OpenStreetMap

OpenStreetMap and Google MapMaker in Haiti

Posted by Harry Wood on 22 January 2010 in English. Last updated on 10 March 2010.

There is, of course, wasteful duplication of effort between the two communities: OpenStreetMap and Google MapMaker. That's a worldwide waste of effort, but there's been a few people pointing this out in relation to the Haiti Earthquake mapping. here and here and here. OpenStreetMap releases data with a sharealike license. Meanwhile Google MapMaker doesn't normally release their data at all, but in Haiti it is released with a non-commercial license. These license terms mean that neither community can import data from the other. Surely there is a solution that would mean all the data comes together on one place or the other? A solution would be for Google MapMaker to drop their non-commercial clauses. This is very do-able, but will google do it? Another solution would be for OpenStreetMap to go public domain in the Haiti region. This could be do-able, but...

Part of me recoils from the prospect of Google MapMaker importing OSM's Haiti data, and effectively claiming it as their own. We've done so well to steal a march on google and all other map providers, so that people are coming OpenStreetMap for the best map data of Haiti. It feels satisfying that all those misguided fools who contribute to the closed MapMaker system, are now looking jealously at OpenStreetMap haiti coverage. But of course we should put all such thoughts to one side at a time like this. As Mikel said on the topic"whatever can be done for the best benefit to the crisis response should be done". So let's think about what could be done.

We could try to ask every single contributor (and there are hundreds in Haiti now) if they would be willing to go PD on their contributions to this region. I think there could be quite a few who do not like the idea for the above reasons, and if we are unable to get agreement from everyone then there will be tricky untangling process to figure out which bits of the Haiti map can be taken, and which bits can't. This could be more complicated than simply extracting the bits edited by that group of users. If we take the idea of "derivative works" to Ordnance Survey style extremes (as is the convention within the OSM community for thinking about ''inputting'' any data) Then a lot of the map could be regarded as a derived work of a few contributors who mapped out some of the base streets.

All said and done, it could be a quite a hassle for OSM to step down to PD in this region. I don't think it's impossible, but it would require some significant motivated effort to engage all the contributors and get them on board, and maybe turning a blind eye on a few legal issues.

Another question though. What would it actually achieve? If google MapMaker was able to import our data then the MapMaker community would be able to continue contributing from a higher level of completion. Are all the OSM'ers who still want to build the map of Haiti going to go and join in over at MapMaker instead? I don't think so. So there would still be parallel mapping efforts. The main benefit perhaps would be to bring OSM data to MapMaker's downstream users, but who are these downstream users? and why aren't they using our resources? As I said in my blog post, I think the key thing at this stage is to get the word out about our great Haiti map resources, and get people using them.

Discussion

Comment from robert on 22 January 2010 at 14:26

I personally have no interest in feeding Google's proprietary data store.

Comment from Snusmumriken on 22 January 2010 at 14:49

Same here.

Comment from amm on 22 January 2010 at 23:10

When heise.de posted a blog entry about OSM help in haiti, several comments were along the line of "Oh, they are only interested in using the catastrophy for their own publicity and push the cause of OSM". Now heise is known for their Trolls and at first I thought those comments were absurde and that people only want to help in a way that they easily can. And I am still convinced that is true for the vast majority of people who have put in a lot of effort into the OSM haiti page to make it a great map. But I do get the impression there is a kernel of truth to those "alligations" and some people are emphazising the "battle" between OSM and MapMaker more than I think is reasonable.

Yes, I generally don't want to just give my data to Google without getting anything back and so yes I am a strong supporter of a share-alike license normally. But the reason I want a share-alike license is because I don't want to work for free and want to get paid for my work. Not with money but normally with more data. However here in Haiti, my payment would be that this work might save lives or at least help make it less devestating for some. This is more than Google can ever give back! So please, can we tone down this whole "battle" between OSM and MapMaker and just concentrate on makeing the best map that can help people as much as possible. Let downstream users who actually know what they need in the field decide which map better suits their needs!

And also we shouldn't forget that a not in-substantial part that made it possible for OSM to have such a good map was Google jumping over their shadow and explicitly giving OSM permission to derive maps in Haiti from their aerial imagery! It would be ironic if the dataset with the most legal hassels would end up being OSM!

So I would support releasing the limited haiti dataset PD for the moment to let all the aid agencies mashup OSM data with what ever other dataset they have and need.

This leaves the technical question of how to do it? I think it will be hard to come up with a good solution quickly, so it is probably to late for haiti, but we should think of structures of how to loosen licenses for a limited amount of data relavant to massive humanitarian crisis. I kind of like the idea recently brought up on the mailing lists of tagging each changeset with either PD or CC-BY-SA/ODbL. In the case of crisis, we could then ask (or potentially mandate?) everyone editing there to only edit in PD mode and thus create a small bubble of PD data for everyone to use as is necessary. This leaves the question of how to determine derivative works and how the viral part spreads. This is a hard question, no doubt, but with the move from CC-BY-SA to ODbL, we will need to develop these tools anyway!

Comment from JohnSmith on 22 January 2010 at 23:21

@amm, even if PD is required after some event happens, what about all the existing data?

Also I was reading a good summary of this situation on slashgeo.org:

http://appdomains.slashgeo.org/article.pl?sid=10/01/22/1936242&from=rss

even if the data was suddenly given out as PD and mapmaker imports it, it will potentially lead to parallel mapping efforts as mapmaker users improve on the new data and existing OSM users don't use mapmaker to continue making their improvements.

Comment from amm on 23 January 2010 at 00:06

Hopefully a disaster doesn't cover that large an area, so that the number of users contributing to the area beforehand isn't as large as making asking those people to relicense in the case of disaster impossible. I don't know if it is feasible, but I think that is one of the questions that should be asked in a proper review of what went right and what went wrong once things have calmed down again and in discussion of how to improve things for the sad certainty of the next crisis.

I don't think PD is necessarily at all about Google vs OSM, but rather can downstream users use either. Dupplicating efforst is annoying and wastefull, but not catastrophic, given the number of uneffected users willing to help is quite large and the limited information from "out of the field" can be shared by both. Not being able to use either by downstream potentially is.

Nearly all of the uses cases listed on the wiki page for use of OSM has been in mashing together variouse data sources (e.g. imagery, road vector data, elevation data, damage assessment data, ...) to create combined maps for distribution. Which if each dataset sticks a slightly different varient of "free" in their license means they are incompatible and noone can use anything.

Of cause, that is a general argument against viral licenses, but whereas normally I would say the advantages of viral outway the disadvantages, in the case of saving lives I think it doesn't.

Comment from Baloo Uriza on 23 January 2010 at 02:53

It'd be nice if google dropped the legal posturing and did the right thing already. They're skirting doing evil as it is.

Comment from robert on 23 January 2010 at 03:49

I am convinced that in the long term, allowing Google to hijack OSM's community (and data) would be a serious detriment to future 'life saving' relief operations.

If it weren't for OSM's existence I can bet you anything Google would not be offering any way of getting vector data from it. In fact I seriously doubt MapMaker would exist at all.

It would be naive of anyone to believe that both sides haven't been using the Haiti work as a means to publicize their platform. At least OSM is using it to further their project, whereas Google is using it to recruit serfs. I'm also not sure about the notion of aiming for publicity is an unsavoury trait. Publicity is not an end in itself. Personally, what puts a smile on my face is having people make use of my work. The more people know about it, the more it gets used.

I don't see how a PD fork would prevent parallel work. Google would not supply us with data in return. We would not be able to incorporate their improvements. We would continue mapping in OSM. It would just be a one way gift to Google.

I also think that at this stage, the merging of the two datasets would be more work than either community could achieve.

Comment from JohnSmith on 23 January 2010 at 07:44

@amm as more data is entered the number of contributors adding data and making changes will increase, although not at the exponential rates I keep hearing.

@robert mixing data sets is what google does best, they probably already have good tested to import OSM data and given half a chance I have no doubt in my mind at all they would have it imported before end of business that day.

Comment from JohnSmith on 23 January 2010 at 07:45

s/good/code/

Comment from Waldo000000 on 24 January 2010 at 22:15

@robert: "I don't see how a PD fork would prevent parallel work.... It would just be a one way gift to Google."

And a gift to the people of Haiti.

Comment from zenfunk on 26 January 2010 at 11:02

Waldo wrote:
"@robert: "I don't see how a PD fork would prevent parallel work.... It would just be a one way gift to Google."

And a gift to the people of Haiti."

The people of Haiti got that gift allready, our license has the least legal obstructions, is freely available and will be "forever".
I'm not much into this "they vs. us" game and I'm very thankful of google to donate their aerial imagery in the first place.
The OP wanted to prevent duplicated efforts. This can only be possible if both licenses are 100% compatible both ways. We would need a constant syncing of both databases if the duplication of efforts is to be prevented. This simply can't be done by just canging our license. This would only mean that mapmaker gets better and the duplication of efforts would be just the same. The easiest way would be if Google would loose the "noncommercial" in their license since our license is allready very "open".

The duplication of efforts is a nuisance but I don't think that it's a fault on our end.
My 2 cents, Christian

Comment from Waldo000000 on 2 February 2010 at 20:22

@zenfunk

"The people of Haiti got that gift already"

But a PD fork would enable Google to import that data and merge it with their own data set. I suspect the result would be even more useful for the Haitians...

Log in to leave a comment