The following are the statistical assumptions we work with in order to generate statistical conversions:

Muscular Strength:

According to a study 50% of men in their twenties are able to squat over 185 lbs and 10% over 250 lbs. For benchpress the corresponding figures are 135 lbs and 170 lbs and for deadlift 210 lbs and 245 lbs.
source

After some calculation and a little bit of rounding off, we get:

Right now this is just some basic notes on conversion. I’m not going to tackle the whole shebang yet:

Attribute distribution:
μ=10; 8-12 is considered human norm. Above 16 is exceptional; above 20 is superhuman. So we may say that 20 is the absolute limit for humans. I’m going to say for now that the “absolute limit” is 5 standard deviations away, so σ=2. I’ll do some comparison on lifting power and stuff later.

Primary attributes:
ST = attributes/physical/strength
DX = attributes/physical/dexterity
IQ = attributes/mental
HT = attributes/physical/health

Derived values:
Basic Lift = ST² / 5 .: μ = 20lbs (over head, one hand, one second)
Max press = 8X BL .: μ = 160lbs

So I’ve decided to do some initial conversions, just to test some stuff out. Currently the XML schema (available on the downloads page) basically only covers attributes, which makes this task a little easier.

Since the actual math of conversion is not yet decided on, I will be doing this conversion using the statistical method I suggested here. Also, I’m going to be ignoring systems without numerical stats for the time being.

d20

D20 attributes range from 0 up, with “standard” stats being rolled on 3d6. We assume that the 3-18 range approximates an appropriate distribution curve with μ=10.5 and σ=3

Therefore 10 and 11 in d20 translate to -1/6 and +1/6 respectively, and an increase of one in a d20 stat represents an increase of 1/3 in RPGVM.

Conversion table:

Converting d20 to RPGVM

The attributes of RPGVM get filled like so:*
SOCIAL = Charisma
MENTAL
  KNOWLEDGE = Intelligence
  MEMORY = Intelligence
  PERCEPTION = Wisdom
  REASONING = Intelligence
  WILLPOWER = Wisdom
PHYSICAL
  AGILITY = Dexterity
  HEALTH = Constitution
  STRENGTH = Strength
SPIRITUAL

Converting RPGVM to d20

The attributes of d20 get filled like so:*
Strength: STRENGTH
Dexterity: AGILITY
Constitution: HEALTH
Intelligence: average of KNOWLEDGE, MEMORY, REASONING
Wisdom: average of WILLPOWER, PERCEPTION
Charisma: SOCIAL

Ex Machina (Tri-Stat)

Character stats range from 1 to 16. Adult human average is 4, so μ=4. “Maximum human potential” is 12. From the description in the Ex Machina sourcebook, it seems like the scale is weighted toward the larger numbers, in other words that the difference between 2 and 3 is not the same as between 5 and 6. I’m going to say that σ is 1 up to 4 and 2 after 4.

Conversion table:

Converting Ex Machina to RPGVM

The attributes of RPGVM get filled like so:
SOCIAL
  APPEARANCE = Soul
  CHARISMA = Soul (Charisma)
  REPUTATION
MENTAL
  KNOWLEDGE = Mind
  MEMORY = Mind(Memory)
  PERCEPTION = Mind(Perception)
  REASONING = Mind(Intelligence)
  WILLPOWER = Soul(Willpower)
PHYSICAL
  AGILITY
    BALANCE = Body(Agility)
    MANUALDEXTERITY = Body(Manual Dexterity)
  HEALTH
    ENDURANCE = Body(Endurance)
    IMMUNESYSTEM = Body(Immune System)
  STRENGTH = Body(Strength)
SPIRITUAL

Some of the above assignments may be modified by things such as the “Less Capable” defect. As a quick call, the stat is reduced by half the amount of the defect’s penalty (simulating a stat drop equal to the defect’s penalty).

Converting RPGVM to Ex Machina

Within any of the three main stats, if one sub-stat is significantly lower than the others the “Less Capable” defect should be automatically applied and the sub-stat adjusted accordingly for the sake of averaging.

The attributes of Ex Machina get filled like so:
Body: PHYSICAL
Mind: average of KNOWLEDGE, MEMORY, PERCEPTION, REASONING
Soul: average of APPEARANCE, CHARISMA, WILLPOWER

Conclusions

This is by no means the end of the conversions. In fact, each system ought to have its own section and discussion thread. This article merely examines a couple of systems (more when I have time) and the way they might be converted. Please feel free to add to this collection with your favourite system.

* when an attribute group has a value assigned to it, that value is passed down to its children. Likewise, when the score of an attribute group is referred to, it is obtained by taking the average of the child attributes. This is sometimes useful in situations where the weight of the group, not the individual values, is to be considered.

Edit 2006-04-30 18:58 to be one possible method rather than the current method
Edit 2006-06-17 06:23 to use slightly more accurate means and deviations

One current idea on ability score conversion is to start out at zero for human-average scores, and rate characters on the number of standard deviations of difference. For brevity’s sake, we refer to the mean (average) score as μ and the standard deviation as σ. Also, I am ruthlessly rounding to the nearest .25σ. This method is conveniently called the statistical method.

For instance, let’s say the average human adult has a maximum dead lift weight of 95kg. This means that μ=95. And let’s say that the standard deviation on dead lifting is 11kg [*]. So someone who can lift a maximum of 120kg will have a “lifting” score (or equivalent) of +2.25 (120 ≅ μ + 2.25σ). Someone who can lift a maximum of 60kg has a score of about -3.25. Note that with these values the minimum possible strength score is -8.75 or so, indicating a character (presumably an incorporeal one) who cannot lift any weight whatsoever — but there is no upper limit to the ability. (source for figures)

So to determine the score equivalent to a given system (assuming there is a directly equivalent attribute) look at whatever charts the system provides for liftable weight and simply convert it to positive or negative numbers of SDs. Then you have the RPGVM value.

D&D conversion example:

Tordek has a strength of 15. According to the chart on page 62 of the Player’s Handbook this means his maximum liftable weight (double “max load”) is 400lb. The difference between 400lbs and the human mean of 210lbs is 190lbs, making 400 about μ+7.5σ. Thus Tordek’s RPGVM lifting score is +7.5. Note that this does not match the variance on 3d6, where the score would be μ+1.5σ. Also, +7.5 is an incredibly unlikely score, pointing out a flaw in my data, in D&D, or both.

The same principle can be applied to most stats, especially physical ones. Mental and social skills are much harder to quantify, and spiritual skills can only be judged by comparison between systems as there is no real-life analog.

Please feel free to add any suggestions for different scoring methods or nitpicky details about statistics. I will update this topic as the discussion evolves.

-Dave

* Note that standard deviation here might not really apply, as something like weight-lifting would probably follow more of a Poisson Distribution-like curve. However, we recognize that not all gamers are math geeks and so we’ll stick with the better-known Normal Distribution’s bell curve.

So a little while ago Torsten and I came up with the idea of having a way to store RPG character information that was system-agnostic. The idea is to behave sort of like Java, where any system with a VM will be able to run any bit of Java code, because it’s all being run in a VM. So imagine this:

You create a character for Shadowrun. Then you decide to switch to D20 modern, but you want to keep the same character. So you search the web and find a Shadowrun->D20 converter. Cool. Then a friend picks up an old CP2020 book at a flea market and says “dude! We should totally play this!” So you hit Google again and eventually manage to find a converter. Fine. But then your GM goes on a minimalism kick and wants to try out FATE. So you look for another conversion. Then someone else says Ex Machina does Cyberpunk really well, so you want to try it out. Then you get so frustrated with the whole process that you either impale yourself on your bardiche or you settle for a system you might not necessarily want to play because, hey, it’s better than running character conversions every few months.

What we want to do is to develop an intermediary “bytecode” character storage format that will allow for transfer between systems. Ideally, all someone would have to do is write a conversion from this format to their favourite game, make it available to the world, and anyone could do a conversion from any system to that one. In the above example I mentioned five different systems. In order for there to be enough conversions to move a character from any of the five to any of the others you would need 20 different conversions (counting conversions both ways). But if there were an intermediate format that can be converted to and from any of the others that’s just 10 conversions. With six systems it’s 12 conversions instead of 30. With seven it’s 14 instead of 42. Yes indeed, we’re talking about 2n instead of n²-n. And the way I see things, conversions will naturally be two-way, reducing the complexity even more.

Plus, this project aims to be (of course) open-source. All the guts of things will be open for discussion and debate. So if you want to create a web-based converter, that’s cool. If you want to write a Java app or tell us about a new paper-and-pencil way of dealing with things, awesome. And certainly, if you have a favourite system conversion to submit, please do let us know!

Speak to you soon,
- The Management