In these early days of planning-by-layers, the foremost danger is the belief that the computer can resolve the age-old problems of planning, by providing a rationalist method leading to inevitable conclusions. McHarg, arguing in favour of planning-by-layers, wrote that:
It provided a method whereby the values employed were explicit, where the selection method was explicit -- where any man, assembling the same evidence, would come to the same conclusion. (McHarg, 1971)
McHarg saw this as "environmental determinism': a method that allows "nature', instead of man, to take development decisions. His title, Design with Nature, cleverly steps back from determinism, implying that nature will simply help man to take better decisions. The text is less cautious, as revealed by the remark that "any man using the same method will come to the same conclusion'.
Later advocates of GIS-based planning have seen the method as a "decision support system'. This implies a method for using geographical information to help planners, rather than a decision-making system. It is an attractive proposition, which takes us to the key question in GIS-based planning: which data should be used?
Consider the case of a developer seeking a site for a new office building. Using a GIS, it is possible to find a site which is:
1. over 2000 square metres in size;
2. located on soils that are suitable for construction;
3. not located on forested land;
4. not located on land of high agricultural value;
5. within 300 metres of an existing sewer line;
6. not within 20 metres of a watercourse.
This example is taken from Understanding GIS: The Arc/Info Method (Environmental Systems Research Institute, 1993). It appears to offer a decision-making procedure that is entirely based on verifiable criteria. If similar procedures were used for schools, housing, industry, transport, and every other land use, one could produce a full land use plan. As every decision would rest upon objective criteria, everyone would support the plan, wouldn't they? Well, they shouldn't. The following points indicate some of the flaws in the decision tree.
1. The size of ownership parcels can be changed, by land assembly.
2. With suitable foundation engineering, construction can take place on almost any substrate.
3. If one area of forest is lost, another can be allowed to grow.
4. With the current food surplus in developed countries, there is no need to preserve land of high agricultural value.
5. New sewer lines can be built.
6. Streams can be protected from pollution by filtration.
The idea that GIS enable a "rational' planning procedure leading to "optimal' land use allocations is wholly misconceived. GIS are no more capable of resolving planning problems than a pocket calculator is capable of telling people how to vote. The contrary view rests on the old enlightenment dream that reason can resolve all conflicts and solve all problems. I wish it could. Computers have merely given the dream a short lease of extended life, partly through the magic of the machine and partly through the concept of layers. "Layers of what?', is the fundamental question. Easy analogies were drawn with layer-cakes and sedimentary formations. Both are unitary structures; both have tops and bottoms (Figure 9). But, if planning is conceived in this way, which is the top layer? Too much twentieth century planning was conducted in a top-down sequence:
1. begin with land ownership;
2. designate the land use;
3. engineer the roads;
4. subdivide the land;
5. design the buildings;
6. arrange the paving and the planting;
7. furnish the interiors
After that, users and maintenance workers were left to make the best they could of a bad job. The whole procedure was tyrannical. And it was no more capable of making good places than dictatorship is capable of providing good government. Both fail through lack of information. Both favour grim concrete jungles with wide roads and boxy buildings. Central planning simply cannot cope with the wide universe of facts, values and beliefs. Democratic planning requires fresh maps, fresh concepts, fresh information and fresh procedures
Fig 5.9 If planning resembles a layer-cake, which layer should be on top?