(Quasi-)Experimental Research in OSM – Transcription
I don’t think our policy makers and decision makers are paying attention to these broader facts, and this has existed for 12 years, we should be able to say something about the amazing data that we have. That’s what I will try to do today. There’s a lot of research on OpenStreetMap, and about 25,000 different papers. Drop in audio.
This is something that no one understood when this happened, but it is in OpenStreetMap. When you go to the mailing list, you see confused contributors saying, my data looks good, this is bad, what is going on. This was linked to the variation in the underlying census data. This is really great, it helps say something about imports. What happens if you imported a half-done map, versus an incomplete import? And looking at the OpenStreetMap history file, you can see how the counties evolved and how it supports the communities in a long-run, which is what I did.
So first the bad news, if you look at contributions, these are editing tags and adding new amenities, places with lower – drop in audio – supports their point of view. You will see that most of the gaps are not attribute. If you look at amenities, these follow on edits that are more visible, imports mean more of those kinds of edits. So what you get is a missing layer of attribute tags, but a higher layer of follow-on tags. This brings the debate of what imports do. They drive people away from the invisible parts of the map, attributes, and move people – we get more contributions in the layers that were not touched. So when we think about designing imports in the future, because they are important to moving the project forward, we should think about surfacing the invisible tags, that is where the most negative affects of imports come up.
A second example is a question I have been wondering, what happens to OpenStreetMap across countries? Why is it really strong in some countries, and not in others?
In particular, I’m struck by the role of competition. So in India, for example, Google maps was launched until 2008 or 2009, and this was true for a lot of the rest of the world. So how does competition from an alternative source affect what happens on OpenStreetMap? Are people getting more mapping, ideological, and going into OpenStreetMap, or are they looking for something to use? When they have an alternative, they leave it? And so I will use the similar design as before. Google enters different countries at dramatically different points in time because they are sorting out the licensing issues.
This is an email that he sent out for two years now, you can see if the initiatives are useful. If this is effort we should be taking, if not, how should we be re-designing, making it shorter or longer, who should we send it to, all of the questions can be answered by an experiment where we contact some users, not others, and compare the activity over time.
We did something like that, not exactly, with the data that was shared. So this is all about a thousand users who he sent messages to, we compared them to the 40,000 users in the U.S. who started editing at this time who did not get an email message. We found that the users who got the message make about 12.5 percent more edits over a two-year period. So a simple email early on has a big affect on how many contributions these users are making. These efforts tail off for a higher number of edits, this someone that makes more than 5 or 10 different edits.
We are seeing nuance, we are finding ways to target the messaging to the people that are not aware of OpenStreetMap and not as committed.
We are doing some more work to figure out demographics and whether new comer messages fixes diversity issues, but that is hard because the data does not exist.
I have a list of this data, and how does imagery affect new contributions? We have imagery in certain places, that means that new people are coming in. Can we design awards and badges? Some of my colleagues have done work in Wikipedia, and showing that awards and badges is important. How can we design awards and badges for contributions? And the list goes on and on, it is interesting to think about these questions from a data-driven perspective, rather than purely just debating about these things. And starting to build a structure within OpenStreetMap where we can test the answers to these questions, both to improve design, and also to start doing the things that don’t work as well.
So a long-term vision would be to have something like Wikipedia research, a physical organization within office, and all they do is run the experiments on the perform and try to inform the policy. So rather than have policy be hashed out on mailing lists, can we bring these little experiment and formalize that in OpenStreetMap is something I would love.
In the short-term vision, are there opportunities to find ways to collaborate with individual academics who are increasingly coming out of the ivory tower? So the new vision of the academic you should have is the academic as a farmer rather than the ivory tower, we are fixing the problems in the world and learning about human behavior at the same time.
There’s a lot of people that look at online communities, and very few look at geospatial data and OpenStreetMap, and there’s ways to look at these collaborations. I have been working on this for the last 3-4 years and want to have more of these conversations.
So I want to end with this experiment, thank you.
Host: Any questions? Yes, this isn’t about the new users, it is about imports. We did a huge building import in Seattle, and we don’t have any hard data, but what we got is the sense that it really helped people add businesses to OpenStreetMap, because the building was there, and they didn’t even have to go and look for it, and a lot of the tools like maps.me, OSM, and you could actually see the outline of the building, the address, and now you can add the restaurant, the bar, the cannabis shop – we have those, too – pretty easily. And we found that seemed to help people. What I don’t know, my sense is that import in that case really helped. And now hearing this morning talking about artificial intelligence, that scares me. They will have a lot of data, and what do we need me for? Drop in audio.
There’s a community of academic out there, like me, that are studying online communities, and pitching to them OpenStreetMap as one case that is similar to the communities they are studying is helpful, because they are interested in the same questions, around how do we engage communities, and questions like that.
And a second level, if you can actually swing it is actually better and broader is what can we learn about human behavior. There are academic questions about why do we volunteer, do things for free, and how does social interaction work in groups. And it is about finding a right lens for these questions that helps to really bring this out to a broader audience. And the final thing that I always do is convince people of the commercial importance of this stuff. So the challenge with OpenStreetMap, you have probably used it, but don’t know it. And we talk about how many organizations – that’s the process of education, and I think that process will be easier over time as well. So all three of those can be useful tools.
Perfect, thank you so much. (Applause).
Live captioning by Lindsay @stoker_lindsay at White Coat Captioning @whitecoatcapx.