Last week there was a QGIS course at our institute. Since the Ilwis-objects work consumes all my times these days, I didn’t attend myself. I did, however, find it interesting to hear what others thought about it. After all, why are we building Ilwis-objects if a big, successful open source package exists in the same niche? Good question, but then again why is Volkswagen building cars when Toyota is doing the same? Functionally, they are the quite similar, aren’t they? They have four weels,  an engine and some controls. No need for others…

Of course it is not that simple and that is also true for the open source world. I am happy that QGIS successfully competes against the dominance of Esri’s ArcGIS. They have quite a lot of developers and apparently quite good PR. For the last five years, they have grown very fast. So why not join them? Well there are several reasons for that, some objective, some subjective.

• First of all, it has to do with the place I work. ILWIS has always been a kind of “poster” for the ITC. It is one of the things that outsiders associate with the ITC and vice versa. Not for all, but it helps as an introduction/selling point. Subsuming our expertise in QGIS would loosen that association. We would become the next, mostly anonymous, contributor. To me, and apparently also to the ITC, this is not that attractive.
• QGIS wasn’t made for education and you can tell. ArcGIS suffers from the same symptoms. It is a bit like a very big toolbox in which tools are seemingly randomly placed and made by different craftsmen. I am exaggerating of course, but I think you get the picture. Eductation needs a well organized system. You want to spend as little time as possible teaching the students which buttons to push. You want to teach them the concepts. And herein lies my third point of critisism.
• There is a lack of conceptualization in the software. What do I mean by this? When I look at QGIS I get the impression that it is a bottom up implementation. Things are implemented on a “necessity” basis and thus bolted on an existing structure that might or might not be suited for it. Now maybe my observation is a bit harsh, but it is the impression I get at least when looking at how things are intergrated in QGIS. They are very flexible in this respect, but I am not sure if that is good. ILWIS has always had a very solid top down structure, which mandated a lot of the things to the bottom up implementations. It is less flexible for sure, but it ensured that everything worked in the same way and looked consistent. I am quite sure that my opinion is biased because I know ILWIS well, QGIS less so.
• Data structures are an issue. Related to the previous point, we have historically put a lot of thought and effort in to the semantics of geographic data systems (see for example domains).  We extended that model with new concepts in Ilwis-objects, but the core remained intact. That is something I don’t want to throw away.
• Last of all, there is a control issue. I work with scientists, PhD candidates and students. They all want different things. Some whishes are simple, some are much more complex. Sometimes I have to add/change things in the core of ILWIS. If I would work on QGIS, adapting the core to our needs would be much more difficult as QGIS core programmers would be (and rightly so) suspicious of any changes in there.

Now QGIS isn’t a closed door to us. We have a plug-in (well we’re busy with it at least, but it doesn’t work for some unclear reason) for QGIS which exposes the map calculator and operations to QGISas far as QGIS can handle them. And there certainly are things we can learn from QGIS. The cartographic output is much better, the hierarchical layerview is probably something I want in ILWIS too, along with a number of other things.

Anyway, it is healthy if there are multiple makers of this kind of software.