Junta_Joe frequently referred
to this thread, but
never provided a link.
... I have only had one totally hopeless game in the two months I have had Tropico. It happened that I had a huge inlet bay. Docks on one side, farms on the other. ... I made a point from then on that I would not let a terrain feature be an unexpected challenge. ... When I want terrain to be a negative, I set it up that way. What I'm saying is that when I start a new game I know the challenges I'm in the mood for. Anything else tends to kill the fun. ...
I'm not really sure what JJ IS
saying about the shape of the island. What he is saying about his style of fun is clear -- he wants nothing unexpected. I have to doubt that the "huge inlet bay" was the result of the map generation program; it was almost certainly the result of his editorial terraforming. The standard for generated islands is that they are almost round whether large or small and never have really deep indentations. That has a direct relationship with the pathfinding system used for the people and for the ocean vehicles. Cafe Dave discovered that ships go crazy and sail across land when there is not a strip of water all around the edge of the map. The discussion seems to indicate that the "Water Coverage" variable may cause larger inlets, and I have not looked for that.
Ole: I've noticed that bays are most common on random maps set to have more than the default amount of water [coverage].
I built a couple of 'land" bridges on an experimental map. I built a long inlet to simulate the mouth and inland course of a large river. My experiment also included small streams and ponds. Believe me, they are left out because they really screw-up all pathfinding. People starved to death even when they were free to go around the inlet\river. My experiment showed perfectly that people having to go any distance beyond 100 tiles will go to a circular band parallel to the outer circumference of the island. There they run great risks of breaking the pathfinding loop and getting stuck in a repeating, endless loop. They will NOT go toward the center of the island to avoid an obstacle.
Long narrow or crazy shaped islands throw a great strain on the pathfinding algorithm. The game engine recognizes only one island on the map no matter the shape. So a series of small islands connected with narrow strips of land will not operate as separate islands connected with bridges, but rather a crazy strain for path finding. I suggest such experiments are a waste of time; there are other, better (more interesting) challenges.