[HOT] Buildings and HOT's reputation in OSM

john whelan jwhelan0112 at gmail.com
Mon Dec 11 18:30:00 UTC 2017


I'm not disputing HOT does a lot of good work but what I'd like to see is
more initiatives and projects being run by local groups.

Developing this expertise is I think very important.  Yes we can run
projects in these areas and they do a lot of good but we can do more and
one of the mores is developing the expertise on the ground.

Kathmandu labs I think is an example where they have developed expertise in
GIS systems.  There are African universities with geography departments.
There are people with the ability to become project managers.

When it's locally run you get more buy in.  They will make mistakes and I'm
certain they'll take their project management experience on to do other
things but I think it is important that they take the responsibility.

I'd suggest no more than setting a target of 10% of new projects be ones
that are at the request of organisations from the country that the project
is based in.

At the moment we seem to be hovering on the brink of local governments
using OSM leveraging on the work of the mapping we've done already.
However it needs someone to play and come up with a tool set if you like.
Hopefully the Universities can come up with some useful projects and
processes.  I think it is something we should support but it is outside our
normal pattern.

Trudy has done some work here including statistical analysis.  There is
interest in government circles and in many ways if we can set something up
as a example others may follow.  Kathmandu labs may have some ideas.

If we can work in this way then I think it is worth a few brownie points on
the reputation side.

The perception at the moment is perhaps we know what is best for you.

Cheerio John


On 11 Dec 2017 12:26 pm, "Tyler Radford" <tyler.radford at hotosm.org> wrote:

Hi John,

Since others have commented on your points regarding buildings, I'll add
one comment to your point on HOT support of projects on the ground.

In short, this is a huge area of focus for us. HOT is currently supporting
12 communities with training, devices for mapping, and basic funding to run
events and collect data. This will increase to 20 communities in Q1 and Q2
of 2018.

Disaster mapping and Missing Maps are two of our most important programs,
and supporting communities on the ground will, over the long term, improve
data quality as well as help communities achieve the goals that are most
important to them.

Some initial statistics from 2017 are below.


Organisation Grant New community members Received training
UdeA (Colombia) Microgrant 73 73
Ger Community Mapping (Mongolia) Microgrant 20 20
OSM- DRC Microgrant 10 5
OSM- Zambia Microgrant 27 82
AUW (Bangladesh) Microgrant 260 260
Mozdevs (Mozambique) Microgrant 15 19
OSM-Niger Microgrant 4 10
Crowd2Map (Tanzania) Microgrant 587 740
West Africa Motorcycle Mapping (Sierra Leone) Microgrant 8 12
OSM Mali Device Grant 36 36
Crowd2Map (Tanzania) Device Grant 48 48
OSM Bangladesh Device Grant 14 14
Totals 1102 1319

*Tyler Radford*
Executive Director
tyler.radford at hotosm.org
@TylerSRadford

*Humanitarian OpenStreetMap Team*
*Using OpenStreetMap for Humanitarian Response & Economic Development*

*Help us #mapthedifference
<https://pages.donately.com/hotosm/fundraiser/help-tyler-and-the-team-map-the-world-s-most-vulnerable-places>
by Dec. 31*
web <http://hotosm.org/> | twitter <https://twitter.com/hotosm> | facebook
<https://www.facebook.com/hotosm> | donate
<https://pages.donately.com/hotosm/fundraiser/help-tyler-and-the-team-map-the-world-s-most-vulnerable-places>

On Sun, Dec 10, 2017 at 8:56 AM, john whelan <jwhelan0112 at gmail.com> wrote:

> I would agree with you on shape visually but the area of a building is
> used by some to estimate population and the question is when we change the
> shape we also change the area albeit slightly.  My scientific background
> suggests this is an additional element of error.  However given the quality
> of the mapping it is probably not significant but I'm not qualified to
> say.  All I can do is identify possible problem areas.
>
> The other problem is round buildings, square them and you do get
> significant errors creeping in.  So the process of squaring them should
> ensure they are omitted from the squaring process.  ie select buildings in
> JOSM then hit q may not give the results desired.
>
> Has anyone done a proper analysis?
>
> Better tools may help in the future but there is still the problem of what
> is there currently.
>
> At the moment I get the impression this is considered not to be a HOT
> responsibility even though much of the poor mapping was done by HOT mappers
> working on HOT projects and that is unfortunately the view held by a number
> of OSM mappers that HOT is irresponsible.
>
> Cheerio John
>
>
>
> On 10 December 2017 at 06:26, Scott Davies <mr.scott.davies at gmail.com>
> wrote:
>
>> >From a validation point of view does it matter if the building is not
>> square?  Is it acceptable to square a building even though we know this
>> will introduce an element of approximation or error.
>>
>> John, just on this one point, I'm not sure this is true. The 'element of
>> approximation' is inherent in the mapping process, and is there whether the
>> building is squared or not. If the building is rectangular in reality, then
>> squaring it will make it better represent the actual shape than a freehand
>> version would. I'm not convinced it would generally make any real
>> difference in terms of its area either.
>>
>>
>>
>> On 10 December 2017 at 00:47, john whelan <jwhelan0112 at gmail.com> wrote:
>>
>>> I have considerable coding experience but not in JAVA script which is
>>> required for iD as I understand it and since it about two years to get up
>>> to speed with a new programming language I'll take  a pass on that.  There
>>> is a building_tool available now in JOSM and new mappers are quite capable
>>> of using it.  I was involved with a maperthon recently mapping buildings
>>> and I just had all the new mappers use JOSM and the building_tool.  Jo has
>>> had a similar experience in Belgium.
>>>
>>> There are times when iD must be used for example when JAVA cannot be
>>> installed or its an Apple computer but for Windows machines if you're
>>> mapping buildings the building_tool plugin does an excellent job.  If you
>>> ask the new mappers nicely to install JAVA on their machines before the
>>> maperthon its quite fast to get them up and working.
>>>
>>> Cheerio John
>>>
>>> On 9 December 2017 at 19:35, Phil (The Geek) Wyatt <
>>> phil at wyatt-family.com> wrote:
>>>
>>>> Hi John,
>>>>
>>>>
>>>>
>>>> Seems to be a few new issues here but I will have a go at them.
>>>>
>>>>
>>>>
>>>> ·         I have never been to a mapathon so can't comment on anything
>>>> to do with them
>>>>
>>>> ·         As I understand it, Missing Maps is a different organisation
>>>> that just uses the HOT tasking manager. Their general mandate seems to be
>>>> to work with local communities. If they are not, then you probably need to
>>>> take that up with them (or their participating organisations). I have only
>>>> worked on their projects via the HOT tasking manager
>>>>
>>>> ·         Do you have coding experience that can help with the ID
>>>> building tool?
>>>>
>>>> ·         I still think some alert to save frequently (regardless of
>>>> tile lock time) would be a better solution than extending tile lock time.
>>>> That might simply mean that more objects have been mapped over a longer
>>>> period and the problem compounds.
>>>>
>>>> ·         Correcting buildings when validating/shortage of validators?
>>>> OK if this is unlikely to be the solution then we are back to better tools
>>>> for initial digitising. Can you help with coding?
>>>>
>>>> ·         Wasn't aware of any scoring for validated tiles.
>>>>
>>>>
>>>>
>>>> Cheers - Phil
>>>>
>>>>
>>>>
>>>> *From:* john whelan [mailto:jwhelan0112 at gmail.com]
>>>> *Sent:* Sunday, December 10, 2017 11:07 AM
>>>> *To:* Phil (The Geek) Wyatt
>>>> *Cc:* hot at openstreetmap.org
>>>> *Subject:* Re: [HOT] Buildings and HOT's reputation in OSM
>>>>
>>>>
>>>>
>>>> I think the London mapperthons expect 30% of their attendees to
>>>> return.  So unfortunately 50% can be expected to be "disposable" mappers
>>>> and they don't search out the information as you do.  If we can keep their
>>>> interest a bit longer they make fewer mistakes.  So one thing might be to
>>>> see what we can do to help them return more than once.
>>>>
>>>>
>>>>
>>>> There are two sides to HOT, one is the disaster side and I think that
>>>> side needs the very organised approach to get things done quickly.  The
>>>> other side is the "missing maps" side and that is where I think we could do
>>>> better in involving the locals. Generally speaking mappers feel more
>>>> commitment and involvement mapping locally.  There are some projects run
>>>> out of Africa and its interesting to see the project managers concerned
>>>> develop their skill sets over time. The first projects can be really not
>>>> well thought out but they learn by experience.  OSM in general likes to see
>>>> local mappers making decisions about imports etc.  Unless we can develop
>>>> the mapping skills of the locals they aren't going to feel involved and I
>>>> think that is important.  It's when you see the cafes and other POIs
>>>> sneaking on to the map that you can be reasonably certain that there are
>>>> locals getting involved and that is where HOT can get a few brownie points
>>>> and at the moment I think it needs every one it can get.  This is politics
>>>> with a small p.
>>>>
>>>>
>>>>
>>>> New mappers first time mapping using iD for buildings you might be
>>>> lucky to see twenty buildings.  Give them a building_tool and you'll get a
>>>> lot more out of them.  When they map the tile is locked for two hours.  At
>>>> the end of that time the tile becomes available again to be mapped.  The
>>>> first mapper may not have uploaded their buildings.  A second mapper now
>>>> maps the same tile and when they both upload that is when I think we get
>>>> the double mapping which is a waste of mapper time and not only that but it
>>>> screws up calculations about how many buildings there are.  With a four
>>>> hour lock we stand a much better chance that two mappers will not map the
>>>> same tile at the same time.  Even uploading every thirty minutes would
>>>> reduce the number of double mappings.
>>>>
>>>>
>>>>
>>>> Correcting buildings when validating?  It takes about three times
>>>> longer to correct a badly mapped building than it does to map it from
>>>> scratch with JOSM and the building_tool.  In Nepal 70% of the mappers
>>>> mapped once.  Their building mapping was exceptionally poor.  When faced
>>>> with large numbers of poorly mapped buildings it seems difficult to find
>>>> validators who are motivated enough to go in and fix the problems.  I'm not
>>>> one of them.  If the mappers are only going to map once any feedback will
>>>> be ignored.  We know that giving feedback within 24 hrs motivates mappers
>>>> and catches early errors so we get better quality work but we don't have
>>>> the validators available to do this.  Validation works best if its done at
>>>> the start of a project as the project progresses.  Validating three month
>>>> old work is much more work, bad habits will have set in.  Instead of
>>>> problem avoid its problem correct and that takes more validation effort.
>>>>
>>>>
>>>>
>>>> TM3 hopefully will improve this by giving a score for tiles validated.
>>>>
>>>>
>>>>
>>>> Cheerio John
>>>>
>>>>
>>>>
>>>> On 9 December 2017 at 18:28, Phil (The Geek) Wyatt <
>>>> phil at wyatt-family.com> wrote:
>>>>
>>>> Hi John,
>>>>
>>>>
>>>>
>>>> I have no formal role in HOT, just a casual OSM mapper so all these
>>>> comments are from that perspective. I have participated in HOT projects via
>>>> their tasking manager. I am also not a coder but am aware of the process
>>>> involved in respect of the ID Editor and Tasking Manager development.
>>>>
>>>>
>>>>
>>>> I don't think building issues are restricted to HOT projects. Indeed
>>>> when I first started mapping in my own neighbourhood I didnt really have a
>>>> clue on how to map buildings but over time I found videos, joined mailing
>>>> lists, found LearnOSM, found tasking managers across the globe and
>>>> generally became a better mapper. At each step I learnt more and hopefully
>>>> became a better mapper. I still don't do any validation as I don't consider
>>>> myself experienced enough in JOSM and lots of the other validation tools. I
>>>> have participated in some map roulette challenges.
>>>>
>>>>
>>>>
>>>> As you have indicated in previous emails, a building tool in ID may
>>>> stop some of the issues you mention and from my investigation this is
>>>> actually underway already (and has been for some time). Any assistance you
>>>> can provide would be appreciated by everyone involved.
>>>>
>>>>
>>>>
>>>> I will try and answer specific points in your email from my personal
>>>> perspective.
>>>>
>>>>
>>>>
>>>> ·         I think all buildings should be polygons rather than points.
>>>> Better to teach people how to map as polygons rather than expect another
>>>> mapper to replace a point with a polygon at some time in the future (if
>>>> ever).
>>>>
>>>> ·         Incorrectly mapped building - I would always try and correct
>>>> the error if I had the required skills. If there were many on the tile in a
>>>> tasking manager project that were poorly mapped I suspect I would
>>>> invalidate the tile if I could not fix them. I would also expect some
>>>> details from the validator, maybe with guidance on where good instructions
>>>> are for splitting the building and maintaining any past history on the
>>>> object.
>>>>
>>>> ·         Likewise for buildings 50% greater than actual - I would do
>>>> the same as above - guide the mapper on what they have done wrong and lead
>>>> them to better resources.
>>>>
>>>> ·         Buildings mapped twice. I am on a crappy Australian fibre to
>>>> the node connection that regularly crashes so I save regularly (20 - 30
>>>> objects). I dont quite understand your comment that a four hour tile lock
>>>> limit would eliminate this problem - seems completely wrong to me. I would
>>>> certainly be saving more frequently than every 4 hours. Maybe a reminder
>>>> popup, after 100 objects, might be a better solution to ensure folks are
>>>> regularly saving.
>>>>
>>>> ·         I don't agree with the view of HOT that "We are the
>>>> professionals and we know best". Having lurked on the HOT slack channels I
>>>> have seen how they size up disasters, deal with local OSM groups and other
>>>> disaster relief organisations before embarking on projects. Indeed on a few
>>>> occasions they have not undertaken any projects when the local communities
>>>> have indicated they have the situation in hand. In those cases they simply
>>>> offer support if required or use their communication channels to direct
>>>> mappers to the other task managers (if desired). There are regular
>>>> references to local OSM groups prior to project commencement.
>>>>
>>>> ·         As for what is acceptable mapping for a building. The best
>>>> we can hope for is improving tools, educating mappers, more validation
>>>> tools plus willing volunteers (or dare I say it, paid workers) to keep an
>>>> eye on things and help the community make OSM an always improving product.
>>>>
>>>>
>>>>
>>>> Volunteer gathered information is a bit of a dark art at the best of
>>>> times and many folks/governments are still coming to grips with how it all
>>>> works and how beneficial it can be. Is it perfect...not really, can it be
>>>> improved...always. I think the HOT (and other) tasking managers and the ID
>>>> editor are always improving with better task details, more links to
>>>> resources etc. I think it's up to all of us to contribute in any way we can
>>>> and put forward ideas, time, funds or expertise to make things better.
>>>>
>>>>
>>>>
>>>> I wasn't aware of the OSMF mailing list so I will join that as well and
>>>> read up what has been happening.
>>>>
>>>>
>>>>
>>>> Cheers - Phil
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> *From:* john whelan [mailto:jwhelan0112 at gmail.com]
>>>> *Sent:* Sunday, December 10, 2017 7:00 AM
>>>> *To:* hot at openstreetmap.org
>>>> *Subject:* [HOT] Buildings and HOT's reputation in OSM
>>>>
>>>>
>>>>
>>>> Recently there has been some discussion of HOT's input into
>>>> OpenStreetMap in the OSMF mailing list.
>>>>
>>>>
>>>>
>>>> Perhaps one of the problem areas is mapping that is less than ideal.
>>>>
>>>>
>>>>
>>>> Basically HOT mainly maps highways, landuse=residential and buildings.
>>>>
>>>>
>>>>
>>>> These shouldn't be difficult to map correctly.
>>>>
>>>>
>>>>
>>>> Buildings appear to be the most problematic.
>>>>
>>>>
>>>>
>>>> I think we need to think about why we are mapping them.  Is node good
>>>> enough?  There would be less room for mistakes.
>>>>
>>>>
>>>>
>>>> If we need outlines and there good reasons why an outline is more
>>>> valuable than a node then we need to define what is acceptable.  Or do we
>>>> even care?  and its the do we even care part that is perceived to be the
>>>> case by some within OSM and that perception is something we should care
>>>> about.
>>>>
>>>>
>>>>
>>>> From a validation point of view does it matter if the building is not
>>>> square?  Is it acceptable to square a building even though we know this
>>>> will introduce an element of approximation or error.
>>>>
>>>>
>>>>
>>>> What should be done with a building=yes that covers more than one
>>>> building?  Do we expect the validator to map each building or just
>>>> invalidate the tile?
>>>>
>>>>
>>>>
>>>> What should be done when the building mapped is more than 50% larger
>>>> than the image?  Invalidate the tile?
>>>>
>>>>
>>>>
>>>> We are still mapping buildings twice.  I suspect some mappers are not
>>>> uploading within two hours.  Getting mappers to upload every 30 minutes max
>>>> would go a long way to reduce this, extending the tile lock to four hours
>>>> would almost certainly eliminate it.  Recently on high priority project
>>>> I've seen in the order of a hundred buildings double mapped.  They have
>>>> been done within the last two weeks so it is an ongoing problem. There is a
>>>> new tool that detects these so they aren't the problem they once were but
>>>> someone has to run the tool.
>>>>
>>>>
>>>>
>>>> If HOT could support a few more projects that were from the community
>>>> on the ground rather than the "We are the professionals we know what is
>>>> best" which appears to be perceived sometimes from the number of projects
>>>> for the RED Cross or other northern hemisphere charities that might also
>>>> help the reputation and relationship.
>>>>
>>>>
>>>>
>>>> So two points here on one message first is can HOT's reputation be
>>>> repaired and I suspect that is longer term problem that will take time and
>>>> a lot of effort rather than a PR job.
>>>>
>>>>
>>>>
>>>> Second would someone care to comment on what is acceptable mapping for
>>>> a building and what guidelines can we give to validators?
>>>>
>>>>
>>>>
>>>> Thanks 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
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20171211/95be5021/attachment.html>


More information about the HOT mailing list