Random Coastlines, Rivers, and Roads

[Man in the Funny Hat]

There is a difference between random generation of terrain and deciding what that result LOOKS like, so to speak. You can randomly determine the appearance of seacoast, lakeshore, rivers, swamps, forest, mountains, sandy deserts, prairie, and windswept and frozen arctic tundra - but random tables can't make that appearance fit sensibly with everything you've already generated or will generate next. That's all that Gygax is saying.

Example: PC's are exploring. Random generation says that when entering the next hex they come upon a sea coast. Well, logic says that to the left and right is more sea coast. That sea coast MUST continue in either direction until it has circumscribed a body of water large enough to be an actual sea. You can't then roll dice and accept a result of lake. You either have a sea coast or a LAKE shore. You, the GM, must disregard the random roll at that point if favor of actually creating a sea.

Example: PC's start in a hex. They know only that hex. It's their starting point for the whole game world that is going to be randomly generated. They have 6 directions to explore in. They go in direction 1 and your random result is... seacoast. Great. You plop down a sea. They go back to the start and go in direction 2. You randomly generate... a seacoast. Fine. You create another sea right next to the first in such a way that the two seas are still separate. They go back to the center and go in direction 3. ANOTHER sea... See where this is going? The random table CANNOT account for what you already generated or what you will generate in the future. YOU have to do that. That is all that Gygax is saying there.

And as for roads, you can have them appear as a random table result, but roads ARE NOT random. Roads lead somewhere - or at least USED to at one point. They go from A to B, either as a planned, deliberately constructed road to make getting through certain terrain easier (or to allow it at all...), or it was a course of travel that was so frequently used that it became a road. Roads are not natural occurrences. They exist for a reason. Yes, it's a fantasy world so you can have any reason you like - including being deposited randomly by the road fairy. But if you're using random generation to build a world you need to have roads lead from a source to a destination. From place A to place B. THAT is what causes roads to make sense and is something that random tables cannot do. YOU have to do that.

Don't get me wrong. I think random generation is fun and I've used it in the past, but it has SEVERE limitations regarding what it can do. It REQUIRES someone to understand when to stop randomly generating and actually make sense of the existing results; to follow the logical inference of the results rather than generate more randomness.

 

[Hriston]

There is a difference between random generation of terrain and deciding what that result LOOKS like, so to speak. You can randomly determine the appearance of seacoast, lakeshore, rivers, swamps, forest, mountains, sandy deserts, prairie, and windswept and frozen arctic tundra - but random tables can't make that appearance fit sensibly with everything you've already generated or will generate next. That's all that Gygax is saying.

While I agree that Gygax is saying that, most notably when he says (also on page 173), "common sense must prevail", I do not think you've captured the full meaning of the quote I have taken issue with, which I'll repeat here:

In any event, the DM must draw in rivers, large lakes, seas, oceans, and islands as these features cannot easily be generated by a random method.​

Taken in context, this quote seems to say that in the service of "common sense" and a "reasonable configuration", the DM cannot randomly determine the presence of "these features", but must draw them in independently of the random determination method, creating an outline of sorts to be filled in randomly. This is the sentiment I've taken issue with, and I've attempted to present a method whereby such features can be randomly determined. You might disagree that such a method is easy to implement, or present in a publishable form, but my premise is that Gygax is using this as an excuse for not giving any method at all to determine such features, and that, in fact, such a system can be devised and presented with the inclusion of only a few extra paragraphs. Nowhere am I suggesting the DM should abandon good judgement in favor of a random method.

Example: PC's are exploring. Random generation says that when entering the next hex they come upon a sea coast. Well, logic says that to the left and right is more sea coast. That sea coast MUST continue in either direction until it has circumscribed a body of water large enough to be an actual sea. You can't then roll dice and accept a result of lake. You either have a sea coast or a LAKE shore. You, the GM, must disregard the random roll at that point if favor of actually creating a sea.

I don't see the problem here. If we assume that the DM is using common sense, which we are, then the situation is adjudicated quite easily. There is no reason why a major lake cannot exist within even one mile (the smallest scale of resolution recommended by Gygax) of the sea. All that is required of the DM to satisfy the needs of common sense in this situation is to establish the existence of some sort of high ground between the sea and the lake, separating the two. Certainly, there is no possibility of venturing into the sea and encountering a lake therein. A lake can only be encountered in the process of land exploration, just as the ocean can only be encountered in the process of land exploration.

Example: PC's start in a hex. They know only that hex. It's their starting point for the whole game world that is going to be randomly generated. They have 6 directions to explore in. They go in direction 1 and your random result is... seacoast. Great. You plop down a sea. They go back to the start and go in direction 2. You randomly generate... a seacoast. Fine. You create another sea right next to the first in such a way that the two seas are still separate. They go back to the center and go in direction 3. ANOTHER sea... See where this is going? The random table CANNOT account for what you already generated or what you will generate in the future. YOU have to do that. That is all that Gygax is saying there.

This example sounds to me like the PCs are on an island or a peninsula. I don't understand the need for the encountered seacoasts to circumscribe separate bodies of water. Nor do I think that is what Gygax is saying when he says that such features cannot easily be randomly determined. I do agree with you, however, that a seacoast, or any other linear feature, must lead into unmapped spaces, as Gygax states, and as I have echoed twice in my last post.

And as for roads, you can have them appear as a random table result, but roads ARE NOT random. Roads lead somewhere - or at least USED to at one point. They go from A to B, either as a planned, deliberately constructed road to make getting through certain terrain easier (or to allow it at all...), or it was a course of travel that was so frequently used that it became a road. Roads are not natural occurrences. They exist for a reason. Yes, it's a fantasy world so you can have any reason you like - including being deposited randomly by the road fairy. But if you're using random generation to build a world you need to have roads lead from a source to a destination. From place A to place B. THAT is what causes roads to make sense and is something that random tables cannot do. YOU have to do that.

I agree that a road only makes sense if it has a destination, but I don't think that destination needs to be known in advance by either the DM or the players. All that is required in my opinion, and as I have suggested a method for in my last post, is that a road, once encountered, extend indefinitely in either direction until it encounters a settlement of reasonable enough size, at which point it either comes to an end, the ultimate destination and purpose of the road having been discovered, or it continues or branches in the direction of further destinations. The existence of an ultimate destination is assumed, but its actual location doesn't need to be established in advance.

Don't get me wrong. I think random generation is fun and I've used it in the past, but it has SEVERE limitations regarding what it can do. It REQUIRES someone to understand when to stop randomly generating and actually make sense of the existing results; to follow the logical inference of the results rather than generate more randomness.

I agree and don't believe I have suggested otherwise. Please read the method I've proposed in my last post for determining linear features (seacoasts, major rivers, lakeshores, and roads) and let me know if you see any problems or have any suggestions.