[HOT] #ValidationFriday - June 7

Fri Jun 7 16:00:33 UTC 2019

John,
There are absolutely many more efficient ways to clean up large areas of
data. I know you've shared a number of times on this list, widespread
issues which you've come across; do you have these collected on a wiki page
or other document somewhere? That could be a good way to share knowledge
related to validating large areas in JOSM.

Personally, I want to see #ValidationFriday as a way to promote validation
however users see fit, be that learning to validate, giving feedback to a
new mapper (on a recently mapped task), polishing up a recent or older
project, or continuing to validate as they have for years...it can even be
sharing tried and tested workflows or tips and tricks.

Happy validating,

Matt

On Fri, Jun 7, 2019 at 8:42 AM john whelan <jwhelan0112 at gmail.com> wrote:

> If you catch the problems early enough then the mapper gets both feedback
> and corrected.  Experimentally the impact drops off very quickly.  So catch
> them within the hour and they change how they map and usually map a bit
> more.  After a week it's not worth the effort of giving feedback.
>
> From a practical point of view there are tools that can be used to clean
> up large areas at a time without using the task manager.  Downloading a
> chunk of OSM then running a script to detect duplicate buildings means you
> can cover an entire country very quickly the same is true for highway=road
> name=Hameau, crossing highways, untagged ways etc.   Scrolling through the
> downloaded data with JOSM <crtl><down arrow> can pick up unmapped
> settlements, unconnected highways, and other oddites.  Don't delete
> anything before redownloading the small area first to make sure you have
> exactly what is on the live map.
>
> Validating mapping that is three months old just so the box gets ticked
> seems a worthless exercise to me but if you really want to waste time and
> put people off validating then ask them to validate buildings.  It can take
> two clicks in JOSM to draw a building with the buildings_tool plugin.  It
> takes a lot longer to correct each misshaped building.  HOT needs to use a
> building tool for mapping.  If you must use iD then spend some money and
> get it added to iD.
>
> Cheerio John
>
> On Fri, 7 Jun 2019 at 08:07, Matthew Gibb <mjngibb at gmail.com> wrote:
>
>> Hi Jean-Marc,
>>
>> Thanks for your thoughts on the subject.
>>
>> I agree with:
>>
>> > the further upstream the quality assurance, the cheaper it is. I feel
>> that an onboarding environment (both technical and social) that glorifies
>> quality rather than quantity might be a step in that direction.
>>
>> There are a whole host of steps that can and are being taken: from
>> improved project manager onboarding, ensuring validation plans like you
>> mentioned, and improvements to the tasking manager.
>>
>> To your point, I agree that feedback to a mapper on a project that hasn't
>> been touched in a year will have no impact on a new (at that point) mapper
>> who hasn't seen it, but if there happened to be lower quality data there,
>> there's still a benefit to making sure it's addressed and improved.
>>
>> I think most would agree that there's not a one-size-fits-all solution,
>> encouraging some more mappers to take a crack at validation if they haven't
>> before is a piece of the puzzle though.
>>
>> Thanks for joining the conversation.
>>
>> Matt
>>
>> On Fri, Jun 7, 2019 at 5:27 AM Jean-Marc Liotier <jm at liotier.org> wrote:
>>
>>> On 2019-06-07 04:52, Matthew Gibb wrote:
>>>
>>> > - Validate! Simply find a project and dive in!
>>>
>>> While http://www.missingmaps.org/validate/ and
>>>
>>> http://www.missingmaps.org/assets/downloads/MissingMaps_validation_josm_en.pdf
>>> offer practical instructions on how to begin about the validation
>>> business, I feel a gap in guidance on what happens after invalidation.
>>> The
>>> advice about how to express constructive criticism in comments is a good
>>> start, but then what ? Even the Organised Editing guidelines only mention
>>> "_plans for a "post-event clean up" to validate edits,
>>> especially if the activity introduces new contributors to
>>> OpenStreetMap_" but omit details.
>>>
>>> The contributor is an ephemeral drive-by account set for a mapathon, the
>>> contributor isn't aware of his Openstreeetmap inbox, the contributor
>>> doesn't care that much about quality, the contributor understands that
>>> his
>>> changeset doesn't satisfy quality expectations but has no idea how to
>>> proceed... There are many reasons but the common result is that a
>>> validation comment will lead to no action at all: most contributors of
>>> bad
>>> data do not clean-up after themselves.
>>>
>>> Some projects, such as
>>>
>>> https://wiki.openstreetmap.org/wiki/Organised_Editing/Activities/Trains_of_Botswana_mapathon
>>> have clear plans: _"A few days after the event, the core team will look
>>> at
>>> the common QA tools (OSMI, Osmose, Keepright) to repair anything that
>>> might have slipped through the cracks_" - but foisting janitorial
>>> responsibilities upon experts doesn't scale: as much as some enjoy
>>> strolling in the garden and pulling the occasional weed, it is not a
>>> popular hobby. Worse, those rare resources spent correcting bad data may
>>> easily make the net value negative.
>>>
>>> I do not have a solution, but I wish to stress one observation: the
>>> further upstream the quality assurance, the cheaper it is. I feel that an
>>> onboarding environment (both technical and social) that glorifies quality
>>> rather than quantity might be a step in that
>>> direction._______________________________________________
>>> HOT mailing list
>>> HOT at openstreetmap.org
>>> https://lists.openstreetmap.org/listinfo/hot
>>>
>>
>>
>> --
>> *Matthew Gibb*
>> mjngibb at gmail.com
>> (518) 791-8505
>> _______________________________________________
>> HOT mailing list
>> HOT at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/hot
>>
>

-- 
*Matthew Gibb*
mjngibb at gmail.com
(518) 791-8505
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20190607/61cfdee7/attachment.html>