## September 17, 2018

### Demoland Demographics

Continuing from here.

I think there are still some issues in interpreting numbers and percentages. After all, as attractive as it could be, I'm not attempting to build a steady state system, which never changes. Instead, this should be reactive and on-the-fly, providing a baseline for the players to interact with.

In the previous post, I developed a model for character levels that might be encountered. In this post, I'll apply that equation directly to the towns in Demoland, but with a few caveats.

 city pop 0 1 2 3 4 5 6 7 8 Llen 698 469 154 51 17 5 2 1 Betryn 698 469 154 51 17 5 2 1 Cadewin 798 536 176 58 19 6 2 1 Gerlin 2295 1541 507 167 55 18 6 2 1 Derl 2893 1943 639 210 69 23 7 2 1 Andox 5587 3752 1235 406 134 44 14 5 2 1 Ffith 1397 938 309 102 33 11 4 1 Norys 499 335 110 36 12 4 1 Malis 2694 1809 595 196 64 21 7 2 1 Kenor 599 402 132 44 14 5 2 1

Firstly, I haven't taken a look at age distribution yet. I think this is an interesting aspect and I'll be tackling it later. But age and level taken together are not so simple. It would be laughable to encounter a Level 5 child of a mere eight years old, and yet the system would allow for such a possibility if I simply multiplied the distributions (as I plan to do with class distribution and racial distribution). If (for example) 10% of all population are children, and there are 100 Level 5s in a city (requiring about 40,000 people total), then is at least 1 a child?

This does a gross injustice to the science of statistics, but I must keep my model simple somehow.

And yet there must be room for true diversity. It doesn't matter if there's one dwarf in a town of a thousand people - he might just be the sole Level 6. It's rare, yes, but the system must allow for the possibility.

Hence the difficulty. These posts, while useful for testing the model, only give the most normal, steady-state situation. A true encounter "table" (really, it's an encounter model) could include the Level 6 dwarf, but discourage the Level 5 child. This is the ideal. But I'll work on age later.

One way to think about this order is that it's emergent, not determinant. The Level 8 does not ensure or build Andox: he is there because Andox will have a Level 8. The numbers are retroactive.

I could have a population "generator" that makes a separate roll for each person. It's a simple enough calculation to do with Python. That would make, say, Llen and Betryn different (currently, they are the same because they have the same population).

 0 1 2 3 4 5 6 7 Llen 698 466 146 61 13 10 5 1 Betryn 698 454 155 55 21 8 3 1 1

Hm. That's way better. However, it takes a bit more elbow grease to implement in Excel. However, it can totally be done. I leave the implementation as an exercise to the reader.

Ultimately, I think what I'm building is a context and location aware NPC generator. But it should also give direction to the DM about what type of place this is. If the $k$ value remains fixed, then this becomes a matter of thinking about how different levels are stratified in differently-sized populations. And if we want the Level 7 guy to help us, we've got to go here (and not there) to find him. I think this will be richer once the classes are added, since it's constrained by geography and the trade network, and not merely a function of population.

Lots to learn. Lots to think about.