OpenStreetMap
Announcing my candidacy for OSMF Board Elections 2017
Posted by David Dean on 22 November 2017 in English.I have been an active OpenStreetMapper since 2007 (and an OSMF member since early 2016), and have been organising Mapping Parties in my local city of Brisbane, Australia since 2008 (although both activities were a bit quiet for a few years in the middle there - sorry about that!). I want to join the OSMF Board to primarily help increase representation from the AU/NZ/South Pacific Region, and indigenous populations worldwide, and secondarily to bring my decades of expertise in machine-learning to the upcoming challenges of integrating automated mapping with the local-community focus that should always drive OSM contributions.
Since coming back into OSM solidly in the last 5 months, I have re-caught the mapping bug big time. I have mapped locally, and across the world for the Humanitarian OpenStreetMap Team, every single day since July! I am working on re-ramping up my involvement in the project significantly, and I am performing outreach in the local spatial community to increase the awareness of both OSM and the Humanitarian OpenStreetMap Team. I particular enjoy teaching and introducing new members to the project, and have run two Missing Maps events, alongside other local Mapping Parties, and I will continue to organise events locally every month.
I am an active participant of HOT mapping efforts, and I believe strongly in the ability of remote mapping to provide a great starting point for local communities to take ownership of their maps. In fact, I believe that this is the first and foremost principle of OpenStreetMap, that if anything is taking away, or discouraging, local ownership of the map it is probably not a good idea.
If elected, I would seek to improve the representation of the needs of Australia, New Zealand and the South Pacific mappers towards the OSMF Board. In particular, I am very interested, and already working towards, how the OpenStreetMap project and communities can extend our community, models and techniques towards indigenous populations around the world. Indigenous populations often look at the world differently than the Euro-centric origins of the OpenStreetMap project, but I believe that some of the valuable lessons already learnt by HOT in how local disadvantaged communities can take ownership of their maps can be extended and improved upon for the indigenous populations of Australia, New Zealand, the South Pacific, and around the world. To be clear, I am not indigenous myself, but I am in already in the process of building relationships to work towards this goal in the Greater Brisbane area, then hopefully across the entire region, and the world!
One other aspect that I would like to bring to the board will be the present and future impact of automation, machine-learning and/or artificial intelligence on OpenStreetMapping, both generally and specifically for Humanitarian Mapping. It might appear that this is antithetical to my belief in local communities owning local maps, but that is why I want to focus on using these powerful techniques in ways that promote, rather than degrade, local ownership.
I have more than a decade of machine-learning experience, and I want to investigate existing techniques in development by Facebook, Development Seed and others, and develop new ones, and I am working with a team at the University of Queensland to begin a study on how these and similar techniques can be opened to all mappers through the Humanitarian OpenStreetMap Tasking Manager and editing tools in a way that ensures that remote and local mappers are assisted directly in our mapping tools, rather than having the output of an algorithm imported widescale without their direct involvement in the process. I am in particular interested in efforts that can focus on getting ML to people on the ground in things like FieldPapers, and in apps like StreetComplete, Kort and OpenMapKit. We need many more of these great simple-survey focussed approaches, and I think ML can help with guiding these efforts.
You can find out more about me at https://dbdean.com, and my OSM profile at https://www.openstreetmap.org/user/David%20Dean contains links to my many OSM-related activities. If you have any questions, or want to get in contact, please don’t hesitate to ask me here, by email at ddean@ieee.org, or at the Election to Board Talk Page at https://wiki.openstreetmap.org/wiki/Talk:Foundation/AGM17/Election_to_Board, where I will begin to wade through the many existing questions shortly.
Thanks, and Happy Mapping!
Discussion
Comment from aharvey on 23 November 2017 at 00:44
Hi David,
A few questions from me.
What do you see as the main issues we face specific to our region? Do you have any details on local matters which you’d like to take higher to the OSMF level?
I’m very interested to hear about your plans, thoughts or ideas on how we can help encourage indigenous populations to map things important to them. On a local level, http://wiki.openstreetmap.org/wiki/Multilingual_names#New_Zealand documents name:mi for Māori names, but we are lacking any documentation around native Australian languages.
What do you see OSMF can help with that we can’t do already at a country level?
I’m glad to see this on your agenda.
I agreed. The most accurate maps come from the people closest to what they are mapping, both in terms of location and domain knowledge (eg. a sydney taxi driver is going to be able to map taxi points very well) and I think OSM should encourage the people closest to the map objects in both aspects to participate.
Comment from David Dean on 23 November 2017 at 03:34
aharvey,
Thanks for the questions.
I think the main issue for our region is organisation, and that is something that the local AU/NZ/PI informal groups (including myself) are currently working on. There is a huge spatial community down here, but we don’t have a local organisation to help drive the adoption of OSM data by these commercial and non-commercial communities. I don’t feel that there i anything particular that the OSMF needs to do differently to help our region out, but that by providing representation from our region to OSMF I will also be able to provide back assistance to local companies, organisations and local community groups in promoting the use of OSM more widely.
Me too! It’s still very early days, and focusing on relationship building at the moment. My main plan at the moment is to talk to indigenous communities and see if there is interest. This should be their project - I just want to spark an interest, and help it grow.
A lot of indigenous communities across the world are largely invisible in their large, often well mapped, countries (in OSM), and often the features important to a euro-centric map may not relate well to their experiences.
I think that a more inward-facing effort with, or similar to that of, missing maps could help put these - often very disadvantaged - communities on the map, and then give these communities local ownership of aspects of their current and former lands that are important to them. I am already (slowly) working on establishing relationships with local indigenous communities here in Brisbane, but I hope that a position on the board would allow me to examine this idea on a worldwide scope, giving respect to the similarities and differences of indigenous communities across the globe.
Finally, thank you for your comments on machine-learning automation. I think that there is a lot to be said for these efforts, but I think a democratisation of these tools is very important. These tools should be designed to map one road, one building at a time, always with a human in the loop, ideally a local human. I am just as skeptical of large-scale ML efforts as much of the OSM community is of large-scale imports.
Comment from philippec on 23 November 2017 at 22:05
Aharvay asked some high level questions. In contrast to the other candidates, do you already answer practical, day to day questions ?
Comment from David Dean on 24 November 2017 at 00:27
@phillippec, yes I will answer the questions on the Talk page shortly (https://wiki.openstreetmap.org/wiki/Talk:Foundation/AGM17/Election_to_Board). I’m planning on getting through them this weekend.
Comment from Stereo on 25 November 2017 at 23:33
Hello David,
That’s very interesting. Considering that the OSMF voters would tend to be passionate craft mappers, can you talk more about how you see this working?
Thank you for your answers, and good luck.
Comment from David Dean on 27 November 2017 at 02:10
Stereo, I’m still thinking this through, but I envisage one approach is tools like MapRoulette saying something like “Our robot over-lords think there are buildings in this area that aren’t mapped. Can you please map them?”. Similar approaches could be taken with StreetComplete/OpenMapKit for on-the-ground confirmation of ML outcomes.
As for mapping directly in JOSM/ID/etc., there might be approaches that allow one-click building/road/landuse mapping, where a editor can click an automatically created feature to confirm it is real before it appears on the map, or ML tools can refined boundaries of manual edits.
Finally, ML might be able to perform a QA role, or identify possible changes in the map, to allow local mappers to confirm and make the changes themselves.
The most important thing is that there will always be humans in the loop, and ideally local humans, confirming the details before they get added to the map.
Comment from amapanda ᚛ᚐᚋᚐᚅᚇᚐ᚜ 🏳️🌈 on 1 December 2017 at 08:05
Indigenous people are currently very marginalised, globally and locally, so this is a worthwhile goal. Can you tell us some of your past history working with indigenous people, or in activism and organisation around indigenous issues?
Comment from David Dean on 1 December 2017 at 23:54
Rory, thanks for the question. I’m really only just starting out on this particular adventure, and I’m still in the establishing contacts stage. While I agree that this is a very important goal for our community, it can only be lead from within the indigenous communities themselves, so I’m establishing contact with some local communities in Brisbane here at the moment and trying to spark interest. Things are progressing, but slowly.
I have noticed that there was an indigenous mapping conference recently in Canada with some OSM representation by Mapbox (https://www.indigenousmaps.com/). More events like that, and more focused on open data would be awesome.
Comment from joel7 on 6 December 2017 at 13:33
Hey David, I wish you luck!
Comment from joel7 on 6 December 2017 at 13:34
EDIT: Formatting failed :(
Hey David, I wish you luck!
Does this mean you wish to make the maps in our region more complete? Or simply AU/NZ/SR’s effect on HOT?
I feel that we need better aerial first, Australia is pretty bad, but NZ is OK. Still sounds pretty good, A one click for buildings is much better than tracing them out. How close are we to having such a tool? Because mapping out all the houses is time consuming, so you certainly have my interest ;).
Comment from David Dean on 6 December 2017 at 23:52
Thanks ClipArtJoel!
I want to increase representation of AU/NZ/South Pacific mappers on the board, particularly indigenous mappers. This should help make our regions maps better, yes, but I want to build the community in this area of the world. Get a SoTM Oceania running, get more involvement with companies in the AU/NZ/South Pacific OSM communities, etc.
There is already published research that can do this in academia, we just need to start moving it over into the OSM community (this isn’t trivial, of course). What is important to me is that we do this with the local communities in mind: bottom-up, not top-down. Facebook has already started doing ML with OSM data, but I have my suspicions about the process, and I think these tools should be available for all mappers, not just those working for big corporations.
Comment from philippec on 7 December 2017 at 07:56
I would like to represent myself.