagging of current function. Someone looking for a church is more likely to be looking for an amenity=place_of_worship. For that reason I don’t think comparing to buildings really helps one way or the other with your barrier problem.

I don’t think it’s necessarily wrong to reflect a certain amount of cultural context in a map. People do quite often say things like “the museum is in a converted church”, and keeping building=church reflects that. To take a different example, the Musée d’Orsay in Paris is well known to be housed in an old train station, so building=train_station has a certain value that would be lost by changing to building=museum. Meanwhile the museum is clearly identified by the tourism=museum tag.

1 Like
20

If barriers are considered by the router, it’s natural for a router to treat a cycle barrier negatively when routing cyclists. Some physically prevent a number of cycles from passing. @Richard probably has some input there. The router could have been adapted to handle the concept of cycle barrier not being a barrier to cyclists, but then how does it know which of them to treat differently?

The negative effects are quite clear. Either cyclists are sent down a route with cycle barriers instead of being routed another way without them, or cyclists are not sent down the best route because of misidentified motor vehicle barriers.

Hence people overtag to reduce the effect of the mistake that they insist on making – barrier=cycle_barrier + bicycle=yes, which is “tagging for the router” to make sure that the misnomer has little practical effect.