Phil Gyford


Saturday 21 August 2004

PreviousIndexNext Euro Foo Camp: Steve Coast - OpenTextBook & OpenStreetMap

Steve Coast (physics student at UCL) on OpenTextBook and OpenStreetMap both of which could do with more collaborators. The latter is especially interesting to me, and it’d be fun to be involved (time allowing).

OpenTextBook is/will be a Creative Commons licensed text book created using open source methods. Text books are expensive and it’s cheaper to import them from the far east. They’re buggy and have a long life cycle which makes it hard to correct bugs. In maths and physics, much of the material hasn’t changed in a long time, and it’s more about the interpretation of these. Mentions Wikipedia and Planet Math [error for me at the moment].

When slashdotted he got lots of emails asking why it wasn’t a wiki. But Wikis are good for small granularity tasks - small entries with quick editing. But text books, especially maths and physics require more time and concentration [I’m not sure this is specific to these fields!]. For these things you need a better editor than a wiki-esque text box, so he uses LaTeX. But wikis are good for allowing multiple people to contribute, and LaTeX creates a barrier to entry. But for a project like this the people likely to contribute probably know LaTeX anyway. (Wikipedia apparently uses LaTeX in the background to output equations etc.)

Where to get material from? Lecture notes? Maybe needs clearance from lecturers. Out of copyright material, which is sometimes OK as material (in maths for exmaple) hasn’t changed for 50-100 years. But with Disney, DMCA etc, copyright terms might always extend. Or you could create new material, which is hard to come across - talk to lecturers again.

OpenStreetMap. The Ordnance Survey map data (in the UK) is not freely available. Can we replace et al with something free as in speech? Current maps have good quality, aren’t free and aren’t editable, holds geolocation stuff back.

Shows the WAAG map of Amsterdam created by people with GPS units wandering around. He got a USB GPS receiver from eBay, an old laptop with large battery life, and put them in a backpack. GPS units like Garmin have their own standards about what data to record and when. An iPAQ would be a good way to do it, talking to the laptop. He’s cycled around an area of London, writing down the streetnames.

His database does OpenGIS SQL (eg MySQL >= 4.1 or Postgres), he has trace software and editing software. Has thought about having multiple users creating and editing data. Display software is hard because people/countries always have different ideas about how to display and label things.

GPS is only accurate to 10 metres - very occasionally they go way out. But streets are about 10 metres across, so maybe this is a reasonable level of error. The end user won’t have a better standard of accuracy, so this may be OK. Altitude data is much more noisy (+/- 200m) than lat/long.

Shows a map and an Expedia map (of Regents Park) that vary differently in the amount and kind of data. Says various parts of the park have been closed for 18 months and isn’t represented, and perhaps an open map could be more temporally accurate.

Why hasn’t it been done before? Hard to start as it requires lots of data becore it’s at all useful to anyeone. Takes a lot of time and effort.

Stef: Maybe courier companies or taxi companies could provide GPS data? Someone else: bicycle courier companies?

Schuyler: Want to demystify cartography as a science. Easy in the US with so much copyright-free data. But in Europe it costs. Maybe iamge processing satellite data? Steve: but this would be a derivative work of the satellite images, so the original photos would need to be copyright-free.



Schuyler: Differential GPS over IP. Trying to cancel error in GPS data, often caused by ionospheric delay [boy, are we living in the future or what?]. To counteract that you can put a device in a known location so you can tell how much the distortion is and use this to correct your mobile GPS unit. You need a DGPS server within a certain range, so we’d have to set one up in the UK. [Just found more here.]

Commenting is disabled on posts once they’re 30 days old.