« Manifesto comparisons | Main | Depression - A Community Resource »

Open, annotated mapping

FROM: Chris Lightfoot

What NEED does this meet?
Many of the other proposals here rely on the availability of high-quality mapping data and supporting applications. Unfortunately collecting this data in a distributed, "open source" fashion would be an enormous undertaking and a duplication of effort; it's not clear how good the results would be, either. Licensing it is expensive, especially for a small project. Of the twenty most recent proposals I can see now, something like a quarter have some GIS component.

My idea is to set up a mapping site (like multimap or streetmap) with the same high-quality mapping data, and an open interface to allow other users to overlay their own data (abandoned cars, volunteers, locations of meetings, houses for sale, whatever) on the site. Obviously this needs funding to license the Ordnance Survey's data -- and their willingness to grant a license on reasonable terms -- but the actual implementation of the site should be pretty easy.

What is the APPROACH?
Either set up a new site to implement a simple API for adding/removing/searching very simple map overlays onto licensed mapping data, or negotiate with an existing provider to add this service. The license would have to allow the uses to which we're putting it, and preferably some degree of republication on other projects' sites.

What are the BENEFITS to people?
It would enable many other useful projects which depend on this sort of simple GIS data, avoid duplication of effort and make possible projects for which the expense of licensing mapping data would be prohibitive.

What is the COMPETITION?
Multimap, Streetmap, etc. offer the mapping data (through a crummy user interface festooned with advertisements...), but no way to exploit the data for third-party projects. Ordnance Survey have the "Get A Map" service which will license small amounts of mapping data for use on other web pages, but the conditions of use are restrictive and you can only re-use data covering small areas. Professional GIS packages like Arc/INFO are expensive and not appropriate for small volunteer projects.

What BUDGET & LOGISTICS are required?
The most significant ongoing costs are licensing the map data and running the physical service -- in particular, serving map images is likely to be bandwidth-intensive. (These costs could be offset by geographically-directed advertising, and perhaps by charging a subscription to the other projects putting overlays on the map, though obviously the point is to make this charge as small possible: preferably zero for most users.) Implementation of the database and overlay API is pretty easy, probably a week's work.

November 3, 2003 in Geographic, User Created Content | Permalink


I like this idea, its usefull, technically feasible and could be of great help when "planning" routes for people. Perhaps even some kind of URL guided system where the URL contains all the element and where on the map they are to be drawn.

Coupled with a good interface such as Java or Flash it could really benfit the masses.

Posted by: Alex Forbes at Nov 3, 2003 4:55:00 PM

The technology is the simpler part of this one. But as you say , for simple overlays, a simple URL encoding would be a nice approach. More complex stuff (which would be too long to pass in the URL) would need more cooperation on the server (e.g. pass a URL which the server queries to find out about overlay information).
I don't think Java or Flash would add much, and it would be preferable to use open standards. Really we would be constrained by the mapping formats available, which are raster and some random vector format, but the raster maps are the ones we're all used to and almost certainly preferable.

Posted by: Chris Lightfoot at Nov 3, 2003 5:56:22 PM

Have you raised this with the OS? I've a suspicion that if you want to make the site really useful - and in particular if mapping data are made reusable - that'll be reflected in the price and (possibly) restricted terms of use.

I've proposed things somewhat like this in the past, and I have existing technology that could be re-used in this if it goes ahead.

Posted by: Nick Kew at Nov 4, 2003 6:50:44 AM

Not yet -- it's something I only thought of while reading the other proposals. From their price list, http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html,">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html,">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html,">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html,">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html,">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html,">http://www.ordnancesurvey.co.uk/products/businessgeographics/pricing.html
it looks as if the base cost for 1:50,000 raster mapping for the whole UK is £5,000 + £1,750/year afterwards, plus another 10% for the gazeteer. This is for "one terminal". Pricing for larger numbers of "terminals" introduces a cost multiplier, ranging from ×1.5 for two "terminals" to ×8 for over 500 "terminals". (Smaller scales are much cheaper.)

Assuming that the web lies somewhere in this range -- i.e., there's no separate pricing structure -- we're looking at up to £44,000 to start up and £15,400/year thereafter including the gazeteer data. In any case I think this is probably indicative of the kind of cost range.

I'm not sure if it's worth contacting them to negotiate a licensing deal if there's no possibility of the other resources to run the site being available. But an informal query might be worth doing....

Posted by: Chris Lightfoot at Nov 4, 2003 9:41:12 AM

One isssue is patents. Multimap own the US and EU patents for displaying points of information on a digital map delivered over the web. Aside from this, the implementation could be done quite easily using the Open Source UMN Map Server software and PHP.

Posted by: andrew at Nov 4, 2003 12:03:07 PM


The European patent EP0845124B, entitled "Computer System for Identifying Local Resources and Method Therefor," relates to Multimap.com's technology for displaying both an image of a map and information data relating to at least one place of interest on the map to an end user

Posted by: andrew at Nov 4, 2003 12:14:27 PM

Hmm. That's tedious. It's unlikely that the patent's valid (there's lots of prior art -- see e.g. http://www.geoplace.com/gw/2003/0306/0306l_col.asp
and http://mapserver.gis.umn.edu/data2/wilma/mapserver-users/0302/msg00367.html -- and the status of software patents is murky in Europe anyway) but it would still cost a great bunch of money to overturn, and therefore it would probably be cheaper just to licence the "technology" from Multimap. Dull dull dull.

(Independently, I note that Streetmap.co.uk (a) now has only two scales of maps; (b) certainly would infringe the Multimap patent, though doesn't say anything about it....)

Of course, it looks like one of the services Multimap resells would be good enough for our purposes, something like http://www.multimap.com/static/business3.htm

No actual price list on the Multimap site, though.

Posted by: Chris Lightfoot at Nov 4, 2003 4:51:11 PM

We're building something akin what you're suggesting, though albeit for a different aim and in partnership with Ordnance Survey (among others). We'll be demonstrating an advanced prototype in London in December that allows participants to annotate information from wireless/mobile devices directly to geo-locations:
Current R&D license terms with OS prohibit us from making the client available to the general public, but we're working with them to open this up in the future. As a small non-profit org ourselves we're well aware of how difficult it is negotiating with large commercial partners when creating socially and culturally useful technologies that don't fit a monetary economic model. But there are ways to work with them and influence access to open systems - Urban Tapestries itself aims to create a model for cross-discipline research with non-profit and commercial collaboration leading to open systems.

Posted by: Giles Lane at Nov 5, 2003 5:19:44 PM

I disagree entirely. RElying on OS data and paying for it is a terrible idea. A system for building an open source map would be much preferable - allowing people to map their own area, so that eventually the maps can meet up. Also some process of version control for improving data, and for allowing different view of the data.

I can't believe you suggested paying £44,000 to OS for data copyright - do you understand the basic ideas of open-source???

Posted by: Mike Tonks at Nov 6, 2003 12:22:32 PM

A contributor map of the type you suggest is unlikely to be of the required accuracy. It would also take a very long time to compile and would duplicate effort already made by the Ordnance Survey. Do you really think such a map could be produced for substantially less than £44,000 per year (counting the contributors' time and everything else)? Note that the Ordnance Survey spends about £80 MILLION per year on its activities, chiefly mapping.

(Now, ideally the Ordnance Survey would release its data to the public. But I don't think that's likely in the near future.)

As for understanding "the basic ideas of open-source", I suggest you try to patronise somebody else.

Posted by: Chris Lightfoot at Nov 6, 2003 2:30:54 PM

To put the various open mapping proposals into a broader perspective, I would like to invite you to read over http://www.webmapper.net/carto2003/. This describes some of the backgrounds, trends and issues related to open source mapping.

Posted by: webmapper at Nov 10, 2003 9:48:58 PM

A fascinating thread which I'd be interested in discussing face to face. Hope to hear from you.

Posted by: Dave Lovell at Dec 2, 2003 8:10:33 AM

We've done something similar at http://www.geowiki.com/ - using maps generated from GPS tracks (major update currently in the offing).

Posted by: Richard Fairhurst at Sep 14, 2004 10:33:02 AM