Overturemaps.org - big-businesses OSMF alternative

Foundation
41

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

I don’t think arguing is gonna help, since the pro-overture folks have been cooking on this for a few years, and see OSMF as an adversary already. Trying to argue is gonna be met with resistance.

Instead, please try to use this to understand how the overture folks think.

From my point of view, it all boils down to a stark difference in our worldview: Us EU liberté-tards see map data as a commons, whereas US freemarketarians see map data as an asset.

And, you know what, if I try for a minute to think of OSM as an asset, they’re right. The OSMF sucks, and gets in my way of exploiting my asset. (If this were to be exaggerated, it’d be similar to the way that those pesky safety regulations get in my way of doing business.)

Since I’m getting old, this looks like GPL vs MIT all over again. And since I remember how Apple gobbled up webkit and the OpenBSD kernel, I’m gonna make a bold prediction a couple of years into the future: Meta+Microsoft+Amazon are gonna be happy in their bubble until Apple+Google get into the fray and start using CLDA data without giving anything back.

15 Likes
42

Suppose someone takes data from OSM and feeds it to some AI model (which does not have a database in the classical sense). And then he feeds this AI with other data that he has collected but does not want to pass on to OSM. What about the licence for the map data that this AI will generate? Is this an abstract situation ? For the corpo giants who are behind Overturemaps this will be a very likely way to go.

7 Likes
43

Sure, besides increasing community engagement I think the super decentralized model in which only the database is centralized won’t scale; ultimately there will need to be some centralization of governance; both with respect to software and mapping itself (i.e. an arbitration to resolve long stalemates just like in Wikipedia).

3 Likes
44

This is definitely not abstract, and it is very similar to what happened in the software world with OpenAI Codex. That model is based on open source code found online and despite the complaints from the Free Software Foundation the model is currently being used without carrying any copyleft or other license constraints from the original code (the model was trained on) to the code generated by the model. In a comment from OpenAI they claim that trainin AI systems constitutes fair use and that institutions should clarify the current legal uncertainty.

7 Likes
45

I am still reflecting on all these recent moves from TomTom and now Overture. Right now, in strategic terms, I think that OSM lost a huge opportunity of being THE geographical standard DB (looking for the past 5-8 years). Given the fact that OSMF was not contacted about these projects, and that this huge blog entry on TomTom does NOT mention OSM at any point, well, that shows a lot the relevance of OSM…

9 Likes
46

Well… What if they actually create a better fork of OSM? Now it’s up to us to decide whether to stay here or moving there, looking for that shiny new things.

To be honest, i’m really interested with their idea of “Global Entity Reference System”, since i’m, too, currently working on this aspect by introducing semantic element to OSM data. If they actually build something better overthere, i probably straight up abandoning my project and move there, if possible.

1 Like
47

The last time I am aware of the OSMF being alerted to something of this order of magnitude going on in advance is more than a decade back. With other words the OSMF being ignored is quite normal, what isn’t normal is OSMF members ganging up and not giving a heads up.

48

@Adamant1 I don’t dispute that there are misguided reverts by DWG and others, or that there is a lot bad HOT data, or that too much done in ad-hoc relationships. What I do dispute is that these factors represent the whole of the project, or even the majority of contribution by these actors, or that 100% nothing has changed in 15 years. OSM would be a total mess if not for the heroic efforts of the DWG. HOT has resulted in high quality and impactful data in many places. This is not to let them off the hook for problems, but we have to look at contributions and dynamics in a balanced way.

I understand your frustrations, take your feedback and fully agree there’s a lot to work on. It’s always been the case that there’s a lot to work on – and we have made a lot of progress. If I have to pull an old man card – I was there in 2009, and things are no doubt very different today. I hope you continue to work on OSM and come to see yourself as part of the community and not an outsider.

To me, the actual problem in OpenStreetMap is absolutism and tendency to cast us vs them narratives. You can see @ivansanchez defaults to this as an explainer for how Overture thinks, and Migurski’s relies on tired tropes in his Twitter thread. Both are cartoonish and unhelpful to make real progress on real problems.

This absolutism exists from the very formation of the project, when the enemy was the Ordnance Survey. It was motivating then but it has long been a pathological part of OSM culture. I have moved against this self destruction for many years, and you might call that “good marketing”, but I call it being practical and open.

8 Likes
49

This is legally largely unexplored territory, but is is highly unlikely that there is going to be anything enforceable here. But as said it is largely unexplored, fwiw there might be even an advantage to OSM data not having copyright protection in the US.

1 Like
50

You shouldn’t be feeding the trolls (I realize that you may have forgotten)…

4 Likes
51

Apparently I’m speaking too much sense for the trolls and they’re marking several of my posts on this thread as spam.

2 Likes
52

The current default is to automatically hide a post after it is reported, but you can still view it by clicking on it. While this generally makes sense it makes it very easy for somebody to be disruptive. I expect the moderators will restore things rsn.

[fixed now]

4 Likes
53

(quoting Michal Migurski)

As @mikelmaron says, this is a bit of a tired trope. I don’t think I’ve been to a pub mapping event since 2008. (I’ve been to OSM London pub meetups, but they tend to be more tech and community-focused.)

The map in Britain has been largely built by “lone rangers” - enthusiasts of many stripes (tech, civic, ramblers, roadgeeks, railfans, cyclists, or just people who like maps), working significantly from survey but also from open data sources. The amount of post-2010 mapping done by people who meet in a pub is single-digit percentage at best, very possibly sub-1%.

There are clearly reasons why the US map hasn’t developed this way, but it’s less clear that it couldn’t. Yes, the US is big, but the corollary of that is that Wyoming and New Mexico are also pretty empty therefore don’t need much mapping. Small towns in Colorado have as many computer literate people with civic pride as small towns in Cheshire. “Too big” doesn’t explain why London is better mapped than NYC.

Which kind of brings me back to the earlier point. I think Overture Maps can, and possibly will, make a success of a map combining AI-derived features with public datasets. It will have roads and rivers and broad-brush landuse. I am not convinced they will get community, ‘soft’ infrastructure (footpaths, micro landuse) of the sort OSM does so well, nor POIs of the sort Google does so well. And until they manage that, they’ve just created a new version of the TomTom basemap with different licensing conditions.

But I’ve been wrong a million times before, so who knows.

21 Likes
54

Hey everyone, nice to see replies to my tweets here. To address one line of critique in this thread, re: old/tired/cliché tropes: literally no one is more exhausted by these tropes than I am but that doesn’t make them untrue. Seeing the same pattern of discourse repeat itself for decades in a project that is otherwise a significant success is really unfortunate!

Continuing to speak for myself as an individual, it makes me happier and less grumpy to put my effort into tangible forward progress on complex or difficult group efforts. Sometimes they include core OSM things, like the 18 months I spent in 2020–2021 getting a single small PR merged into the website repo or the year-plus I put into collaborating with Maggie, Heather, Arnalie, Steve, Rob, Rubén, and many others on the LCCWG moderation recommendation. Mostly though I find that energy and reward in efforts like Overture. I hope this sheds some additional light on why a bunch of smart and informed people at these partner companies decided to create a new organization in the LF instead of starting at OSMF.

5 Likes
55

No, that doesn’t shed light for the smart and informed people here on why a new organization was created. You’re still painting this in us-vs-them fashion. A pattern of discourse is not a reality, especially when you feed that discourse. Not to say that you do not have strong points about how much work a PR takes or the challenge of moderation. It’s just incredibly selective of viewpoints and work in OSM/F.

The more I learn about Overture through conversations with people directly involved now, the more I see it’s about doing something necessarily different than and complementary to OSM. I’m sure more light will come out about that soon.

5 Likes
56

As practical action: I believe we do solutions, and fast (already by 2023, their release date), to mitigate the technical gaps and put this under OSMF infra, not outside. Don’t need to be perfect, but needs to happen, even if read-only generated version from real data/metadata. Some things might be easier than others, but continuation of gap not only massively will make new volunteer collaborations outside these companies go to Overture Map Foundation (because of usability), but even part of existing collaborators directly on OpenStreetMap will move to Overture Map.

Bit more context

They do say they do not try to be a community, and are focused on production-ready data use (normalize schema, cleaning for vandalism, etc). However, while they might be perceived here as mere competition to OpenStreetMap, they seem to be advertised by news media as open data competition to Google Maps related services. I even noticed here in Brazil, notices like this one (which, by the way, are the type of tech influencer I would ask to promote OSM for developers):

Captura de tela de 2022-12-16 16-40-11
Captura de tela de 2022-12-16 16-40-11720×1125 370 KB

Triva: "mapas apertos"@pt = "open maps"@en.

So, even if they don’t say they’re trying to be a community, the fact they will start of with small part of heavy collaborators from OpenStreetMap (the ones from roads) and have an normalized output to please developers of geo data (not just map), assuming the obvious fact that to compete with Google Maps, they would need to start collecting amenities from massive user base out there’s, then, how much collaborators they actually would have? My argument here is that the fact that “OpenStreetMap collaborators” not appear (and it was a way to advertise OSM exist) is actually less problematic than the likelihood of them making it easier to accept data directly on them. Their license already is free like OSM, and is backed by a trustworthy organization, Linux Foundation. And they are even likely to provide very well documented APIs with low fees at negative margins of profit to target Google Maps and related services (not OpenStreetMap).

I didn’t do the math, but outside Europe, the coverage of OpenStreetMap is lower, and they’re going for it. What I know is that Brazil is over 80% of the area of the entire Europe, yet data dumps are less than Germany, so while my account is new (5 October) I was already discussing here with well planned ahead, imports and synchronizations could easily get top contributions. The way this Overture Map Foundation seems to be aiming is making a similar approach, both at their working groups (data schema and conflagration), and seems to be targeting at world level. So, by inference, even with less perfect care like tooling for a country like mine, they’re in very few years will be able to not need to show “OpenStreetMap collaborators” at all in special in counties in Global South like India, but before this they’re likely to stop all together to get amenities from OSM (as the data model might shift over time).

So, without making this overlong, yes, I do believe we should make the technical gaps fast. Their “Global Entity Reference System” likely to provide at least identifiers for abstract concepts (like Wikidata Qs for an specific amenity that will not change over time), so this cannot be done without add something into data model, however other parts like normalized version would be feasible to do with existing tools without need to change the original data

1 Like
57

it makes me happier and less grumpy to put my effort into tangible forward progress on complex or difficult group efforts […]

I’m sorry, @migurski, but I interpret that as “Long ago I got fulfillment with OSM, but now it bores me; and now Overture fulfills me”. Which doesn’t shed any light on the motives of MAMT (Meta+Amazon+Microsoft+TomTom) to band up together and antagonize OSM(F).

3 Likes
58

:open_mouth:
Sounds very different from Sweden. I have never been accosted by anyone during all my mapping adventures.

6 Likes
59

My guess is that the hype and forwardness will be in the AI companies favor. Training with OSM data plus other data on highres orthography and outputting to a database will most probably not be protected by obdl if it is ever challenged in court.
Paying editors to feed the AI incrementally improving training data seems like a good idea to me :man_shrugging:

60

You’re right, I apologize for making my post sound too much like a personal fulfillment story.

Individuals and orgs who participate in open mapping are doing so to make something tangible happen to advance their goals, like improving map quality in a region/category for FB or improving some slice of tech stability and community conversation for me. Many companies beyond MAMT have spent years making OSM work for their needs. With engineering staff communicating and changing jobs between them, the various efforts all start to look the same after a while and cry out for a shared utility/commodity standards approach. We find that OSMF has not prioritized these areas in its actions so we create other avenues to be involved in open mapping.

6 Likes