Mapping Hacks

This was an insight offered by Alex Robinson and others at the Open Geodata Forum; why do we, as enthusiasts, citizens and consumers need the very highly accurate data which the Ordnance Survey prides itself on providing?

Working with GPS traces, Steve Coast’s maps at openstreetmap are only accurate to within 5 metres. But that’s perfectly good enough for route planning, or street level mapping; though not for digging electricity cables or deciding on the impact of bay windows on planning permission.

90% of our use cases don’t need that kind of very accurate data. Medway council was asked to remove the images of electoral boundaries from its website, because they showed some polling positions in erroneous places. But for electoral analysis mapping of the kind indymapper provided, of which we saw so much in early November 2004, highly generalised outlines are adequate. Generalised files are an acceptable part of state level geodata provision in the US, and plenty of algorithms for performing the generalisation are in the public domain (well, some of them have patents crossing them, but that’s another story.)

From generalised ward boundaries and streets we could map the politics of London. One barrier we keep coming up against, is our ability to geocode addresses or postal codes a-la geocoder.us. Postal codes, which cover a block or two, perhaps 14 houses, provide a good-enough generalisation. Yet the ownership of this data set is obscured - some way between the Ordnance Survey, and the Post Office or Royal Mail as it is now known. Address-based geocoding for the UK is a big job! and one it may take a while to build into openstreetmap.

The release of low-density version of street, addressing and administrative boundary data would accrue most of the public service benefits in projects like those that MySociety runs. It would enable us to build the kind of collaborative annotation, space archiving and maintenance which we’re having to build services like openstreetmap to be able to create.

How significantly would this impact on the Ordnance Survey’s remit to operate a cost recovery policy in order to maintain as much accuracy and currency of data as possible? Possibly, not a great deal; Local councils and local transport planners and operators would still need the ‘bay window’ level of mapping detail, as will the phone and water companies with their drills. A significant proportion of the entrenched market would be unaffected by a free licensing of generalised OS data for non-profit use. The applications developed in a research and startup context would drive data sales as well as economic activity in general.

Posted in policy |

You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.