One of the classic asymmetrical curves is the Pareto Curve, which has been popularized as the 80-20 rule, where in management 80% of your problems are caused by 20% of your people; or 20% of your people do 80% of the work. Helpdesk ticket closure times are another good example, where the bulk of the them are closed in a fairly short range of time, but others, especially those requiring equipment replacement, extend the overall time range far to the right. Back when I was at GE Fleet Services, I would check the 3rd quantile, or the mark at which the technician had closed 75% of their tickets, rather than their average time to close.
3d6 distributions give me the opportunity to illustrate that you don't need to stay with the same size die when creating a long tail. You have a 83.8% chance of rolling less than a 14 on 3d6. So to create the long tail, I would roll another die whenever the 3d6 is 15 or more and add it to 14. If I use a d6, it only gives a 3-20 distribution, not much of an increase over 3-18. By using a d10 instead, I create a 3-24 distribution range, producing a significant tail at the high end.
I used anydice to look at the probabilities. Their calculator will allow you to see the percentages for any combination of dice.
I haven't come up with as simple a way to do a long left hand tail, but I don't have a use case either. If I get one I'll probably just flip the scale at the bottom and read from right to left.
As I said, the use case I identified is random NPC levels, I'd be interested in hearing of any other use cases you might come up with.
No comments:
Post a Comment