Crowdmapping That Works
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
Title |
| |
Title of Series | ||
Number of Parts | 156 | |
Author | ||
Contributors | ||
License | CC Attribution 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
Identifiers | 10.5446/68451 (DOI) | |
Publisher | ||
Release Date | ||
Language |
Content Metadata
Subject Area | ||
Genre | ||
Abstract |
| |
Keywords |
FOSS4G Europe 2024 Tartu33 / 156
6
33
35
53
55
59
61
67
70
87
97
99
102
103
104
105
107
111
121
122
123
124
125
126
127
128
134
144
150
151
155
00:00
MappingMultiplication signEntire functionLevel (video gaming)Projective planeOpen setLecture/ConferenceComputer animationMeeting/Interview
01:25
Multiplication sign
01:49
FreewareMappingComputer animation
02:12
Process (computing)Entire functionLevel (video gaming)Point (geometry)1 (number)MappingOpen setProjective plane
03:15
ResultantGoodness of fitMappingComputer animationDrawing
03:52
Projective planeProcess (computing)WebsiteBuildingComputer animation
04:33
MeasurementDifferent (Kate Ryan album)SatelliteProjective planeNoise (electronics)View (database)
05:13
MappingLevel (video gaming)HomographyBus (computing)MereologyBuildingMultiplicationPlanningProjective planeMedical imagingMobile appInteractive televisionPoint (geometry)SatelliteComputer animation
06:25
MappingData managementMultiplicationSatelliteMereologyTask (computing)2 (number)LaptopPlanningAttribute grammarComputer animation
07:40
Software developerMobile appMultiplication signOpen setCovering spaceComputer animation
08:07
MappingField (computer science)Digital photographyKey (cryptography)Point (geometry)Computer animation
09:16
NumberMobile appRevision controlProjective planeDigital photographyEvent horizonPlanningSoftware testingMappingSoftware developer
10:44
Mobile appComputer animation
11:12
FacebookMobile appEmailMappingDatabase
12:03
ResultantMultiplication signXMLComputer animation
12:25
Revision controlMapping2 (number)Projective planeVideo gameOnline helpLocal ringComputer animation
13:12
Computer configurationNoise (electronics)BuildingComputer animation
13:40
CASE <Informatik>Online helpBuilding1 (number)Service (economics)Projective plane
14:17
Mobile appMereologyProjective planeConnected spaceGame theoryImage resolutionComputer animation
14:57
SurfaceRankingLevel (video gaming)Mobile appLevel (video gaming)Open setGoodness of fitArithmetic meanSurfaceText editorState of matterBuildingMultiplication signPoint (geometry)Expert systemStatisticsStudent's t-testType theoryComputer animation
16:22
NumberCountingMobile appGame theoryPoint (geometry)Projective planeNumbering schemeCryptographyPhysical systemDiagram
17:33
MappingOpen setProjective planeStudent's t-testInformationLevel (video gaming)Physical systemStatisticsProper mapComputer animation
18:47
VideoconferencingOpen setProjective planeAlgebraic closureMultiplication signBlogLevel (video gaming)Building
19:22
MappingComputer animation
19:53
Open setProjective planeGame theoryCommitment schemeRule of inferencePlanningStatement (computer science)FreewareMedical imagingMappingTrailTerm (mathematics)outputMereologyLevel (video gaming)Matching (graph theory)Source codeProcess (computing)Group actionPresentation of a groupRoutingMathematicsInformationDoubling the cubePhysical systemCompilation albumSelf-organizationVirtualizationProper mapMathematical optimizationComputer animationLecture/Conference
25:41
Special unitary groupComputer-assisted translationComputer animation
Transcript: English(auto-generated)
00:00
So, crowd mapping. Crowd mapping is making people who don't know do weird things. I come from OpenStreetMap, you know this, actually amazing project, yeah, because it's a map of entire world that is made for free by volunteers who spend their time
00:22
doing, collecting the data for no reward, nothing. And the data is great, you all know that, and people from government, from commercial companies, they look at that and they think, oh, I want this for people, unpaid people to make the
00:43
data for us. Like, come on, that's like the best thing. And I'm here for OpenStreetMap, will I tell you how to do that? Well, yes, but the first advice regarding OpenStreetMap is that, no, stop, no, you don't know what you're getting into.
01:01
Like, yeah, it looks attractive, you might think, yeah, I want that, but you don't know the whole story. You will regret it, believe me. Like OpenStreetMap took like four years to just take off. You don't have that kind of time.
01:23
So how do you do crowd mapping? First try not to, yeah, try just buying things, it works, it will save you time, it will save your nerves, it will save you money actually. Just look around, maybe the data you need, you hear, like, because you need the data
01:45
probably, I don't know, or because this is fun, try buying. You can buy a lot of things nowadays, you can buy shops, data about shops, you can buy, I don't know, land use data, you can get it for free.
02:01
You're at the conference about free data, so probably there's somebody who knows where to get things you need, without crowd mapping, come on. If there's no data, you can pay people to gather it. They're professional surveyors, they're OpenStreetMap members. Very few people pay OpenStreetMap members, and too bad, because they do get the job
02:27
done. Well, people actually pay OpenStreetMap members, there are thousands employed at different companies, but still, you get my point. Find a way to spend money so you don't get into crowd mapping. It works better than you expect.
02:43
But yeah, sometimes there is no other way. You need data for entire country, for entire city, you need some specific data, like, I don't know, street lamps, nobody has data on street lamps. So yeah, crowd mapping, how to do that?
03:00
There haven't been actually a lot of successful projects in crowd mapping, I will tell you about the good ones, but there is kind of common process to crowd mapping. Yeah, but you all need to know, like, the problems.
03:21
Like, crowd mapping produces really, like, uneven results, uneven coverage, that's like the main thing about this. This is the result, yes. There was a really good research paper coming out in March this year about, like, the impact of crowd mapping, about companies ordering that, and every single company complained
03:46
that the coverage was worse than they expected. So, yeah. So, the process. How do you start? You start with planning. If you don't start anything with planning, you're doing it wrong.
04:03
So, and that means answering the question, what exactly do you need? What is the project that you're doing? Do you need, like, to collect benches, collect street lamps, bicycle lanes, illegal parking spots, illegal dumping sites, entrances to buildings, anything?
04:24
So you think really hard, what exactly do you need? Because the simpler you made the process, the better, the more people participate. Like, you might want people to go outside, like, in this project where people map noise levels.
04:40
They made people go around the city and measure noise, but how many people will get outside? It's really hard. Like, look at us, we're indoors. Getting people outside is hard. So, but it's 2024. We got satellite imagery, we got streetwise imagery, Google Street View, Mapillary,
05:06
CartoView, Panoramix, different ways of doing things from home. So, like, look at the most important, most popular crowd-mapping project, MapSwipe. Some of you have probably used it.
05:21
Really simple. It shows you satellite imagery and asks you to look for buildings, look for roads. If you see one, you tap. If imagery is better, you tap twice, swipe, and the next part of imagery. And you just swipe through a lot of parts of imagery, and it's really addictive.
05:40
You get points for that, you level up, and it's fun. You can do it from everywhere because it's just imagery. I have done it from glass. I have done it from bathroom, anywhere. Hundreds of thousands of people have used the app, they completed hundreds of projects,
06:03
and it's really popular because it's so simple. Just two interactions. But you will say, that's not mapping, right? We're here for crowd-mapping. It's just tapping on imagery. Come on. And that's the secret to proper planning.
06:22
It doesn't have to be a single campaign. It can be multiple steps. So MapSwipe is just the first step. People just annotate the imagery. It's first step in humanitarian mapping. There are also next steps. Like, that data is fed into humanitarian tasking manager.
06:43
And people behind laptops, they get narrow scope to map because the scope is usually the entire country. You cannot map the entire country. Like, you know that. But you can map parts that are interesting. So as a second step, people do the actual mapping,
07:04
tracing from satellite imagery. And of course, there is a third step, when that data is shared among the people who actually go into, like, get in the car and get in the place and add the missing attributes, like names,
07:22
like clinics and stuff like that. Multiple steps, each different, each reasonably simple. You don't have to do it all in one. So planning might. And the first step is the most important step. And it should be as simple as possible, maybe even without giving hope.
07:42
And now that you know what you need, it's time to develop. And developing, you know, development takes money and time. It takes some specific knowledge that some of you don't have, especially when you need to develop an app to go outside.
08:01
So when you use an app, you got it covered. You're the open conference. And we got merging maps. We got cool fields made specifically for crowd mapping. There are an exist collector from my company.
08:21
I won't tell you about this. Ushahidiya for collecting points, which had been synonymous with crowd mapping. So many tools that are already present. There can be drawbacks. They're really complex to set up, to use. And you will have to teach people how to use them.
08:44
And when your crowd mapping is insolvent, it's hard to teach everyone. So you can make it simpler. You can collect the photos. Like you're collecting broken benches to fix city infrastructure. Just make people go around and collect geo-referenced photos
09:01
with Mapillary or just sent to the messenger. Like here's the bench I want fixed. It continues. It's really simple. It doesn't need to install anything. You get more people contributing. Getting more people is the key. There was this project in Italy recently about collecting, about crowd mapping illegally parked cars,
09:28
which is kind of funny, but it's really important for city infrastructure. And they resorted to their own app. Why did they make it?
09:40
Because it was a really intense and focused project. First, it required a lot of planning because the entire project was meant to be done in just four hours in one evening. So they made a beta testing version of the app.
10:00
They tested it with 50 people and it shows where you need to go, what to make photos of, all the numbers. Really simple, few clicks. And the actual event took 2,000 people who over four hours mapped, made photos of 60,000 cars all over Milan.
10:25
Really massive project. Come on, 2,000 people in four hours. And for that, really my solution won't work. That kind of thing can be done only in custom things.
10:40
So yeah, development is important. And developing sometimes hard, the same paper I mentioned, here's the name, has some charts about how to properly develop and test and publish the app. Really useful stuff. And you have published the app and nobody comes.
11:04
Why? Because the next step, marketing. People need to know about the app. How do you do that? And of course, that's obvious. You publish articles in the news, you reach out to communities, you just share it as wide as you can.
11:23
How did that Italian app get 2,000 participants to go outside? Like, come on, the harder thing. There were activists, so they basically had an email database and asked everybody to sign up. So activists work.
11:41
They're called activists for a reason. So try to reach to interested people. If you're doing crowd mapping in Tartu, go to Tartu Facebook communities that are active, that post things every day.
12:00
But marketing, first and foremost, is about answering to why exactly are you collecting? What is the benefit for people? You know what you need, what people need. And this is a time to be truthful with yourself. Like, what is the end result? It's not to get the data.
12:21
It's never the goal, actually. It's to, I don't know, improve stuff to make better. The best version is in missing maps of the answer to this question. Missing maps is a second step in humanitarian mapping. They gather the people in the room and they make them map like African cities or Asian
12:43
or kind of like that, affected by natural disasters. And they basically say, we map to save lives. And you can't beat that. Like, that's the ultimate goal, to help somebody's life with it. And you might want to use, you shouldn't use that.
13:04
That's like one in a kind opportunity. But your project is important locally. So you should answer, who am I helping with this? You're not just collecting broken benches. You're helping tired people to spend time outside
13:22
and improving general health. You're not collecting bicycle lanes. You're improving city infrastructure, reducing noise and air pollution, helping children get safety to schools and people to have options of getting to work. You don't collect building entrances, just you're doing it
13:43
to help emergency services because there were cases when there was a fire and the car just didn't know from which side to come to the building and they lost a minute. And that was very bad. So try to answer, who are you helping? Because people who will help you gather the data,
14:03
they need to internalize your goals. So they need to know that they will be the ones that will benefit from this project. So people have got to a project,
14:21
they've installed the app that made one edit. What makes them do the second edit? How do you keep people inside? They already know why they're doing this but they're lazy, everyone is lazy, I am. So why do they keep doing this, called engagement?
14:41
And actually, this is a solved problem, like look at games, gamification, look at social networks, like connection. People become activists in part because they want community. So, but yeah, apps generally have means.
15:01
I mentioned a score in MapSwipe, there is a really good opposite map editor, street complete, that asks you questions, like what the surface, how high the building. Every time you answer the question, you get points. You get many points, you level up. More type of quests are opened.
15:20
At some point, you become an expert. People love to be called experts. Yeah, my friend, Letwin, she manages Youth Mappers in Zimbabwe and Youth Mappers are students, and she says students want good titles.
15:42
Like at the beginning, you're a beginner mapper, nobody wants to be beginner mappers. They want to be intermediate or expert mappers. So when you give students those achievements, they are really happy and they do more work. So yeah, street complete has that. Achievement scores, leaderboards.
16:01
Leaderboards are important because people love to compete. Some people, not me, but I heard that people love to compete. So, and calculating stats for leaderboards is hard, but it's possible and you should do it because it drives engagement, it gets you more edits.
16:23
There was a project 15 years ago from New York-based start-up about collecting POI, points of interest, shops and stuff like that. And they made several successful projects, including like 60,000 shops in Indonesia
16:42
and hospitals and stuff like that. And what they did, they had leaderboards, but also they awarded money to most prolific contributors. It was kind of crypto scheme, I don't know, but it works, people were really engaged, 60,000.
17:02
But it's gamification. When you add games to your app, people start gaming the app. So you're measuring counts, people will start adding wrong fake edits because they want to see a number go up. So this app required two, three confirmations
17:23
for a single POI and that reduces the number of POI you can trust. That's also the problem, gaming the system. So the project is complete. Finally, it's been a week or a day or four hours.
17:42
Everything's done. What do you do? You want to just get the data and run? You shouldn't. You should close the project properly. You should, I don't know, post articles, collect final stats. If you're sharing the awards for mapping, do the awards.
18:03
This was a project a couple years ago, Open Street Map Mapping Competition, and we collected 5,000 kilometers of streets with lane information and thousands of crossings and stuff like that. We took a lot of effort to measure stats properly
18:23
and did the final article about the competition. We went the extra mile. We contacted every winner, got some story from that because people like to be presented. People like to show themselves.
18:44
This also works. Same with students, actually. This is a project we made in 2011, early days of Open Street Map, everything was empty. And we mapped this in two days. The entire city, 20,000 buildings,
19:01
all the land uses, all the roads. And when we finished it, we didn't just show before and after. We made a video. We made several articles. We published a big article to a country-wide tech blog. So this was the entire celebration. It was so fun. We did it several times more.
19:22
Closure is important. And then you're done. You can get the data. You can continue your work. You can help. But also you got the experience. Very few people in the world have got experience of running a successful crowd-mapping campaign.
19:42
And maybe you could work on that. So now you're unique and maybe help somebody to do it even more. And yeah, it sounds really simple, five steps. I'm from Open Street Map. When I apply these steps to Open Street Map,
20:00
it doesn't work. There was no plan. Developing is really bad. In Open Street Map, there is no marketing, no engagement, and no end goal. But still it works, I don't know. So those are like guidelines, they're not rules. So really, few people know how to do that exactly. But that's an overview.
20:23
And it helps. And some campaigns were successful. So if you want to run one, we can talk and do some brainstorming about how to get the crowd-mapping properly. And that's all. Thank you. I mean, commitment also of the communities and everything.
21:02
So any questions from your side? Yeah. Oh, wait, wait, no. I'm going. Yeah, you need to check back. Miki doesn't have a microphone. What is your approach to vandalism with user inputs if you have to have an image or a free text?
21:23
Yeah, what about vandalism? It's a problem in Open Street Map. This talk is not actually about Open Street Map, but I mentioned gaming the system, getting fake edits to get a higher score or get monetary price or something.
21:40
Just double-check everything. Like StreetCred did, just require not one, but two, three people validating the same thing. Yet it usually works. It gets you less data, but it gets you higher quality and more edits, I guess. So it evens out.
22:02
Just more people works. Other question? That was a lot of info, so yeah. I have one question, Ilya. What are your thoughts also for some organizations who are using a crowdsource data such as virtual maps?
22:26
How do you see that complementary at what was happening with crowd-mapping? Is that something that the communities can also embrace, or is more about how they can collaborate together?
22:40
So what about companies that's using crowdsource data? Again, this is not exactly about Open Street Map, and Overture Maps is the entire topic I can make a separate talk on. But generally, seeing your data used,
23:01
seeing the thing you worked on as a part of crowd-mapping company, you went outside and collected broken benches, and seeing those benches fixed because of your action is like the greatest reward you can have. So these projects using crowdsource data, they're doing the job of getting your work
23:23
to people who can do something with that. And this is really important. It's not only important to collect the data. It is important to make the change that you intended to do with the data to fix benches, to optimize fire brigade routes,
23:42
to remove illegally dumped waste, and stuff like that. So don't forget about actually doing the thing. Crowd-mapping is not a goal in itself. Thanks so much, Ilya. One last question. We have still a few minutes, or, okay, yeah.
24:02
Sorry, we need another microphone. Thank you for interesting presentation. There was a very pretty guideline, how to make a crowd-mapping that works. And you mentioned that OpenStreetMap doesn't match for this guideline.
24:23
Do you know any project that also done much but was successful? Frank, do I know other projects that's unusual in terms of those guidelines? No, because if they don't do marketing,
24:42
I cannot know about that. But also kinda, because there are local projects that short-track on some of the items. You don't need marketing if you need 50 people, and all those people are part of your community. Yeah. But again, if you miss any of those steps,
25:04
actually then your project won't take off. Come on, there are pretty obvious steps. I did them in like five minutes, because, come on. You have to share your project. You have to engage so people don't leave.
25:20
You have to develop stuff. So, I don't know. OpenStreetMap is unique. I started my talk with that. Don't look at it. Don't wish for that thing. Just do your small project really good. Maybe one quick last question before the next talk,
25:40
and then, okay, all right, so. Thanks, Ilya. Thank you, everyone.