Deutsch Français English Italiano
GEOWebforum Header
www.sogi.ch www.swisstopo.ch www.sik-gis.ch www.kkgeo.ch Logos Founders
Anmeldestatus: nicht angemeldet


Diskussion «OpenStreetMap: Thread or Complement of Google and Official Base Maps?»
Artikel 1-6 von 6



Jean-Luc Miserez
31. Januar 12 (08:46 Uhr)
Beitragsnummer: 2456
Adding my 2 cents on this interesting discussion, I'd like to point to a thesis comparing the positional Accuracy of OSM data in relation to British OS Data:
[discovery.ucl.ac.uk/1318053/1/1318053.pdf](mostly on pages 172-179, but the rest seems also interesting).

Ralph Straumann
24. Januar 12 (08:44 Uhr)
Beitragsnummer: 2450
[English below]
Diese Diskussion (danke an die Initiatoren) hat uns veranlasst, noch etwas mehr über unsere Erfahrungen mit Crowdsourcing und Daten aus Crowdsourcing nachzudenken. Auf unserem Blog haben wir einen Artikel verfasst, der die Chancen und Risiken von Crowdsourcing für die öffentliche Verwaltung skizziert: [geo.ebp.ch/2012/01/24/chancen-und-risiken-von-crowdsourcing-fur-die-offentliche-verwaltung]

This thread (thanks to the initiators) has made us think a bit more about the crowdsourcing phenomenon. Reflecting upon our experiences with crowdsourcing and crowdsourced geodata we have tried to sketch the opportunities and risks of crowdsourcing from a public administration perspective (in German): [geo.ebp.ch/2012/01/24/chancen-und-risiken-von-crowdsourcing-fur-die-offentliche-verwaltung]

Ralph Straumann
3. Januar 12 (16:06 Uhr)
Beitragsnummer: 2439
It's indeed interesting to see how Google's announcement to charge heavy users of their maps has caused some stir in the Geoweb community – despite the fact that most website owners won't even have to pay for using the Maps API under the current pricing scheme (see [bit.ly/viTRHY] [German]).

But the question at hand is: Where does OpenStreetMap (OSM) stand compared to other "map vendors" (i.e. Google Maps, TomTom, Garmin, national mapping agencies like Swisstopo, Ordnance Survey or IGN, etc.)? What are the goals of OSM compared to other the actors’ ones? Does OSM challenge Google Maps and official base maps or does it not?

Cédric Moullet [abbreviated as (C)] has raised some interesting points ([bit.ly/vgqcY1]) and later added some clarifications ([bit.ly/vJtcoB]). Stefan Keller [abbreviated as (S)] commented on, and objected to, many of Cédric Moullet's points (see below post).

Here are our two cents [Ralph Straumann and Stephan Heuel from Ernst Basler + Partner ([geo.ebp.ch], [www.ebp.ch]), abbreviated as R and SH, respectively]:

The original question was "Why OpenStreetMap fails to replace official or proprietary base maps in a sustainable way". The assumption is that OSM exists to replace base maps. But in our opinion, OSM is simply a different beast with partly overlapping and partly differing applications than the others. In his post, (C) quite certainly didn’t want to express that OSM failed as a project at large – on the contrary: it’s a big success story and the discussion shows that it has potential to be as disruptive as Google Maps was to the GIS industry in 2005.

Our point of view: Despite several caveats we think OSM is a great resource for many applications. However, OSM is not a full replacement for official base map data. The detailed discussion follows below. We have used OSM data in various projects and are convinced that there can be a fruitful exchange between crowdsourcing initiatives like OSM, Geonames and others on one hand and official authorities on the other hand. In fact, we would argue that OSM and others are part of our national spatial data infrastructure NGDI. We have submitted two articles to the GeoSummit conference ([www.geosummit.ch]) on this topic and hopefully – fingers crossed – can give a talk about it this spring in Bern.

In any case and regardless of how valid the original question is: it is a good starting point to compare OSM with official or proprietary base map data providers. So we continue with the individual items raised by (C) and (S):

Why OpenStreetMap fails to replace official or proprietary base maps in a sustainable way

(C) Because the finances are not secured on long term.
(S) No; but are they elsewhere? As far as I know it seems that the near and midterm future of OSM is rather well established.

R+SH: Two things to note: Google's revenue streams, business outlook and finances look very solid, more solid than those of OSM. As for official base maps: In the current economic crisis the financial situation of countries bears different shades of grey. However, for Switzerland, the national mapping agency Swisstopo has a sound legal basis (in the form of constitution articles, laws and ordinances) and a periodically renewed mandate by Swiss authorities (see [bit.ly/tgtTjq] [German]). Thus, even in times of financial austerity it would be everything but easy to economise by shutting down national services.

------------------------------------------------

(C) Because the data model is not defined.
(S) This is a well debated issue. It's Key-Value-based and this has quite some advantages.

R+SH: Key-value-based storage models have their merits, but for more sophisticated applications a complex and explicit data model is unfortunately necessary (just think of the discussion on minimal data models in Switzerland). A debate with regards to different data storage philosophies is a bit like the “wars” between proponents of NoSQL and those of relational databases, Android and iOS or Linux, Mac OS and Windows.
In our opinion, both alternatives – key-value-based storage and explicit data model – are valid and have their advantages and disadvantages, depending on the application, see also the following point.

------------------------------------------------

(C) Because the data are difficult to extract.
(S) Why this? You can get it in several ways beginning with XML extracts right from the map view up to whole planet!

R+SH: The extraction of data is mainly a technical issue and there exist more than one solution for your favorite GI software package, see [bit.ly/vENgeB] (and don’t forget FME and gdal/ogr). Services like Geofabrik.de let you download current shapefiles of OSM data. For GIS professionals, it may not seem straightforward to use the native OSM data (and the key-value pairs), but for experts outside the GIS domain, the OSM data format and model may seem better suited than any detailed GIS data model in which the data is not available.
Granted however, the OSM data model, the OSM data structure, the access points for obtaining data as well as the means to use OSM data in everybody's off-the-shelf (commercial or open-source) software are not straightforward for non-expert, occasional users of geodata.

------------------------------------------------

(C) Because the precision is heterogeneous.
(S) From what I know, precision (around 1m to 5m) is good enough for most purposes unless you need official map information like in civil engineering.

R+SH: We assume that we are in fact talking about accuracy here, not precision. We agree that accuracy of OSM is good enough for a wide set of applications. However, it's indeed true that OSM accuracy is spatially heterogenous and this can pose problems to some applications. For UK, Haklay (2010) found strong variations in accuracy between different regions. The situation in France is similar (Touya and Girres 2009). Also, attribute accuracy in France was found to be not very good.

------------------------------------------------

(C) Because the completeness is heterogeneous.
(S) Well: So you agree that completeness is sometimes better than Google and Official Base Maps? OSM contains many map features which don't exist in official maps (nor in Google). And OSM will typically be much more up-to-date than official maps and data. OSM is updated daily whereas official maps have a delay of months. Anyway: Why not free official base maps, cooperate and integrate OSM?

R+SH: Clearly there are examples where OSM is more complete than Google Maps (and updated more quickly than Google Maps, see for example the mapping efforts after the Haiti earthquake). In some regions of the world, OSM is also more complete than available official base maps. Furthermore, OSM maps certain types of features which the official base maps won’t contain.
However, OSM completeness varies by country and also region. For France it was found that OSM contained only 10% of the objects in the official (very detailled) BD TOPO dataset (Touya and Girres 2009). For UK, Haklay (2010) found that 30% of the area is sufficiently well covered by OSM. 45% of Britons live in that area, which hints that densely populated areas are mapped in more detail than sparsely populated ones (for good or bad). Furthermore, it's worrisome that OSM in UK also maps economically well-off regions better than poor regions (Haklay and Ellul 2010). Conclusion: OSM completeness is spatially heterogenous. An interesting question is, if (and possibly when) the completeness can be improved by OSM’s crowdsourcing model. However, at the moment this question remains open.

------------------------------------------------

(C) Because it requires attribution.
(S) I wish, OpenStreetMap would be Public Domain but I can live (and work and deliver cool apps) with the new license model (ODbL).

R+SH: Attribution may be a problem, but isn’t it only fair to the users and contributors and transparent to the users to, in fact, merely require attribution? For official or proprietary base maps there is always attribution and often more complex terms of use than with OSM. We think it's great that a resource like OSM can be used freely in exchange for mere attribution. Also, with the adoption of ODbL OSM has managed to do away with uncertainties and ambiguities around Creative Commons (CC) licences for data.

------------------------------------------------

(C) Because no one takes the responsibility about the data.
(S) Agreed: Taking responsibility is one of the main raisons-d'être of official maps, of OEREB cadastre and of patented surveyors. (Sidenote: you know well, that they don't really do if you read the small print in contracts - And neither does Google!)
(C) Because it lacks a QA step by an accountable body.
(S) That's an important issue. It's true that there's no accountable body. But there are processes and tools which establish a quite working Quality Assurance (see [OSM Wiki]). This was reported in several recent scientific publications.

R+SH: We agree, if you seek accountability on part of the map (data) provider, OSM and similar initiatives are clearly not where you should start your search. While there is no accountability, OSM has put forward a lot of quality assurance measures and tools like for example OpenStreetBugs or NoName Map, as Stefan Keller (S) mentions. It remains to be seen how effective these will be, see also our remarks on completeness.

------------------------------------------------

(C) Because first acquisition is fun and data update is boring.
(S) Yeah; can't deny this from my observation. But does not apply for all mappers - and there are around 500'000 Mappers around the world (~5'000 in Switzerland)...

R+SH: This is a somewhat open point. We are not clear about how many users engage in updating activity, despite mapping terra incognita being more exciting. "So far it seems as if processing within the OSM project is closely related to visual factors, meaning that most data is collected in areas where there are white spots on the map, and thus no information is available." as Neis et al. put it ([www.mdpi.com/1999-5903/4/1/1]). Missing attribute data is not as readily visible as missing geometric data!

------------------------------------------------

(C) Because Google Map Maker Workflow is for the broad public and OpenStreetMap workflow for the map enthusiasts.
(S) Can't comment on Google Map Maker: Does anybody use this, while taking responsibility for the data and giving all rights to Google?

R+SH: (R) has tested Google Map Maker (available in 180+ countries) months back. From his (brief) impression Google Map Maker does have an easier workflow than OSM. Google Map Maker has one single point of entry and an easy, clean, (and due to its resemblance to 'normal' Google Maps) for most users familiar GUI. Google Map Maker also has a reviewer system, where user-edits are checked by a senior/expert user. OSM, again due to its crowdsourced and decentralised nature, has various ways one can get into the project and contribute and for example does not have just one editor for editing map data.


PS: It would be great to have a better and more interactive way to discuss these issues. It is hard to track down the individual inputs. What about a face-to-face discussion in Bern? Or more immediately: would Google+ be a (alas, proprietary ;) ) way to discuss this?

Marc Wick
2. Januar 12 (14:13 Uhr)
Beitragsnummer: 2438
I think we should differentiate between maps and data.

While OSM maybe an interesting mapping project it is pretty useless as a data source due to its un-open share-alike license. If all you want to do is drawing maps then you can probably live with this restriction but as soon as you want to do more interesting stuff the license will be a show stopper.

At the other hand national mapping agencies traditionally had a focus on maps, but they now have to change focus, move away from maps and evolve into national geodata agencies. Maps have become a commodity product and there are many different providers of map products and services. It does not make sense that governmental organisations compete in a market where there is already enough other competition. The official national geodata agency should focus instead on providing the underlying base data without restriction and free of cost.

Examples:
- The very moment a new road is opened to traffic this information should be available on a data feed provided by a national agency.
- Traffic jams, special road conditions (caused by weather, construction, accident, etc) are included in a special feed.
- why not offer a dataset with public transport information and a real time feed with changes to schedule?
- newest population numbers are included as attribute for populated places, links to other bfs attributes (this will require a national place id, just like the bfs gemeindenummer but for all feature types)
- hiking tracks etc
- possibly a feedback mechanism to allow for contributions. This is also the only reason that could possibly warrant an online map on the national geodata agency website.

The transition to a national geodata agency will not be easy and requires cooperating with a lot of different other governmental agencies on several levels.

Marc
GeoNames.org

Stefan Keller
30. Dezember 11 (17:18 Uhr)
Beitragsnummer: 2437
Cédric asked me to spend more thoughts about the issues he mentioned. I'll try to give some answers coming from my experiences with the OpenStreetMap scene and running several web applications based on OpenStreetMap data and services:

These are the items from the [blog] on "Why OpenStreetMap fails to replace official or proprietary base maps in a sustainable way":

1. Because it’s not possible to make a map for all zoom levels
=> Why not? To me level 18 (20m) from OSM seems much clearer than the similar scale (zoom 12) at [map.geo.admin.ch/]

2. Because the finances are not secured on long term
=> No; but are they elsewhere? As far as I know it seems that the near and midterm future of OSM is rather well established.

3. Because the data model is not defined
=> This is a well debated issue. It's Key-Value-based and this has quite some advandages.

4. Because the precision is heterogeneous
=> From what I know, precision (around 1m to 5m) is good enough for most purposes unless you need official map information like in civil engineering.

5. Because the reliability is heterogeneous
=> What reliability?

6. Because the completeness is heterogeneous
=> Well: So you agree that completeness is sometimes better than Google and Official Base Maps?
=> OSM contains many map features which don't exist in official maps (nor in Google).
=> And OSM will typically be much more up-to-date than official maps and data. OSM is updated daily whereas Official maps have a delay of months.
=> Anyway: Why not free Official Base Maps, cooperate and integrate OSM?

7. Because it requires attribution
=> I wish, OpenStreetMap would be Public Domain but I can live (and work and deliver cool apps) with the new license model (ODbL).

8. Because the data are difficult to extract
=> Why this? You can get it in several ways beginning with XML extracts right from the map view up to whole planet!

9. Because no one takes the responsibility about the data
=> Agreed: Taking responsibility is one of the main raisons-d'être of official maps, of OEREB cadastre and of patented surveyors. (Sidenote: you know well, that they don't really do if you read the small print in contracts - And neither does Google!)

10. Because it lacks a QA step by an accountable body
=> That's an imporant issue. It's true that there's no accountable body. But there are processes and tools which establish a quite working Quality Assurance (see [OSM Wiki]). This was reported in several recent scientific publications.

11. Because it is not multilingual
=> What do you mean? To me it's multilingual in data the and user interface.

12. Because first acquisition is fun and data update is boring
=> Yeah; can't deny this from my observation. But does not apply for all mappers - and there are around 500'000 Mappers around the world (~5'000 in Switzerland)...

13. Because Google Map Maker Workflow is for the broad public and OpenStreetMap workflow for the map enthusiasts
=> Can't comment on Google Map Maker: Does anybody use this, while taking responsibility for the data and giving all rights to Google?

Stefan

Stefan Keller
30. Dezember 11 (15:34 Uhr)
Beitragsnummer: 2436
When Google recently announced that it will begin to charge for its maps, obviously some map websites moved to OpenStreetMap (OSM) services (like its Web Map Tiling Server, "Mapnik styled"). Then lively discussions came up if Google underestimated the power of the biggest crowdsourced mapping project (see ["'Wikipedia of Maps' Challenges Google"], 26 Dec 2011).

On the other hand, in a recent blog Cédric stated ["Why OpenStreetMap fails to replace official or proprietary base maps in a sustainable way?"] (27 Dec 2011). So, does OpenStreetMap challenge also official base maps?

My recommendation to many discussions that will occur is: Let's try to understand first what OpenStreetMap is! And what not!!

To me it seems that OpenStreetMap - and other open data projects like [Geonames.org] - currently are often misunderstood, underestimated or completely ignored. I think it's important to keep in mind the different goals and 'products' of the three projects, OpenStreetMap, Google Maps and Governemental Base Maps. For a comparison between OpenStreetMap and administrative maps and data see this presentation from Frederik Ramm entitled ["OpenStreetMap: Übersicht – Chancen – Grenzen" (PDF)] (January 2011).

Of course, there will never be an 'officially agreed' definition on "what OpenStreetMap is" since it's a crowdsourced, 'democratic' project. To me (as a researcher and GIS coach) it's mainly a freely accessible, accurate enough, very up-to-date and astonishingly complete, massive and raw geodatabase! I think it complements the official base map data - given the latter finally become accessible too. Some OSM services like from MapQuest complement also Google Maps (= services). But OSM is and remains to be mainly a database - and not a ready-made map service.

Yours,
Stefan

P.S. (de): Antworten sind auch auf deutsch willkommen.
P.S. (fr): Vous pouvez également répondre en français.



  1