NPC random statistics roll chart?

[Captain Joy]

Originally Posted by Fish

This is the chart I created.

Wow, your chart is most useful. Do let us know if you make any additions.

[Fish]

Thank you. I know that I would like to finish the specific occupation randomizers, as I said upthread. What other statistical variants can you think of? I can think of a number of categories but I don't know which would be most useful.

For instance:
age
weight
height
political party affiliation
smoker/non-smoker
religion
owns a gun?
criminal record?

It might also be useful to have regional distributions:

South, Southeast, New England, Northwest, Southwest, Great Lakes, Texas, Florida, California, New York, Alaska, Hawaii.

If anybody could point me at statistics for other places (and times!) I would like to make several other variations: United States WWII-era, United States 1900, United States 1850, Great Britain modern, Great Britain 1900.

[Captain Joy]

Originally Posted by Fish

What other statistical variants can you think of? I can think of a number of categories but I don't know which would be most useful.

For instance:
age
weight
height
political party affiliation
smoker/non-smoker
religion
owns a gun?
criminal record?

It might also be useful to have regional distributions:

South, Southeast, New England, Northwest, Southwest, Great Lakes, Texas, Florida, California, New York, Alaska, Hawaii.

The first three are a must, but height and weight should be based off ST. So, you'd need to randomly determine a ST first.

I would add that your list appearance--hair color, eye color, skin color, that sort of thing--before political party affiliation (assuming your list is in order of desired things to add next). I would also add names appropriate to their ethnicity at the top of the list.

[Fish]

I only listed those items in the order that they occurred to me, not in any particular order of preference.

As for hair color and eye color, that's really based on ethnicity, which is not the same as nationality. That might a little trickier; I could generate a few basic hair/eye tables like "Mediterranean complexion," "African complexion," "Caucasian," "Germanic," "Slavic," and so forth, and let the user figure out when to apply each.

I suppose if someone were feeling really ambitious they could convert the randomizer into an Excel-compatible spreadsheet with lookup tables and probabilities; that would make it easier to build a bunch of interrelated tables.

[Ragitsu]

The next best helpful chart would be Europe, I figure.

[Fish]

Hmm. Tricky. Europe is a pretty broad category, and its population is dominated by the really big players, leaving the small ones out in the cold. Europe, even more than the United States, varies greatly from region to region.

What did you have in mind, specifically? A chart for "Random Person Who Comes from Europe" won't look quite the same as "Random Chart for Person on the Street in London."

[Captain Joy]

Originally Posted by Fish

I could generate a few basic hair/eye tables like "Mediterranean complexion," "African complexion," "Caucasian," "Germanic," "Slavic," and so forth, and let the user figure out when to apply each.

I suppose if someone were feeling really ambitious they could convert the randomizer into an Excel-compatible spreadsheet with lookup tables and probabilities; that would make it easier to build a bunch of interrelated tables.

Indeed, such is my intent.

[Fish]

It would actually be easier to do a randomizer spreadsheet with raw data, rather than with dice tables, to be honest. For instance, take the employment tables. A 3d distribution means that I can have, at maximum, 16 different results; the least-likely result would be about 0.5% and the most likely about 12.5%. This is a poor way to model the sheer number and concentration of workers employed in state jobs, where there is one HUUUGE department (education), a few very large categories (state patrol, corrections, welfare), a pile of minor agencies (revenue, parks, transportation) and a whole massive list of so-small-you'd-miss-it agencies (governor's staff, state auditor, veterans administration). I might be able to model this accurately with a 5d distribution, but it would be kind of ugly — for one thing, state agencies vary. What we in Washington call the Department of Licensing, California calls the Department of Motor Vehicles. Likewise, Washington has a Liquor Control Board because the state runs the liquor stores; New York does not.

Also, take names. It'd be difficult to come up with a list of potential NPC names of any length and model it to a multiple-dice probability distribution. A 5d distribution only has 26 distinct results, far too few for a realistic list of names. It'd be much easier to use a spreadsheet (or percentile dice) for a list of that names — dozens if not hundreds of names, ranging from a maximum of 1% to a minimum of 0.001%.

It's also a lot easier to model age as a percentile-dice distribution, too, because age distribution is not a bell curve — it starts high, dips and peaks with baby booms, and drops like a rock starting about age 50.

Would you mind if I used percentile dice for some of these random tables, or do you insist upon d6?

[Fish]

Wait: it just occurred to me (belatedly) that I could get a larger number of discrete results by rolling 6 dice in sequence (46,656) than rolling 6 dice and adding them together (31). I don't need percentile dice after all; I just need a creative way to display the table.

[Fish]

All right, I've been thinking about how a more complete randomizer will work, and in particular, in which order the results must be generated for maximum realistic action.

Ethnicity can be determined randomly by the existing charts.

Age can be generated randomly, either from a chart of "everybody of all ages" or a chart of "all people aged 16 or more," depending. Age isn't dependent upon other factors.

Gender can be determined randomly from Age, since women tend to live longer than men.

Using Ethnicity, Gender and Age, you could generate a modifier for Education, which gives you the proper modifier for determining your Wealth.

HT can be generated from a combination of a random table (to determine your baseline), your Age and your Wealth. As you get older your HT is likely to drop; and wealthy people are likely to have better nutrition than the poor. The HT charts can include random HT-related diseases (Unfit, Terminally Ill, Hemophilia).

From your Ethnicity and Gender, you can generate a baseline Height, and adjust for factors like HT and Age. The chart can include the (suitably rare) chance for Gigantism and Dwarfism.

Weight can be generated randomly from Height, using Age as a modifier. The chart can include a chance for Gluttony.

ST can be generated randomly, and adjusted for Height, Weight, Gender, and Age. This chart can include the possibility for ST-related disorders (Lame, One Arm, Bad Grip).

IQ and DX can be generated randomly and adjusted for Education and Age. The charts can include IQ-related disorders (Delusions, hallucinations, seizures) and DX-related disorders (Klutz, Bad Back).

Hair Color and Eye Color can be generated from Ethnicity (and possibly modified by Gender and Age — for instance, because some women dye their hair).

Appearance could also be generated randomly, I suppose.

Of course, that assumes that a randomizer this comprehensive is even necessary or desired. :)