importance=* tag (for transportation etc)

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
42 messages Options
123
Reply | Threaded
Open this post in threaded view
|

importance=* tag (for transportation etc)

Daniel Koć
I have just read on WeeklyOSM that OpenRailwayMap may start to use
importance=* tag for ranking railway stations instead of
railway:station_category=* :

http://lists.openrailwaymap.org/archives/openrailwaymap/2016-March/000408.html

The proposition is 7 years old:

http://wiki.openstreetmap.org/wiki/Proposed_Features/Importance

and is quite generic. I think it would help for example with rendering
airports without resorting to less clear properties like size:

https://github.com/gravitystorm/openstreetmap-carto/pull/1734

and it would be good to have universal scheme instead of
railway:station_category=* or flights_range=*. Looks like it is being
already used and quite popular:

http://taginfo.openstreetmap.org/keys/importance

What do you think about making this scheme official?

--
"Завтра, завтра всё кончится!" [Ф. Достоевский]


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

Re: importance=* tag (for transportation etc)

Chris Hill-6
On 18/03/16 17:38, Daniel Koć wrote:

> I have just read on WeeklyOSM that OpenRailwayMap may start to use
> importance=* tag for ranking railway stations instead of
> railway:station_category=* :
>
> http://lists.openrailwaymap.org/archives/openrailwaymap/2016-March/000408.html 
>
>
> The proposition is 7 years old:
>
> http://wiki.openstreetmap.org/wiki/Proposed_Features/Importance
>
> and is quite generic. I think it would help for example with rendering
> airports without resorting to less clear properties like size:
>
> https://github.com/gravitystorm/openstreetmap-carto/pull/1734
>
> and it would be good to have universal scheme instead of
> railway:station_category=* or flights_range=*. Looks like it is being
> already used and quite popular:
>
> http://taginfo.openstreetmap.org/keys/importance
>
> What do you think about making this scheme official?
>
There are no official tags. Only tags that are used and / or documented.

--
Cheers, Chris (chillly)


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

Re: importance=* tag (for transportation etc)

Daniel Koć
W dniu 18.03.2016 18:50, Chris Hill napisał(a):

> There are no official tags. Only tags that are used and / or
> documented.

I understand it, just used informal wording. I meant "accepted by voting
and documented as such on Wiki", which is - well - longer.

--
"Завтра, завтра всё кончится!" [Ф. Достоевский]

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

Re: importance=* tag (for transportation etc)

John Willis
In reply to this post by Daniel Koć
I was told point-blank by the head of OSM-carto on github That (as I remember it) 


A) "importance" is unverifiable, so it is useless for OSM. 

Gravitystorm:
"Importance' and related concepts fails the absolutely vital verifiability test, so it's not a suitable concept for OSM."

B) I assume then it will not be supported in -carto renderings. 

My example us d mountains in the post, as I had pictures to illustrate the issue. 

I wanted importance=local-city-regional-national-international or similar tag to put on mountains to control when it is rendered (So Mt Fuji and Mt Everest is rendered at a very low zoom level, mt Rushmore is rendered at a little closer zoom, San Gregornio (in Los Angeles) is rendered at a regional level, Mt Helix on a city level, and My tiny, unimportant little peak next to Mt Helix is rendered at a very local level (Grossmont Peak). 

This allows someone to say "these little named points all over the top of Mt Fuji (there are 7) should only be rendered locally (high zoom) and the Mt Fuji volcano icon should be rendered at z8 or something.  

OSM wants local knowledge, per this idea, but not the kind that could lead to better rendered maps or better routing. 

Javbw

On Mar 19, 2016, at 2:38 AM, Daniel Koć <[hidden email]> wrote:

I have just read on WeeklyOSM that OpenRailwayMap may start to use importance=* tag for ranking railway stations instead of railway:station_category=* :

http://lists.openrailwaymap.org/archives/openrailwaymap/2016-March/000408.html

The proposition is 7 years old:

http://wiki.openstreetmap.org/wiki/Proposed_Features/Importance

and is quite generic. I think it would help for example with rendering airports without resorting to less clear properties like size:

https://github.com/gravitystorm/openstreetmap-carto/pull/1734

and it would be good to have universal scheme instead of railway:station_category=* or flights_range=*. Looks like it is being already used and quite popular:

http://taginfo.openstreetmap.org/keys/importance

What do you think about making this scheme official?

--
"Завтра, завтра всё кончится!" [Ф. Достоевский]


_______________________________________________
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: importance=* tag (for transportation etc)

Michael Reichert
In reply to this post by Daniel Koć
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Hi,

Am 18.03.2016 um 18:38 schrieb Daniel Koć:
> and it would be good to have universal scheme instead of
> railway:station_category=* or flights_range=*. Looks like it is
> being already used and quite popular:

You know that you usually have more train stations than airports per
square kilometre, don't you? :-) That's why creating on tagging scheme
for both might be difficult.

> http://taginfo.openstreetmap.org/keys/importance

importance=* is mainly used in France. Usage in Germany was brought to
Germany by French mappers who mapped international TGV route
relations. It is only (with some few exceptions) used on railway
tracks, but not on stations.

http://overpass-turbo.eu/s/f6Q

> What do you think about making this scheme official?

Have you read the objections at OpenRailwayMap mailing list? Especially:
http://lists.openrailwaymap.org/archives/openrailwaymap/2016-March/00041
0.html
http://lists.openrailwaymap.org/archives/openrailwaymap/2016-March/00042
4.html
http://lists.openrailwaymap.org/archives/openrailwaymap/2016-March/00042
6.html

The old importance proposal might fit well for small countries like
Belgium, the Netherland, Austria or Switzerland. But it is not
suitable for larg countries (Germany, US) and islands (UK). How much
stations in UK or US have international trains?

Best regards

Michael


- --
Per E-Mail kommuniziere ich bevorzugt GPG-verschlüsselt. (Mailinglisten
ausgenommen)
I prefer GPG encryption of emails. (does not apply on mailing lists)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=d/T7
-----END PGP SIGNATURE-----

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

Re: importance=* tag (for transportation etc)

Michael Reichert
In reply to this post by John Willis
Hi John,

Am 18.03.2016 um 22:36 schrieb John Willis:
> I was told point-blank by the head of OSM-carto on github That (as I remember it)
>
> https://github.com/gravitystorm/openstreetmap-carto/issues/323
>
> A) "importance" is unverifiable, so it is useless for OSM.
>
> Gravitystorm:
> "Importance' and related concepts fails the absolutely vital verifiability test, so it's not a suitable concept for OSM."

I agree that an importance tag for mountains is not a suitable concept
but a importance tag for train stations (or airports) is surveyable and
suitable for OSM. Just take the timetable or go out and stay one day on
the platforms, count and note down all stopping trains. In addition,
more important stations often have better/more facilities for
passengerts like a ticket shop (smaller ones only have vending
machines), a toilet, backeries, fast food stores, waiting rooms, etc.

If OSM would free of any importance-like tags, we would not have the
highway=* tag as we have it now. Tagging is highway=primary vs.
secondary, secondary vs. tertiary, tertiary vs. unclassified is often a
question of importance, not only width, paving and lane count.

Best regards

Michael


--
Per E-Mail kommuniziere ich bevorzugt GPG-verschlüsselt. (Mailinglisten
ausgenommen)
I prefer GPG encryption of emails. (does not apply on mailing lists)


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

signature.asc (836 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: importance=* tag (for transportation etc)

John Willis




Javbw
On Mar 19, 2016, at 7:15 AM, Michael Reichert <[hidden email]> wrote:

I agree that an importance tag for mountains is not a suitable concept

So displaying more important train stations that:

- are more well known, so people would look for them.
- are popular points for people (for stations, more amenities and trains), so marking the ones more likely to be the ones people want (AKA the big station on the big line vs the little unimportant station on the nearby line that has a similar name)

-are better to be mapped because they provide spatial awareness because their location is widely known 

Applies equally to mountains or stations. 

In Japan, there is a national joke about people showing up in my (tiny) town town to climb my Fuji because the name of the station in my town is "below Mt Fuji" in Chinese. The tiny 6 story tall mountain is named my Fuji, as are hundreds of other little hills and mountains in Japan. 

Google used it in a national ad campaign to promote "ok Google", showing a bunch of Americans showing up at the wrong station, and some people could help them (thanks to google translation or whatever). 


I'm sure the popular train stations near Fuji and the little local stop for grandmas and students to go to the grocery store would have different importance rankings. 

Similarly,

Hiding the Mt Fuji red volcano icon under orange little icons for its little labeled points Obscures the more important label of Mt Fuji with labels and icons 99% of people don't care about, right?

Hiding the rendered label of a very important station because it has some other nearby local stations that crowd out its rendering seems like the same thing to me. 

Javbw 

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

Re: importance=* tag (for transportation etc)

dieterdreist
In reply to this post by John Willis


sent from a phone

> Am 18.03.2016 um 22:36 schrieb John Willis <[hidden email]>:
>
> OSM wants local knowledge, per this idea, but not the kind that could lead to better rendered maps or better routing.



importance is relative and it depends on your criteria which things you consider more important than others. I agree that we should better map the properties that make something important rather than stuff it all in a generalized "importance" tag. For mountains this will likely mean to look at prominence and topographic isolation, for train stations it can be amount of passengers or trains per time, or size (amount of platforms), availability of highspeed trains, etc., for airports similarly you could look at amount and size of runway/s, passengers/time, amount of connections etc.
the problem with rendering is that these calculations are typically too expensive to do them on the fly, but we could use precomputed data and integrate it in a coastline like way (external shape file or pre-computed attribute)


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

Re: importance=* tag (for transportation etc)

John Willis

On Mar 19, 2016, at 10:13 AM, Martin Koppenhoefer <[hidden email]> wrote:

prominence and topographic isolation,

Neither are good measures of mountains, besides for record holders. 

- There are bigger volcanoes than Mt Fuji  in Russia, just north of Japan, that no one knows the names of (internationally). They are equally isolated. Klyuchevskaya Sopka is over 4200m (fuji is 3776), and equally as isolated as Mt Fuji - and no one outside of that region knows it’s name. 

Who can name one of the other 12 peaks in the Sierra Nevada Mountains in California that are over 14,000 feet tall, within 500 feet as tall as Mt Whitney (14,505)? I can name Mt Langley, but that is about it. 

- Mt fuji is climbed by 100X (?) more people during climbing season than Everest. So should Mt Everest be rendered later? What About Denali? Few people climb it. 


But these are all record holding - international, national, or regional mountains - this idea of mapping mountains via prominence or topography completely and utterly fails at a provincial level. 

Regionally and provincially important mountains are often more important than their taller neighbors due to their proximity to towns, or odd shapes - not any height or isolation. Their proximity to the towns ingrains them into the culture, through naming, religious significance, or tourism reasons.

Right next to Mt Fuji is a collapsed volcano and caldera called Mt Hakone https://goo.gl/maps/hNSC9NwsHg42 . it is very short now, and not nearly as prominent as nearby Mt Ashitaka or (of course) Mt Fuji. But Hakone is a very famous place - though it’s height and prominence would say otherwise. People all over Japan (and many international tourists) come there buy eggs cooked in sulfurous vents and enjoy the hot spring resorts inside the caldera. 

In my region, Mt Akagi is famous. https://goo.gl/maps/8A5STm9VwAs & https://en.wikipedia.org/wiki/Mount_Akagi . WWII buffs may recognize the name, as the Carrier Akagi (lost at the battle of midway) is named after it. It is the namesake of hundreds, perhaps thousands of places and things (I drink "Mt. Akaki” Sake).  However, Mt Kessamaru is higher than Mt Akagi nearby. Most people don’t know of it, nor care. This mountain, and two other visible, but low mountains are called the “three mountains” of my provience - though they are surrounded by taller ones. And the little points around the caldera (some of which Google renders alongside Mt Akagi’s label) are only locally known, and shouldn’t be rendered except at high zoom. 

OSM is for gathering data - lots of lots of locally based knowledge of things. Mountains are no different. Trying to decide what mountains are worth labeling at different zooms via some GIS data is ridiculous. 

So we render them all equally - which is equally as ridiculous. 

So we will never have a better map / map data than the random GIS data that everyone already has and already uses to make inferior, confusing maps - which is what I’m trying to fix in OSM.  

Javbw


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

Re: importance=* tag (for transportation etc)

dieterdreist


sent from a phone

Am 19.03.2016 um 08:41 schrieb johnw <[hidden email]>:


first, those internationally unknown volcanoes in Russia won't compete with your Japanese mountains, because they're too far away, what I suggested was aimed at deciding locally what to show/label, not necessarily compare significance on a global level.



Right next to Mt Fuji is a collapsed volcano and caldera called Mt Hakone https://goo.gl/maps/hNSC9NwsHg42 . it is very short now, and not nearly as prominent as nearby Mt Ashitaka or (of course) Mt Fuji.


the fact that it is a volcano might already come into play when deciding what to render/label


But Hakone is a very famous place - though it’s height and prominence would say otherwise. People all over Japan (and many international tourists) come there buy eggs cooked in sulfurous vents and enjoy the hot spring resorts inside the caldera. 


I'd say: tourism=attraction for this volcano (admittedly another way of saying important=yes), the presence of the resorts also indicates importance. Is this checkable automatically? Not sure (of course you can check for this if you know what you're looking for, but a different place might have completely different reasons to be "important")

cheers,
Martin 


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

Re: importance=* tag (for transportation etc)

Andy Mabbett
In reply to this post by John Willis
On 19 March 2016 at 07:41, johnw <[hidden email]> wrote:

> OSM is for gathering data - lots of lots of locally based knowledge of
> things. Mountains are no different. Trying to decide what mountains are
> worth labeling at different zooms via some GIS data is ridiculous.

It's nowhere near as ridiculous as trying to render them according to
some arbitrary and subjective "importance" (Importance to whom? The
people who live near them? Tourists? Mountaineers? Ornithologists?
Aviators? Geologists? Climatologists? Oil prospectors?).

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

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

Re: importance=* tag (for transportation etc)

Richard Fairhurst
Andy Mabbett wrote:
> It's nowhere near as ridiculous as trying to render them according
> to some arbitrary and subjective "importance" (Importance to
> whom? The people who live near them? Tourists? Mountaineers?
> Ornithologists? Aviators? Geologists? Climatologists? Oil
> prospectors?).

Exactly.

Here's an example: https://en.wikipedia.org/wiki/Sg%C3%B9rr_Dearg

This is the notorious "Inaccessible Pinnacle". If you're a mountaineer, specifically a Munro bagger, it's a highly significant peak: it's the hardest Munro (Scottish peak over 3000ft) to get. If you're a tourist looking for pretty mountains, though, it's probably not significant; it's just, well, a bit of rock. Wider cultural significance? I couldn't tell you. Somewhere between the two: certainly less than Ben Nevis, but how do you decide the "importance" of the hardest peak to climb in Scotland which just happens to be an anonymous lump of rock?

Importance means value judgements. One of the reasons OSM is so successful is that our data doesn't make value judgements. This allows people to make their own maps with their own value judgements. This is why OSM has become, from nowhere, the world's pre-eminent geodata source for walking and cycling - because every other dataset is car-biased. Let's not close off future uses of OSM by imposing centralised value judgements on its data.

John Willis wrote:
> Trying to decide what mountains are worth labeling at different zooms
> via some GIS data is ridiculous.

It's only ridiculous, to be blunt, if you're no good at GIS. I show identically-tagged pubs at different zoom levels on cycle.travel based on my own criteria, not some importance scale that someone else has decreed. It takes me about three lines of PostGIS and two lines of CartoCSS. It isn't hard at all.

Richard
Reply | Threaded
Open this post in threaded view
|

Re: importance=* tag (for transportation etc)

Andy Townsend
In reply to this post by John Willis
On 19/03/2016 07:41, johnw wrote:
OSM is for gathering data - lots of lots of locally based knowledge of things. Mountains are no different.

Great!  Let's gather lots of data about each place...

Trying to decide what mountains are worth labeling at different zooms via some GIS data is ridiculous.

No, as Andy and Richard have already pointed out it's the _exact opposite_ of that.  Richard's already mentioned how he gives pubs different prominence depending on where they're located; many of the "specialist maps created with OSM data" use some other data (relevant to that map) to decide what to render and when (e.g. historical tags, railway tags, whatever).


So we render them all equally - which is equally as ridiculous.

Who's this "we"?  There are lots of maps made with OSM data; there are 5 different ones on osm.org.  A related issue that I've been thinking for a while now how to make natural=peak render sensibly when there are lots of them together, like here for example:

http://www.openstreetmap.org/#map=13/52.9348/-3.5334

Even by the "does it look impressive from the road" measure, some of those "natural=peak" I suspect are really very prominent, some of them are not, and some probably aren't really peaks at all (we've had a bit of an issue in GB with a keen but somewhat misguided mapper adding spot-heights from historic maps as "peaks", some of which have since become quarries).

What isn't going to work is deciding that some of them are, by some measure, "important".  We got to that problem (via a different route) with individual trees and lots of problems ensued.  According to http://taginfo.openstreetmap.org.uk/search?q=natural%3Dpeak there are over 14k natural=peak in GB; it's very likely that be any global "importance" measure _none_ of those in that map view are.  Even some local measure isn't going to work as I'm certainly not going to review 14k bits of data and with no local knowledge try and come up with some "importance" value.

What we need to do instead is come up with a way of using other verifiable data, perhaps from OSM, perhaps from other sources, that allows maps to decide whether it is, in their eyes "important".

Cheers,

Andy


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

Re: importance=* tag (for transportation etc)

John Willis
In reply to this post by Andy Mabbett


> On Mar 19, 2016, at 9:18 PM, Andy Mabbett <[hidden email]> wrote:
>
> It's nowhere near as ridiculous as trying to render them according to
> some arbitrary and subjective "importance" (Importance to whom?

All of the examples I have given are all sourced in local culture.

Usually having things named after it, noted on road signs, depicted in paintings, noted in historical documents, included in historic lists (of famous peaks), and other *easily understood* things that provincial or regional people would be able to define - but be extremely difficult for a person who doesn't live in the country or region (or speak the language) to verify.

Is someone supposed to make a "importance" database, detailing the hundreds or thousands of references that a mountain's name is used in - or can some local mappers just tag it to influence the renderings below z15 so it doesn't look like a useless jumble of unordered garbage labels?

I think it is very obvious to people here that giving an icon and label to all of the little points on the lip of the crater of Mt Fuji the same rendering priority as the entire volcano of Mt Fuji is wrong.

I also think it is obvious that rendering an icon for a 25m AGL hill in a city park with the same icon and labeling as a 2500m mountain is wrong.

Doing nothing to rectify the situation - and using an argument against subjective data - when so many other types of data in OSM are ranked and rendered with subjective, but verifiable by locals - seems obtuse, and produces a objectively and unarguably inferior map.

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

Re: importance=* tag (for transportation etc)

Andy Mabbett
On 19 March 2016 at 17:47, John Willis <[hidden email]> wrote:

>> On Mar 19, 2016, at 9:18 PM, Andy Mabbett <[hidden email]> wrote:
>>
>> It's nowhere near as ridiculous as trying to render them according to
>> some arbitrary and subjective "importance" (Importance to whom?
>
> All of the examples I have given are all sourced in local culture.
>
> Usually having things named after it, noted on road signs, depicted in paintings, noted in historical documents, included in historic lists (of famous peaks), and other *easily understood* things that provincial or regional people would be able to define - but be extremely difficult for a person who doesn't live in the country or region (or speak the language) to verify.

So far as "importance is concerned, that's not "sourced", that's your
*subjective* interpretation.


> an argument against subjective data

You have yet to give evidence that there is any subjective data
relating to the very vague concept of "importance".

--
Andy Mabbett
@pigsonthewing
http://pigsonthewing.org.uk

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

Re: importance=* tag (for transportation etc)

Alexander Matheisen
In reply to this post by dieterdreist
Am Samstag, den 19.03.2016, 10:28 +0100 schrieb Martin Koppenhoefer:

> > But Hakone is a very famous place - though it’s height and
> > prominence would say otherwise. People all over Japan (and many
> > international tourists) come there buy eggs cooked in sulfurous
> > vents and enjoy the hot spring resorts inside the caldera.
>
>
> I'd say: tourism=attraction for this volcano (admittedly another way
> of saying important=yes), the presence of the resorts also indicates
> importance. Is this checkable automatically? Not sure (of course you
> can check for this if you know what you're looking for, but a
> different place might have completely different reasons to be
> "important")
Correct, there are so many different aspects that influence the
importance, that it is nearly impossible to determine the importance so
that it fits to the importance in reality and how people see it.


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

signature.asc (484 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: importance=* tag (for transportation etc)

Alexander Matheisen
In reply to this post by dieterdreist
Am Samstag, den 19.03.2016, 02:13 +0100 schrieb Martin Koppenhoefer:

>
> sent from a phone
>
> > Am 18.03.2016 um 22:36 schrieb John Willis <[hidden email]>:
> >
> > OSM wants local knowledge, per this idea, but not the kind that
> > could lead to better rendered maps or better routing.
>
>
>
> importance is relative and it depends on your criteria which things
> you consider more important than others. I agree that we should
> better map the properties that make something important rather than
> stuff it all in a generalized "importance" tag. For mountains this
> will likely mean to look at prominence and topographic isolation, for
> train stations it can be amount of passengers or trains per time, or
> size (amount of platforms), availability of highspeed trains, etc.,
> for airports similarly you could look at amount and size of runway/s,
> passengers/time, amount of connections etc.
> the problem with rendering is that these calculations are typically
> too expensive to do them on the fly, but we could use precomputed
> data and integrate it in a coastline like way (external shape file or
> pre-computed attribute)
If you have a look at the highway=* tagging: This scheme is subjective,
but there is no alternative. Without such categorization by mappers, it
would be necessary to calculate the type of road just by absolute
values such as the width, the surface, the number of cars per hour, the
number of lanes, etc. Everybody should agree that that would not work.

As the person who created that station importance draft, I will focus
on stations, but for other features like mountain peaks the situation
should be similar: It is not possible to calculate the importance of a
station just by some values. Values like the number of platforms or
passengers are just absolute values, but the importance of a station is
a relative information which is influenced by neighbouring features.
Without any importance-tag, it would be necessary to analyse many other
feature, which is very difficult. The importance of a station is
influenced by so many aspects that it is nearly impossible to calculate
it, especially in a reasonable time.

I also see problems in getting some of the proposed values. For
example, the amount of passengers or trains per time is difficult to
measure for a mapper and is not easy to be checked by other mappers. I
do not see a possibility to map such values for a larger number of
stations.

I also see the problem that calculating the importance by a complex
algorithm might be very intransparent. There will be situations where
such algorithms produce results that does not fit to reality, but
mappers will not have any possibility to influence the results.
Currently one strength of OSM, compared to many commercial services, is
that we use the local knowledge of the mappers. With subjective tags
like highway=* they can map information that helps applications to
produce the best results.


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

signature.asc (484 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: importance=* tag (for transportation etc)

John Willis
In reply to this post by Andy Mabbett




Javbw
> On Mar 20, 2016, at 3:30 AM, Andy Mabbett <[hidden email]> wrote:
>
> So far as "importance is concerned, that's not "sourced", that's your
> *subjective* interpretation.

Go google search for:

赤城    (Generic images for Akagi)

赤城神社. (The shrine and related shrines)

赤城山 (the mountian itself)

All the results are for things named after that mountain. There are my sources.
Sources locals understand, because they recognize the connection the mountain and the things named after it.

I can't scoop up images of all the road signs using "mount Akagi" as a control point for direction, paintings depicting the mountain.

I can point to it being one of the "3 mountains of Jomo" and on the list of the 100 famous mountains of Japan" - and to the lack of other mountains being on the list.

I can point to it being labeled as a visible  object on the observation deck map for the Tokyo Sky Tree.

But I can't aggravated this into some buzzfeed style listicle "11 mountains in Gunma you should see" - or a GIS database.

Nor can I point to the lack of images for Kessamaru. Do you expect me to somehow meta-aggregate The Internet to show you why some mount and I some region should be labeled more or less prominently than others? Or can we use the power of locally sourced familiarity  that OSM is supposed to be drawing from?

~~~

This entire subject about mountains is the most infuriating topic I have ever dealt with as an OSM mapper.

Q: Can we have some kind of sub-peak tag-relation so we can say "this is a small subpeak of a larger mountian, instead of the subpeaks competing with the main peak for rendering?

A: no. It will get too confusing. And people will want to tag climbing prominence, and that is a big can of worms we don't want to open.

Q: can we use a "hill" tag, so we can separate out these ~100 foot AGL little lumps that are named but shouldn't be rendered as a mountain peak?

A: no, we can't decide where to draw the line, so a 25m AGL mountain and mt Everest get tagged and rendered the same.

Q: then can we use this "local information" for locals to influence when  peaks are rendered, maybe using some kind of "importance" tag or something?

A: No, because We don't consider sources in aggregate - we're hoping for some magical, impossible GIS information solution that has never existed and will never exist, because local opinion is "subjective" and we want to be myopic that we are totally dependent on this local subjective nature for a myriad of other tags.

~~~~~

Getting the maps to render this kind of data has always required the opinion of the mapmakers. And OSM takes mappers' opinions with so many other tags - but for this it is unacceptable. for some unexplainable reason and hand wave saying it is somehow not empirical enough.

I can only conclude from these discussions - and comments from people controlling the rendering - that an *objectively* inferior or substandard map is good enough for the group, when it comes to mountains, because people want a data set to improve it that will never, ever exist.

Javbw.



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

Re: importance=* tag (for transportation etc)

Janko Mihelić
ned, 20. ožu 2016. 04:55 John Willis <[hidden email]> je napisao:

This entire subject about mountains is the most infuriating topic I have ever dealt with as an OSM mapper.

You actually already have all the data you need, and it's on Wikidata. Just look at the number of articles about each peak, and render them according to that. More articles=rendered at lower zooms. Problem solved, and you don't have to put vague tags in OSM. 

Janko

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

Re: importance=* tag (for transportation etc)

Daniel Koć
W dniu 21.03.2016 21:32, Janko Mihelić napisał(a):
> ned, 20. ožu 2016. 04:55 John Willis <[hidden email]> je napisao:
>
>> This entire subject about mountains is the most infuriating topic I
>> have ever dealt with as an OSM mapper.
>
> You actually already have all the data you need, and it's on Wikidata.
> Just look at the number of articles about each peak, and render them
> according to that. More articles=rendered at lower zooms. Problem
> solved, and you don't have to put vague tags in OSM.

1.

Interesting idea worth testing, IMO. However I suspect in practice there
will be lot of problems to make it really fly.

Let's look at the similar simple idea (with scoring based on city type
and population) used to render city labels in osm-carto in a more sane
way:

https://github.com/gravitystorm/openstreetmap-carto/pull/1461

Almost 5 months of discussing and hand tweaking the algorithm tells me
just having the hard (numerical) data might be not as easy as you depict
it. Raw population data is far from having world-range city ranking we
needed.

2.

We have a big problem with peaks. Once I have tried to add peak subtypes
(with mountain peak as default for backward compatibility), but in
general it was rejected and is rarely used (see
http://taginfo.openstreetmap.org/keys/peak#values - most of them are
strange service=peak+peak=school for some bus routes in Germany,
anyway). Elevation alone may be not the only important thing - relative
height can be also relevant property. And, as John points out, there are
also some "soft" cultural references which make some peaks much more
interesting than other, even similar in the physical sense.

Using Wikidata as a base for peaks scoring is worse than basing it on
population, because it's less universal and relies on one particular
website, but I don't reject it at this moment.

3.

Let's not forget about transportation. I have read discussion about
problems, especially about small near-border train stations (being,
well, "international") and stations in USA (most of them are national,
despite the country occupies about half the continent). I don't have a
solution for it yet, but we could craft the definition to catch such
corner cases, if they are rare enough.

Still I think "international airport" in the name hints us something and
is worth using this way or another to indicate importance.
International/domestic/local fares are rather useful and popular
description of importance level for railway (and bus!) stations and with
military/private distinction I guess it could also work with the
airports.

Maybe the key name should not be "importance" - I just took an
existing/proposed scheme with some nice defined values and tried to
extend its scope outside just the railways. While for example "range"
may be good for transportation, it wouldn't work with peaks, of course.

--
"Завтра, завтра всё кончится!" [Ф. Достоевский]

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
123