mass iD validation arrives in NYC

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

mass iD validation arrives in NYC

Jmapb
See yesterday's changesets:

https://www.openstreetmap.org/changeset/70676813 (
https://nrenner.github.io/achavi/?changeset=70676813 )
https://www.openstreetmap.org/changeset/70676888 (
https://nrenner.github.io/achavi/?changeset=70676888 )

I believe this is just a casual user browsing around in iD and making
the suggested changes it advises -- to about 1000 objects. These giant
changesets are nearly impossible to review. My fear here is that iD's
new validator will make QA extremely challenging in dense areas.

Scrolling through the tag additions, these changesets look almost
identical to the behavior of a bot... or rather like 6 or 7 bots
operating at once. If they *had* been made by a bot that was following
the mechanical edit guidelines, they could be comprehended and reviewed.
But the various tagging changes are all mixed up together in a single
changeset, along with whatever the mapper reidpelton's *actual* changes
were -- if any.

So how do I even begin to do QA on this? I don't see any options other
than 1) mass-revert or 2) skip review of all large changesets that
appear to be triggered by iD validation. Any other suggestions?

Jason


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

Re: mass iD validation arrives in NYC

SimonPoole
1st thing to do is to ask the napper to slow down in a change set comment.

Am 28. Mai 2019 17:13:49 MESZ schrieb Jmapb <[hidden email]>:
See yesterday's changesets:

https://www.openstreetmap.org/changeset/70676813 (
https://nrenner.github.io/achavi/?changeset=70676813 )
https://www.openstreetmap.org/changeset/70676888 (
https://nrenner.github.io/achavi/?changeset=70676888 )

I believe this is just a casual user browsing around in iD and making
the suggested changes it advises -- to about 1000 objects. These giant
changesets are nearly impossible to review. My fear here is that iD's
new validator will make QA extremely challenging in dense areas.

Scrolling through the tag additions, these changesets look almost
identical to the behavior of a bot... or rather like 6 or 7 bots
operating at once. If they *had* been made by a bot that was following
the mechanical edit guidelines, they could be comprehended and reviewed.
But the various tagging changes are all mixed up together in a single
changeset, along with whatever the mapper reidpelton's *actual* changes
were -- if any.

So how do I even begin to do QA on this? I don't see any options other
than 1) mass-revert or 2) skip review of all large changesets that
appear to be triggered by iD validation. Any other suggestions?

Jason
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk

--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit Kaiten Mail gesendet.
_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: mass iD validation arrives in NYC

SimonPoole


Am 28. Mai 2019 17:44:25 MESZ schrieb Simon Poole <[hidden email]>:
>1st thing to do is to ask the napper to slow

 ... mapper :-)

--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit Kaiten Mail gesendet.

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

Re: mass iD validation arrives in NYC

Mateusz Konieczny-3
In reply to this post by Jmapb



28 May 2019, 17:13 by [hidden email]:
Any other suggestions?
Suggest user to split edits into smaller chunks.

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

Re: mass iD validation arrives in NYC

Jmapb

On 5/28/2019 12:25 PM, Mateusz Konieczny wrote:

28 May 2019, 17:13 by [hidden email]:
Any other suggestions?
Suggest user to split edits into smaller chunks.

Yes that would be far preferable, and I did message this user. That doesn't address the immediate question of how to attempt QA on these two changesets. And the larger problem of other users making similar changesets, in this city and thousands of other densely mapped places on the planet, is still looming.

I can (and do) plead with people to keep changesets small, but the nature of the process has changed. Before, a casual making hundreds of changes in a single changeset was pretty rare, because they'd actually have to *make* hundreds of changes. Now they just have to agree to them -- and as far as they can tell they're being asked to do so by OSM itself, so it's highly likely they will.

(And sidebar -- I like the idea of the https://wiki.openstreetmap.org/wiki/Good_practice page including some advice about keeping changsets to a manageable size -- both number of changes and geographic area. Something to point to when trying to make the case for smaller changesets.)

J


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

Re: mass iD validation arrives in NYC

General Discussion mailing list
In reply to this post by Jmapb
I notice these changesets were completed in 30/60 seconds respectively.
I don't use iD. How is this possible? Does it have a JOSM like mass edit
ability?

   I don't see asking users to split the changesets as a solution to
what is the clear problem of mass adding/amending tags to
unknown/undocumented ones.

DaveF


On 28/05/2019 16:13, Jmapb wrote:

> See yesterday's changesets:
>
> https://www.openstreetmap.org/changeset/70676813 (
> https://nrenner.github.io/achavi/?changeset=70676813 )
> https://www.openstreetmap.org/changeset/70676888 (
> https://nrenner.github.io/achavi/?changeset=70676888 )
>
> I believe this is just a casual user browsing around in iD and making
> the suggested changes it advises -- to about 1000 objects. These giant
> changesets are nearly impossible to review. My fear here is that iD's
> new validator will make QA extremely challenging in dense areas.
>
> Scrolling through the tag additions, these changesets look almost
> identical to the behavior of a bot... or rather like 6 or 7 bots
> operating at once. If they *had* been made by a bot that was following
> the mechanical edit guidelines, they could be comprehended and reviewed.
> But the various tagging changes are all mixed up together in a single
> changeset, along with whatever the mapper reidpelton's *actual* changes
> were -- if any.
>
> So how do I even begin to do QA on this? I don't see any options other
> than 1) mass-revert or 2) skip review of all large changesets that
> appear to be triggered by iD validation. Any other suggestions?
>
> Jason
>
>
> _______________________________________________
> talk mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/talk


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

Re: mass iD validation arrives in NYC

dieterdreist


Am Di., 28. Mai 2019 um 19:56 Uhr schrieb Dave F via talk <[hidden email]>:
I notice these changesets were completed in 30/60 seconds respectively.
I don't use iD. How is this possible? Does it have a JOSM like mass edit
ability?

   I don't see asking users to split the changesets as a solution to
what is the clear problem of mass adding/amending tags to
unknown/undocumented ones.



indeed, I see no way to judge whether the iD suggestion to change objects from crossing=zebra to crossing=marked makes sense, because there is no documentation of crossing=marked. Going by the words, probably any zebra crossing can be seen as a marked crossing so it may not be introducing errors, just reducing specificity/detail.

Cheers,
Martin

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

Re: mass iD validation arrives in NYC

Andrew Hain
Just out of idle curiosity, do we know of any data consumers that understand crossing=marked?

--
Andrew

From: Martin Koppenhoefer <[hidden email]>
Sent: 28 May 2019 19:00
To: Dave F
Cc: osm
Subject: Re: [OSM-talk] mass iD validation arrives in NYC
 


Am Di., 28. Mai 2019 um 19:56 Uhr schrieb Dave F via talk <[hidden email]>:
I notice these changesets were completed in 30/60 seconds respectively.
I don't use iD. How is this possible? Does it have a JOSM like mass edit
ability?

   I don't see asking users to split the changesets as a solution to
what is the clear problem of mass adding/amending tags to
unknown/undocumented ones.



indeed, I see no way to judge whether the iD suggestion to change objects from crossing=zebra to crossing=marked makes sense, because there is no documentation of crossing=marked. Going by the words, probably any zebra crossing can be seen as a marked crossing so it may not be introducing errors, just reducing specificity/detail.

Cheers,
Martin

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

Re: mass iD validation arrives in NYC

Philip Barnes
In reply to this post by dieterdreist
On Tue, 2019-05-28 at 20:00 +0200, Martin Koppenhoefer wrote:


Am Di., 28. Mai 2019 um 19:56 Uhr schrieb Dave F via talk <[hidden email]>:
I notice these changesets were completed in 30/60 seconds respectively.
I don't use iD. How is this possible? Does it have a JOSM like mass edit
ability?

   I don't see asking users to split the changesets as a solution to
what is the clear problem of mass adding/amending tags to
unknown/undocumented ones.



indeed, I see no way to judge whether the iD suggestion to change objects from crossing=zebra to crossing=marked makes sense, because there is no documentation of crossing=marked. Going by the words, probably any zebra crossing can be seen as a marked crossing so it may not be introducing errors, just reducing specificity/detail.

Just had a play with iD in my local High Street. 

First edit was a minor one to tidy up a spurious line, iD made no attempt to change any other objects.

Second edit was to add a tactile paving tag to one of the zebra crossings, it warned me that a marked crossing has outdated tags and wanted to loose valuable information. which I declined but it is easy to see how an inexperienced mapper could be coerced into making such a change.

As you say marked is undocumented, and zebra is documented.

Phil (trigpoint)

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

Re: mass iD validation arrives in NYC

SimonPoole
In reply to this post by General Discussion mailing list
The times in the changeset do not reflect the length of the associated editing session except if the changeset was opened on purpose at the beginning which IMHO no editor does.

Am 28. Mai 2019 19:53:22 MESZ schrieb Dave F via talk <[hidden email]>:
I notice these changesets were completed in 30/60 seconds respectively.
I don't use iD. How is this possible? Does it have a JOSM like mass edit
ability?

  I don't see asking users to split the changesets as a solution to
what is the clear problem of mass adding/amending tags to
unknown/undocumented ones.

DaveF


On 28/05/2019 16:13, Jmapb wrote:
See yesterday's changesets:

https://www.openstreetmap.org/changeset/70676813 (
https://nrenner.github.io/achavi/?changeset=70676813 )
https://www.openstreetmap.org/changeset/70676888 (
https://nrenner.github.io/achavi/?changeset=70676888 )

I believe this is just a casual user browsing around in iD and making
the suggested changes it advises -- to about 1000 objects. These giant
changesets are nearly impossible to review. My fear here is that iD's
new validator will make QA extremely challenging in dense areas.

Scrolling through the tag additions, these changesets look almost
identical to the behavior of a bot... or rather like 6 or 7 bots
operating at once. If they *had* been made by a bot that was following
the mechanical edit guidelines, they could be comprehended and reviewed.
But the various tagging changes are all mixed up together in a single
changeset, along with whatever the mapper reidpelton's *actual* changes
were -- if any.

So how do I even begin to do QA on this? I don't see any options other
than 1) mass-revert or 2) skip review of all large changesets that
appear to be triggered by iD validation. Any other suggestions?

Jason
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk

talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk

--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit Kaiten Mail gesendet.
_______________________________________________
talk mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: mass iD validation arrives in NYC

Michael Reichert-3
Hi,

Am 28.05.19 um 20:28 schrieb Simon Poole:
> The times in the changeset do not reflect the length of the associated editing session except if the changeset was opened on purpose at the beginning which IMHO no editor does.

A better method to guess the length of the editing session is to look
when the previous changeset was uploaded. (It assumes that the human
editing OSM does not have to open editing sessions in parallel)

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)


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

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

Re: mass iD validation arrives in NYC

General Discussion mailing list
In reply to this post by General Discussion mailing list
Oh Good Lord, that's dangerous.

'Everywhere' in the 'Where' option accumulates more warnings (1000+) as
you pan around
Even JOSM doesn't go that far.

DaveF

On 28/05/2019 19:17, Markus wrote:

> On Tue, 28 May 2019 at 19:56, Dave F via talk <[hidden email]> wrote:
>> I notice these changesets were completed in 30/60 seconds respectively.
>> I don't use iD. How is this possible? Does it have a JOSM like mass edit
>> ability?
> Yes, in the issues pane on the right side there are options to check
> "Everything" and to "Fix All". "Fix All" automatically does changes
> without informing you. In a densely populated area it's easy to change
> 200+ elements with just one mouse click.
>
> As i wrote in the other thread (Re: Remove validation rule asking to
> add highway=footway to railway/public_transport=platform) this
> violates the Automated Edits code of conduct.
>
> Regards
>
> Markus


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

Re: mass iD validation arrives in NYC

Clifford Snow
In reply to this post by General Discussion mailing list


On Tue, May 28, 2019 at 10:53 AM Dave F via talk <[hidden email]> wrote:
I notice these changesets were completed in 30/60 seconds respectively.
I don't use iD. How is this possible? Does it have a JOSM like mass edit
ability?

Yes - JOSM does allow mass fixes through the validator. I've even seen suggestions to fix objects on ways that are outside of the downloaded area. For example, missing power poles on power lines. I don't "fix" those because the validator is just looking at a node without a power pole. Often their isn't a pole at that location according to the imagery. 

Best,
Clifford 

--
@osm_washington
OpenStreetMap: Maps with a human touch

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

Power pole validation (was: mass iD validation arrives in NYC)

Mateusz Konieczny-3



28 May 2019, 22:10 by [hidden email]:
I don't "fix" those because the validator is just looking at a node without a power pole. Often their isn't a pole at that location according to the imagery. 
Sometimes proper fix for that report is to delete node placed where there is no pole.


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

Re: Power pole validation (was: mass iD validation arrives in NYC)

Clifford Snow


On Tue, May 28, 2019 at 1:59 PM Mateusz Konieczny <[hidden email]> wrote:



28 May 2019, 22:10 by [hidden email]:
I don't "fix" those because the validator is just looking at a node without a power pole. Often their isn't a pole at that location according to the imagery. 
Sometimes proper fix for that report is to delete node placed where there is no pole.

It's actually more complex. Often there are numerous poles that haven't been identified but show on the image and the power lines can be many km away from where I was mapping, in an area I'm not that familiar with. 

Automated fixes need to be done with caution. Some are simple, for example, zip codes on highway which are likely artifacts from an import. Since we don't add postal codes on highways they can be deleted.

I was just trying to point out that JOSM has had automatic fixes built in for some time where iD is just catching up. 



--
@osm_washington
OpenStreetMap: Maps with a human touch

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

Re: mass iD validation arrives in NYC

Marc Gemis
In reply to this post by Clifford Snow
I assume that that way (powerline) was downloaded because it also has
nodes in the area you downloaded. JOSM will never complain about
objects that are not downloaded. Powerline ways tend to be long, so
the warning can easily be in another state, that is true.
Furthermore, I thought that it is not allowed to have nodes in a
powerline that do not have tags. So the suggestion to fix it, is a
valid one.
And it is not because JOSM warns about nodes without tags in a
powerline, that it allows you to solve that problem by clicking a
magical "Solve" button. You still have to add the tags yourself.

There is a magical "Solve" button in JOSM, but its capabilities are
limited (e.g. removing vehicle=yes on highway=residential and the
like), or merging 2 nodes that are lying on top of one another (at
least when the nodes have no tags).

regards

m.

On Tue, May 28, 2019 at 10:14 PM Clifford Snow <[hidden email]> wrote:

>
>
>
> On Tue, May 28, 2019 at 10:53 AM Dave F via talk <[hidden email]> wrote:
>>
>> I notice these changesets were completed in 30/60 seconds respectively.
>> I don't use iD. How is this possible? Does it have a JOSM like mass edit
>> ability?
>>
> Yes - JOSM does allow mass fixes through the validator. I've even seen suggestions to fix objects on ways that are outside of the downloaded area. For example, missing power poles on power lines. I don't "fix" those because the validator is just looking at a node without a power pole. Often their isn't a pole at that location according to the imagery.
>
> Best,
> Clifford
>
> --
> @osm_washington
> www.snowandsnow.us
> OpenStreetMap: Maps with a human touch
> _______________________________________________
> talk mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/talk

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