So, I just got REIGN by mail earlier today, and I'm totally in love. Beautiful game, beautiful book, etc, etc. I have only given in a quick read-through thus far, but it seems like a very solid game.
So, why am I writing this on the First Thoughts forum? Well, recently I've mostly been buying books to mine for ideas, either plot ideas for current or upcoming campaigns, or ideas concerning how you handle specific mechanical issues when designing a system.
The reason I brought REIGN was basically, that I'm building a system which supports a wide variety of playstyles. Everything from post-apocalyptic scavenging, Firefly-ish smuggling hijinks, courtly intrigue and so on. The main thing is that most, if not all, of the supported playstyles sooner or later (much sooner in the case of courtly intrigue), evolve into a campaign where the PCs are really, really, powerful. I mean, who haven't played in a fantasy campaign where the players gathered so much money and influence that it made little to no sense for them to continue adventuring? What's this got to do with REIGN? Well, I bought it to look into a system for handling faction vs. faction conflict, management of domains, etc. I don't really have huge amounts of experience with this type of play or the rules supporting it, so I thought I'd just come here and ask for help.

So, what do I need help with?
Well, in REIGN, the system works on an exponential scale. A Company with Might 4 isn't the same as four companies with Might 1; but rather Might 4 = 2x Might 3 = 4x Might 2 = 8x Might 1. This seems to be a great way to handle it, as it enables you to handle mercenary bands with the same system as you handle a duchy, a nation, an empire or even a federation of planets. The reason this works in REIGN, as far as I've understood, is that the One-Roll Engine works on a similarly exponential scale. You're vastly better off if you roll a pool of 4 dice, than you are if you roll 3.
The thing is, running a faction isn't the main point of my game, and the system is mostly finished (needs some tuning, of course, but it's mostly all there). Why is this an issue? Well, because I can't use a system similar to REIGN, because my basic resolution system doesn't work on an exponential scale* - it's basically a roll-over DC mechanic. Thus, I can't really have a system where Might 1 is a quarter of Might 3 and an eighth of Might 4, because it isn't supported by the system (and establishing one thing in the fluff and then not supporting it with the system, sucks). Thus, I need to find a way of handling mercenary companies or bandit bands on the same scale as as nations - some of them rivaling China in size.

Some different options include:
-A huge scale. This might prove highly impractical, and due to the static range of the randomizer (see footnote below), it would make it impossible for, for example, Vietnam to win a war against the US.
-A slightly exponential scale. This might be practical. I'll have to adjust the scale to make it impossible for a band of mercenaries to defeat an empire in open conflict, but possible for a smaller state to beat a larger one by the use of situational modifiers.
-Some kind of scaling mechanic, where Might 10 for a type D faction, translates as Might (10/X) for a type C faction, where X is either 2 or 10 depending on what seems reasonable. A type D faction might be a mercenary band, a type C faction a duchy, a type B faction a nation and a type A faction a federation of planets.

Thoughts? Other ideas?

*if it makes any difference, the particular of the system is: You play a card from a hand of five tarot cards, add your skill/generic value and try to beat a difficulty level determined by the GM by drawing a card from the deck and adding a difficulty number.

My main exposure to an exponential system attempting to really use its mathematical features and interact with the "real world" is from DC Heroes, aka the Mayfair Exponential Game Systems (MEGS) I think.

Short example: If I have run 20, and need to go distance 25, it takes me 5 time. (They plugged in real numbers for these things.

So why not have a function where you use the values to determine odds. EG Acting Value-Opposing value correlates to a Difficulty Number?

If you keep your exponential scale shallow, this doesn't get too bad that fast. MEGS, like Hero Damage Classes uses a scale were you double every step. For your type of scale, you might want something that double in size every 4 or 5 steps. In the end you want the difference between a force and a force twice its size to give you room to hit as many Difficulty numbers along the way, if that makes sense?

either

1) draw more cards
2) play more cards

for example, China (3 might) vs. McDonalds (5 might) vs. The Forge (1 might)

1)  China draws 8 cards, McDonalds draws 10 cards, The Forge draws 6 cards. (better hands...)
2)  China plays 4 cards, McDonalds plays 6 cards, The Forge plays 1 cards