Hi all,
I propose to remove all ‘memorial:conflict’ tags and replace them with ‘subject’.
The reason for this is that subject does cover this in a more general way. Or are we also gonna make a memorial:person, memorial:animal, memorial:event,… tag?
In tandem, I’d like to propose a mechanical edit to change them all in bulk and - if possible - to use the wikidata version for some common values (e.g. WW2)
Not sure about this.
Here for example a memorial I tagged myself: Node: Monumento alla Resistenza (2012212126) | OpenStreetMap
I tagged the subject as the italian resistance movement, which operated during WW2, which I put in memorial:conflict.
Another example, this memorial has the soldier as subject, and the conflict he fought in as memorial:conflict: Node: С. П. Васильеву (6694676764) | OpenStreetMap
ecc.ecc.
Sorry, but I’m not convinced about your examples - even though they are an unexpected use case and means that special care should be given when retagging all of them.
The memorials indeed remember someone who fought during the world war, and the memorials are placed because of that.
However, the wikidata entities often state that they were ‘resistance fighters’ during this war; hence the fact that it is an implicit WW-memorial can be deduced by combining the datasets.
An alternative option is to use subject:wikidata=Q706633;Q362, which then indicates that the memorial is both about the person and WW2
that means that you now need to use Wikidata, which is currently not needed with memorial:conflict | Keys | OpenStreetMap Taginfo
Also, you can have memorial dedicated to someone who fought in more than one war.
that means that you now need to use Wikidata
Which is implicitly already the case. I’ve encountered a few memorials with memorial:conflict=wk1;wk2 in Germany, which I assume is an abbreviation for Wereldkrieg - or - World-War - which is why memorial:conflict:wikidata is in use as well. Using Wikipedia/Wikidata solves this neatly.
Also, you can have memorial dedicated to someone who fought in more than one war.
Which means that ‘memorial:conflict’ is used as ‘this is the context for this memorial’. I understand there is some need to give context, but we should then move this to a proper tag and not abuse it, using values as Holocaust and Stalinism. I mean no disrespect to the respective victims, but calling ‘Stalinism’ a conflict is a stretch, to say the least.
At last, it is my opinion that trying to state a context for a memorial is - in some cases - subjective. If a memorial lists a number of names with as text “In memory of people deported by the Nazi’s between 1940-1945”, it is very hard to now if these were people deported due to political reasons (jews, gays, gipsies), due to military reasons (soldiers) or due to economical reasons (e.g. forced labourers). Would the context then have to be WW2, Holocaust, Nazism or something else? What if the people listed were deported for various reasons? How can we objectively verify this, while only seeing the memorial itself?
For this, I propose we only map the physical attributes of memorial (e.g. inscription, type, …) and leave the context building to WIkidata - where there is also room for references to sources and links to other entities.
holocaust has a few more uses (205), but “stalinism” has actually 7 out of 6554 current uses, don’t pick stuff from the long tail in order to justify deprecation.
IMHO, the answer to your initial proposal “I propose to remove all ‘memorial:conflict’ tags and replace them with ‘subject’.” was already given, with examples where “subject” was already used for a different purpose on the same object as memorial:conflict.
This also applies to the second proposal: “In tandem, I’d like to propose a mechanical edit to change them all in bulk”.
I’m really struggling to understand the harm of this tag. I don’t like the idea of relying completely on another dataset (like Wikidata) when we already have our own data. Now, if I want to search for WW1 memorials, I can simply run an Overpass query. If I were to combine the two datasets, I honestly wouldn’t have a clue how to do it.
Not to mention that Wikidata is not always more complete than OpenStreetMap. Sometimes the OSM element has more details, sometimes the Wikidata one does. So having both can be an advantage.
I don’t know, but that’s not a problem of ours at the moment since we don’t have a context=* tag. The memorial:conflict=* value should be =WW2 instead. I agree with you that holocaust, stalinism, etc., should not be used as a memorial:conflict=* values. However, in my opinion, that should be a reason for discussing cleaning up the values or better documenting the tag, not deprecating the whole tag.
I’m really struggling to understand the harm of this tag
The harm is that two tags are used (subject and memorial:conflict) for (at first glance) the same thing: namely what the memorial remembers.
Some have pointed out that this tag should rather be understood as a context to amend the subject tag - but a quick overpass-query pointed out that only a handful of items have both memorial:conflict and subject (like not even 10 items) or subject:wikidata (<500).
However, in my opinion, that should be a reason for discussing cleaning up the values or better documenting the tag, not deprecating the whole tag.
I just see no added benefit of having a tag specifically for conflicts as we have a more general tag that fills this need, but does it better.
During the retagging, a cleanup should happen as well. But, for this cleanup, directly linking wikidata would help as well.
(with a data consumer hat on) that really isn’t a problem, unless both have values and they contradict. It’s trivial to primarily look for “tag A” for a feature and if not set, then look for “tag B”.
With MapComplete, I’m a data consumer and contributor.
And it is not really a problem, but we can do better then this, which benefits us in the long run. My instincts are to remove unneeded complexity as much as possible, which this tag is.
Consider that I would propose a tag memorial:natural_disaster with values like Tsunami 2011, Earthquake 2019,… what would your opinion be of that?
I don’t think it would be a problem. One could completely ignore it if
these specifics are not of interest. In some way, one could also just
use the key and say, oh, it has a memorial:natural_disaster tag, so it
is a memorial for a natural disaster, and if you were interested in
memorials about a specific natural disaster, this tag would make it
very easy to find them. Go for it 
I’m also not sure subject and memorial:conflict are quite the same thing, even beyond cases where the memorial is for a particular person who served in a conflict. For the Civil War memorial in my hometown (OSM Node 8276536201), the memorial is more specifically to people from the town who died in the conflict (and who are listed on the sides of it, though I’ve haven’t entered those sides as an “inscription” into OSM), not just the war in general.
Now, maybe the “right” approach would be for there to be something like a wikidata item for “soldiers from Charlton who died in the American Civil War”, which would be the subject:wikidata, and that wikidata item would itself somehow link to the information on the war itself.
But even for more general war memorials that aren’t intended for a specific subset of people, I would think that their subject is usually more “the people who died in a conflict” rather than “the conflict itself”. So I think there’s a bit of nuance that makes just moving memorial:conflict to subject not quite right, though it may be close enough for many purposes.
Further to Talk:Key:memorial:conflict - OpenStreetMap Wiki , whereas Proposal:Subject - OpenStreetMap Wiki has this format
memorial:subject=Infanterie-Leib-Regiment Großherzogin Nr. 117
memorial:subject:wikidata=Q1662273
artwork:subject=Hessian Lion
subject=World War I
I suggest following other historic= eg historic:subject= to be used for the background, to emphasize it refers to the view of in the grand scale of history. historic:period= may be too wide, and requires historic:civilization= awkward in modern globalized history. subject= can then be used for individual battles or events.