Best practices regarding implied tags

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
12 messages Options
Reply | Threaded
Open this post in threaded view
|

Best practices regarding implied tags

François Lacombe-2
Hi all,

This proposal is currently in RFC

It proposes among other points to make man_made=utility_pole + utility=power implied by power=pole (for sake of consistency with telecom utility poles which won't get a telecom=pole because they're not a telecom feature)

A good point is raised in Talk regarding this implication.
It's clear that implied tags won't be mandatory on osm features but what about actually adding them ?
Is that completely wrong or mappers could eventually add implied tags if they want to?
The proposal currently states they are optional and it won't raise an error if mappers add them beside mandatory tags.

Let me know if a common practice is already established.

All the best

François

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

Paul Johnson-3
On Wed, Sep 16, 2020 at 5:20 PM François Lacombe <[hidden email]> wrote:
Is that completely wrong or mappers could eventually add implied tags if they want to?
The proposal currently states they are optional and it won't raise an error if mappers add them beside mandatory tags.

No, it's not wrong to add implied tags explicitly.  It's actually encouraged in some cases where the implicit tag is not consumable by automated system (such as the "none" default for turn:lanes tends to be ambiguous between "you can't turn from this lane" and "you can't use this lane" and "there's an implicit but unspecified implication that isn't painted on the ground"), or access defaults (such as in the US where bicycle=* and foot=* varies a lot on highway=motorway)

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

Matthew Woehlke
On 16/09/2020 18.32, Paul Johnson wrote:
> No, it's not wrong to add implied tags explicitly.  It's actually
> encouraged in some cases where the implicit tag is not consumable by
> automated system (such as the "none" default for turn:lanes tends to be
> ambiguous between "you can't turn from this lane" and "you can't use this
> lane" and "there's an implicit but unspecified implication that isn't
> painted on the ground")

Pedantic: wouldn't "you can't turn from this lane" be correctly
specified as turn:lanes=through? As I understand turn:lanes, "none"
would be "you can't use this lane". (Also pedantically speaking, a blank
value would mean there are no specific markings. I think the only
ambiguity here is that it's unclear if the tag is simply missing — in
which case the truth on the ground could be *anything* — or if there are
no markings. Sort of like how a missing oneway could mean oneway=no, or
could mean oneway but not tagged.)

(Incidentally, I tend to add oneway=no whenever possible... or at least
I did in iD, which made it easy. I can't recall now how well I've been
keeping that up with JOSM.)

--
Matthew

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

Kevin Broderick
In reply to this post by Paul Johnson-3
+1.

Explicit tagging indicates a level of confidence not generally associated with implicit tagging. While there's certainly an 'ad nauseum' level of doing so (e.g. adding surface=paved, motor_vehicle=yes to highway=motorway in the U.S. would be kinda silly, IMO), there are plenty of cases where a primary tag generally implies something about the tagged object but doesn't guarantee it. I'd point to the recent discussion of access= on driveways as an example; while most driveways allow for certain types of access by default, it's far from guaranteed—there may be a no-trespassing sign or a locked gate, and explicitly indicating the lack of such in the access tagging is helpful. (Adding the implied value without survey or other definitive knowledge is not, as then you express a higher degree of confidence than actually exists in the data).

On Wed, Sep 16, 2020 at 6:34 PM Paul Johnson <[hidden email]> wrote:
On Wed, Sep 16, 2020 at 5:20 PM François Lacombe <[hidden email]> wrote:
Is that completely wrong or mappers could eventually add implied tags if they want to?
The proposal currently states they are optional and it won't raise an error if mappers add them beside mandatory tags.

No, it's not wrong to add implied tags explicitly.  It's actually encouraged in some cases where the implicit tag is not consumable by automated system (such as the "none" default for turn:lanes tends to be ambiguous between "you can't turn from this lane" and "you can't use this lane" and "there's an implicit but unspecified implication that isn't painted on the ground"), or access defaults (such as in the US where bicycle=* and foot=* varies a lot on highway=motorway)
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging


--
Kevin Broderick

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

François Lacombe-2
Thank you all for replies

Then the current proposal sounds to be ok regarding what is said upside.
I admit to automatically adding implied tags when importing data covered by the proposal, so no apparent problem is mappers add them explicitly.

All the best

François

Le jeu. 17 sept. 2020 à 15:11, Kevin Broderick <[hidden email]> a écrit :
+1.

Explicit tagging indicates a level of confidence not generally associated with implicit tagging. While there's certainly an 'ad nauseum' level of doing so (e.g. adding surface=paved, motor_vehicle=yes to highway=motorway in the U.S. would be kinda silly, IMO), there are plenty of cases where a primary tag generally implies something about the tagged object but doesn't guarantee it. I'd point to the recent discussion of access= on driveways as an example; while most driveways allow for certain types of access by default, it's far from guaranteed—there may be a no-trespassing sign or a locked gate, and explicitly indicating the lack of such in the access tagging is helpful. (Adding the implied value without survey or other definitive knowledge is not, as then you express a higher degree of confidence than actually exists in the data).

On Wed, Sep 16, 2020 at 6:34 PM Paul Johnson <[hidden email]> wrote:
On Wed, Sep 16, 2020 at 5:20 PM François Lacombe <[hidden email]> wrote:
Is that completely wrong or mappers could eventually add implied tags if they want to?
The proposal currently states they are optional and it won't raise an error if mappers add them beside mandatory tags.

No, it's not wrong to add implied tags explicitly.  It's actually encouraged in some cases where the implicit tag is not consumable by automated system (such as the "none" default for turn:lanes tends to be ambiguous between "you can't turn from this lane" and "you can't use this lane" and "there's an implicit but unspecified implication that isn't painted on the ground"), or access defaults (such as in the US where bicycle=* and foot=* varies a lot on highway=motorway)
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging


--
Kevin Broderick
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

Joseph Eisenberg
The previous responses are focusing on the benefit of adding explicit tags in situations where the current tagging is ambiguous.

Certainly there is a benefit of adding "oneway=no" on all two-way roads and "oneway=yes" on motorways to make the situation explicit.

But the original question was about whether or not we should add "man_made=utility_pole" + "utility=power" to current power poles.

These are currently tagged "power=pole" which is clearly defined as a power utility pole, so adding the two other tags does not provide any information.

Does anyone think that it is a good idea to add those two new tags in this particular situation?

-Joseph Eisenberg

On Sun, Sep 20, 2020 at 9:46 AM François Lacombe <[hidden email]> wrote:
Thank you all for replies

Then the current proposal sounds to be ok regarding what is said upside.
I admit to automatically adding implied tags when importing data covered by the proposal, so no apparent problem is mappers add them explicitly.

All the best

François

Le jeu. 17 sept. 2020 à 15:11, Kevin Broderick <[hidden email]> a écrit :
+1.

Explicit tagging indicates a level of confidence not generally associated with implicit tagging. While there's certainly an 'ad nauseum' level of doing so (e.g. adding surface=paved, motor_vehicle=yes to highway=motorway in the U.S. would be kinda silly, IMO), there are plenty of cases where a primary tag generally implies something about the tagged object but doesn't guarantee it. I'd point to the recent discussion of access= on driveways as an example; while most driveways allow for certain types of access by default, it's far from guaranteed—there may be a no-trespassing sign or a locked gate, and explicitly indicating the lack of such in the access tagging is helpful. (Adding the implied value without survey or other definitive knowledge is not, as then you express a higher degree of confidence than actually exists in the data).

On Wed, Sep 16, 2020 at 6:34 PM Paul Johnson <[hidden email]> wrote:
On Wed, Sep 16, 2020 at 5:20 PM François Lacombe <[hidden email]> wrote:
Is that completely wrong or mappers could eventually add implied tags if they want to?
The proposal currently states they are optional and it won't raise an error if mappers add them beside mandatory tags.

No, it's not wrong to add implied tags explicitly.  It's actually encouraged in some cases where the implicit tag is not consumable by automated system (such as the "none" default for turn:lanes tends to be ambiguous between "you can't turn from this lane" and "you can't use this lane" and "there's an implicit but unspecified implication that isn't painted on the ground"), or access defaults (such as in the US where bicycle=* and foot=* varies a lot on highway=motorway)
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging


--
Kevin Broderick
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

dieterdreist


sent from a phone

> On 20. Sep 2020, at 18:59, Joseph Eisenberg <[hidden email]> wrote:
>
> Does anyone think that it is a good idea to add those two new tags in this particular situation?


while I am personally not unsatisfied with power=pole I could understand that people who want to deprecate this tag in favor of man_made=utility_pole (17200 as of yesterday)
would want to add it.

I could see some benefit from such a switch (better supports multiple utilities on the same pole, no need to decide on kind of utility if you don’t know it).

Cheers Martin



_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

dieterdreist
In reply to this post by Joseph Eisenberg


sent from a phone

> On 20. Sep 2020, at 18:59, Joseph Eisenberg <[hidden email]> wrote:
>
> Does anyone think that it is a good idea to add those two new tags in this particular situation?


utility=power seems to be a redundant concept in general (you can see which kind of lines are attached - if they are mapped), here all the more.

Cheers Martin
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

Paul Johnson-3
In reply to this post by Joseph Eisenberg


On Sun, Sep 20, 2020 at 11:58 AM Joseph Eisenberg <[hidden email]> wrote:
The previous responses are focusing on the benefit of adding explicit tags in situations where the current tagging is ambiguous.

Certainly there is a benefit of adding "oneway=no" on all two-way roads and "oneway=yes" on motorways to make the situation explicit.

But the original question was about whether or not we should add "man_made=utility_pole" + "utility=power" to current power poles.

Well, does narrow it down from a neighborhood pole that might have cable television or carry the PSTN.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

Janko Mihelić
There are a lot of big power towers that carry an optical communications line together with the power lines. Would that be utility=power;communication?

Adding specific implied information is not wrong, but data consumers shouldn't rely on them. If someone changes utility=power to utility=communication;power, or if someone outright deletes the utility tag, that power pole is still correctly tagged.

Janko

ned, 20. ruj 2020. u 20:30 Paul Johnson <[hidden email]> napisao je:


On Sun, Sep 20, 2020 at 11:58 AM Joseph Eisenberg <[hidden email]> wrote:
The previous responses are focusing on the benefit of adding explicit tags in situations where the current tagging is ambiguous.

Certainly there is a benefit of adding "oneway=no" on all two-way roads and "oneway=yes" on motorways to make the situation explicit.

But the original question was about whether or not we should add "man_made=utility_pole" + "utility=power" to current power poles.

Well, does narrow it down from a neighborhood pole that might have cable television or carry the PSTN.
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

dieterdreist

sent from a phone

> On 21. Sep 2020, at 15:36, Janko Mihelić <[hidden email]> wrote:
>
> or if someone outright deletes the utility tag, that power pole is still correctly tagged.


if it is a power pole, why would you remove the utility tag?
When there’s a highway=track and you remove the tracktype tag the object also will still be correctly tagged :)

Cheers Martin


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Best practices regarding implied tags

Janko Mihelić
On Mon, Sep 21, 2020, 16:00 Martin Koppenhoefer <[hidden email]> wrote:
if it is a power pole, why would you remove the utility tag?
When there’s a highway=track and you remove the tracktype tag the object also will still be correctly tagged :)

You're right, I meant the whole information is still there. Oneway=no says "I was there, and it's definitely twoway" so if you remove that tag, you lose that. But I feel you lose nothing when you remove usage=power, because it adds no additional meaning or even meta information. Everybody knew its usage was power.

I would never remove the tag, I was just talking about tags in general.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging