Tao of Administrative Districts, Computing the Terrain

Last summer, I finally visited Norway and saw the fjords and the landscape, and had some fuzzy thoughts I’m still working through.


Streaming over mountain passes, through tunnels, around fjords .. administrative divisions are something like terrain computation, both at best obeying some kind of Tao across many domains of constraint.

Driving through Nordland, the kommune boundaries were surprisingly frequent, much more frequent then seemed practical for any kind of administration. The first order administrative district in Norway is called a fylke, something like a county, and below that the kommuner. From Wikipedia, kommuner “are the atomic unit of local government in Norway and are responsible for primary education (through 10th grade), outpatient health services, senior citizen services, unemployment and other social services, zoning, economic development, and municipal roads.” Seems like a lot for every tiny, low populated, northern Norway kommune to cover.


These tiny divisions are the logic of the landscape applied to administration, but don’t sit well with modern travel habits. The computation is an anachronism.

Kommunes were originally each set up with an administrative center one days travel from the next, by boat, horse or foot. The crumpled terrain of Norway and the coast would lead me to think that each kommune was one days sail around the fjord to the next place. Suppose these were organized by the church, as they were here in California, with the Spanish mission system.

Now when driving through Nordland, the kommuner fly by. With good roads and tunnels and bridges, the one day rule doesn’t fit. The kommune divisions remain, though they’re under constant review and debate in Norway. Again from Wikipedia

The consolidation effort is complicated by a number of factors. Since block grants are made by the national government to the municipalities based on an assessment of need, there is little incentive for the municipalities to lose local autonomy. The national policy is that municipalities should only merge voluntarily, and studies are underway to identify potential gains.

There’s competing domains over which the logic of the landscape boundaries are measured. National financing works against consolidation, yet most every public service might benefit from joined forces. Take snow clearance. I assumed that this must be organized along a different landscape logic, that maintenance stations would be distributed along roads according to some practical distance they can plough in a single day .. the domain constraint of modern plowing of modern highways. But that’s not the case, each kommune is responsible for ploughing, and it often happens in the winter that driving along a national highway, at the kommune boundary you’ll suddenly find a unploughed road.

The early days of increasing road travel worked in the opposite direction, towards increasing kommune. Lyngen kommune in Troms was split into Lyngen, Storfjord and Kåfjord. People began to travel overland, even though it took longer than traditional sailing, and as a result the one day was reapplied, splitting the kommune in three.


How are administrative districts like a computation? We start with a landscape over which to ask questions — the non-human geography, the terrain and resources in an area, over which one government has control (in contrast to borders, which are divisions negotiated by adversaries, historically by killing each other, while administrative boundaries are usually less adversarial and bloody). The questions we ask are generally about distribution of resources .. where to build towns, where to farm, over which path to build roads. The answers are a combination of top down decisions (one days travel), development processes (locales grow/shrink due to many factors), and naturally the landscape itself (richer resources lead to a better settlement).

Landscape and terrain are terms also used in computation, as a means to visualize the space of all solutions to a problem. A fitness landscape. Problem solutions which are better, according to some criteria, are represented as peaks in the landscape. To find those solutions requires some traversal of the terrain. A brute force search doesn’t take any account of the landscape, in preparing the solution search or in response to what’s encountered. An evolutionary computation responds to peaks and troughs of the landscape. A solution engineered entirely in the mind of a programmer is imposed on that landscape — it’s an interaction but one we’d consider outside the computation domain.


Lines can be drawn in any way over the earth. Code can be organized in any fashion. Only same of those lines and code are going to make any sense. Usually those lines and codes are not arbitrary.

In post-revolutionary France, the map above was proposed to rid France of the ancient province system, which no doubt incorporated some ancient injustice. This rational solution held no account for old territories or any fact of the landscape itself. Truly fair? Never adopted of course.


But then look at the American West. This huge territory had ancient division of use and administration (I’m not aware by what method or agreement, but I’d be very curious to hear how boundaries worked among native Americans) .. but those ancient divisions were completed ignored. And in the name of expediency of settlement, the huge territory was divided along rational, geometric means. For the US to maintain it’s claim, it needed to be settled fast and needed to be settled “fairly”. There was no time survey the land and adjust the size of plots according to division of resources. The domain restrictions weren’t the landscape, and there were plenty of problem solvers to through at it. The American West was settled by a brute force search.


Is there some natural way these divisions should fall? They are difficult to change. In Norway it’s under constant discussion, but there is a huge amount of feeling, and with time human feeling overruns any efficient or rational organization of the landscape .. perhaps the longer time people put into a place, the less likely they are to approach any reorganization rationally (or perhaps that’s a symptom of the European system of real estate).

The map above was alarmed by the Daily Mail in the UK. I don’t know the context of the map, but it shows divisions of the UK and northern Europe along some non-nationalistic lines. The EU is of course not reorganizing national boundaries, as the Daily Mail is alarmed, but perhaps for some domain constraint such borders make sense. Cornwall probably does share more in common with communities along the Bay of Biscay then it does with Kent and Sussex.


Tao, by Lao Tsu (Lao Tzu, Lao Zi) (by way of Wikipedia) “taught that the wisest approach was a way of ‘non struggle action’ (“Wuwei”or “wu wei”) – not inaction but rather a harmonization of one’s personal will with the natural harmony and justice of Nature”.

I have no idea what this means practically. Somehow I think there’s some natural harmony of computation, and some natural harmony of organizing ourselves along this planet, and there’s some deep principles common to computation and the landscape which inform both and will lead to more sensible decisions.

Thanks to Magne, einarr, bokare on #osm for discussing Norwegian kommuner


  1. Morten said,

    January 7, 2008 @ 9:44 pm

    Actually ‘Kommune’ is a lot closer to the meaning of ‘Municipality’ (if not exactly the same) than ‘Fylke’ is to ‘County’.

  2. Ed Davies said,

    January 7, 2008 @ 10:19 pm

    Yay – back to the Danelaw!

RSS feed for comments on this post · TrackBack URI

Comments (2)