GPS and Cryptonomicon

April 28, 2005 7:53 PM

In the book _Cryptonomicon_, the author's generated emails from Randy Waterhouse, a main character of the book, contain GPS information which refers to a "Nearest Geographical Feature". This mythical NGF seems to work worldwide. How would you practically implement such a feature using today's technology - are there any free datasets that would do the job? Having just purchased a GPS receiver I am very interested in locating data sets that can accomplish a similar set of functionality.
posted by arimathea to technology (5 comments total)

Implementation would be easy -- just a database containing the coordinates of prominent geographical features. Compare the user's coordinates to the database and return the nearest feature. I don't know of any GPS apps like that, though.
posted by forrest at 8:08 PM on April 28, 2005

Thumbing through my copy...

(And in keeping with the book, I may get a little geeky here...)

I see only two instances in which his email signature contains this. (I would have guessed more.) The first several emails shown are exchanges between [e-mail] and Randy using [e-mail] (to be somewhat anonymous, and thus is not tagged with his name and location). There are also a couple from Cantrell regarding the Pontifex cryptosystem, but none (quoted) from Randy to him.

The first gives the NGF as "Palawan, the Philippines". I am going to guess-sume that this may be one of the locations that Randy physically mapped when doing his GPS work to determine lines of sight for the microwave towers.

The second gives the NGF as "the Bonin Islands". Given the location of the Bonin Islands and the fact that at this time, Randy is on a plane for LA, I don't think he mapped this one.

This gives us an awfully small dataset to work with. However, if we look at these locations:

Palawan is relatively large; Randy's GPS has a degree of accuracy much higher than having to just say "I'm somewhere on or near Palawan". On the other hand, a database of major geographical locations intended to be tied to GPS data, if it covers the whole world, is not likely to have a resolution fine enough as to list Palawan by itself rather than the Philippines as a whole. If that were the case, the database would have to be huge. But perhaps it only covered that area of the world - in which case it may have been custom-built by Randy, or culled from a larger database.

The Bonin Islands, on the other hand, are relatively very small. Again we have the question of the resolution of the database - what is the size cutoff for the geographic features listed? On the other hand, the Bonins are the only thing worth mentioning for miles and miles around them. This would argue for them to be included in a database with a low enough resolution to otherwise exclude them. Then again, they aren't very close the places Randy is mostly working in - but they are roughly halfway between the Philippines (or Japan) and the west coast of the US, so he passes over them on his intercontinental flights. Would they interest Randy enough for him to include them in a database, if he compiled it?

We also have the possibility that Randy's sig automatically includes the GPS coordinates, and probably "Nearest geographical feature:", but leaves the actual feature blank for him to fill in. He could certainly identify and fill in Palawan, and would almost definitely pick that, as he was apparently just off the coast at the time. In the middle of the Pacific Ocean (on his flight path), the only thing to put would be the Bonin Islands (assuming he could see and identify them, or just looked at an atlas).

So, for an answer more relevant to your question - I don't know if there are any free datasets extant. If not, I would encourage you to start one for your area (or whatever area you would like to have data for). Find someone who knows what the best format for this data would be to merge it with GPS coordinates. They, or you, or someone, will also have to work up some code to actually do the merging. Sounds like a great open source project, with many collaborators contributing data on worldwide geographical features, landmarks, points of interest, etc., perhaps in a Wiki format.

Or, just use Google Maps.
posted by attercoppe at 9:19 PM on April 28, 2005

If you really wanted to implement this you'd need access to geographical data. There's was a set of data, that at least at one point in time, was downloadable that included various types of spatial information. I think it was called Tiger something or other and one of the prominent Open Source guys provided access. I remember it was structured in layers (I played with it a bit) that included roads, lakes, rivers etc. I don't remember if it had a layer for things like hills and mountains.

To build something like this (assuming you can find the data) you could use a USB GPS receiver and pass the co-ordinates into a database query.

It'd probably end up being a pretty damned big database unless you did it for small geographic regions but it seems possible.

The database would have to be able to sort of wildcard around a geographic area and then sort by proximity to pick the most useful feature.

Here's a link to the Tiger stuff by a web interface.

PostGISadds the ability to query a PostgreSQL database geographically, i.e. by longitude and latitude.

You can download Tiger census data from here.

Unfortunately it doesn't look like Tiger has actual geological information in it other than lakes (great for Minnesota but probably piss-poor for Arizona I guess)
posted by substrate at 6:56 AM on April 29, 2005

Maybe the USGS can help.
posted by substrate at 6:58 AM on April 29, 2005

You'd use The Getty Thesaurus of Geographic Names. It maps down to city quarter, and has gps coörds. Here's Palawan for example. With that dataset, it'd be easy.
posted by DangerIsMyMiddleName at 7:30 AM on April 29, 2005