OpenStreetMap
We all know the OSMF has a small membership compared to the mapping community. Worse, it is skewed towards certain countries. You might hear people say Germans and US Americans dominate the membership.
In a perfect world, all countries have a similar participation rate to OpenStreetMap, and mappers from all countries participate in OSMF by the same degree. The first is not something that is easily changed, but the second should clearly be our ambition. In France, a discussion about this led to a recruitment campaign with the explicit goal of rebalancing. This piqued my curiosity. Oh, and it increased membership with 90 people (from a mere 42).
But numbers without context are of little use. Guillaume (user Stereo) supported the French with membership statistics, and hooked me up with a list of membership by country for all countries. Now I’d say OSMF membership would ideally reflect the mapping community - skewed slightly for countries with a bigger data-user community. Unfortunately, no thorough statistics about mappers by country is available. Fortunately, there is something close to it: “the number of daily mappers” in a country. It is not a perfect measure:
-
hdyc has a de facto “estimated home location”. This would allow to take some of the random noise out of the equasion
-
“active mappers” (people with at least three months with a mapping day) might be a better measure
-
in some cases, it’s impossible to find the country of people who only map abroad
But it’s close enough, and it’s available. Pascal Neis, deserving his last name as usual, sent me a list of all countries with the avarage number of daily mappers over the last 12 months.
Numbers with context
The easiest useful measure in this context is the membership rate, which allows to compare countries. For the world, you get 26 OSMF members for every 100 daily mappers. This rate varies from 0 in Tanzania to 94 in the US. On a map, it looks like this:
For greater detail, I made a table:
First we have the number of active mappers and the number of OSMF members.
Then we look at the membership rate. The column “expected membership” gives you the number of members you would expect if all countries were the same.
The representation rate is 100 when a country is perfectly average. Germany scores 156, which means that for every 100 expected OSMF members, they actually have 156. A country like Russia is heavily underrepresented with just 29 OSMF members for every 100 we would expect.
Since a day without learning a new thing is a lost day, I played around a bit with interactive tables and found an example on CodePen. It was surprisingly easy (even without really understanding JavaScript) to adapt it to need. So you can now play with the table yourself! You can use it to easily identify over- and underrepresented countries, including your own.
For privacy purposes, countries with less then 1 or 2 OSMF members are shown with “-1” instead of their actual values.
One more thing
I should probably have been working on my position statement for the OSMF Board election, since I am one of the candidates. But since participation, growth and diversity are some of my core interests I just couldn’t stop myself.
- Update: corrected the table screenshot
Discussion
Comment from Stereo on 20 November 2018 at 13:56
Hah, we’ve been procrastinating with the same things! I’ll post about my analysis soon too.
Comment from imagico on 20 November 2018 at 14:34
Thanks - i was contemplating doing a similar analysis but i already suspected this is something others will look into so no need for me to invest time. :-)
My suspicion already was that the traditional over-representation of Germans in the OSMF has reduced significantly over the past years. But this is of course not due to the German community being less represented in the OSMF but because the US kind of raised the bar for everyone else. Your numbers also confirm an impression i already had before that in terms of absolute numbers Russia, Japan and Poland are the countries most severely underrepresented in the OSMF.
What i would really like to see (and i hope maybe Pascal will be able to provide a better look at this at some point) is how the same analysis would look like not for OSMF members per mapper but OSMF members per hobby mapper. As Simon for example pointed out a much larger portion of mapping activities in the US seem of commercial nature compared to most of Europe so the US overweight would probably be even more extreme if you take into account that SEO spammers are not really the kind of mapper you want represented in the OSMF.
Comment from joost schouppe on 20 November 2018 at 15:27
Exactly the point of statistics: confirming (and occasionally disproving) suspisions :)
I think that’s a bit of a broad conclusion to base on the stated evidence. There is also a significant drop in other countries. And it might just be that US Americans have a greater tendency to map at work or busier weekends. Making a more sensitive indicator of “probable hobby mappers” vs “probable commercial mappers” would of course be useful.
For further analysis, I would personally be most interested in the correlation between representativeness and English proficiency.
Comment from imagico on 20 November 2018 at 15:40
Yes, even if there are also other indicators that broadly speaking the contributor structure differs significantly between the US and Europe this is something a closer look at is required, especially if you want to make quantitative statements.
Comment from Nakaner on 20 November 2018 at 16:43
imagico wrote: > What i would really like to see (and i hope maybe Pascal will be able to provide a better look at this at some point) is how the same analysis would look like not for OSMF members per mapper but OSMF members per hobby mapper. As Simon for example pointed out a much larger portion of mapping activities in the US seem of commercial nature compared to most of Europe so the US overweight would probably be even more extreme if you take into account that SEO spammers are not really the kind of mapper you want represented in the OSMF.
I would not draw that hard line between us (hobby mappers) and them (paid mappers). There is not the stereotypical paid mapper who believes in satellite imagery only and whose quality is poor. There are paid/commercial mappers who work in a specific region only and have similar local knowledge to an average hobby mapper in that area as well. [1]
If you have a look at the list of new contributors in the U.S. on a normal workday, you will see dozens and dozens of accounts who add a single business – often without a meaningful tag like
shop=*,amenity=*oroffice=*(often wrong keys likekeywordsorPhoneas well). The list of new contributors in Germany contains business edits as well but much more “normal mappers”. While the business accounts don’t upload more than one or two changesets, normal mappers are more likely to upload more changesets or contribute later (not very likely but more than 0%).It would be interesting to use the list of average active contributors per day who have edited on at least two days (or five days) to remove one-time accounts from the list.
[1] The observation refers to German speaking countries.
Comment from Heather Leson on 20 November 2018 at 19:35
There is large potential to reframe the discussion. What is the future of OSM? Consider that the youth of the world are in Asia and Africa. What if we planned and grew for those audiences? What if we supported the incredible communities in other continents? How would the quality of mapping and the beauty of engagement grow?
State of the Map Asia just concluded (Congrats to the team). State of the Map Africa is being planned for next year. Let’s all get behind the needs and support this.
There are real technology, cultural differences, and financial hurdles to growing. However, the opportunity to really be global by supporting the spirit of OSM is key. http://www.bizcommunity.com/Article/196/813/184370.html
Welcome all, I say.
Heather
Comment from joost schouppe on 20 November 2018 at 19:52
Exactly: let’s not forget to talk about the lack of representation from Tanzania, Ukraine, Iran, Iraq, Thailand and Vietnam - to just name a few countries with a lot of mappers and not a single OSMF member…
Comment from imagico on 20 November 2018 at 19:54
So Heather - does your statement mean that you think proportional representation of the OSM community in the OSMF membership is not important?
Comment from MarTintamarre on 21 November 2018 at 12:41
Thanks Joost for the analysis! Interesting to see that the French community “surge” has easily allowed to move from a severely under-representation for being one of the (proportionally) well represented communities. This will of course not apply for countries where cost of membership or means of payments are a problem (cf. the ongoing discussion) but it means that a simple message encouraging people to join (like Christian did) can already bring a lot of new members in “Northern” countries. I’m not on the OSM US list but maybe that’s just what they did at some point? (I’m not good at conspiracies when I can find simple explanations, sorry :-p )
Ah and good luck with the election Joost ;-)
Comment from RobJN on 21 November 2018 at 15:06
That’s not the impression I got from Heather’s response. I took here comment to mean, let’s not just look at the proportionality across USA versus Europe but let’s discuss also how we lift membership in Asia and Africa.
Heather: Do you have any ideas how best to achieve this? Countries that are well represented have strong communities so maybe some push on community building in these places is the first step. How can other OSMers and the OSMF help?
Comment from LivingWithDragons on 21 November 2018 at 15:52
Here is the list again, but country names are followed by “LC” if they are an OSMF local chapter* (can someone make it highlight the row?). https://codepen.io/anon/pen/MzQerz
All except Italy are “over-represented”, but those they are not exclusive in being over-represented. This is interesting, as OSMF members in this country may find themselves choosing to be a member of one or the other, or both (and paying dues for both).
Comment from Seandebasti on 21 November 2018 at 16:51
first things first, well to start this topic/discussion!
primarly, I wish the priority is more cleaning the gaps to get a community and an osmf representative for each country or specific regions in under-representative areas (disputed contries as well). Thats a better aim to get every country and their needs representated.
secondly, we could start pointing out “local communities” to a map to identify the work of range of their community. For example, every mapping community could make an easy map where they point out the recent boundary of their activity. to sort out which areas of their community is under-representative. There could be categories of activity work. “global” work, thematical fields,
starting at different admin level, starting with countries (admin_level = 2), sub_level 3, etc.
to adress we could count - the wikiprojects in osm.wiki, mostly country - splitting in “local” internal mappers, and external mappers - e.g. for the german mapping meetups (Stammtisch) its mostly the urban area of the cities. - every mapper has his AOI (area of interest) with different themes they map - …
to support we could use sotm “attendance” statistics, for realizing which countries taken part at which conf to address the interest and activity range
more ideas to contribute?
greetz from the world
Comment from imagico on 21 November 2018 at 18:47
While still waiting for Heather’s answer i wanted to point out that the subject of Joost’s analysis was comparing per-country numbers of OSMF members with per-country numbers of mappers or in other words: The representation of mappers in the OSMF.
I can’t help but notice that the comments from Brits and Americans so far seem oddly unrelated to this subject in a whataboutism kind of way. I know that being made aware that you are on the privileged side of decreasing diversity trends in the OSMF is not the most comfortable experience but just ignoring this and concentrating on other more pleasant things is not going to change anything. So all of your should ask yourselves the same question i have asked Heather: Do you think proportional representation of the OSM community in the OSMF membership is important?
Comment from RobJN on 21 November 2018 at 20:31
Really? That’s not what I read. Whilst some worry about the last column having values well in excess of 100, others worry about the last column having numbers far less than 100. Personally I have no issue with come countries “raising the bar” - it encourages others to get involved too. The more the better :-)
Comment from imagico on 21 November 2018 at 20:49
q.e.d.
Comment from SimonPoole on 21 November 2018 at 20:55
@joost note that my statement that a large number of the new mappers in the US are SEO spammers is -not- based on the different workday - weekend patterns. It is based simply on the findings of investigating the large increase in incoming new mappers in the states starting end of 2016, which are in the 1000s of SEO spam accounts. If these were filtered out (many different ways possible) you would find the US even more overrepresented as it is.
Comment from joost schouppe on 22 November 2018 at 11:51
@seandebasti: there have been tries before of what you seem to want. The latest project was quite succesful and is even integrated into the ID-editor. It’s this project: https://github.com/osmlab/osm-community-index/ Jonathan Beliën turned it into a website: https://community.osm.be/ And there’s talk about making it into a map too: https://github.com/osmlab/osm-community-index/issues/79
Comment from joost schouppe on 25 November 2018 at 13:50
imagico, does your questions mean that you think that stating “A is important” implies “B is not important”? Color me surprised.
Thanks for clarifying, Simon.
That’s a bit happy-go-lucky, Rob. I think the relative advantage (native) English speakers have in having their voice heard is something serious. I’d love to see more research investigating the possible link between mapping community size and level of English in a country. And as well the impact on the correlation between mapping community size and OSMF membership. The privileged need to understand their privilege if they want to lift the rest to their level. Otherwise they risk building something that is only relevant to people like them. I count myself among the privileged BTW, since my native language Dutch is (on a global scale) extremely close to English. But being Belgian, the language issue is of course something you are confronted with on a daily basis. This is part of a wider issue I think. The medium is the message, right? For example, maybe part of why OSMF is such a “political” place, is that we choose a medium (mailing lists) which are excelent for people who like deep thinking, but aren’t great for involving people with a more occasional involvement (e.g. the easy participation of upvoting in Reddit).
That’s an interesting addition, Gregory. In Belgium, there’s no membership fee, but straight from our chatroom I saw a nice example. Someone saying they rather pay a donation to OSMbe rather than OSMF, because OSMbe are people they know and trust, and OSMF is seen as a more “political” organisation. Anyway, i wouldn’t expect a perfect correlation between LC and representation, since large part of their being a LC or not is the result of individual action (e.g. having people crazy enough to build an organisation)
Comment from imagico on 25 November 2018 at 14:24
No, my question was expressing a genuine interest in Heather’s priorities. As i explained your analysis is specifically working out the degree of proportionality between geographic distribution of mapping activities and geographic distribution of OSMF members. Since Heather’s statement was in my eyes sidestepping the discussion with the “welcome all” i became curious about her opinion on the main topic.
In the abstract form you ask - there are of course potential pairs of goals A and B that are mutually exclusive. For example the goal to finance the OSMF through individual membership fees and the goal for proportional representation would be fairly hard to equally pursue.
Comment from RicoElectrico on 9 November 2019 at 11:36
Hi Joost, do you have membership stats by country for 2019? I’m curious about Poland in particular.
Comment from joost schouppe on 11 November 2019 at 19:44
Hi Rico, I’m completely swamped at the moment, but I’ve been hinting for some of the other MWG members to release some stats when the election deadline has passed.
Comment from JanetChapman on 6 July 2020 at 14:42
Very interesting. Is there any updated information anywhere please? Thanks Janet
Comment from imagico on 6 July 2020 at 15:57
Last year update is on
https://www.openstreetmap.org/user/imagico/diary/391322
No newer data is publicly available so far (not even overall numbers - see: https://wiki.osmfoundation.org/wiki/Membership/Statistics)
I had been bugging the MWG several times to set up regular automatic publication of statistics and also offered to help but nothing came out of this so far. If you ask for current data you will probably get it though.