armchair mappers putting errors back into the map

classic Classic list List threaded Threaded
15 messages Options
Reply | Threaded
Open this post in threaded view
|

armchair mappers putting errors back into the map

Richard Welty-2
we've talked about this before, i guess we will never stop talking about it.

i recently noticed in my OSMAnd display that an error got put back
into the map about 1 mile from where i live. a short dead end
was put in as a residential road (it's really an unnamed driveway)
based on NYS GIS data. i removed it years ago (it was in tiger 2007).
the mapper's source data (at least he put that in) implies he is
an armchair mapper.

i'm not calling out the mapper here by name (just yet); i sent him
a message about 5 minutes ago telling him that he's putting errors
back into the map and that NYS GIS data is not sufficiently
reliable. but i don't know for sure if he'll even see the message.

what can/should we be doing about this sort of stuff? i'm really
at a bit of a loss here.

richard

--
[hidden email]
 Averill Park Networking - GIS & IT Consulting
 OpenStreetMap - PostgreSQL - Linux
 Java - Web Applications - Search

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

Re: armchair mappers putting errors back into the map

Mateusz Konieczny-3
Mar 3, 2019, 5:09 PM by [hidden email]:
we've talked about this before, i guess we will never stop talking about it.

i recently noticed in my OSMAnd display that an error got put back
into the map about 1 mile from where i live. a short dead end
was put in as a residential road (it's really an unnamed driveway)
based on NYS GIS data. i removed it years ago (it was in tiger 2007).
(...)

what can/should we be doing about this sort of stuff? i'm really
at a bit of a loss here.
When I remove things visible in aerial images or other high quality sources
I often leave geometry with note.

For example on removed power tower (I made edit as crane lowered it,
I just happened to be looking at it outside window) I changed

power=tower node

to

note="power tower visible on aerial images was removed on <DATE>"
survey_date=<DATE>

to keep people from remapping it based on high quality aerial images available for my area.

---

I also don't make risky armchair edit in areas with active local mappers.

---

Is "NYS GIS data" something that can be considered nearly error free? If not then
it should not be used for edits without verification with at least high-quality aerial imaged
and this mapper should be reminded about it.

If yes then leaving geometry with note is a good idea.

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

Re: armchair mappers putting errors back into the map

Richard Welty-2
On 3/3/19 11:47 AM, Mateusz Konieczny wrote:

> I also don't make risky armchair edit in areas with active local mappers.
>
> ---
>
> Is "NYS GIS data" something that can be considered nearly error free? If
> not then
> it should not be used for edits without verification with at least
> high-quality aerial imaged
> and this mapper should be reminded about it.
>
> If yes then leaving geometry with note is a good idea.

it is certainly an area with an active local mapper (me) which can be
seen easily by the removal of tiger:reviewed tags and by looking at
edit histories.

i have not reviewed NYS GIS data because there were, in the past
at least, licensing issues. i do not know if those have been resolved
so i'm not pitching a fit about that. it might be ok now. i just don't
know. but from this edit, it appears that NYS GIS contains things that
were wrong in 2007 when we took in the TIGER data.

richard

--
[hidden email]
 Averill Park Networking - GIS & IT Consulting
 OpenStreetMap - PostgreSQL - Linux
 Java - Web Applications - Search

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

Re: armchair mappers putting errors back into the map

Andy Townsend
In reply to this post by Richard Welty-2
On 03/03/2019 16:09, Richard Welty wrote:
> what can/should we be doing about this sort of stuff? i'm really
> at a bit of a loss here.
>
Aside from the excellent technical suggestion that's already been made,
I'd suggest a polite changeset discussion comment, explaining that the
imagery that they're using there is out of date or wrong for some other
reason.

Changeset discussion comments have an advantage over direct OSM messages
in that they are visible to all, which tends to mean that (a) everyone
is that much more nice to each other and (b) other mappers may know that
there is a potential problem.  Also, if you get no reply and need to ask
the Data Working Group for help we can easily see the details of the
problem, without have to ask you to repeat what you've already sent in PMs.

Best Regards,

Andy (from the DWG)




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

Re: armchair mappers putting errors back into the map

Richard Welty-2
On 3/3/19 1:06 PM, Andy Townsend wrote:

> On 03/03/2019 16:09, Richard Welty wrote:
>> what can/should we be doing about this sort of stuff? i'm really
>> at a bit of a loss here.
>>
> Aside from the excellent technical suggestion that's already been made,
> I'd suggest a polite changeset discussion comment, explaining that the
> imagery that they're using there is out of date or wrong for some other
> reason.
>
> Changeset discussion comments have an advantage over direct OSM messages
> in that they are visible to all, which tends to mean that (a) everyone
> is that much more nice to each other and (b) other mappers may know that
> there is a potential problem.  Also, if you get no reply and need to ask
> the Data Working Group for help we can easily see the details of the
> problem, without have to ask you to repeat what you've already sent in PMs.

> Andy (from the DWG)

Good suggestion; i have added a changeset comment describing my concerns.

a review of the bio of the mapper in question shows that he is not
just an armchair mapper, he is a corporate armchair mapper. i don't
know if that makes it better or worse.

richard
--
[hidden email]
 Averill Park Networking - GIS & IT Consulting
 OpenStreetMap - PostgreSQL - Linux
 Java - Web Applications - Search

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

Re: armchair mappers putting errors back into the map

Nick Bolten
In reply to this post by Mateusz Konieczny-3
I like the idea of having information indicating the lack of certain features, when necessary - such as the geometry with a note on it about what not to map. It would be great if it could be done more formally, for purposes of QA, such as how the construction tags work for buildings, streets, etc - the tags say "this thing will exist, but it doesn't right now, so don't map anything else in it".

I run into similar issues with pedestrian paths and wonder whether a more formalized "this feature does not exist" tagging strategy (beyond free-form text) could be useful. Example from personal experience: how do we know which areas are in need of having "X feature" added, where the feature requires a new geometry (like a sidewalk line)? You can't differentiate between "nobody has attempted to add the sidewalks to this block" from "there are no sidewalks on this block" without worldwide manual reviews - but you could if there was a tag called "highway:exists=no" (e.g.). I know that there are some potential hacks for sidewalks in particular, but it's part of a general issue of the determining nonexistence vs "unmapped" for features requiring their own geometry.

This is also something that could be helped via editors if there were a "find similar historical" API to the OSM backend so that users could tell if they were replicating / negating historical changes. For the case of the driveway, it should (theoretically) be possible to ask, "has there been a similarly-located + shaped geometry with similar tags been added here in the past?" and show the results + changesets.

Apologies for the potential derailing.

Best,

Nick

On Sun, Mar 3, 2019 at 8:49 AM Mateusz Konieczny <[hidden email]> wrote:
Mar 3, 2019, 5:09 PM by [hidden email]:
we've talked about this before, i guess we will never stop talking about it.

i recently noticed in my OSMAnd display that an error got put back
into the map about 1 mile from where i live. a short dead end
was put in as a residential road (it's really an unnamed driveway)
based on NYS GIS data. i removed it years ago (it was in tiger 2007).
(...)

what can/should we be doing about this sort of stuff? i'm really
at a bit of a loss here.
When I remove things visible in aerial images or other high quality sources
I often leave geometry with note.

For example on removed power tower (I made edit as crane lowered it,
I just happened to be looking at it outside window) I changed

power=tower node

to

note="power tower visible on aerial images was removed on <DATE>"
survey_date=<DATE>

to keep people from remapping it based on high quality aerial images available for my area.

---

I also don't make risky armchair edit in areas with active local mappers.

---

Is "NYS GIS data" something that can be considered nearly error free? If not then
it should not be used for edits without verification with at least high-quality aerial imaged
and this mapper should be reminded about it.

If yes then leaving geometry with note is a good idea.
_______________________________________________
Talk-us mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-us

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

Re: armchair mappers putting errors back into the map

Mateusz Konieczny-3



Mar 3, 2019, 7:26 PM by [hidden email]:
You can't differentiate between "nobody has attempted to add the sidewalks to this block" from "there are no sidewalks on this block" without worldwide manual reviews
sidewalk=none, 270k uses worldwide

In general this difference appears far often in context of StreetComplete (a quite unusual
and interesting editor).


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

Re: armchair mappers putting errors back into the map

Nick Bolten
As I mentioned, I'm aware of hacks regarding sidewalks - though I can absolutely guarantee that strategy is insufficient, particularly when the issue is, "this non-expert mapper needs to know what work has been done more easily" (I've tried using those tags many, many times). This issue will occur any time someone might expect to map X feature and can't tell whether it doesn't exist or isn't mapped.

On Sun, Mar 3, 2019 at 10:32 AM Mateusz Konieczny <[hidden email]> wrote:



Mar 3, 2019, 7:26 PM by [hidden email]:
You can't differentiate between "nobody has attempted to add the sidewalks to this block" from "there are no sidewalks on this block" without worldwide manual reviews
sidewalk=none, 270k uses worldwide

In general this difference appears far often in context of StreetComplete (a quite unusual
and interesting editor).

_______________________________________________
Talk-us mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-us

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

Re: armchair mappers putting errors back into the map

Andy Townsend
On 03/03/2019 18:43, Nick Bolten wrote:
> As I mentioned, I'm aware of hacks regarding sidewalks

I wouldn't describe sidewalk=none as a hack - speaking as someone who
walks a lot, any verifiable tag that says "you're allowed to, but you
probably don't want to walk down this road" is really useful.

Best Regards,

Andy

PS: I'm not trying to make any point about how to map the presence of
sidewalks here (we don't need to dredge that up again) I'm just saying
that mapping the absence of them is useful too.


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

Re: armchair mappers putting errors back into the map

Mateusz Konieczny-3



Mar 3, 2019, 8:20 PM by [hidden email]:
On 03/03/2019 18:43, Nick Bolten wrote:
As I mentioned, I'm aware of hacks regarding sidewalks

I wouldn't describe sidewalk=none as a hack - speaking as someone who walks a lot, any verifiable tag that says "you're allowed to, but you probably don't want to walk down this road" is really useful.
Also for me - in my opinion it is not a hack, it is a well working solution.

Something even better would be interesting, but I consider it very unlikely that it can be improved.


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

Re: armchair mappers putting errors back into the map

Kevin Kenny-3
In reply to this post by Richard Welty-2
On Sun, Mar 3, 2019 at 11:58 AM Richard Welty <[hidden email]> wrote:
> i have not reviewed NYS GIS data because there were, in the past
> at least, licensing issues. i do not know if those have been resolved
> so i'm not pitching a fit about that. it might be ok now. i just don't
> know. but from this edit, it appears that NYS GIS contains things that
> were wrong in 2007 when we took in the TIGER data.

Assuming that you're talking about
http://gis.ny.gov/gisdata/inventories/details.cfm?DSID=932 , that
particular data set is OK. The other one that's recently OK (there was
an announcement) is the address point data at
http://gis.ny.gov/gisdata/inventories/details.cfm?DSID=921. They're
both part of a Street and Address Maintenance (SAM) project at NYSGIS.

Generally speaking, NYSGIS data are effectively public domain, but
some data sets were created by contractors or county governments that
licensed the data to the state on restrictive terms. Several of the
data sets that presented difficulties - such as NYS Public Lands -
have been taken down in recent years.

The awkward one is parcel data, where the horrendous Suffolk v
Experian case set an awkward precedent. It left undecided some major
points because the parties settled after the Second Circuit remanded
for further proceedings, but left open at least the possibility that
tax maps are copyrightable. It is the only Federal appellate court
that has reached such an opinion.

Nevertheless, twenty counties http://gis.ny.gov/parcels/ have agreed
to let the state re-release their parcel data. Moreover, the state has
compiled a database (from county records) of the lands that it owns at
http://gis.ny.gov/gisdata/inventories/details.cfm?DSID=1300 - and is
claiming that it has the unquestioned right to release the plats of
its own holdings and daring the counties to sue.  That data set
replaces the no-longer-released "NYS Public Lands" dataset, which was
copyrighted by the contractor that produced it.  (Alas, it no longer
includes county and municipal facilities.)

Incidentally, a year or two ago all NYS parks, historic sites, and
recreation area were compared against that data set and many were
updated accordingly - including contacting the original mappers in
case of conflict. In all cases, the original mappers responded that
their data were traced approximately from aerials and welcomed the
updates. I didn't call this job an 'import' since the lines eventually
drawn were gleaned from a variety of sources and not drawn directly
from the data set.  An extreme example is
https://www.openstreetmap.org/relation/6440291 (note in particular the
'note' tag there).  Notes like this are an important trail to build a
case that data were not directly copied, and hence copyright was not
infringed. A copyright holder has no monopoly over facts, merely the
expression, selection, structure, sequence and organization - and
notes like these provide evidence that our criteria for selection and
organization are different from those of our data sources.

You're right that the SAM data has obsolete data, errors, and streets
that were platted but never built. In some cases, I think they stay in
the database because the state or a municipality still own the
rights-of-way.

A few weeks (months?) ago, there was an announcement of a MapRoulette
project to add streets from
http://gis.ny.gov/gisdata/inventories/details.cfm?DSID=932 that were
not in OSM.  I suspect that's where your offending driveway came from.
I tried a few in Schenectady and Saratoga Counties, but wound up
passing on most as "too hard" because they were things like a
subdivision platted in a wood, where the most recent NYS Orthos Online
were three years old, so it could be new construction needing a field
survey. I don't think I wound up adding anything from the couple of
dozen things that I tried - where I did have local knowledge, OSM was
right and NYSGIS was wrong.  Which was unsurprising, since I'd already
mapped my local knowledge.

I'd classify http://gis.ny.gov/gisdata/inventories/details.cfm?DSID=932
as "best available third-party data" for NYS streets - it's not nearly
as hallucinatory as TIGER. But it's still surely imperfect. I'll
copy-n-paste from it if I have corroborating evidence that it's right,
and I'd investigate significant discrepancies between it and OSM, but
I'd not import anything from it blindly.

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

Re: armchair mappers putting errors back into the map

Nick Bolten
In reply to this post by Mateusz Konieczny-3
> Also for me - in my opinion it is not a hack, it is a well working solution.

I want to use this exact info to figure out which areas need more mapping of sidewalks - and I plan to do so eventually. But it's not easy!

Here's a friendly and annoying challenge: tell me which sidewalks need to be mapped (because they are separate geometries, I mean highway=footway, footway=sidewalk) in Graz, Austria. We will ignore thinking about stale road way information. Use any tools you'd like! Graz is very well-mapped and was chosen to be, theoretically, one of the easier places from which to ask this question.

As an example, I've written up the overpass queries to get the appropriate street ways, but now we have to find the separate geometries - that's no fun for anyone, and especially not non-experts trying to find things to map. This will pop up any time a feature requires a separate geometry but its presence/absence would be expected by context. Another example is street crossings: sometimes it's illegal or dangerous to cross a street somewhere, but there's no tag for creating a crossing way that shouldn't be used (and the idea of drawing nonexistent things ostensibly violates best practices in tagging). How do we know a street crossing doesn't exist / should not exist vs. someone forgot to map it? It'd be nice to have a strategy for "I reviewed this area and the thing you think should exist at this exact place, does not".

On Sun, Mar 3, 2019 at 11:33 AM Mateusz Konieczny <[hidden email]> wrote:



Mar 3, 2019, 8:20 PM by [hidden email]:
On 03/03/2019 18:43, Nick Bolten wrote:
As I mentioned, I'm aware of hacks regarding sidewalks

I wouldn't describe sidewalk=none as a hack - speaking as someone who walks a lot, any verifiable tag that says "you're allowed to, but you probably don't want to walk down this road" is really useful.
Also for me - in my opinion it is not a hack, it is a well working solution.

Something even better would be interesting, but I consider it very unlikely that it can be improved.

_______________________________________________
Talk-us mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-us

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

Re: armchair mappers putting errors back into the map

Nick Bolten
In reply to this post by Andy Townsend
Absolutely! I may have miscommunicated.

I want the exact same thing: for it to be easier to tag "missing" infrastructure. 

I'm just remembering my many attempts to evaluate sidewalk network completeness (that's highway=footway objects), and it's not easy. But I don't want to launch another discussion about pedestrian tagging standard! I'm just saying that having a dedicated way or point that says, "this thing does not exist here" can be useful, but there are no current standards for doing so that I'm aware of. 

It's subtle: sidewalk=no/none on a road says, "this road does not have sidewalks", whereas a dedicated way for a sidewalk (footway=sidewalk) that said, "this sidewalk does not exist" displaces the expected location of a sidewalk. The latter has its own (big) issues, but I think the problem is worth exploring. For the sidewalk case: how can we related metadata of one thing (in this case, a street) to another (a sidewalk) help mappers create consistent and accurate data? For the nonexistent street case: how can we specify that this piece of infrastructure does *not* exist when there is no metadata from which to guess (as we had in the case of sidewalks) and the current appropriate schema is, "this is a blank area of the map with no data".

On Sun, Mar 3, 2019 at 11:22 AM Andy Townsend <[hidden email]> wrote:
On 03/03/2019 18:43, Nick Bolten wrote:
> As I mentioned, I'm aware of hacks regarding sidewalks

I wouldn't describe sidewalk=none as a hack - speaking as someone who
walks a lot, any verifiable tag that says "you're allowed to, but you
probably don't want to walk down this road" is really useful.

Best Regards,

Andy

PS: I'm not trying to make any point about how to map the presence of
sidewalks here (we don't need to dredge that up again) I'm just saying
that mapping the absence of them is useful too.


_______________________________________________
Talk-us mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-us

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

Re: armchair mappers putting errors back into the map

Mateusz Konieczny-3
In reply to this post by Richard Welty-2
Mar 3, 2019, 10:30 PM by [hidden email]:
> Also for me - in my opinion it is not a hack, it is a well working solution.

I want to use this exact info to figure out which areas need more mapping of sidewalks - and I plan to do so eventually. But it's not easy!

Here's a friendly and annoying challenge: tell me which sidewalks need to be mapped (because they are separate geometries, I mean highway=footway, footway=sidewalk) in Graz, Austria. We will ignore thinking about stale road way information. Use any tools you'd like! Graz is very well-mapped and was chosen to be, theoretically, one of the easier places from which to ask this question.
I agree that it is harder than it should be - in my city I started also adding sidewalk=separate
on roads where sidewalk are mapped as separate highway=footway.

As for going from ready data - it is possible to take all roads, exclude ones with sidewalk tag,
exclude ones with footway=sidewalk within N meters but no matter chosen N there will be
some false positives and/or false negatives.

Tagging sidewalk tag on all city roads (sidewalk=separate, sidewalk=none) is required to
solve that, though it will still not catch simply outdated info.



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

Re: armchair mappers putting errors back into the map

Andy Townsend
In reply to this post by Nick Bolten
 > tell me which sidewalks need to be mapped (because they are separate geometries, I mean highway=footway, footway=sidewalk) in Graz, Austria

Locally I'd do that by going there and having a look.  The map style I use mostly on Web maps shows sidewalk=both/left/right/separate, the Garmin maps I use also show that.

I haven't been to Graz for 30 years so my recollection of it is unlikely to be useful, but I bet there are plenty of local mappers who might be interested in a Maproulette/Streetcomplete challenge or similar, provided that they can see that their efforts are in a good cause.

Best Regards,
Andy


_______________________________________________
Talk-us mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-us