[Osmf-talk] Country-level hosting of addresses overlay layers and more detailed imagery for mappers (was: Consultation on fundraising strategy)

[Emerson Rocha]

*TL;DR: by providing hosting when is unlikely to have alternative, both
increase awareness OpenStreetMap Foundation as public good at country
level, yet on the same action, provide more detailed content to mappers
(e.g editor-layer-index options)*

I'm writing here not as if I would have to implement it. While I do not
have experience with the very specific task of converting orthophotos to
tiles, in addition to help with conversion scripts for the vector files to
tiles, I do have expertise in infra and could help keep it online cheap as
a volunteer. Anyway, some places I'm interested in do have reference data,
and aren't friendly yet as a layer, but this seems to be more common. I
will break it into 5 parts.

1. Some existing practical implementations (not yet an OSMF dedicated
strategy)

To give an example of country level of both types:

   1. [overlay data layer] With iD editor (or any other editor which uses
   editor-layer-index) try to edit any city in Brazil and select the available
   "IBGE Nomes de Ruas" overlay layer (this was done by others here). This
   heavily helps mappers here to update OSM (unless history of an element
   shows the name already was updated, sometimes changesets even have
   reference to local laws that make the new name official). It was based on
   these shapefiles <
   https://www.ibge.gov.br/geociencias/organizacao-do-territorio/malhas-territoriais/28971-base-de-faces-de-logradouros-do-brasil.html?=&t=downloads>
   and current version is based on 2019 (not the 2021 latest year).
   2. [background imagery] From editor-layer-index <
   https://osmlab.github.io/editor-layer-index/>, South Africa is the only
   non-global layer available in Africa. Give a look at <
   https://community.openstreetmap.org/t/new-25cm-aerial-photos-available-from-chief-directorate-national-geo-spatial-information-cd-ngi/8452>.
This
   is a background layer (e.g imagery, but much better level of detail),
   however is quite heavy on server storage compared to overlay layers.


(By the way, others here with examples of layers which show addresses in
your regions, feel free to explain how you do)

2. Examples of good reference data and imagery which exist, but not ready
to use for the mappers (or just undocumented)

The streets layer in Brazil (this one is not provided as a layer by the
government, so I guess someone setup Mapbox and let it free for mappers) is
working, but in general the idea when doing it is keep up to date. My
argument is that OSMF could be the owner of infra to host also such kinds
of reference layers, in particular with addresses data (at least when it is
known to exist). This is also better than mappers copying from private
services when the government already has and is open license.

On recent calls on the Steve thread <
https://lists.openstreetmap.org/pipermail/osmf-talk/2023-May/008566.html>,
I did some research. To my surprise, for example, Zambia does actually have
very high detail imagery (10cm) and some other information which could be
overlayed (check <https://www.map.gov.zm/arcgis/rest>). It's slow on very
higher levels of zoom, but definitely far better to use than generic
Bing/ESRI/Maxar (but unlikely to handle if more people keep accessing it at
the same time). In Kenya, the open data server (which was also cited as a
potential source by UGA-ITOS, which also helps UNOCHA with the Common
Operational Datasets)  is currently requiring authenticated access, so I
think maybe they changed (check <https://www.opendata.go.ke/>). I started
to look at other countries, but while obviously some of them just don't
have the imagery online (might likely do have on disks if asked) the
mappers from several work regions aren't using the tools with the same
potential their government already has.


In case of regions that truly do not have addresses in vector files which
could be converted to data layers (often major roads and big cities aren't
a problem, but are for residential roads), it would be a different topic
than I'm focusing on here, but in general is better assume that may have
addresses, just not easy to find it. But I already found different places
on OSM where someone (likely from government or paid by the government)
updating addresses of their near small region, which in my honest opinion,
unless they have some conflict with DWG or licensing (e.g. how they get
back the data), it could be documented/encouraged. But in the practical
side, the on-the-ground rule could be used for the local government to be
sure to add the plate with the street names and know that it could be asked
to prove with pictures (or trully only use names that are known by the
locals).

3. On serving imagery/data layers be or not relevant to OpenStreetMap
Foundation (considering is open to donations from government)

I know a major reason for OSMF not taking some roles is to avoid competing
with the ecosystem around. However the hosting of very high detailed
imagery and (which requires help from local users to re-run the scripts to
re-generate updated dada layers over the years) have better up to date
reference layers (like the addresses which a mappers could consider if
adding of not to OpenStreetMap) is already one task which is non-profitable
to offer for free to mappers. And lack of access directly affects data
quality and level of details. I mean, sometimes Bing and ESRI can be more
up to date, but in my experience as a mapper, features that are older (but
higher detail and extra care to be precise on coordinates) often are very
useful.

Very few mappers would have additional skills to know how to set up a
custom background layer or how to convert the scripts to load on their
editors. In Europe more often there's people from the community doing this,
but with some planning, it could be done in other regions.

Also, the mere fact of just hosting higher levels of detail imagery and
being a place for volunteers (or government itself) suggest more up to date
data layers is sufficient to OSMF not make any promises on what the general
mapper community would do or not with this. This kind of thing is "boring",
but somewhat tends to be repetitive across countries. The difference of
today's editor-layer-index is OSMF (in special as owner of donated
hardware) be willing to also host such content which today either is locked
in someone else's shapefiles in a computer or simply the server used at
regional level would be made private access if starts to have too much
bandwidth use (which likely be the case if suggested for mappers).

>From a fundraising perspective, since the suggestion is keep initial and
continuous cost low to OSMF scale regions, I would suggest be better just
not have "package prices'', and leave donations to be fully voluntary by
governments and focus on mappers (which likely will have not only generic
local mappers and overseas hobbyist, but actually employers from
government, which would naturally become stakeholders to support in future
OSMF relevance). This is also good because it can take time to find
potential content sources, prepare the data and then prepare the hardware,
and then expect the feedback.

4. On keeping it cheap the costs make it online (so OSMF can use donations
justified by this to other areas)

For hosting the hardware, typically a university with support for on own
data center (in Brazil https://ix.br would be the place to search for
candidates) or worst case, any friendly Internet Service Provider (similar
to how Netflix already would use as cache server
https://openconnect.netflix.com) could do it.

On the hardware itself, unless others have better alternatives, my
suggestion would be we look for donations of servers which are being
decommissioned from datacenters (even between different countries, the
tendency of specifications is similar). Either with a recommendation letter
from OSMF (or, if it wouldn't be of interest to OSMF, me and others just
ask Steve Coast to do it outside), this is viable. Often at least the disks
will need to be changed for bigger new ones (look as example the strategy
used in South Africa), but this is unlikely to be done by the same donor of
the server(s). These extra parts in which original hardware cannot be used
likely require calling other sponsors (again, some kind of letter of
recommendation or better).

People to fix/upgrade/prepare the hardware will depend on where the country
will be. But if the source or destiny of a hardware would be in Brazil,
this wouldn't be a problem and my peers here (most nor related to OSM, but
either working as sysadmins or hackerspaces) could help other regions
depending on the language they speak.

What is granted to involve costs is shipping and smaller non predicted
issues (e.g. if you need some cables or some piece of hardware too cheap to
ask for a donation, but not for who is doing the repair/upgrade).

The major permanent costs to OSMF would be the paid employers to keep the
infra after it is online (this might be a good reason to decide with them
the minimum acceptable donated hardware), including having a disaster
recovery plan ahead. Both some spare hardware and more disks than what goes
on the production server for imagery would be needed.

Anyway, it is possible to do this, even without OSMF, but at a lower
scale/focus. But OSMF owning donated hardware and making sure content very
focused to mappers is kept available (when sometimes not even the original
may be), is one strategy of preparing itself to be more welcomed directly
from the government without compromising what mappers would do it not.

5. Final comments

I edited the title, but this was mostly a more focused reply to email weeks
ago on fundraising (because could make additional reasons for local
governments get closer to OSMF) and, obviously, the recent discussions on
addresses and the fact that without such layers this make less likely
mappers to improve street names on the map. The background imagery likely
would be the most demanding on hardware (and need far more planning), but
the idea of looking for potential reference addresses vector files and
putting them online (and we try to do our best to keep them updated when
source updates them) as a layer seems a win-win sooner.

As for content not already imagery, I do believe in special data relevant
for geocoding (street address, but if available house numbers) as long as
log overly fragmented to automated update of the layers, it could be done.
A counter example is when the government also has vector dada for points of
interest, but this could create too many layers on each region, while it
might be better to focus on addresses or features used for routing.

Also, from the drafted https://wiki.osmfoundation.org/wiki/Cluster_A, one
idea already was an attempt to make it friendly to have more volunteer
sysadmins, which would fit perfectly with host additional country and
province level content.

Att.
Rocha

---------- Forwarded message ---------
De: Mikel Maron <>
Date: ter, 21 de mar de 2023 10:49
Subject: Re: [Osmf-talk] Consultation on fundraising strategy
To: OSMF Talk <osmf-talk at openstreetmap.org>


Olivier asked
> Do we have materials that explain in a quick/effective way what is OSM
and how the money they will provide will be used?

Yes, developing new communication materials is part of the campaign. This
is an area where help from writers and designers, from all over the world,
will be key.

Rocha suggested
> propose to the board vote about being willing to accept government
donations

No need to vote, we'll gladly take government money. This fundraising
campaign will focus on diversification. Expect public sector money will
take a long term effort to produce results. There are opportunities with
governments and multilaterals like UN OCHA, mainly as programmatic grants.
For example, there are some programs in the EU that support open source. If
anyone comes across something that might fit, please let us know. We will
look into these opportunities, and see what might fit.

-Mikel




On Tuesday, March 21, 2023 at 02:12:14 AM EDT, Emerson Rocha via osmf-talk <
osmf-talk at openstreetmap.org> wrote:


I would propose to the board vote about being willing to accept government
donations: several are directly and indirectly using OpenStreetMap data for
years, even if to revalidate their own geodata.

Examples of how is used:

1. Going on use country by country varies. It's not straightforward to
summarize, however it does exist, just not explicitly organized editing. I
think they're likely to focus on fixing/improving features which are
relevant for what department they work in, which both non international
administrative boundaries (often the very first edits) and roads network be
a common trend. Likely places with far more content, there's more heavy
use. (this is something that could eventually be documented upfront on the
OSM wiki)
2. But at international level, just to give an idea of tip of the iceberg:
OpenStreetMap data is the second major dataset provider on UN OCHA data
portal https://data.humdata.org/organization?sort=datasets%20desc, just
after the World Bank (which is mostly for statistics, not what goes on
OpenStreetMap). And from a significant amount of datasets by organization
uploaders, quite often OpenStreetMap data and directly related ecosystem of
tools are used in part of their data workflow (not hard to think,
considering the world-level alternative tends to be proprietary or
shapefiles). Also, the idea of "Humanitarian OpenStreetMap Team United
States Inc" be some kind of a bridge between humanitarian sector and
OpenStreetMap is erroneous: first, by far, most used data from
OpenStreetMap are from non organized editing at all, and then, from what
HOTUSI actually brings for data on OpenStreetMap, buildings without any
metadata (not even if they if they're a house) while take space on map and
is viable have large numbers, have no use at all in emergency response (one
starting point for what is used:
https://humanitarian.atlassian.net/wiki/spaces/imtoolbox/pages/242090029/Natural+Disaster+CODs);
not even UN OCHA have any documented use for these generic buildings
promoted by HOTUSI as "humanitarian mapping", but they have for so much
more which OpenStreetMap have. I could talk much more here, but I'm keeping
it short.

> Frederik Ramm said:
> I'm not advocating for either, just pointing out that the need for funds
> is not god-given. Frequently on these mailing lists, a knee-jerk
> reaction of people to various problems is not "how can I help" but "the
OSMF should pay someone to do it" (...)

I'm anxious to help on this, and do it for free, pro bono publico! Then we
could go for others to use their contact network with governments, but even
without more people with me, it is feasible I get some sorts of "letters of
recommendation" from trusted professionals on how OpenStreetMap data is so
essential. It's also a "low risk, high reward" approach, not just because
it's a niche which OpenStreetMap is become the open alternative without
replacement (even comercial alternatives which, for example, can deal with
"world views" on disputed borders depend on OSM data), but because is would
be very, very weird go for government donations (which can sometimes even
be predictable commitment 3 to 5 years ahead) while would be public know
someone would take a %. The logic is similar to why individual OSMF
membership donations would cause trouble as part of this paid fee job, but
in case of going through this kind of donations, a bigger network of
contacts is better.

> Steve Coast says
> One of the many advantages of this is that companies often find it easier
to fund something if there is a reason, something they get in exchange,
like conference slots and so on, rather than throwing money into an eternal
black hole, with nothing to show for it.

No idea how others here have about the government, but to say upfront, how
the government spends money is different from commercial companies. This
thinking is partially applicable. Assuming one is able to prove
OpenStreetMap as a public good (and I personally would focus as
country/province/municipality level, not as foreign aid) then implies is
can suffer Free-rider problem <
https://en.m.wikipedia.org/wiki/Free-rider_problem>, which means the
government can do it, if not profitable from a market perspective. However,
the reason I partially agree is, similar to how companies as donors would
try to act in self interest, a government could naively attempt to try to
influence things related to disputed territories and the default place
names, so it may totally be worth reinforcing things upfront (but this
might already be ready, on this document from 2013
https://wiki.osmfoundation.org/w/images/d/d8/DisputedTerritoriesInformation.pdf).
Governments also often are a source of reference data, so it can be
stimulated to explicitly make it compatible to add to OpenStreetMap, even
if it is not granted if it will be imported. Maybe there's other things
which could be acceptable in exchange (even if it means those who do it are
also volunteers), without any new compromise than already is possible. I
also think that this is a moment of going after governments, even if lower
values are given to OSMF, and then use the contacts to get rights with a
more formal way to import data *to OpenStreetMap* (it's better than letting
potential future competition do it alone). It's up to suggestions of others
any other point I'm missing.

As last comment, others may see less problematic, but for government (if
the contacts are the ones who would think as "aid", not internal use), I
believe is better not accept "earmarking donations'' (WikiMedia Foundation
is successful in avoiding it, but charitable organizations often not) and
while is obviously good multi-year commitment, consider limit how my any
single government could donate per year (this reduces incentives to make
threats of stop donations to force some decision). While (at least if
considering foreign aid) it is easy to find massive numbers, by going with
a lower average, it simplifies use of contact networks to make more
countries/provinces/municipalities get engaged and reduce the need for
higher justification on those which could pay more.

Att.
Rocha
-- 
Emerson Rocha
Full stack developer at Alligo
Transdisciplinary researcher at Etica.AI
Member of  The IEEE Special Interest Group on Humanitarian Technology (IEEE
SIGHT)
Member of The IEEE Global Initiative on Ethics of Autonomous and
Intelligent Systems
_______________________________________________
osmf-talk mailing list
osmf-talk at openstreetmap.org
https://lists.openstreetmap.org/listinfo/osmf-talk
_______________________________________________
osmf-talk mailing list
osmf-talk at openstreetmap.org
https://lists.openstreetmap.org/listinfo/osmf-talk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmf-talk/attachments/20230522/bf99c613/attachment.htm>