Talk:Trolltag

Page name/term too offensive?

The idea behind this page is great, too much additional tags have poped to change the original meaning, and explaining the problem with examples about data usage is great. But using the term "trolltag" is quite offensive and not really friendly (Though I find it fun and easy to remember, the goal of this page could be to link from problematic tag pages, and might not be warmly welcome with such a name !) . Would you mind if I change the page's name/title to make it more acceptable in that regard ? What about (few, I'm short on ideas) : Meaning changing tags ? any other ideas are welcome. sletuffe (talk) 01:22, 15 February 2016 (UTC)

Not opposed and not sure it is needed to find a less offensive name either. The term "trolltag" has been discussed in the mailing list quite hotly so it has a certain value as an easy to recognize term. Perhaps asking the osm-talk ml could help to find a friendlier name and also make people aware of the issue.

The difficulty of saying something is a trolltag lies more in the exact circumstances: "end_date" is a problematic tag if not used with great care but there are enough cases where it can be used without any problem. RicoZ (talk) 11:01, 15 February 2016 (UTC)

Whether intended or not, people do take offence when they see the term used. In the interest of more effective communication it'd make sense to use a different term. Perhaps "tagging 'false friends'" (as per https://en.wikipedia.org/wiki/False_friend )? --SomeoneElse (talk) 21:23, 1 August 2016 (UTC)

"False friends" may also be considered offensive by someone not familiar with the term and the well defined meaning of it doesn't fit well here. Finding a better name is not quite easy... something describing the technical problem would be nice but would seem too lengthy for a title. "Treacherous Tags"? "False attribute"? RicoZ (talk) 12:24, 2 August 2016 (UTC)

How about "turnabout tags"? Since one key-value pair leads a data consumer to one understanding about the situation of the tagged feature, but another pair turns that about and leads to a different, in some way contradictory (rather than merely clarifying) understanding. And in this case, turnabout is not fair play... :) Arlo James Barnes (talk) 19:45, 28 February 2019 (UTC)

"using the term "trolltag" is quite offensive" - for whom? Mateusz Konieczny (talk) 10:49, 7 August 2023 (UTC)

"Broken tags" might be OK. Something B (talk) 17:46, 26 February 2025 (UTC)

"Proposed tags are a really poor idea..."

RicoZ, I know it took me five years to address this, but I must take issue with your "tags like proposed=yes are a really poor idea." The USBRS in OSM uses state=proposed in a controlled, deliberate way, with excellent results. Its use has good merits. Yes, "proposed" can be abused, but when its use is well-defined and used within parameters, it makes sense to both use it and render it. Stevea (talk) 23:53, 28 September 2021 (UTC)

I think you miss the point. Trolltags are not a problem because "proposed" information is "always worthless", it is a problem because they invert the meaning of the primary tag. So, it makes all data unreliable unless every data consumer (of which there is unknown number) understands all possible tags that anyone can invent at any time (of which there is also unknown number). Making all data consumer understand all possibly existing tags is totally unattainable in practice, and thus using trolltags necessarily makes every data consumer worse at doing its job.

Sure, state=proposed may work your one preferred data consumer and you might not care that 90% of others are going to give invalid results, but even that one data consumer you prefer doesn't support all of possible tags, like operational_status=under_construction, proposed=yes, active=no, inactive=yes, usability=no, proposal=yes, etc. -- simply because it is impossible to support all new tags (of which there might be infinitely many), and even if you looked into all database and understood and supported all existing tags today (which is impossible), tomorrow there could be a million new tags for you to support, so it create possible infinite (and completely unnecessary) workload. And just because something is documented somewhere in the wiki does not mean all world will know about it and implement it.

Thus, proper way of dealing with such "proposed" stuff would be using lifecycle prefix: e.g. instead of using route=bicycle + state=proposed, use proposed:route=bicycle instead. That way, same information is recorded, but no data consumer is ever going to be confused by mistaking invalid features as valid ones (because searching for "route=*" won't find "proposed:route=*"). And it also allows for recording more data in unambiguous way, e.g. if route is valid, but it is only new ref=* for it that is being proposed, you could use route=bicycle + proposed:ref=* which is unambiguous (as opposed to route=bicycle + state=proposed + ref=* which is unclear whether is only new reference number that is being proposed, or it is the whole route that is being proposed). In other words, even when data consumer doesn't support some lifecycle prefix, it will always fail in a safe way (as opposed to using trolltag which, when unknown, will fail in an unsafe way). --mnalis (talk) 10:40, 23 April 2025 (UTC)

Out of order

There needs to be a tag that says a e.g., drinking fountain, still very valuable as a landmark, is currently out of order, but has hopes of one day being repaired.

That way users could still find it on a map, but would know not to expect water.

It is different that one permanently broken.

OK, Key:operational_status#Values_in_use still has

operational_status=needs_maintenance
operational_status=broken

I can use.

Jidanni (talk) 07:06, 7 August 2023 (UTC)

We have disused:amenity=drinking_water for "would know not to expect water" - and operational_status=broken is a problematic trolltag inferior to that solution Mateusz Konieczny (talk) 10:40, 7 August 2023 (UTC)