GeoWeb 2009 was my first trip to the conference and Vancouver. Both were outstanding and I’m looking forward to having a chance to visit again. There were lots of great talks across the board and perceptive coverage by James Fee, Peter Batty and our own Andrew Turner. A persistent theme I saw emerging was the tension between top-down vs. bottom-up approaches. Top down being standards developed by committees (W*S, GML, CWS etc.), data sharing initiatives in the form of Spatial Data Infrastructures (SDI), and implementations built around protocols like SOAP. On the other end we have bottom-up approaches where de facto standards are created around iterated implementations (KML, GeoRSS, SpatiaLite etc.). Data sharing takes the form of indexed geodata that is directly Web accessible (e.g. Jason Birch’s work with Nanaimo). Protocols for implementation in the bottom-up category typically follow a RESTful philosophy.
Over the course of the conference it became apparent there were two camps/factions that people fell into. Although to be fair there are a good number of folks straddling the two. Although I’d argue most of the straddling is tops-down folks moving into bottom-up approaches. The two are definitely not mutually exclusive. With GeoIQ/GeoCommons we support OGC and de facto standards and we’ve been working on integrating with SDI style architectures. That said from a philosophical stand point we are firmly in the bottom-up camp. While there is an undeniable top-down legacy that has merit at many levels. I think we are going to see the bottom-up approach continue to gain momentum and become the prevailing approach. There is no doubt a need for both. We’ve probably just swung too far in the top-down direction in the past.
Why are we such big fans of bottom-up? In short it works. We’ve seen the cool animations of a year in edits of OpenStreetMaps from 140,000 and growing contributors, we’ve seen the animations of 500 million KML points from Google, and no we can see the animation of 14,000 contributed datasets to GeoCommons:
Visualization of Uploads to GeoCommons from GeoCommons on Vimeo.
Each blue square represents the bounding box of a data set uploaded to GeoCommons.
The best thing about bottom-up; it is here today, it works and there is tons of content.
2 Responses to Top Down vs. Bottom Up at GeoWeb 2009
Leave a Reply
New GeoCommons Maps- NYJ city barsone
- Israel Outdoors: Where our applicants are from carine
- jets by state cluster barsone
- Maissade Milko5571
- T-Mobile gulyi01
- AOD MODIS gianluca
Sean,
I did a Top Down talk at GeoWeb and had the pleasure of defending SOAP in the SOAP vs REST vs PERR2PEER. Well the SOAP vrs REST was really quite fun at GeoWeb (it even got me my first appearance on Geek TV). Anyway, the Neogeographers were out in force defending REST at GeoWeb and their view seems to be if you can’t access it in the browser its not in the GeoWeb, in my mind this doesn’t hold up. What about B2B? Are people really saying the GeoWeb is not about B2B data exchange. This got me thinking that there are two distinct communities in the GeoWeb. Taking a mathematical approach I think the GeoWeb can be defined as: GeoWeb = B2B + B2C (the C being customer or citizen). Refining that further, GML over SOAP is best methodology to orchestra business data exchange and KML over REST is the simplest method to get data to the citizen. So I came up with this formula GeoWeb = (GML+SOAP) + (KML+REST). Now I do agree with you that the REST and Bottom Up solves about 80% of the use cases (and that this is only going to grow) but there are scenarios where Top Down and rigours of SOAP are the only option. Aviation is a great example of this, an complex data model, safety critical, cross business and country data exchange. Its clear that strong standard bodies like ICAO need to set international standards from the top down in order to create safe environment. It maybe that the Top Down approach gets less popular but it will never go away.
Ian
Hi Ian,
I really enjoyed your talk at GeoWeb and you make a very strong case. I don’t think it is an either or, and definitely plenty of room for both. I’m assuming business-B2B means the enterprise – fortune 1000 etc. If so I’m curious where the adoptions have been for GML+SOAP. I might just be ignorant of it, but largely I don’t see the majority of enterprises doing either (GML+SOAP or KML+REST). Not convinced anyone has come up with the solution that makes a geospatial implemtnation palatable to the vast majority of enterprises. A big greenfield that needs to be figured out no doubt. Maybe GML+SOAP is the answer I’ve just not seen a large market adoption of it or anything else for that matter in the enterprise.
best,
sean