Overturemaps.org - big-businesses OSMF alternative

I think ultimately this is a recognition that OSM itself is not going to be the map that Facebook, Microsoft et al want it to be. They want a map database where they can host large AI-derived datasets without having to argue with a bunch of crusty craftmappers like us, and the OSM community has made it clear (…insofar as the OSM community ever has a settled view on anything…) that OSM is not that place.

And that’s fine.

It remains to be seen where the less automated contributions from/via the Overture partners end up. Will Amazon Logistics keep adding driveways to OSM, for example, or will they add them to an Overture db instead? Will Facebook have a “POI feedback” tool which goes straight into Overture and not into OSM?

It’ll also be interesting to see what happens with the software projects supported by the Overture partners - particularly RapiD and MapLibre.

35 Likes

I’m not assuming that they want to do anything that is illegal, just distasteful.

4 Likes

then I misunderstood your comment.

1 Like

To be clear: I didn’t see anywhere that they were intending to distribute OSM data on anything else than ODbL terms, just that -their- data would be CDLA licensed. So IMHO that is actually something that shouldn’t cause any concerns.

7 Likes

If they can do that effectively that might be of benefit Currently OSM (certainly the DWG, and to a lesser extent the LWG) gets lots of complaints that are basically about Facebook and similar products (my current location is shown incorrectly, [some data sourced from within a Facebook product] is incorrect, that sort of thing). However, based on previous experience, I’m not hopeful.

2 Likes

yelp (source of most of the dismal POI data FB uses) should be concerned not OSM. There is very little non-community use of OSM POI data for a number of reasons, so while it would be nice to have corporate contributions to that I’m not holding my breath and it is unlikely anything is going to change (our road data is in general what has spiked corporate interest).

3 Likes

I’d be happy if they just had a clearly visible way for people to report POI problems so they didn’t all come moaning to us when Facebook/Instagram/Whatsapp puts them in the wrong place…

5 Likes

So I sit down & map this footway through the Park, then that change goes off into cyberspace, gets “incorporated” then “validated”, but “somebody” decides, based on data from somewhere else, that my path is wrong, so it gets wiped.

From where is it wiped from?

Do “they” come back into OSM & delete the path; or does OSM still show a path running through here, but Overture doesn’t, as it’s under the trees so not visible to aerial imagery?

1 Like

Copy-pasting the response from Michal Migurski from a twitter thread, since I think it offers some insight on the reasoning of pro-overture people.

This is in reply to a “What priorities of the new initiative could we identify that are not achievable with OSM?” question:

In very general terms, OSM currently believes that community precedes the map. This emerged around 2009ish to explain OSM’s then-success by way of its tactics and actions in Europe. Paraphrased: good community is both necessary and sufficient for a good map.

Therefore the top priorities of the OSMF are internal conflict avoidance and boundary policing via its MWG and DWG working groups. The density of the map in a place is just an in-group signal of local community strength. The map doesn’t need to be fit for any particular purpose.

You can see these priorities in action as the OSMF board spends its time tightening membership requirements, restricting voting, and generally worrying about takeover protection. Good choices if you believe in a narrow model of community-building that worked for e.g. Germany.

The Euro-centric community-precedes-map idea has only ever translated well to places where you can make a good map with the same bicycle/pub/yellow-vest approaches that got Northern Europe done: U.S. is too big and the global south lacks similar disposable income or hobby time.

From the perspective of Overture partner companies a good map in non-U.S./non-E.U. geographies is essential. In FB’s case the map needed be display-ready for our ~2018 growth areas; you can see those priorities in @ MapWithAI and Daylight Map. Other companies have other needs.

Another way to say this is that partner companies believe the map precedes the community. Now I’m just speaking as some dude online here, but I did work on this effort through March 2022 and map fitness-for-purpose has always been presented as the real goal and interest.

18 Likes

I think there’s an obvious risk here.

TL:DR: I think Overture is an explicit move by corporations to displace OSM by appropriating its value and adding it to a platform under their control, gutting the copyleft license, using OSM as a base layer mixed with unshared proprietary enhancements, and eventually return mapping to being an endeavour dominated by commercial interests in which most people must pay to participate. The only way to prevent this is by vigorously defending and enforcing the copyleft provisions of the ODbL license on OSM data.

There’s a risk if we want OSM to be widely used and be the default basemap for most applications. This is what I want, though I’m aware that not everyone agrees; some people are happy for OSM to be a niche project by and for enthusiasts, which is perfectly OK but not how I feel.

The ODbL’s copyleft provisions have been, I think, instrumental in ensuring that OSM is not buried by proprietary maps. If OSM data can be admixed with proprietary data, OSM will always be literally the least complete map database available. Every corporate map will contain the OSM data plus the proprietary stuff, making every corporate map OSM+whatever; OSM will be the only one that’s only OSM data. With copyleft, corporates can’t do that; they are forced to either go it entirely alone (Google) or at least make a token gesture toward sharing back with OSM (Apple, Mapbox, TomTom, etc). OSM is more complete and useful than Google Maps in many places, and the corporates that use OSM directly, such as Apple, don’t have anything in their map that OSM doesn’t (they add and/or correct data all the time, but they share it back into OSM).

I firmly believe that Overture intends to break this model.

Overture’s corporate members have openly announced their intention to blend open and proprietary data, mixing the two but keeping the proprietary data to themselves. “TomTom’s Maps Platform will leverage the combination of the Overture base map, a broad range of other data, and TomTom’s proprietary data in a continuously integrated and quality-controlled product…"

They also intend to use OSM data, saying “This will include channeling data from long-established projects such as OpenStreetMap…”

They’re telling us what they’re going to do. They’re going to pull OSM data into another “open” ecosystem, strip it of copyleft, and make it a base layer that every corporation can use, add to, and not contribute back to. If they manage this, every single one of them will have all of the data that OSM contributors have created, plus their own proprietary data.

It will be very difficult to persuade people to use OSM when TomTom and Microsoft have literally everything OSM does, plus more.

Initially, I’m very confident that the corporates will provide their offerings at no cost to most users; they know that they can’t compete with OSM if they’re charging a fee while OSM isn’t—at least, not as long as their data isn’t much better and OSM is still alive and well-known—but they’re patient. They can wait until their proprietary datasets accumulate an edge in quality and completeness over OSM (remember, in this scenario any improvements to OSM are also added to their maps, but not vice versa), and until OSM slips out of the public eye and ceases to be an obvious alternative. Then they’ll jack up the price.

And OSM will die, or become a small niche project by and for a handful of enthusiasts in wealthy countries.

It doesn’t have to go that way. The ODbL license is, in theory, sticky. Overture can’t just take the OSM data and add it to a new repository with a non-copyleft license as long as the OSM community effectively defends and enforces the copyleft on the data.

So yeah, maybe cue up the lawyers.

27 Likes

In high school I had to read The Fire Raisers (play) - Wikipedia and I’ve always tended to take what people or entities say at face value and not to rely on rationalizations that boil down to “they didn’t mean it that way” since. As a result I tend to agree with some of the sentiments, but any strategy to combat this based on legal action is fraught with risk, if not to say doomed, in particular with the risk of an outcome that doesn’t leave us with any thing we can wave around for protection at all.

@ivansanchez thanks for the text from Michal. For those that don’t know, while Michal is no longer with Facebook, he is essentially the thought leader for the groups that think overturemaps is a good idea and the posting reiterates tropes from years back that were already old and tired back then.

Anyway, now is my turn to reiterate old stuff :slight_smile:

The thing overturemap is claiming to do is to provide a one stop shop for open geo data playing on the usual fears with respect to validation and so on. Can we provide something similar without giving up our governing principles? Sure we can! Will there be a monetary and staffing impact? Yes, sure there will.

The OSMF has long refused to become engaged in anything outside of core OSM, for example we don’t serve as a distribution point for elevation data in its many forms leading to users that want to use such data having to shop around among other projects, potentially reinventing the wheel for the 10’000th time by reprocessing raw DEM data, there is no reason we can’t serve as the distribution platform for such data. And there are other low impact no brainers that would just make things easier for people wanting to build on open data. If anybody is, we are the actual experts in curating open data in to a useful form.

Then there is more aspirational projects which could be looked at, for example actually open aerial/sat imagery (notably missing from the overturemaps announcement BTW). Not something that we are likely to be able to do on our own, but given the wealth of open data available now, an imagery curation and open mosaic project would seem to be substantially more viable than it was when WMF-DE last attempted it. Or the long overdue actually open alternative for Mapillary and co.

25 Likes

This is easy to understand because companies don’t need to maintain a community. In OSM the map is made by a community. If there’s no community then there’s no map. Companies, OTOH, can just pay people to map. If one particular mapper doesn’t like the job they can leave, the company will find a replacement.

4 Likes

Can someone elaborate if ODbL allows mixing ODbL licensed data with proprietary without contributing it to OSM/making it publicly available?

I believe it do not.

2 Likes

I mean, you can just look at the low quality of the map in most places in the United States and the chronic lack of any kind of community building here to see that most of what Michal says in their post is true. The whole community based bicycle/pub/yellow-vest approach that worked in Europe clearly doesn’t work here.

I’ve been an editor since late 2016. The only real mapping community in the United States is focused on a small area of Northern California. There’s never been and probably never will be the same kind of bicycle/pub/yellow-vest mapping efforts here that worked in Europe. As far as I know no one is even trying to create local mapping groups here anyway. Let alone would we be able to mobilize and map the country in the same way that guys did with Europe.

I live in a pretty safe area and I can’t even add a telephone pole to the map while on a walk without getting questioned by the cops or being accosted by a concerned citizen. I can’t image how on the ground mapping here would be tenable anywhere even slightly dangerous. Let alone at any kind of scale that would matter. So as far as I’m concerned a purely, or mainly, community based mapping system in the United States like they have in Europe is a non-starter. That’s one the main reasons we embrace presets and AI assisted mapping. In most cases they are the only options we have.

You can’t just take what worked in Eastern Europe and expect it to work everywhere else though. The excessive over focus on local mapping communities at the cost of everything else will really be the death of OpenStreetMap IMO. Things like OvertureMaps are really just the natural outcome of an inability by certain interest groups within the project to allow for other opinions or cede any ground on how things are done in certain countries. The priorities of mappers in the United States and how we prefer to do things are always on the backburner and seen as infer to the whole “community based mapping” thing. So of course companies from here that want to utilize OpenStreetMap are going to create their own maps and ways of mapping instead of endlessly being on the losing end of things.

6 Likes

Mixing is a very fuzzy concept and the question can’t really be answered conclusively without a concrete use case.

You can find further information on practical interpretations here Licence/Community Guidelines - OpenStreetMap Foundation

To be very clear substantial amounts of OSM extracted data always remain ODbL licensed as long as they are used as a database.

8 Likes

I can’t really say I blame them considering how badly Bryan Housel was treated by people in the community over the years. There’s really no upside at this to them allowing the community to have even a minimum of control over the project. There should be some give and take when it comes to the community having any kind of control over anything and there just isn’t any. From what I’ve seen it’s always been completely one sided. I don’t know if the OSMF involvement in the project would be the same way, but it’s understandable that they would want to be as independent of the project as they can be. Especially considering how hostile some people in the community are towards AI mapping and the fact that it’s at the core of their editor.

3 Likes

Dang it @ivansanchez you posted that and it almost drove me back to Twitter to argue with Migurski over every misrepresented point.

It’s simply not the case that OSM has stagnated on the methodology developed in late 2000s in northern Europe. High quality satellite imagery is widely available. Human in the loop review of automated data and imports is not uncommon. Corporations have made significant amount of edits. OSM data is significant in every country of the world, and how it’s done has adapted everywhere. Figuring out none of this is easy, and I’ll readily admit the debates have frequently frustrated me, but the trend has been to considerately expand to new methodologies that retains what is the essential high value of OSM. People everywhere working in cooperation with the right amount of engineering.

11 Likes

I am hopeful that OSM community will prove that it’s not their last word. We just have to double down on what makes us unique, the community itself.

My observation is that there are two stalling factors: by sheer numbers hardly anyone new to OSM knows the unwritten rules and values of OSM; also hardly anyone bumps into contact with another mappers - I checked some accounts which just achieved their 10th mapping day and most of them has had none of the changes commented on.
This is due to many factors: experienced mappers themselves not realizing that this work is important - and the UX of OSM.org website (for starters, some kinda notification center is long overdue, e-mail notifications can drown in other spam; also many countries are mobile-first which degrades UX quite a lot)
Thankfully both of these barriers are not insurmountable. I am mildly optimistic, we just have to recognize the pain points and work around them.

My feeling is that current community building events such as mapathons won’t cut it at large - the return on (time) investment is measly due to mapper attrition no lower than for organically signed up mappers.

6 Likes

100% it has. People from Europe routinely revert “automated edits” made in the United States that no one here cares about simply because of the European mentality that presets and AI mapping are evil or whatever. A day doesn’t go by where I don’t see someone from the DWG reverting someone adding brand tags to stores using NSI presets just because they can, not because the data was bad. Same goes for MapRoulette challenges. I can’t edit more then two POIs that have the same tag in the same week of each other without someone jumping down my throat about me doing undiscussed mass edits. The idea that constant reverts of otherwise acceptable data and knee jerk tantrum throwing about “undiscussed mass edits” (whatever that means) doesn’t lead to stagnation is just laughable.

To quote @ZeLonewolf’s comment from another discussion “I continue to beat the drum that the project is governed entirely by ad hoc human relationships when it comes to getting anything done. What may look like “process” to an uninitiated observer is really just smoke and mirrors.” That’s literally all it is. A few people from Eastern Europe who are lucky enough to have the human relationships with each other dialed in dictating how everyone else does things. There isn’t really even any kind of methodology behind it. It’s 100% impromptu reaction to random, mostly harmless things in the moment.

That’s simply wrong for most of the world. There’s plenty of places in Africa that were badly mapped years ago by HOT contributors that are still that way. Personally, I’ve spent most of my time over the last six years mapping in a hundred mile radius around where I live. I’m still struggling to clean up imports from 15 years ago even after 6 years of editing in the same area. No one has ever come along to help me or review the data. Just look at the Tiger import from 15 years ago. We’ll be lucky if it’s cleaned up by 2030. The main reason the Tiger data was ever cleaned to the degree that has been is because of paid editors. It has nothing to do with people in the community cooperating together to make the map better with the right amount of engineering or whatever though. Sentences like “the high value of OSM is people everywhere working in cooperation with the right amount of engineering” are good marketing, but that’s about it.

Look, at the end of day I’m just a random nobody. I don’t work in the industry. I’m not a geographer or whatever. I just do this as a past time. That said, it can sometimes help to get an outsiders opinion. And in my opinion as an outsider there’s only two ways this whole thing can turn out. Either the community can do some serious reflection on what brought it here and evolve to meet the moment, or it can dig it’s heels in by doubling down on the same old things that got it here and eventually die out. That’s it. There isn’t really a third option here. Personally, I’d like to see the project do some serious reflection and evolve to meet the moment. That’s just me though :man_shrugging: But not taking feedback from experienced editors seriously isn’t going to help anything regardless. @RicoElectrico’s suggestions are a good starting point, but there needs to be more behind whatever changes are made then just doubling down on the same old tired clichés about how the community is what makes OpenStreetMap unique or whatever. There’s nothing wrong with it, but it’s ultimately an outdated paradigm that most mappers never have any experience with anyway.

17 Likes