we are not calling it a “railway”, we are calling it a former railway, visible remains of a former railway. The analogy to your example would be: just the skeleton and not the body.
This aside, I think nobody would be questioning that a finger is part of a body. If police were looking for a body, would they reject a finger because it is too few?
If I was to map the fingers found, I would create a way or node per finger and a relation to gather them all and say they apparently belong to the skeleton of a homo sapiens. And I would sketch the live human in another document.
To be honest, I have never seen a railway line that has been completely overbuilt by a building 
It would either have to be a very short line or a very long building.
Joking aside: can we all please make an effort to be more precise about what we mean? There have indeed been cases where 20 km of railway line have been cancelled because of a single building.
That (albeit with teeth not fingers) is pretty much a standing joke in mammalian palaeobiology 
Representing a former railway with a relation using the extant pieces is definitely preferable to mapping a continuous line feature, say extrapolating 20km from a 2km remnant that was found in a ground survey.
I would be fine with the standard used by @Ferrocarriles_de_México
Folks that accept the usage of railway = razed, could you please give examples of situations where you would and not use this tag for railways in various levels of decay? How ‘complete’ can the railway remains be to continue using this tag?
… and the OSM community is not well equipped for validating such extrapolations, while the OHM community is.
Another implied railway hypothetical: there areas in rural agricultural areas of US where the former railway embankments have been completely removed except at the road crossings. There is approximately 1-2km of farm field between each of these former crossings. Mapable?
[I hope that deepl will provide a meaningful and understandable translation.]
This is the other side of the coin. In my view, such extrapolation is not a good idea.
But I think a missing bridge or a single building built on the former railway line is perfectly acceptable in order to clarify the overall context.
And there is one point for which a razed must be preserved in any case: as a normal lifecycle atribute like for all other OSM objects. Namely, if it has only recently been (really completely) demolished and rebuilt with buildings, but no updated aerial images are yet available.
Of course you can replace railway=razed with razed:railway=*, but that doesn’t solve the problem of those who want to abolish railway=razed altogether.
Everyone uses the seemingly same words and yet it seems that everyone imagines something and means something different. That’s fine, because everyone has had their own experiences. But this way we will never find a solution that - if never for everyone - can at least do justice to many mappers.
In addition to a few other things that need to be regulated and clarified first (I’ll go into this later), I think it’s important that there is one (and only one!) page in the wiki where as many clear cases as possible, as well as borderline cases, are explained with pictures. I think it’s a bit of a shame, because until now OSM has also thrived on having a little room for interpretation. But that no longer seems to be possible with historic railways.
I expect both sides to approach each other: to concede to one that not everything that existed 150 years ago belongs in OSM, and to concede to the other that the visible remains should be mapped as what they are: railway=abandoned (even with a bit of razed in between).
as what they are, which might sometimes be highway=cycleway or something to denote an embankment, with additional tags to explain that it once was a railway. I’m not clear at all with the nuances you guys discuss for the various states of railway lifecylce, but when it has become something else it must be clear even to historic railway beginners that firstly it is that something other now.
I was wondering about the following approach:
Yeah and delineating the boundary between what should be mapped and what shouldn’t is the hard work. Everyone is definitely working with their own mental model.
Instead of arguing about tags in a general sense, maybe we should debate specific application examples. We could start with a dozen examples where the tags have been used in different contexts from the OSM database. Expand from there (especially railway = razed). The folks here could defend/protest the submissions.
It’s good to work on concrete examples, but examples of what? Is our goal to find the perfect tags for us (hence examples of mapping situations on which we can agree)? Or is it to find the tags that will reduce undesired tagging by e.g. beginners (hence examples of errors on which we can try and understand what could have been clearer for the contributor)?
I mean discussion of mapping examples where the tags railway = razed and railway = dismantled have been both abused and applied smartly. This will be a bit subjective, but maybe by examining use cases we can work out more detailed, limited conditions for their use.
I myself am in favor of deprecation, but maybe it is more practical to strive for something halfway - defining specific conditions to limit ‘damage’ to the map. we can also better clarify how the different railway tags relate to each other while we are at it.
Haven’t been following the discussion really but I’m quite in favor of having railway=razed on OHM and to mark railways that have been demolished but still are visible (because they’ve been overbuilt with cycleway, etc.) with railway=abandoned.
Talking about things as much as we do is a good direction. There is a lot of listening here, too; good.
There is a lot going on, substantially one of the dimension of time, that it is important we “get right.” So far, so good, I say. It’s tedious at times, yes, that’s OK.
There is a path building to do this. For “beginners” and more-precise mappers alike. It might start with a sketched and agreed-upon ideal and pilot projects and movements towards that, it might be something else.
It’s a lot for our project to grok (understand in a shared way) at once. Visually, in our language, in our centuries of shared technology (such as rail), in how we put the pieces and building blocks together. By we, I mean OSM.
Yet, we’re doing it. I appreciate the shared patience everybody shows, especially.
Keep talking.
Yes, Mateusz has a strong opinion. I would disagree about “extremely” in #1 because I also have a strong opinion.
Might we find a consensus that railway=razed and railway=dismantled mean exactly the same thing?
“Razed”, “removed”, “dismantled” and “demolished” all mean the same thing.
there is a distinction made between 1 remove just the tracks, 2 keep tracks and let everything rot without doing anything, or 3 remove tracks and re/move ballast and move the ground.
I have always understood it that way:
I would summarise all three of the above points today in the liefcycle scheme under abandoned
Fourth point and completely different:
And razed is - as with other applications of the lifecycle scheme - to be used as long as there is a risk that armchair mappers could redraw this from old aerial photographs.
Translated with DeepL.com (free version)