[talk-au] Tracks flagged as missing from government data

[Warin]

In NSW tracks in;

National Parks are usually access=private. Some are access=yes (thinking 
of tracks in the Sturt NP - Mt Woods etc).

State Forests are usually access=yes.


Note public roads can be closed form time to time e.g. Sydney Harbor 
Bridge was closed for some yank president to cross it in his car... 
Similarly State Forest tracks and road can be closed for rallies. We 
don't tag the Sydney Harbor Bridge access=permissive/private... so we 
don't need to tag for the seldom exception.

On 22/8/21 2:51 pm, Andrew Harvey via Talk-au wrote:
> On Sun, 22 Aug 2021, at 12:08 PM, Little Maps wrote:
>> Joe, when you talk about reaching consensus, I assume you mean consensus among the Australian osm community, not the global osm community. Is that right? The highway=track wiki states: "highway=track does not imply any particular access=* value", and different countries have adopted different positions on the issue, as described here (thanks for the link Andrew): https://wiki.openstreetmap.org/wiki/OSM_tags_for_routing/Access_restrictions
>>
>> In most countries that have reached consensus, the absence of an access tag on tracks is implicitly taken to mean that access=yes for motorcars. This includes Austria, Belarus, France, Finland, Italy, Netherlands, Spain, Switzerland and the Ukraine. (Brazil, Germany, Iceland and Belgium have partial/proposal variations on a yes approach). In contrast, two countries went with access=no (Finland and Denmark). Many countries haven't reached consensus, including Australia. Have I interpreted this correctly?
> Yeah
>
>> This message isn't meant to be part of any process for reaching a consensus in Australia. That discussion needs to be approached far more deliberately and inclusively. I'm trying to understand whether, to some degree, Andrew and I have been talking past each other because we have different views on this fundamental issue.
>>
>> About 92% of tracks in Australia do not have an access tag (TagInfo Australia), about 5% have access=private, 1% have yes and 1% have no. If I understand things correctly (and this is the whole point of this email), this means that:
> Yeah 7.85% of tracks in Australia have access=*, 3.95% have motor_vehicle=*, 5.33% have foot=*, 4.15% have bicycle=*.
>
> However if I count access OR motor_vehicle OR motorcar then it's 11.1% of tracks have vehicle access set.
>
> While we are at it, in this way 12% define pedestrian access and 11% define bicycle access.
>
> Breaking this down, if I consider (based on a selection of actual values used)
>
> public access includes the tags: designated, destination, permissive, customers, seasonal, public, dry_weather, yes, 4wd, 4WD
> no public access includes the tags: forestry, emergency, agricultural, private, no, official, military, delivery, prohibited, permit, restricted
>
> Then 1.98% of tracks have public vehicle access and 8.7% of tracks have no public vehicle access (of all tracks). So where we know the vehicle access then 18% are public and 81% are not public access.
>
>> - Strictly speaking, since no consensus has been reached in Australia, 1% of tracks are known to be open to the public, 1% are known to be closed, and access is considered to be unknown on 92% of mapped tracks.
>>
>> - If the community accepted a default position of access=no (i.e. all tracks without an access tag are implicitly assumed to be closed to public vehicles), 1% of tracks are known to be open to the public, 1% are known to be closed, and access is assumed to be closed on 92% of mapped tracks.
>>
>> - If the community accepted a default position of access=yes (i.e. all tracks without an access tag are implicitly assumed to be open to public vehicles), 1% are known to be closed to the public, 1% are known to be open, and access is assumed to be open on 92% of mapped tracks.
>>
>> Have I got this right? If I have, then would this lead to the following implications for tagging decisions?
> I don't like the idea of a default at all for tracks. It'll lead to a lot of wrong assumptions, and then you'll have some mappers discouraging setting the access value at all since it's the "default". It makes it impossible for data consumers to know what really is yes/no and what is incomplete data. So to this point I say we should not promote any kind of default value and instead say a missing access/motor_vehicle tag is incomplete data in OSM.
>
> In my view, data consumers should treat incomplete access/motor_vehicle tags as no access because I'd rather it miss out on a potentially available route then route down a private track, but that's a decisions for each data consumer. This is not the same as promoting a default value within the tagging documentation in OSM.
>
>> - Under the current position of no consensus, a "maximilist" tagging approach is required, and it's equally important to add access tags to all tracks, whether open or closed on the ground.
> Yes that's my opinion.
>
>> - If the community accepted a default position of access=no (i.e. all tracks without an access tag are implicitly assumed to be closed to public vehicles), then it's most important to add access tags to all tracks that we know are open on the ground (with evidence of course). It's less of a priority (but still useful) to add tags on tracks where access is known to be no/private/etc.
>>
>> - If the community accepted a default position of access=yes, then it's most important to add access tags to all the tracks that we know are closed/restricted on the ground. It's less of a priority (but still useful) to add tags on tracks where access is known to be open.
>>
>> - Alternatively, as Andrew suggested, some editors may prefer to adopt the maximilist approach (tag 'em all) regardless of what decisions we adopt, because many data consumers (routers etc) may not be aware of, or have the capacity or interest in basing their products on our consensus.
> Yeah I think I've made my point already above.
>
>> Thank you all for your tolerance, I'm keen to make sure I haven't misunderstood any of these principles and their implications. Best wishes Ian
> So I can understand, do you think we should have a default value and mappers should not set the access tag if it's the "default"?
>
> PS. Here is a overpass turbo wizard query to find tracks without vehicle access set:
>
> highway=track and (access is null AND motor_vehicle is null)
>
>
> # queries for above stats below
> -- public access
> SELECT
>    count(*)
> FROM
>    planet_osm_line
> WHERE
>    (
>      motor_vehicle IN ('designated', 'destination', 'permissive', 'customers', 'seasonal', 'public', 'dry_weather', 'yes', '4wd', '4WD')
>      OR (
>        motor_vehicle IS NULL
>        AND
>        access IN ('designated', 'destination', 'permissive', 'customers', 'seasonal', 'public', 'dry_weather', 'yes', '4wd', '4WD')
>      )
>    );
>
> -- no public access
> SELECT
>    count(*)
> FROM
>    planet_osm_line
> WHERE
>    (
>      motor_vehicle IN ('forestry', 'emergency', 'agricultural', 'private', 'no', 'official', 'military', 'delivery', 'prohibited', 'permit', 'restricted')
>      OR (
>        motor_vehicle IS NULL
>        AND
>        access IN ('forestry', 'emergency', 'agricultural', 'private', 'no', 'official', 'military', 'delivery', 'prohibited', 'permit', 'restricted')
>      )
>    );
>
>
> _______________________________________________
> Talk-au mailing list
> Talk-au at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-au