Retention Revisited

[songqu88@gmail.com]

It's not unnatural or uncommon human behavior for the established elite to tend to be exclusive, and desire to maintain that status at the expense of newcomers. Look at the early 20th century American conflicts between the established "aristocracy" and the new millionaires. Well, look anywhere that such a thing occurred and you'll find that in more cases than not, the old rich families weren't very happy with no-names having comparable, sometimes even greater, wealth than they had.

[Anaris]

Or if the dukes intentionally give small oaths in order to do manual transfers after. I hate such micro-managing. (note: not the same as giving everyone decent oaths and then giving a supplement to units that got significantly damaged compared to others, or to finance a takeover unit or two per army, which is perfectly legit imho)

Hm. That gives me an idea for another possible metric to display for people looking for a realm to join: gold handed out to non-lords via oaths, as compared to gold transfers from lords to non-lords.

[Indirik]

"Average Oath Income"

("Average Estate Income" for the new system...)

[vonGenf]

"Average Oath Income"

("Average Estate Income" for the new system...)

Nice hint dropping you did there....

I wonder if the median income wouldn't be better? After all, if the problem is income inequality, the average won't tell you anything; the non-Lord average gives you a very specific information that may not paint the whole picture. In many realms, city estates could skew the average up, but could be reserved to a closed elite.

Would it be thinkable to code the Gini coefficient of a realm?

[Anaris]

Would it be thinkable to code the Gini coefficient of a realm?

Given that the one who would (most likely) be coding it hasn't a clue what you're talking about...probably not 

[vonGenf]

Given that the one who would (most likely) be coding it hasn't a clue what you're talking about...probably not 

http://en.wikipedia.org/wiki/Gini_coefficient

Basically, it is a measure of inequality of income. In a realm where everyone has equal income, G=0; in a realm where a single person has all the income and everybody else zero, G=1. It is independent of sample size for large N (so may be skewed in very small realms) and of average income (so poor realms could have a good score) and is easily calculated if you have a list of the incomes of the realms (the 3rd formula on the wiki is probably the one you want to use).

[Indirik]

Seems only marginally useful. If you have to send someone to a wikipedia page to explain the math behind a term that only a statistician has ever heard of, then it seems a bit pointless.

Median may be better. But again, how many people will understand the difference between average and median? I'd say keep it simple, and use terms that any player wanting to join will understand.

[Anaris]

Seems only marginally useful. If you have to send someone to a wikipedia page to explain the math behind a term that only a statistician has ever heard of, then it seems a bit pointless.

Median may be better. But again, how many people will understand the difference between average and median? I'd say keep it simple, and use terms that any player wanting to join will understand.

Well, if I were to implement the Gini coefficient, I'd probably just display it as "Income Distribution: Very Uneven/Uneven/Moderately Even/Very Even" or something like that.

[Indirik]

Would any realm really have anything better than "Uneven"? I'd imagine only the smallest, poorest realms would approach "Moderately Even".

[Anaris]

Would any realm really have anything better than "Uneven"? I'd imagine only the smallest, poorest realms would approach "Moderately Even".

I don't know; it'd be an interesting thing to investigate.

[Indirik]

So, fire up your SQL-fu and check.

[vonGenf]

Seems only marginally useful. If you have to send someone to a wikipedia page to explain the math behind a term that only a statistician has ever heard of, then it seems a bit pointless.

Median may be better. But again, how many people will understand the difference between average and median? I'd say keep it simple, and use terms that any player wanting to join will understand.

You may be right for the Gini coefficient, but I refuse to think that we can ask people to understand the difference between a liege and an officer, but not the difference between median and average.

[Vellos]

Gini coefficient would be interesting, but irrelevant for BM.

It requires a large sample so that the "top portion" is a meaningful group as is the "bottom portion." Assuming a realm of 30 nobles, comparing the top 10% to the bottom 10% is comparing 3 dukes to 3 new knights with no oaths. Comparing 2 samples of 3 is not very useful.

It could be interesting if set on a distribution of sorts with categorical labels, so that "comparatively equal" realms would get more.

But, better measures would be:
Average or median ratio of a lord's income to his/her knight's income (I think this would be an excellent measure, as it basically shows the oath shares being offered)
Ratio of gold given manually vs. through oaths (as Tim suggested)
A "haves and have-nots" Gini coefficient, wherein the income of the top 50% is compared to the bottom 50%

Economics is my course of study. If I think about this for a while, I can find us probably at least a dozen reasonable measures of inequality. And a way to weight and aggregate them. I might even color-code it. You have no idea how fun that would be for me. Give... me.... data!

[Chenier]

Nice hint dropping you did there....

I wonder if the median income wouldn't be better? After all, if the problem is income inequality, the average won't tell you anything; the non-Lord average gives you a very specific information that may not paint the whole picture. In many realms, city estates could skew the average up, but could be reserved to a closed elite.

Would it be thinkable to code the Gini coefficient of a realm?

I favor medians when it comes to wealth indicators. Averages are so biased.

Case in point, GDP per capita is higher in the US than in Québec. However, if you exclude the top 1%, it becomes higher in Québec than in the US. What does this mean? The US has more filthy rich people than Québec does, but unless you are among the 1% richest of your country, you are likely to live better in Québec than in the states.

All this to say, "averages" don't represent your "average" joe is people tend to imagine him (as they are rather judging the median joe with the averages).

My rant of the day against averages.

[Vellos]

Case in point, GDP per capita is higher in the US than in Québec. However, if you exclude the top 1%, it becomes higher in Québec than in the US. What does this mean? The US has more filthy rich people than Québec does, but unless you are among the 1% richest of your country, you are likely to live better in Québec than in the states.

But you speak French, so the standard of living drops significantly.

/nationalism