Points of Interest Generator - Design Blog (Patreon)
Content
Greetings patrons, my players will soon be getting their hands on an enchanted wagon and heading out into the wide world of Ashk as an adventuring company.
I've given my players access to my LegendKeeper project so I can start sharing information with them and get them hyped, and I want to populate the map with evocative points of interest that they want to explore.
I believe the reason so many people struggle with the exploration pillar of D&D is because they aren't making exciting locations the players *want* to explore.
Now, I could sit down and make a list of locations by hand, but I'm trying to take a more hands-off approach to my weekly prep that will help speed prep up in the long term, so I decided I wanted to make a generator.
Why Do This?
So why make a generator instead of finding one online or just making something up on the spot?
By making my own generator, I control all the data that goes into it, and am able to make as sure as possible that all the outputs are suitably interesting and evocative.
I'm choosing to build a generator for this so that I don't have so much pressure on myself during prep and at the table.
If my players go in a direction I don't expect, I can generate a location and that prompt will be enough to spark my improve brain and get me going.
It's the same principle as just writing something on a page to conquer the blank page and get your brain firing.
The Data
The biggest hurdle for this task is the data. A generator is only as good as what you feed it; put rubbish in, get rubbish out!
I know that a tried and true method for naming things is [simple adjective] + [concrete noun], so things like the "Black Forest" or the "Lonely Mountain".
This was my starting point. I scoured the internet looking for lists or generators for descriptive adjectives and concrete nouns, and imported about 1,000 of each into Excel.
I quickly realized that this was the wrong approach!
There was too much rubbish being fed into the system, and I was spending too long trimming the lists.
So I would need to manually feed the machine, which is more time consuming at this stage, but will result in better outputs.
This process is familiar to me, as it's essentially what I did when I worked on the character generator for Here Be Taverns (which is a great generator platform that is worth your time checking out!).
Because I wanted this to be a location generator, it made the most sense to me to start with the concrete nouns side of the data, now named "Location", and I started inputting various geographical words like forest, mountain, river.
This got me a few decent results (Nauseating River was my favorite!), but it highlighted how weak my adjective data set (now called "descriptor") was.
I think for the best results, I need to vastly increase the size of the descriptor data set. I plan on also including some verbs in there for more variety.
My thinking behind this is if you have the howling woods, and the howling mountains, if they are not in the same place then that is going to seem strange, but it's much less strange to have the howling woods and the old woods. Those are two places that could reasonably exist on the map and not be related.
Getting to work
I started off my new descriptor data set with colors, as that seemed an easy place to begin.
I inputted the colors and then started using synonyms of those colors. Doing this resulted in outputs like "Ruby Caverns", "[The] Amber Shrine", and "Silver Hollow".
These are all perfectly serviceable and wouldn't look out of place on a map, so I knew I was heading in the right direction.
After that, I started expanding the list with more adjectives, which started returning results like "Voiceless Ocean", "Scorching Plains", and the "Woeful Valley".
These are much closer to what I'm looking for and carry with them an immediate question or story.
Why is the Voiceless Ocean called such? There's a story there and I want to find it!
Even this system won't produce perfect results every time, but it should produce results that can be adapted to something useful.
For example, the generator might output "Dagger Labyrinth" which isn't very strong. But if we take that result and change it to "The Labyrinth of Daggers" we have something that fits the evocative criteria.
My current version has 134 descriptors and 134 locations, giving almost 18,000 unique combinations for results, but I want to increase that significantly.
I want to get to a minimum of 300 descriptors and 200 locations, as that will give us around 60,000 unique combinations for the output.
So I'd love you hear your suggestions for words that can be added to the generator!
Much love
Anto