What does Overture Map mean for the future of OpenStreetMap?
blog.andygol.co.ua
blog.andygol.co.ua
I wrote a post on my blog about my concerns for the project, and some ideas for improvement:
https://blog.emacsen.net/blog/2018/02/16/osm-is-in-trouble/
It was not met well. I was (and still sadly am) subject to personal attacks by certain members of the OpenStreetMap community.
I don't know much about Overture Map, but standardization of tags, and a more sophisticated data model would both be very welcome, and make a map built on this data easier to work with.
More to the point, it's never been the technical issues that have held OSM back. Instead, as my post tried to point out- the technical issues stemmed from issues with leadership.
The post here mentions issues with the OSMF. At the time I wrote my post- the OSMF was largely symbolic, and certain community members simply held the key to certain aspects of the project- everything from servers, to software used, and more.
It was a huge mess, and I'm cautiously optimistic in seeing something new come around.
That fear came from lack of vision. Without a vision to to work towards, forces tended to see the best path forward as "No change", since no change is safest.
I hope that this new entity can do something good with this data, and use their vision to build something really great!
Very common, in any human community, but pure poison, in a technical one.
Insistence on structure, is often derided as “fear of change,” but it’s a very different animal.
Sorry to hear about your experience, but I have dealt with similar behavior. In my case, I’m actually a fairly cantankerous, stubborn bastard, and bulled my way through.
OSM might also feel more pressure to standardize things like tags for features, which would resolve a considerable problem in using OSM data. Both tag names and value formats vary from region to region, and that's probably why almost any proprietary maps software is more useful than OSM apps. Having personally had to use OSM data, writing heuristics to parse tag names and values is a lot of work.
As for criticism of Overture on OSM community forums, there seems to be a lot of vague gesturing at "big tech bad, OSM community good." A more constructive place to look for what Overture means for OSM is the OSM Foundation's blog/press statement - https://blog.openstreetmap.org/2022/12/22/views-from-the-ope.... Which, to my mind, reads as saying that OSM data is free to use for corporations, but OSM hopes that Overture improvements flow upstream.
If my tone seems a bit negative, it's because I'm a bit tired of the negative reaction open source communities tend to have when being challenged or offered a collaboration opportunity with big tech. I feel like the approach Blender takes with big tech (like NVIDIA) is much, much more productive.
That's a pretty bold statement. There is certainly lots of variation in usage, but commonly used tags are used pretty consistently, the variation is largely in things that "almost any proprietary maps software" doesn't even bother with.
Let's say I want my maps application to know whether a building is commercial or residential. Some buildings have building=residential, others have building=yes or building=oui, and others just have landuse=commercial (or landuse=residential, suburban, quarter, suburb, neighbourhood, or residential=yes, or residential=apartments). This is not the worst-case scenario; we can write an interpreting adapter for these tags. But somewhere in the world, there will be edge-case tags the adapter won't handle. And it is challenging to use that approach for applications that tens of thousands of users will see and scrutinize.
The solution is straightforward for OSM - standardize the tags. It's been a long debate in OSM, but I hope the commercial influence will make them do it. There are practically no downsides and significant benefits.
There are 435,057,783 building=yes, and 1 building=oui.
The OSM community has standardized.
https://taginfo.openstreetmap.org/tags/building=yes https://taginfo.openstreetmap.org/tags/building=oui
Maps is better because it has all the shops and opening times… which might be out of date by a year or two.
Sure map users now need to do cleaning and quality control, but this is always application-specific and cannot be forced to the mappers. I feel OSM balance between structure and freedom is good enough.
Now some map applications are quite standard: eg nice base map and navigation, so for these quite many companies have built own data cleaning/structuring/rendering stacks, where only some pieces are easily available as OS, rest is kept as business secrets. I hope Overture will provide usable OS full stack here. Certainly this will be also specific to the contributors own interests, and it does not try to “fix” OSM into well structured database.
One way to gauge the impact might be to keep any eye on what Meta does with its Daylight Map Distribution project (a clean OSM), https://daylightmap.org/.
> “Overture is a data-centric map project, not a community of individual map editors. Therefore, Overture is intended to be complementary to OSM. We combine OSM with other sources to produce new open map data sets. Overture data will be available for use by the OpenStreetMap community under compatible open data licenses. Overture members are encouraged to contribute to OSM directly.“
Apple is also a heavy contributor to OSM: https://wiki.openstreetmap.org/wiki/Organised_Editing/Activi...
OSM maps are superior to the new maps in some places like Australia, although maybe in irrelevant ways - there’s not really value in having every single building drawn on the map.
I don't have any special insight, but I expect that Overture is moving that cooperation outside of any of the players.
Overture's sponsors seem pretty focused around big company needs, like coordinating efforts between mapping orgs and standardizing metadata. They're each pledging $3M + 20 FTEs per year to accomplish it [1].
But I would argue it's more productive to coordinate enterprise+individual contributions through a central organization: the OSM foundation.
When io.js forked from Node.js in 2014, it was more of a call for attention than a desire to split the ecosystem in two. They merged the next year. (I joined the core dev team after the split.)
Maybe instead of asking what OSM will become after Overture, we should ask, why can't enterprises accomplish their goals within the framework of the OSM Foundation?
Disclaimer: my startup has a product built on OSM [2].
[1] https://overturemaps.org/become-a-member/
[2] https://buntinglabs.com/solutions/openstreetmap-extracts
Isn't it a perfect business model for corporations?
1a) Launch an open source aka crowdsourced project at a relatively small cost, or 1b) Just wait until it comes out of the woodwork on its own.
2) Watch the data of interest flowing in, for free, in a self-organizing way.
3) Wait until it reaches the stage when bringing it up to a production-ready state takes again a relatively small investment, then take it over.
4) Now offset your small investments orders-of-magnitude-fold compared to what you would without crowdsourcing - by selling the data which people hand-picked and gathered for years, by training your AI on their code, by debugging a prototype of future blockchain money at the expense of enthusiasts and profiteers etc.
Would this work? Hell no, except maybe on Flat Earth! You forgot
0) To set the stage by preaching how noble and virtuous it is to give away the fruits of one's work away for free for the greater good, providing for a better future for all of us on the principles of freedom, predominantly as in beer, all over the Internet, especially in places where young and idealistic graze on the fields of IT.
Overture Maps Foundation - https://news.ycombinator.com/item?id=33999873 - Dec 2022 (28 comments)
Meta, Microsoft, AWS and TomTom launch Overture Foundation for open map data - https://news.ycombinator.com/item?id=33998438 - Dec 2022 (56 comments)
Linux Foundation Announces Overture Maps to Build Interoperable Open Map Data - https://news.ycombinator.com/item?id=33997856 - Dec 2022 (2 comments)
I would imagine a map equivalent of an RSS feed would be nice, although companies like Starbucks and McDonald’s files would probably need to be compressed due to the sheer number of locations.
https://www.alltheplaces.xyz/ scrapes location maps and digests them into consistent line delimited GeoJSON.
The rest is supported by Google Maps and Apple Maps, where businesses can submit their location, hours, etc, and do that ASA matter of course.
The predominant pattern for a customer is to use a (really the) search engine, feed it the name of the business, and see the details in a snippet, or jump to a (really one if two) maps app.
I don't see what would be the upside for Google and Apple to interpret such files, instead of forcing business owners register with their sites, prove identity, contact info, etc, and keep them within their silo along the way :-(
I don’t think Google or Apple are capable of locking in another brand like Home Depot like that though.
In the near future I predict an AI like GPT 3 could interpret unstructured screenshots of websites to answer queries like "what hours are they open"
Though Google canned Duplex just a couple weeks ago: https://www.engadget.com/google-duplex-on-the-web-shutdown-a...
[0] I'm aware that "just" having everyone converge on even a handful of standard formats is, in the real world, a high bar. I do think we could do it - yes, probably by Google applying pressure - but adoption remains Hard.
Big tech created a new foundation because this is the way they can exercise control.
This new foundation will work for big tech interest, but with a facade that they are doing what they are doing for the public interest.
Have you ever been oppressed by an online map?
I’m excited about a group of people improving that process and the data storage format. I’m not excited about it being these people.
I had been seeking this confirmation. Both can, and will be using one another's data.
And Parent indicates Overture like an administrative fork of OSM.
I'm all for standardisation and collaboration improving the foundations of mapping data, but I reserve judgement about the scale of impact until I see more concrete detail.
My comment in another Overture thread: https://news.ycombinator.com/item?id=34035938
- things that work sooner or later will prevail, BUT only when some have found the way to overturn their functioning from "good for all" to "good for some, mostly against all the rest";
- the "free market" model of classic liberalism was built because those who have built it was LESS powerful than governments, then they successfully penetrate government, reach a certain level of power than their propaganda change back to the classic pro-governance propaganda, in software terms this is simple: before many pushed their proprietary model BUT when some succeed really others who do not can't destroy the new monopoly, so they turn to a mocked freedom move.
Google Maps is simply a proprietary de-facto monopoly, the only way to crush it is being better and the only way to being better is being open. The very thing they deny until the day before, then they falsely embrace to create their walled gardens, now the embrace more crushing the walls since they can't survive internally. Unfortunately people do not realize that.