I have spent the last 10 years organising, analysing and presenting spatial data in a variety of projects. My research has led me into the dark corners of transport modelling, climate statistics, energy modelling and a whole host of other fields. Spatial data underpins all these disparate activities (and we also make cool maps and use cool tech). However, I am increasingly frustrated by the effort that goes into finding and looking for spatial data. In the UK we are sort of OK as far as it goes as we have the Ordnance Survey and a few commercial and academic portals that provide (at a price) one stop shopping for most things. I suspect the situation is similar in most first world nations but that certainly is not the case in many other places.
I have now sat through a number of presentations on INSPIRE, have looked through geoportals both open and closed, battled with OGC metadata specs, geoserver config and openlayers and come to the realisation that we should just let Google do it. Think about it, they have the financial clout, the technical clout and they are probably (definitely in my case) smarter than us anyway. I envisage a Google geometry search so that I type a place name and I find spatial data stored as a file of geometry (raster or vector) as the result.
Ah, but will it be the right one – well surely this is just another good use for Page Rank? Is it not the case that more likely than not in the UK we will use an Ordance Survey Opendata admin boundary. Surely their metrics will show this? I mostly want the best information for the job and this could give it, quickly, easily and cleanly. Page rank for spatial can do what ISO 19115 will never do and provide a filter of perceived utility.
Ah, but “spatial data is more complex”. Is it really? What is complex about it that big compute and big tech cannot sort? Take projections – why can’t we just reproject our data a la Google translate (easier as it is just maths not language). What about layers of geometries that interact spatially – well this is resolved using a spatial index in a database – I suspect having a massive spatial index of all the objects in the world would not be a stretch for a company whose starting point is an Index. And format conversion – that is just compute – shouldn’t be an issue (or perhaps just index and cache call the converted versions for everything). Linking spatial data to a rich keyword based and natural language processing means we should be able to find things – alternative geographical names fall out of the system (and as a not insignificant byproduct the number of neogeography talks may dwindle).
Imagine a world where I want data on Nigeria Rivers and I type into Google Geometry such a query term – the result, peer reviewed by the vast caucus of humanity (or those special ones interested in spatial anyway) is a link to a file that I can visualise online, download and add value to or combine in a web based “mash up”. One suspects our tools would be simplified as well when data is easy to find, view and retrieve.
We can then ditch the metadataverse and multiple WMS/WFS endpoints and catalogue services that make up the current crop of portals and the INSPIRE vision and do what we do best – use spatial to help understand the world and make better decisions. If this is the way that users expect spatial data then there is incentive to put your data out there (and let Google find it and index it). There’ll need the spatial index anyway to do sensors properly (but that’s another story).
GeoAnorak. Changing the world one vertex at a time.