[HOT] Getting organised

john whelan jwhelan0112 at gmail.com
Tue Jan 5 18:30:46 UTC 2016


The post was basically some ideas and there are quite a few in there even
for me.  There are others such as modifying the task manager so that once a
tile has been marked done only josm can be used to validate.

Validation is difficult basically look through the projects and you'll see
very few tiles get validated.  We don't have enough validators.  Even the
tools we have available or josm validation aren't used josm is too
difficult to teach new users.

Also for some time now we've taken the view that it isn't important and can
be left to the end.  Yes sure then the validator has to wade through and
correct the same error by the same mapper 25,000 times its boring to say
the least, and one reason validation isn't popular.

Should we run a job once a day that checks the tiles, any done by polygot
or Ralph get automatically validated?  The risk of their work being below
standard is very low and it would reduce the effort needed for validation
there are very few people I'd put in this class by the way.

Another idea is some sort of way to detect a new mapper who has worked on a
tile for more than ten minutes then validate their work.  Validate the
mapper rather than the tile?  It probably needs a mixture of the two.

Mapathons, I learnt a new French word when I heard someone talk about data
quality and maperthons.  I forget the drop out rate but is it worth sending
feedback to someone who will never map again?

I don't think realistically we can do much on the validation front we don't
have the resources.

Should we use Google mailing groups for niches to engage people?  I find
them useful in joining with other content creators to discuss technical
problems in making content for a train simulator.

Perhaps my mention of reserving a couple of slots for MSF and red cross was
misunderstood.  Are we in danger of a top down approach where other
organisations find their projects get lost by the sheer number of projects
put up by these two organisations should we give some shelf space to others
such as Bangladesh?

The ideas need reflecting on, simpler projects improve data quality by
being more easily completed by new users.

Not every idea can be implemented tomorrow even if it just makes people
stop and reflect this is probably a good thing.  Some of these ideas are
major changes to the way we work and major changes need thinking about and
the implications understood before making them. Hi

Time to put the kettle on.

John
On 5 Jan 2016 12:27 pm, "Dale Kunce" <dale.kunce at gmail.com> wrote:

> John,
>
> I agree that HOT is evolving. I've been doing a lot of work personally to
> clean up the Tasking Manager. Most of the discussion has been happening on
> the Tasking Manager and Activation lists so it hasn't made its way to the
> main HOT list. I reached out to the appropriate TM user and after checking
> with them archived and deprioritized a lot of projects that had not
> received any love recently. Blake also brought up the work and we had a
> good conversation in the (Projects on the HOT OSM Tasking Manager - lots
>  of them) thread.
>
> I agree that we need to do a better job of managing and promoting our
> projects. I don't agree with your proposal for doing it however. I
> appreciate your proposal to reserve a couple of spots for MSF and Red Cross
> but I think that ultimately it wouldn't have the effect you are hoping for.
>
> MSF and Red Cross are driving tons of new users to HOT and OSM through the
> Missing Maps projects. Having our projects on the front page isn't
> mandatory to help complete our projects for us because our mappers are
> mostly coming from mapathons or are searching for Missing Maps tasks. We
> make a huge effort to finish all of our tasks, even for the Red Cross
> spending lots of staff time to ensure they are completed. Lastly, I ensure
> you we don't needlessly create tasks where we won't be doing field work.
>
> I love the idea of a community spotlight somehow. Someway that we could
> focus our collective attention to a not as well resourced project.
>
> Basically all of this is to say that ensuring that when you create a
> project in the Tasking Manager *you* are taking responsibility to ensure
> that it is completed. In the past project creators have not effectively
> managed resources and thus many left over forgotten half finished projects.
>
> I'm happy to keep this conversation going as I'm very keen to get the TM
> working the best way possible for the entire community.
>
> Dale
>
> On Tue, Jan 5, 2016 at 10:56 AM, Leon van der Meulen <
> leonvandermeulen at gmail.com> wrote:
>
>> Nice email John, I totally agree with this idea. What I hear from new
>> mappers is that they find it difficult to pick a task themselves. Something
>> I found difficult myself in the beginning was the fact that tasks for
>> experienced mappers are also shown in the first screen, e.g. imagery
>> adjustments, road checks. Often you'd only see the note when the task is
>> already opened. It would be nice to have some kind of filter (maybe even
>> automatically set) for new, intermediate and experience mappers in
>> combination with the prioritising you proposed.
>>
>> About the validating Jim was talking I also strongly agree. Especially
>> for mapahtons it is crucial to have validators who are there at the same
>> time. Personal and immediate feedback really helps a lot.
>>
>> Best,
>> Leon
>>
>>
>> 2016-01-05 15:24 GMT+01:00 john whelan <jwhelan0112 at gmail.com>:
>>
>>> HOT is evolving, we started off with mainly experienced OSM mappers
>>> mapping in response to short term emergencies.  Now do do a wider range of
>>> tasks with a wider range of mappers.
>>>
>>> We're seeing organisations such as MSF and the Red Cross guiding people
>>> who want to help them to mapping with HOT, so our mappers are changing.  It
>>> helps engage volunteers.
>>>
>>> We're starting to see more GIS and mapping being organised locally but
>>> still not as much as I'd like to see.  The locals are better at reading
>>> street names than people mapping from imagery.
>>>
>>> Currently http://tasks.hotosm.org/ is our splash page if you like.  At
>>> the moment we see two or three high priority projects followed by medium
>>> priority projects in date order, the most recent last.  So the medium
>>> priority projects get their five minutes of glory then drift down into the
>>> mass of medium projects.
>>>
>>> We end up with a large number of projects that get 25% done, 5%
>>> validated and drift on for years.
>>>
>>> I suggest we have ten slots set at high priority, that way we stand a
>>> chance of getting a few projects completed.
>>>
>>> Of these I suggest that MSF, American Red Cross get given two permanent
>>> high priority slots.  That would concentrate their minds on which of their
>>> projects are most urgent.  At the moment I suspect mappers are seen as a no
>>> cost resource and some projects that are still around where the ground
>>> workers have finished their tasks but we're still mapping.
>>>
>>> Of the others, I'd give three to other organisations for a period of say
>>> three months.  For example Cameroon has a mapping department, Bangladesh
>>> has an active community, Nepal again has an active community and there are
>>> others.  Let them decide which projects will do them most good.  They may
>>> make some mistakes but I hope it will develop a sense of ownership.
>>>
>>> The other five I'd let the board decide.
>>>
>>> Then we have the dead wood projects that clutter up the medium priority,
>>> these are projects where no one has mapped or validated for more than a
>>> month.  It maybe the imagery is poor, the instructions too complex, or
>>> whatever.  I'd downgrade these to low priority, that way the active
>>> projects would stand out more and again stand more chance of completion.
>>>
>>> It might even be worthwhile having a weighting attached to each project,
>>> the more active projects or highly validated projects move up the list the
>>> less active ones move down.  It just needs a tag for the task list to
>>> order.  The actual computation can be done off line say once a day and
>>> someone else can sort out the algorithm.
>>>
>>> Can we get more out of our mappers?
>>>
>>> I think we can.  On project 1390 I validated as people mapped.  New
>>> mappers got feedback, the advantage is they get a bit more involved and
>>> feel engaged.  So we got more tiles out of them.  We got 6 or more tiles
>>> out of a number of new mappers or ones who had mapped less than six tiles
>>> before, typically I'd expect one or two tiles from them.  On the data
>>> quality side I think it was up.  By correcting problems early on the new
>>> mappers made less mistakes that need intervention later on from the
>>> validation side.  However it takes effort, the faster the feedback the
>>> better the results.
>>>
>>> Validation is a tricky.  HOT isn't traditional OSM where individuals
>>> like to map park benches or waste bins.  There is a service level to be
>>> met, and if the building is five feet out, well that's probably good enough
>>> although we  should strive for accuracy.  It's a bit lets get something
>>> useful done with the resources we have.  Normal good management technique
>>> is catch them doing something right.  So invalidating a tile is the last
>>> thing you want to do.  I normally correct then leave a note, such as added
>>> fourteen settlements. Where possible third party it JOSM validation picked
>>> up the following: African wiki suggests higway=living street in an African
>>> small village may not be appropriate and I must confess I normally delete
>>> any zebra pedestrian crossing I come across in Africa.  There is another
>>> issue with experienced OSM mappers HOT has its own conventions, such as
>>> everything that looks like a building is mapped as an area and tagged
>>> building=yes.  Some OSM mappers use a node tagged building=hut etc.  Giving
>>> feedback is delicate because in the OSM world what they are doing is
>>> acceptable.
>>>
>>> The worst validation is by someone who has mapped two tiles.  I recently
>>> double checked one and JOSM validation picked up a dozen problems and
>>> visually there were a few more.  This is a data quality issue and it
>>> becomes a matter of are any of the validated tiles to be trusted.  In Nepal
>>> this was a major issue.
>>>
>>> We don't have enough good tactful validators.
>>>
>>> 1390 was also interesting in that I came across an MSF volunteer, new
>>> mapper who mapped accurately, did a fair number of tiles but when it was
>>> complete wanted to know which MSF project to map next.  So project managers
>>> need to have something in the pipeline.
>>>
>>> Project managers need to understand the mappers.  We have more
>>> inexperienced mappers than experienced ones.  Projects that demand only
>>> buildings and have documentation on how to map them work well.  Projects
>>> that ask for landuse=residential and connecting highways to be tagged
>>> unclassified work well.  Ones that ask for forests, and everything else
>>> don't work as well.  New mappers get confused and give up and take up
>>> basket weaving instead which is not what we'd like.
>>>
>>> Mapping should be fun.
>>>
>>> Anyway there is enough to mull over.
>>>
>>> Cheerio John
>>>
>>>
>>>
>>> _______________________________________________
>>> HOT mailing list
>>> HOT at openstreetmap.org
>>> https://lists.openstreetmap.org/listinfo/hot
>>>
>>>
>>
>> _______________________________________________
>> HOT mailing list
>> HOT at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/hot
>>
>>
>
>
> --
> sent from my mobile device
>
> Dale Kunce
> http://normalhabit.com
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20160105/e61dd68e/attachment.html>


More information about the HOT mailing list