The OSMF AGM is over and the results of the board elections and resolution votes are in. With the broad spectrum of candidates that was available for election this year the election results give us interesting insight into the political orientation of the OSMF membership.
In advance i want to apologize to the blog readers not so much interested in elaborate discussions of OSMF details. There will be other posts on different topics again in the future. But i have the impression the current developments in the OSMF could be quite significant for the future of the OpenStreetMap project so i give them a bit more attention this year.
The results of the votes on resolutions – which i discussed in detail before the elections – are relatively simple. All of them have passed, except for the resolution on strict term limits. That means the OSMF will from now on have Russian style term limits for board members, meaning that board members can be resurrected indefinitely to continue being on the board as long as they pause from time to time to reset their term limits. This is unfortunate and opens the door for all kinds of informal deals and agreements for influential interest groups to maintain a presence on the board.
The reason for this – and i have mentioned this before the election – is that British corporate law puts up fairly massive hurdles to changes to the Articles of Association of the OSMF. So while strict term limits had a huge majority of >2/3 among the normal members who were allowed to vote on these that was not sufficient to pass the 75 percent hurdle required by law. But having term limits at all is a big step forward of course.
I would not necessarily consider a high hurdle for AoA changes a bad thing – there are similar hurdles in German Vereinsrecht for Satzungsänderungen. The problem in case of the OSMF is that the AoA as the basic constitution of the organization exclusively contains procedural rules but does not codify the need for the OSMF to serve the interests and values of the OpenStreetMap project. This means that it is much easier to re-purpose the OSMF to serve special interests of certain people or organizations than it is to reform the procedural rules to prevent them from doing that. This imbalance is a huge problem and the AoA changes that have been approved this year and the modifications these implement in the procedures are likely insufficient to counter the attempts of commercial actors to influence the OSMF to serve their business interests. Luckily at least some of the newly elected board members are well aware of this problem so there is hope we might see some initiatives in the coming year attempting to better address this issue.
The board elections
This brings me to the board elections. Overall the results – with one exception – match my recommendation before the election so i should kind of consider this a good development. But it is not that simple and i will explain this in more detail later.
This year’s elections where a very good demonstrator for how STV works. Based on the first choices of voters alone we would have a very different board – consisting of Guillaume, Mikel, Michal and Steve. That is hypothetical of course since voters most likely would have voted differently in many cases in such a scenario. With the STV voting system the surplus votes for Guillaume (in other words: Those votes he did not need to win his seat) were distributed to the second choices of those voters who voted for Guillaume. This brought Allan into the game because he was a very popular second choice among voters of Guillaume, about half of them had Allan as their second choice. In subsequent rounds the candidates without chances were eliminated one by one and the votes for them distributed according to the second choices as well. This first got Allan and Mikel elected after a few rounds and finally Rory – who won their seat by a margin of less than one vote to Michal in the 12th round.
Now the first thing i want to do is to look at the popularity of the candidates beyond the specific mechanisms of STV. STV gives any voter just one vote and your second, third and further choices only come to count if your first choice is eliminated or is elected with surplus votes. But the ballots also give you an idea of the overall popularity of candidates. Someone putting Guillaume as second choice makes a statement of preference for him even if – because they have chosen Rory as first choice for example – the second choice does not come to bearing in the election.
I did this by weighting all the ballot choices with a weighting factor based on their position on the ballot. The first choice has a weight of 1.0, the second choice 0.5, the third choice 0.25 and so on. This weighting function is somewhat arbitrary, you can make the same analysis with a different weighting but the point i am trying to make here does not very much depend on that. The overall absolute popularities you get with that are as follows:
Like in the actual vote Guillaume, Allan and Mikel were the first three winners of the election the fourth place goes to Jinal in this scenario. In a way you could say that Jinal was more popular among voters than Rory when viewed in isolation (and also more popular than Michal who lost to Rory by a very small margin) much of the support by voters she got came from voters who voted for Mikel and Michal as first or second choice and since these got more first choice votes than her this ultimately did not help her to win the election. So in a way you could say an STV vote like this what counts is how many voters prefer you over the direct opponents in the election.
What makes this year’s election fairly interesting is the broad spectrum of candidates regarding their political views and agendas. This allows us to get a pretty good picture of the structure of the OSMF membership at the time of the election. For that i produced an illustration showing the relationships of the first and second choices of the members relative to each other. Since in STV voting you only have a single vote the first choice on your ballot is the most important, the second choice less so and further choices even less. That means reducing the data by just looking at the first and second choices is a reasonable start into the analysis. Based on these i produced a chord chart showing for all voters who voted for candidate A as first choice how many voted for candidate B and the other way round – in size proportional to and sorted by the number of votes.
The subjective choice i made is re-arranging the order of the candidates around the circle to preferably put candidates that have been popular pairs in first and second choice among voters close to each other. Doing this on a one dimensional circle is a very rough approximation so different orders would be possible but als you will see in broad strokes this is already defined by how people voted.
To make you familiar with reading this illustration have a look at the largest connection in the diagram, the one between Guillaume and Allan. As already mentioned above about half of the voters of Guillaume had Allan as their second choice and this results in this broad ribbon on Guillaume’s side. Among Allan’s voters six had Guillaume selected as second choice – which is also the largest group among Allan’s voters so the ribbon starts at the left side of Allan’s section of the circle since the order of the ribbons is by size clockwise. And a broad ribbon indicates a large number of voters has chosen a certain pair of candidates as their first and second choices. There is a lot of interesting things that can be observed in this illustration if you take the time to look around it.
Political factions in the OSMF membership
One of the things this diagram can be used for is identifying the political factions in the OSMF membership based on how they have voted. When i analyzed the candidates’ answers and manifestos before the elections that was of course influenced by my personal perspective on things. Now we have the results and can try to look at this through the eyes of the actual voters.
Guillaume was not only by far the most popular first choice candidate among voters, he was also the most popular second choice among voters of many of the other candidates. Most of these candidates have been outspoken proponents of the philosophy of craft mapping and the values of egalitarian cooperation and local verifiability of data in OpenStreetMap. I outlined this faction of craft mapping supporters in a very conservative fashion including only those voters who voted as first and second choice for two of the following candidates:
- Guillaume Rischard
- Allan Mustard
- Rory McCann
- Gregory Marler
- Dietmar Seifert
- Nuno Caldeira
This is not meant to indicate that members who have voted for other candidates are not behind the ideas and values of craft mapping – as said, this is a very conservative estimate. Because it is on the other hand with the choices we had in these elections unlikely that anyone who voted for two of these candidates as first and second choice is not a supporter of these values. I marked this segment of voters identified as part of the craft mapping supporters faction in red color on a second circle around the main diagram. All together this amounts to about 220 voters or about 41 percent of the valid votes in this election.
There is another large faction of voters you can identify in the diagram due to broad connections, that is the group of
- Mikel Maron
- Michal Migurski
- Jinal Foflia
Most of the voters of each of these three have voted for one of the other two as second choice. This is a pretty remarkable and unique feature in the voting data because if you read the agendas and answers of these three to the questions there are not that strong commonalities between what they write, especially not between Jinal and the other two. The thing they have clearly in common though is that all three of them work for big corporate OSM data users. Therefore it is quite logical to conclude that for those who vote for two of these on first and second choice, this aspect plays an important role. This corporate and professional interest faction is probably less than the craft mapper support faction bound by common values but more by shared interests. It is likely that this faction includes for example employees of corporate OSM data users, paid mappers or people otherwise with a professional career or economic interests in organized activities around OSM and interested in an OSMF in support of these goals.
What is interesting is that Steve, who is also a corporate employee, is quite clearly not considered an equally desirable candidate within this faction. Steve’s style of campaign was different from the other three and his employer (TomTom) is not part of the inner circle of large corporate OSM data users and OSMF corporate members the other three are working for.
I marked the voters identified to be members of this faction in blue color on the outer circle. In total – again with a very conservative definition – this amounts to 85 voters. Compared to the 220 voters identified as members of the other large faction that is a ratio of approximately 1:3 – which leads to the near-tie between Rory and Michal on the last seat available.
One obvious question is of course how many of these have the third candidate of the trio as third choice. The answer is 37. Among the voters of these three Clifford has also been a popular third choice with a total of 22 votes.
Looking at the votes of these 85 voters in more detail you can see an interesting pattern of a number of votes being very similar with small permutations:
8 11 5 10 7 2 3 4 1 9 12 6
8 11 5 7 10 3 2 4 1 9 12 6
8 11 7 5 10 2 3 4 1 9 12 6
11 8 5 7 10 3 2 4 1 12 9 6
11 8 5 7 10 3 2 4 1 9 12 6
11 8 5 7 10 3 2 4 1 9 12 6
11 8 5 7 10 3 2 4 1 9 6 12
11 8 5 10 7 3 4 2 1 12 9 6
11 8 5 10 7 3 4 2 9 6 12 1
11 8 7 5 10 2 3 4 1 12 9 6
11 8 7 5 10 2 3 4 9 1 12 6
11 8 7 5 10 3 2 4 1 12 9 6
These are all minor permutations of the same base sequence
11 8 5 7 10 3 2 4 1 9 12 6
which translates into the following names:
Now while it is not impossible it is pretty unlikely that 12 OSMF members have decided to vote for this sequence with small permutations without any coordination. I made a fairly prominent public recommendation for a certain vote before the election and looking for this in the voting data gets this:
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10
1 2 6 7 12 4 9 10 8
1 2 6 7 12 9 10 4
1 2 6 7 12 9 4 10 8
1 2 6 7 12 10 9
1 2 6 12 7 9 10 4
1 6 2 7 9 10 12 4
1 6 2 7 9 12 4 10
1 6 2 12 7 4 9 10
As you can see there are twelve votes exactly matching my recommendation and 8 further small variations. And that with 8 entries on the ballots to match instead of 12. It is therefore very likely that the 12 votes above (and possibly further votes with somewhat larger permutations of course – that is difficult to tell) have been coordinated in some way, possibly with the deliberate introduction of permutations to disguise that. Given the main beneficiaries of these are employees of Facebook and Grab it is likely that this is based on a voting recommendation, possibly as part of a candidate analysis like the one i posted in public, on internal communication channels.
Note this kind of voting coordination is not illegal nor has the OSMF so far made clear statements condemning such actions. Still it is interesting to see this actually happening in the wild and not just in theory.
Now apart from the two main factions i have identified you could also try to delineate other smaller factions within the membership. But the number of voters and the amount of data is not really large enough for other reliable conclusions about such so i will skip that.
Overall popularity of candidates
Now i can imagine that some readers might criticize the identification of distinct factions in the OSMF membership as divisive. But as i tried to point out these are not a figment of my imagination, these are clearly visible in the voting data.
What i however want to look at a bit more is how polarizing and how popular different candidates are across the political spectrum. For this i looked beyond just the first and second choice votes but looked at the average position of candidates on the ballots separated by the first choice candidate.
Here is the illustration for that. It shows, separately for each candidates, the popularities of this candidate among the first choice voters of the different other candidates. This popularity is measured by the average position of the candidate on the ballot. So a higher value means lower popularity because the name is father from the top of the ballot. If a name is not on the ballot it is assumed to be on the 12th position for this purpose.
Note these are the average positions so we are talking about relative numbers. The total number of votes a candidate gets does not have an influence on that. This allows us to look at the popularities and polarization between candidates independent of the strength of their position within the OSMF membership in terms of the number of votes.
We can see here again that Allan was very popular among the voters of Guillaume – like we have already seen before. Similar popularities are achieved by Jinal among the voters of Michal (which includes the above mentioned likely coordinate votes) and of Michal among the voters of Mikel. On the other hand:
- Mikel and Michal are very unpopular among the voters of Guillaume, Rory, Nuno and Dietmar
- Steve and Jinal are very unpopular among the voters of Guillaume
- Dietmar is very unpopular among the voters of Michal and Jinal
- Guillaume, Rory, Nuno and Dietmar are very unpopular among the voters of Mikel and Michal
We can average the popularities of the different candidates among all voter groups and get the following ranking:
This shows that Allan is the candidate with the most support across the whole political spectrum independent of the strength of the political factions. Gregory is also pretty good in that regard. On the lower end of the spectrum we have Dietmar and Nuno who were the candidates with the least elaborate answers and manifestos and and who were therefore probably not positioned that well for support even from voters who would politically align to their positions.
What we can also determine from that data is the standard deviation of the rankings among the different voter groups which is a measure how polarizing a candidate’s view is and how strongly the opinions among the membership differ depending on their first choice on the ballot. Large bars here mean more polarizing. As you can see Gregory has the lead here, he was the candidate who created the least extreme reactions of like and dislike among the different segments of the voting membership. On the other end of the spectrum we have Mikel and Michal who are strongly supported by parts of the membership while being strongly disliked by others.
As you can see there are a lot of interesting details that can be found in the results of this year’s election. In the introduction i mentioned that while the results align to 3/4 to my recommendation before the election i none the less have mixed feelings about the development.
First of all the only candidate this year who came close to achieving majority support within the first four positions on the ballot – in other words: Which a majority of the voters had in their top four on the ballot – is Guillaume. Only Guillaume and Allan achieved this within the first five positions. This can be illustrated by the last diagram i want to show – which depicts the accumulated support for the different candidates over the position on the ballot. The horizontal gray line indicates the 50 percent majority among the participants in the election.
The left end of the graphs here is the number of first choice votes while the right end indicates how many voters had the candidate on their ballot at all. In between are the numbers for a cutoff on the ballot at different positions. On the other end of the spectrum Mikel only achieved majority support from the voters if you include the tenth position on the ballot. This is quite natural for an election with four seats available. But it turns the election into something more like a parliamentary election and less like the election of a government.
In the future with the changed election system with fixed term lengths for the board members we are going to have either three or four board seats up for election every year. Next year this is going to be the seats of Paul, Tobias and Joost, in 2021 the seats of the four board members elected this year are going to be up for election again. This means we will have a similar situation as this year with the chances for a candidate being elected even without majority support being pretty high – as long as a candidate has solid minority support they can be elected even if strongly disliked by a majority of voters. This is good for minority representation in a parliament like body (although there you would usually want more than seven seats for being able to represent small minorities within a diverse community) but for a functioning executive body like a board this is problematic. With the current strengths of the two factions i identified in the OSMF membership there will likely be one or two board members aligned to the goals of the corporate interest faction and two or three members from the craft mapper support faction. If there are in the future candidates who can achieve broad support from those members who are not firmly aligned to one of the two large factions they could elect at least one board member per election who does neither appeal to the corporate nor the craft mapper faction of course. On the other hand consensus candidates in the current constellation will have a hard time because for the two large factions they will usually be less appealing that a candidate strongly in support of the faction’s interests and values.
But of course that is all based on the assumption that the current composition of the OSMF membership stays the same and that the political views and interests of the members stay the same. Neither of these is likely, especially since in the future active mappers will be able to join the OSMF as members without paying a membership fee and this way the membership numbers could significantly increase.
This year we had with Eugene a candidate who cannot be clearly connected to either of the large factions and who ran on a platform of regional identity politics to a large extent. He did not have significant success this year because there were apparently not many voters who put these interests above either professional interests as members of the corporate faction or craft mapping values – which were more convincingly represented by other candidates. With a widening and more diverse membership this kind of candidacy could become more common in the future – but such candidates will likely have a tough time anyway since both of the large factions are largely agnostic to identity politics of any kind and it will be hard for any candidate not to position themselves regarding those factions and their interests and values. What is more likely is that we will see candidates who are running as part of these factions to try appealing to specific identities to gather additional support. This already happened this year with Jinal who managed to gather more approval outside the core of the corporate interest faction than Michal and Mikel. It did not help her much though since she was ultimately not able to provide enough people with clear and convincing ideas to put her on top of their ballots.
On the other hand the values of the craft mapping supporters, in particular with the focus on egalitarian cooperation, tend to speak inherently strongly to community members world wide. The key here is to bridge the language barriers and break the dominance of the English language in the OSMF which tends to systematically favor the corporate interests because the world of corporate OSM data users is inherently and massively English language dominated.
The most important thing for the immediate future is to get the free OSMF membership for active mappers on the way – this has been decided on in this year’s votes with an impressive 93 percent majority but it needs to be practically implemented. Once this is done a lot depends on if this will actually appeal proportionally to all active mappers or if certain groups of people (like native English speakers or paid mappers for example) will sign up in larger numbers than others. This is something that we will be able to follow probably and should follow and analyze quite closely.
The other more long term goal would be to reform the government structure of the OSMF. The double role of the OSMF board as both legislative and executive body creates serious problems – not only because of the voting system but also for other reasons. Separating these two functions would be very good but rather difficult. The board simply delegating one of these won’t work – it has already been tried with the management team in one direction and it is unlikely to work any better the other way round. If the board just voluntarily delegates things to others but is at any time able to take these rights back that is not going to work. It seems however that unfortunately the British corporate law does not provide a way to change this more fundamentally. So addressing this issue would be a larger endeavor requiring to change the fundamental legal basis of the OSMF as an organization.
There would be different possibilities to implement policy decisions in a fundamentally reformed OSMF – either through direct democracy, through some parliament like body or through a federal system of representatives of local communities.
Apart from these more abstract considerations my hope is that the new board – which like in 2015 consists of three completely new members, one re-elected old timer and three members from the previous board – will be able to work productively on the important questions and topics the OSMF is confronted with and that unfortunately have been stuck for the last few years in many cases. I in particular hope that we will finally remove the last vestiges of the culture of secrecy and in-transparency that has prevailed till today despite the significant improvements in that regard initiated in 2015 in the OSMF – something i have criticized extensively during the past years. The chances are good given the current majorities on the board if members are bold enough to move forward and initiate changes.
I have explained and illustrated above that in terms of election dynamics the OSMF board is now with three or four seats being up for election every year much more like a parliamentary body than it was before with usually two seats available every year. The board members need to embrace this change by acting more like a parliament. That means open discussion about disagreements of the members on decisions and arguments about what is the right approach on things. In the past the board has often tried to present itself as speaking with one voice and has shied away from admitting internal dissent publicly – most likely out of fear that this makes them look weak as an executive body. But this of course never really worked because knowledgeable people in the OSM community were always aware of where the lines of dissent in the board are and the pretense these do not exist made the board look ridiculous rather than strong and unified.
Embracing the more parliamentary nature of the board with a more open and less conflict averse work culture could have a significant effect on the next elections because support by voters will largely also be based on if they see their elected representatives work for and make decisions in their interest and according to the values they have promised to represent before the elections. In the past this has unfortunately often been thoroughly disappointing. Always keep in mind that in this year’s election more than 40 percent of the members have not voted in the elections. There is a lot of potential in that for improvement.
It is no secret that i am politically on the side of the craft mapping support faction and the values of egalitarian cooperation and local verifiability of data in OpenStreetMap. This is not because this faction somehow caters for my personal interests and privileges as some people occasionally imply but because – as i have explained in detail on several occasions – these values are essential for a functioning OpenStreetMap project able to achieve its goal of being the best map of the world. Hence i am happy that we have a board now with a clear majority of outspoken supporters of these values (with three out of four newly elected board members – Guillaume, Allan and Rory as well as Tobias who remains on the board and Paul also has been an active supporter of these values in the past).
But i also want to express a bit of a warning that if the board in this composition cannot productively work and make and enforce beneficial guiding decisions for the OSM community during the next year and demonstrate that these values can help bring the project forward and make the OSMF more open and more inclusive for other languages than English and mappers from all over the world this majority on the board might be very short lived. How corporations and their representatives are going to react, if they continue to try exercising influence through the working groups and try finding more inroads into the OSMF membership in some ways or if they can learn to accept that OSM will not adopt data usefulness for short term corporate profits as its highest goal and that in the long term it is in everyone’s best interest that the OSMF is in control of people without immediate business interests in the project remains to be seen. Also here the best incentive for a constructive attitude is transparency in all deliberations and arguments.
Finally i also want to strongly voice an encouragement to all OSMF members and everyone in the OSM community to give their support to the new board by volunteering to help them – either in one of the working groups or outside by contributing to a broad and open discussion on OSMF matters during the next months, offering feedback in board meetings, on document drafts etc.