This month for our Nestoria Interview we chat with Muki Haklay a Senior Lecturer at University College London’s Department of Civil, Environmental and Geomatic Engineering. A long time participant in, follower and commentator on the ‘neogeography’ trend, Muki recently published about OpenStreetMap, and its explosive growth. He is a fellow of the Royal Geographic Society and a member of the Association for Geographical Information (AGI).
Muki, many thanks for speaking with us 1. Why is that OSM and other neogeography projects have grown so rapidly in the past few years?I think that around 2005 there was a fundamental shift in some underlying technologies and social trends that allowed it. Of the wider driving forces, I think that there are 5 significant ones: GPS availability, Web standards, Broadband, Data storage and a new paradigm in programming APIs. Quickly explaining what I mean: 1st May 2000 should be the birthday of Neogeography, because that was the day when President Clinton removed the selective availability of GPS signal so we can capture locations easily and with good accuracy. Without this decision, we wouldn’t have the huge explosion in location data. The second thing is web standards. I still remember when I first read on Dave Winer’s blog (although it wasn’t yet called that) about the release of SOAP, and the whole development of XML and other standards. Without the evolution of XML,SOAP and the gradual adaption of browsers that are capable of running AJAX which took some years, we wouldn’t be able to run geographical applications which usually require more processing than an email application. The third is broadband access – as Tom Friedman notes in ‘The World is Flat’ – because of the Dot.Com crash and the infrastructure that companies like Global Crossing and WorldCom created at the end of the 90s, we now got cheap and fast broadband across the world. Think about piping lots of tiles to the user’s browser and you realise that you can’t do it over a dialup. The fourth point is data storage. We tend to forget that as late as 2000, the cost of gigabyte storage was about $10, which again limits the ability to tile the world. By 2005, it was 50c. Finally, with all these things, a new paradigm of light APIs emerged. The learning curve of Google API is much shorter than, say, understanding how to program a typical GIS mapping sever, such as ArcIMS or MapServer. Take all these together, and you have an explanation of why Neogeography happened in 2005 and not 1995. It’s interesting to notice how, as in many other aspects of geographical computations, GIS and Geography is more complex/voluminous/computationally intensive so it lags behind other applications by about 10-15 years. You can show similar lags in the development of geographic technologies in previous generations of computing platforms. 2. What are the implications for the professional GIS community?
I’ve noted on my blog the implications of all these are far reaching. For too many employers, the justification in employing GIS professionals is that the software used to create maps is very complex, so having specialists who produce maps is justified. But if it is possible to create maps with a simple API instead of buying an expensive and difficult to maintain Internet mapping server software, or if it is enough to analyse the data by creating a point map on Google Earth – then why keep the expensive professional? This is a good thing in the long run. GIS analysis and clever use of data and analysis techniques are the real benefits of using a professional GIS package, and these are things that an API can’t handle (for now) so GIS professionals should move up the chain and focus on services where their geographic knowledge really adds value. It’s much more exciting and interesting to use GIS to do proper analysis and not just to produce maps… 3. You recently posted about a comprehensive evaluation of comprehensiveness of OSM geodata versus professionally gathered data. What did you learn?
The report is quite extensive and covers lots of aspects of the OSM data, so it’s worth focusing on some highlights. One of the most striking aspects is that OSM provides a different type of data in terms of accuracy – and that for many applications this is absolutely fine. Not everyone need very accurate data – so OSM teach us what is good enough in geographic information. In terms of quality, just as in the case of open source code, we need to think about the importance of Raymond statement ‘Given enough eyeballs, all bugs are shallow’ For mapping, this should be translated as the number of contributors that worked on an area and therefore removed ‘bugs’ from it. And this is an aspect that the OSM crowd need to figure out and that will improve quality. Too many places are covered by a single person. I’ve also understood the completeness is more important than high accuracy – as you think about fitness for purpose, you understand that once OSM will get to 75-80% coverage of a place, the data will be very valuable and very usable. One of the motivation for another part of the analysis was a matter-of-fact comment from Steve Coast (ed: see past Nestoria Interview with Steve Coast) about mapping less attractive locations (council estates) and I wanted to check if this tally with government measures of deprivation. Indeed, more deprived areas are far less mapped then wealthy areas. For me, that’s one of the overlooked aspects of Neogeography – it doesn’t really happen in digitally excluded place. Even in places such as India where Google do their Map Maker activities, they do it with students, who are naturally coming from a wealthier background and not with socially excluded people… 4. What effects will crowd sourced geodata have for the the web community in general and start ups, like Nestoria, in particular?
Well, because we’re just three years since it really took off, we need to wait and see which sources are becoming stable and reliable. For example, as we get more geotagged images, and with tools like photosynth and the developments in image recognition, I think that we will see lots of applications that use visual imagery to extract geographic information – think of getting data on all the shops in an area by looking at their signs. As I’ve noted, we also need to wait until some places will have a robust coverage – once OSM got proper coverage of a major city such as London I’m sure that we will see exciting applications from CloudMade or similar ventures that build on it. It’ll enable many more start-ups in the area of geographic information. For example, UCL and London 21 Sustainability Network are currently setting up a community mapping social enterprise (that’s like a start-up, but all profits go for social goals). It’s called ‘Mapping for Change’ and it will focus on making places more sustainable. Getting the data is a major issue for us. Using OSM as our source of geographic information will enable us to spend our effort and money on engaging communities, not on buying data. I’m sure that there will be many more examples. Thanks Muki. We recommend anyone interested in tracking Muki’s work learning from his experience subscribe to his blog. We look forward to watching OpenStreetMap’s continued growth in coverage and comprehensiveness, and finding ways for Nestoria to work together with the academic community. Readers may be interested in our past cooperations with UCL). past Nestoria interviews: Brad Inman, Simon Baker, and Tim Youngman.
