OpenStreetMap Usability – Converting More Sign-Ups To Active Contributors

Yet Another WhereCampEU 2011 / Berlin Report

From Data To Stories

Chris Osborne did a great job making the second European WhereCamp happen. He drew the curtain on the two-day unconference with an inspiring talk on big data, which his company ITO World is in the business of crunching and visualizing. Showing some of their impressive visualizations, ranging from personal location history to the effects of the London congestion charge, Chris drove an important point home very convincingly:

We don’t need data, we need stories.

Data Usability

In a world that is increasingly overflowing with big data – be it from the government open data movement or from crowdsourcing initiatives such as OpenStreetMap – we need to be able to tell the story that is hidden in that swamp of raw data. Stories are a language that everyone understands. Speaking that language means that almost everyone can engage with big data in a meaningful way. I like to look at what Chris and his colleagues do as data usability – creating a meaningful interface between people and data.

Two of my contributions to WhereCampEU also dealt with this topic – but in a very different way. Data usability has always been a challenge for OpenStreetMap, in more than one way. On the consumption side of things, OpenStreetMap data is not very accessible for those who want to do something other than making a map. I addressed a part of this challenge in a talk about dealing with historical OpenStreetMap data. More on that a little later.

Churn Rate

Shifting the focus to contributing to OpenStreetMap, adding your own knowledge to the map, the numbers speak volumes. Almost 70% of those who sign up for an OpenStreetMap account never proceed to make any contribution to the map. As there really is no other compelling reason to sign up for an OpenStreetMap account, there is something missing in the input interface between those 270.000 (and change) people and the OpenStreetMap data. That interface is comprised mainly by the OpenStreetMap editors – there’s a whole range of them carrying an eclectic array of names: JOSM, Potlatch, Merkaartor, MapZen, Vespucci, to name a few. Some are web-based, others desktop applications or apps for smartphones. They are elaborate pieces of work and I use some of them every single day. What they all fail to do however is compel those 70% of new sign-ups to OpenStreetMap to actually start making contributions. They do not trigger the initial motivation to become an active OpenStreetMap contributor. It’s hard to say exactly Why they fail to do that, but WhereCampEU offered a number of suggestions.

Address Hunter

Skobbler – the German-Romanian company that builds innovative things using OpenStreetMap data – gave us a sneak peek of a spare time project of theirs. It is called AddressHunter and it is a multi-player location-based fast paced action game. The goal? Hunt down as many unmapped addresses as you can before your opponents do. When you find an address, you shoot a picture to prove that you were there, confirm and move on. When all the addresses are hunted down, the game ends. The group or individual with the most hunted addresses wins the game and gets bonus points that go towards your overall rank.

We got to try AddressHunter out in teams of 3-5 WhereCampers, each team under the supervision of a Skobbler team member. We played using iPhones that they provided – but the game is all in HTML[1] and thus by nature multi-platform. Even though the game was still in beta stage, with a planned release later this summer, it looked awesome and already feels very polished and well thought out. There were a few glitches in gameplay but nothing serious. The version we played had a 17th century Great Explorer kind of theme; Skobbler plans to offer different versions of the game appealing to specific target groups. And they have some more cool plans for it.

What makes this game not great but awesome? All hunted-down addresses are added to OpenStreetMap directly! AddressHunter picks up the tedious and arguably boring task of adding addresses to OpenStreetMap and transforms it into something a) anyone can do and b) fun. With these ingredients, AddressHunter gives OpenStreetMap a part of that missing interface between people and data, and may very well help bring down that enormous churn rate of 70%.

But what about data quality? Will the address contributions through AddressHunter be as good as those from real OpenStreetMappers using real editors like JOSM?

No. But is that a bad thing?

Yes, but it’s not the end of the world. Let’s take a closer look at the contribution metrics. OpenStreetMap now has just over 400,000 contributors. If the churn rate figure of 68% is still valid – this was based on data from early 2010 – only 128,000 of those will have made any contributions to the map. Of those, a further breakdown shows, only 19% is considered active, that is, has edited in the last three months. That is only about 24,000 people. While that seems like a lot, it is not. Going with the very rough estimate that 10% of Earth’s 150 million square kilometers of land mass is populated, that leaves an area about the size of Madrid for each active contributor to map. What I’m trying to convey here is that OpenStreetMap could do with some more active contributors, and that the challenge seems to not be in attracting them, but rather in retaining them – converting sign-ups to contributors. That is the road that needs paving, and cool initiatives like these provide some of the paving stones.

Tweet your node

I put forward another idea that could help improve this conversion rate. It had been sitting in the back of my head for a while as I was working on a project to get more CCTV cameras into OpenStreetMap. To allow casual contributions, I discussed a few ideas for low-barrier single POI contributions to OpenStreetMap. We came up with the idea of using Twitter as an interface to OpenStreetMap. On the U-Bahn on my way to from Kreuzberg to the WhereCampEU venue, I decided to do run a session about the idea to get some feedback.

The basic idea is very simple: make sure your phone adds a geographical coordinate to your tweets – a standard feature in most mobile twitter client nowadays – and tweet something like

amenity:pub name:Bellman Bar #osmadd

This would then be picked up by a twitter scraper that would parse the content into OpenStreetMap tags and add the POI. This system could be used directly through a Twitter client, but also by third party applications.

The Return Of Anonymous Edits

The main drawback of this is that the contributions could never be traced back to an individual, which is Bad Behaviour in OpenStreetMap. Anonymous edits were abandoned in 2007 for that reason. Most of the discussion after my short introduction of the idea was around this issue. I have thought about the idea some more since and came up with an idea that would also stimulate conversion. It is too big to fit in the margin though.

Tweet Finesse

I also got valuable feedback on the format of the tweet. Firstly, encoding tags like this hardly makes for a casual contribution, because some intimate knowledge with the OpenStreetMap tag definitions is required. This could be addressed by having different hashtags and allowing only a name as the content of the tweet, like

Bellman Bar #osm-pub

, or venturing into the dangerous lands of natural language processing.

Secondly, addressing a hashtag pollutes the timeline of the twitter user with meaningless – at least to human followers – tweets. This can be addressed by having an actual twitter @username as the magic word, like

@osm-pub Bellman Bar

. This way, the tweet will not show up in human follower’s timelines if they do not follow @osm-pub as well, reducing timeline pollution.

I started implementing this idea, incorporating all the ideas gathered at the WhereCampEU session – so more on this soon! I wanted to include it in this post because I believe it can be another example of paving the first mile leading from sign-up to active contribution to OpenStreetMap.

OpenStreetMap Usability, Status Quo

The last WhereCampEU session I want to address in this context of OpenStreetMap usability is Patrick Weber’s. I noticed his blog post describing initial results of the extensive OpenStreetMap usability research he and his colleagues at UCL are carrying out a few weeks before WhereCampEU. The study also got picked up by the Strategic Working Group and (re-)sparked a debate on the layout of the home page. As it turned out, people unfamiliar with the OpenStreetMap web site took more than 6 seconds to find the geographical name search box. For a web site that is all about places and maps, that is a Bad Thing.

In his session, he talked more in-depth about the still ongoing usability study. Using eye tracking technology and more traditional usability study methods, Weber c.s. observed new OpenStreetMap contributors finding their way through the sign-up process and towards their first meaningful contribution. It was interesting and somewhat unsettling to hear about all the snags that new contributors can – and will – hit on their way. A simple task like zooming in on the map in the Potlatch editor proved to be daunting enough for some to either give up or screw up their first editing effort.

Patrick’s study is ongoing, and he will be presenting more results at the upcoming State of the Map in Vienna. This is an important effort that I believe will expose many challenges on the road towards a healthier conversion rate from sign-ups to contributors.

WhereCampEU produced a few good ideas for paving that first mile, and what I am writing about here is only a tiny fraction of everything that was going on. I cannot wait for the next edition. See you somewhere in Europe!

[1] The app is wrapped in the PhoneGap framework to create a native platform app, necessary for continuous GPS and camera access.


6 thoughts on “OpenStreetMap Usability – Converting More Sign-Ups To Active Contributors

  1. You are quite right, most users are either terrified of changing OSM or don’t know that they could.

    The Twitter post method is interesting, but are smartphones assisted GPS accurate enough to actually add a new point to OSM? I’ve been using my iphone4 to photograph post boxes for OSM mapping. I normally need to use the satellite imagery to get them accurate as they are generally +/- 30m which is sometimes the wrong side of the road.

    Our site is partly about verification of geodata. I’m starting to think about creating microtasks for contributors so that they might be prompted with a variety of questions, e.g. “here is a photo of a plaque, what colour is it?” or “From your location you should be able to see a plaque, please photograph it”.

    1. Jez – GPS accuracy would be an issue. It has been suggested that POIs submitted through a mechanism like this should go to a separate data layer. Existing OpenStreetMap bug tracking solutions such as MapDust and OpenStreetBugs (no longer being maintained?) were mentioned. From a data quality perspective this does seem like an attractive idea, however, the instant gratification of adding something to the actual map is lost. I believe that is very important for a tool like this, because it would be squarely aimed at casual contributors who have no or very little intrinsic motivation.

      The microtasks idea sounds extremely compelling to me. I have actually been thinking about something similar as a mobile app for smartphones, based on either known bugs (pulled from one of the bug submission systems I mentioned, or on some other detection algorithm for potentially weak or incomplete data in OpenStreetMap. Are you getting anywhere implementing it?

  2. Yes, we have a set of “todo” queues which are only visible to admins. As soon as each queue has a custom edit view then it will be ready for users.
    I like the idea because the UI can be tailored to the particular task and not just a general ‘edit this record’ view. Also, users could select the type of task they are happy with.

  3. The front door thing from bing looks like a good microtask interface: Lots of potential with this kind of thing, but I’m puzzled why they have chosen to set that particular task. Who really cares where exactly the front door to the address is? and more importantly you can’t actually see that information in the aerial imagery!

    I like the tweeting idea, but Jez is right about the GPS accuracy problem. Maybe it could contact you 12 hours later to ask you if you want to launch Potlatch to check the positioning of your new node. …or as you say, the boring solution is to triage the data or just submit them as bug reports.

    I was wondering if there’s any reason why you’ve gone for colon instead of equals character for the key value separator? In my opinion it would be more consistent (e.g. with wiki tag documentation) , less problematic (e.g some tags have a colon character in the key) to go with an equals character in the tweet e.g. :

    amenity=pub name=Bellman Bar #osmadd

    But the other “tweet finesse” suggestions, to not include full key value pairs, would solve that too.

Comments are closed.