For those of you that are not aware, Torchbearer is my game of mythic fantasy, in a world where symbols and themes are obvious, tangible, and powerful. Three threads most relevant to this discussion (nonessential):
The First Torchbearer Thread
Thoughts on Mechanics
Character Death and the Metagame Conflict

To summarize, Torchbearer characters have some number of Traits. Conflict occurs when the goals (Goals) of two characters are mutually exclusive. The player that wins the conflict decides what occurs.

Mechanically:
Characters count appropriate Traits. These may grant or take away one die from a character.
Then the resulting quantity is multiplied by the character's Contrast (a deprecated term, but useful nonetheless; Contrast indicates how "good" a character's positive Traits and how "bad" a character's negative Traits are, at the moment).
The difference between pools is found. The character who had more dice has the Advantage.
And the difference is multiplied by Scene Contrast to find the number of dice rolled.
Torchbearer dice work like six-sided World of Darkness dice; 4 and above is a success and 1 cancels successes (but is not unusual in any other manner). If at least 1 success appears, the character with Advantage wins narration; the other otherwise. The winner must narrate consistent with one of the Goals.

This is a bloody mess; there are numbers flying every which way, too much math for this game to sustain.
Characters also, incidentally, have a quality called Myth, which is ranked on a ladder and means more or less "immunity to Conflicts with beings of lesser Myth", as well as indicating a level of 'over-the-topness'.


I have two ideas for a solution:
Zeroth: This happens regardless; implement a 'degrees of success' scale that actually responds to the number of successes rolled, so that exceptional rolls have a bit of extra flavor. A math request: The numbers I give for the result scale are off-the-cuff; I'd like to have a scale such that each higher level is about half as probable as the one below, and after a few levels can just keep going up incrementally. At the incrementality point, it doesn't matter as much what the probability patterns are.
First: Throw out personal Contrast, leave in only Scene Contrast. Takes out two potential multiplication steps and some complex power-games, but is still a trifle mathy.
Second: Cut it out completely; Environmental Myth instead takes the place of Contrast; it scales results up and down on the 'result scale' instead of changing the dice mechanics. It affects the results of Conflicts in this way: the rank of a Conflict result is shifted as many steps as the difference between the Myth of the Conflict and the Environmental Myth (lower-Myth Conflict > lower-myth result). So characters of low Myth in high-Myth Environments will rarely do anything very amazing.


The Result Scale: The number of Flames rolled on a result indicate its "intensity" in relation to the characters involved, and its enduring, not-necessarily-related consequences. (The Myth scale tops out at Mythic, incidentally, and Flame is a term I use to mean 'success'.)

Fewer than Zero Flames: A tear in an ocean, a result two levels of Myth below the parties involved. The Goal cannot be completely fulfilled. (This can only occur by Environmental Myth staging down a result)
Zero Flames: A trivial result, a level lower than the parties involved. The Goal can be fulfilled, but with some flaw.
One Flame: An unremarkable result, the same level of Myth. The Goal can be fulfilled.
Three Flames: A significant result, one Myth rank higher. Six Flames: A surprising result, two Myth ranks higher.
Ten Flames: A staggering result, three Myth ranks higher.
Fifteen Flames: A shattering result, four Myth ranks higher.
For each additional 5 Flames, and for each level a result is pushed past Mythic, raise the Myth rank of a consequence or add a Heroic consequence.

Example: Fjalar Two Hands in Darkness and Árichesja Raven-Swallows-the-Sun are trying to kill or maim each other in a town square; each has permission to narrate the death of the other. Their myth is Extraordinary (the second-lowest level of five). The decision favors Árichesja; this is how the result might be narrated:

Fewer than Zero Flames: Árichesja (hereafter known as Ari) throws a wild, uncontrolled punch at Fjalar. He moves out of the way, but this puts him in a slightly uncomfortable position. They are pulled apart by bystanders before they can do anything further. (This is not even an Ordinary result.)

Zero Flames: Ari gives Fjalar a good hard punch in the stomach, and another blow to the temple, which knocks him out. She gets dragged to the gaol by guards for the day, and fined for 'rowdy conduct.' (Ordinary.)

One Flame: After a lengthy knife-fight, Ari cuts deep down Fjalar's sword arm with her boot knife; his blood writes her name in the sand as his life spills away. (Extraordinary.)

Three Flames: Ari slices apart a kebab vendor's booth with a thrown knife, sending the booth's contents flying. Another well-placed knife throw sends the smouldering contents of a charcoal brazier flying at Fjalar's chest; he is too stunned to block it, and sustains terrifying wounds. (Legendary.)

Six Flames: Ari throws her knife into the root of a fountain, which shatters and tears into Fjalar with its flying, water-trailing shards.

Ten Flames: Ari throws her knife into the sun and they continue the fight barehanded. A few moments later, drops of the sun's flaming blood fall onto Fjalar, who dies screaming. (Heroic.)

Fifteen Flames: Ari throws her knife into the sun; it falls down and destroys much of the city square, including Fjalar. (Mythic.)

Twenty Flames: Ari kills the sun when she knocks it out of the sky. It does not rise again until after the next new moon. (Mythic with a Heroic consequence.)

the system certainly seems to work if you replace all the Contrast steps with rescaling using Environmental Myth, so the only real question is: does dropping Contrast change the feel of the game? I mean, here, the feel you are looking for; to me, the 2nd option seems very mythic and I don't notice a difference between the two options (or making no changes at all) other than mechanical complexity.

I can't remember who it was (Mike Holmes?) who posted a list of the average person's level-of-dislike for math in a game... I remember it started with simple matching, going up to addition, addition+subtraction, and multiplication. multiplication was definitely Bad on this scale, in terms of what players were willing to put up with. this would mean that players would be less adverse to the second option.

Wasn't me, John, but I remember that post.

First, I'd say that I'm not sure that for your game that the current system isn't good as is. There are a lot of clarifications that we need.

Characters count appropriate Traits. These may grant or take away one die from a character.

Then the resulting quantity is multiplied by the character's Contrast (a deprecated term, but useful nonetheless; Contrast indicates how "good" a character's positive Traits and how "bad" a character's negative Traits are, at the moment).

The difference between pools is found. The character who had more dice has the Advantage.
And the difference is multiplied by Scene Contrast to find the number of dice rolled.

I remember that thread too.

Issues:
The added layer of complexity lent by Contrast is interesting to me in a mathematical way. But I don't think it really serves any thematic purpose in its current state, and to make it do so would require an unnecessary proliferation of rules. So my better design sense tells me to take it away.

Attempt at clarification:
Traits have no range; they are either + or -, that is all. On the other hand, the number of Traits that can be brought to bear on a particular situation is theoreticaly unlimited.

The range on Contrast is (artificially restricted to) 1-9.

I object to the multiplication not because it's difficult or cumbersome, but because it breaks the elegant and obvious Trait=die symmetry. If I could come up with a way to preserve Contrast while at the same time removing it from resolution, I would do it at the drop of a hat. Contrast is similar enough to Myth, though, that simply lifting it out of resolution simply won't do; they would become redundant.

Attempt to clarify old conflict resolution process, with result scale added:

• Two characters have conflicting Goals; a Conflict results.
• Each character counts up his relevant Traits, cancelling positive Traits with negative Traits. This is his 'pool'.
• Each character multiplies his pool by his Contrast.
• The difference between pools is found.
• The difference is multiplied by Scene Contrast to generate the number of dice to be rolled.
• The character who had the larger pool in step 3 rolls the dice. He wins narration if any uncancelled Flames are rolled; otherwise, the other character does.
• Count Flames to determine the Myth rank of the event that results. Narration must be consistent with one of the Goals.


The difficulty with this lies not only in the multitude of steps, but also in that there is no obvious and easy way to keep track of the many fluctuating numbers. The revised procedure would look something like this:

• Two characters have mutually exclusive Goals; a Conflict results.
• Count up Traits that support each character (undermining one character is equivalent to supporting the other).
• Traits supporting opposing characters cancel each other out, one-for-one, until only one character has Traits remaining (This should, in play, blend together with step 1).
• The character with uncanceled Traits rolls that number of dice. He wins narration if any Flames are rolled.
• Count Flames and look up result on the scale.
• Adjust for Environmental Myth to find the Myth rank of the resulting event. Narration must be consistent with one of the Goals.

Here the Trait totals are never multiplied, so a Trait can be straightforwardly represented by a die.

There are other problems with the system that you haven't mentioned. First, if two characters have similarly sized pools (even if they are huge in size), you're not going to get very dramatic results, because the amount of dice to be rolled is going to be small (and the resulting number of flames, as well).

Personally, I think many of your contrast/multiplication issues could be solved by the use of tables like the one for number of flames rolled. What if you had a level of "Conflict Myth" (need a better term) that was determined by examining the Myth level of all entities involved in the conflict? So a mortal is challanging a demi-god on the site of a legendary battlefield. This means the resulting Conflict Myth is X (perhaps you just average all the factors or have a base Environmental Myth that gets shifted by modifiers). Conflict Myth could be the base result, the result that would normally occur, without Fate. Then the resulting number of Flames or Waters rolled serves to push the result up or down the scale. So two Flames would give you a resulting Myth of X+2 (a lot like Fudge, actually, without the funky dice).

What do you think of something like that? It's definitely simpler, but I don't know if it still incorperates all the factors you want or supports the same kind of feel.

Another thought I just had...

Say a mortal has a Myth level of 1 and the demogod has 5. After adding in all the factors, the resulting Conflict Myth is 3. Now, because of the way Torchbearer character creation works, it's likely that both characters might have the same number of applicable traits in the conflict, and since we're not multiplying by character Myth, how do we figure out who rolls and give the demigod a proper advantage?

How about this: a system where either player can roll.
-- If you win, you can narrate your victory or your opponent's loss (they narrate whatever you choose not to).
-- If you lose, you can narrate your loss or your opponent's victory (and they narrate the other).

Works like this:

Say the Demigod decides to roll, and they have 3 applicable traits. Because their Myth is 2 levels higher than the Conflict Myth, they roll 3 dice and add two Flames to the result.

Say the mortal decided to roll, and they have 3 applicable traits. Because their Myth is 2 levels lower than the Conflict Myth, they roll 3 dice and add two Waters to the Result.

Not a perfect system, but I thought it might give you some fresh ideas to consider.

Hm... you've pointed out one issue; to simulate actual mythic-stuff-happens-at-mythic-places conventions, I should have the effect of Environment reversed, or have Place and Story act at cross-purposes (this only makes sense if you posit that any Entity outside a Conflict can influence the mythicity of its result... that could lead to interesting results.)

It's important to preserve the rule of "Only Conflicts between Entities of like Myth." On the other hand, allowing result scaling by Outside Factors rather than by environment alone would be a sensical and more general way to allow Entities of another kind to influence the results. The rule would be something like, "An Entity of greater Myth than the participants in a Conflict, who is nonetheless somehow involved in that Conflict, can move the result rolled up or down the scale a number of ranks equal to the difference between its Myth and that of the Conflict's participants." Then this makes it easier for the hero carrying the Gorgon's head to turn a city to stone, or for Frodo to destroy the Ring at Mount Doom and nowhere else.

You point out another interesting condition involving the number of dice rolled. I don't know yet how to resolve that, but it's possible that making Conflicts all-opposed (everyone rolls his own pool, the winner gets to narrate and his dice determine magnitude) could not only simplify the process, but also alleviate this.

I'd also rather not give Water any identity other than "the antithesis of Flame". The term dilutes my metaphor badly enough without mechanics sloshing around underneath it. But I can see how you could extend the Conflict rules to handle Conflicting beings of differing Myth; you simply require that the result narrated has a base Myth of the Entity the narration favors. Then the demigod has no math-mechanical privilege over the mortal, but he does have cooler narration privileges.

So, a decision that I've come to is this:
Use the cancelling mechanic as before, but each cancelled pair contributes to the magnitude of the result as though it were a Flame. Then closely matched opponents will still tend to have dramatic Conflict outcomes, but it reduces the number of dice rolled, which consequently makes the result more likely to swing in either direction.

Jonathen, looking at your idea again:
This seems to require that both players narrate at every Conflict juncture, which is a great departure from the old mechanical core of Torchbearer, the idea that the Conflict determines who gets to tell you what happened.

This thread is giving me a lot of ideas that I could apply to Exquisite, though, in addition to clearing up Torchbearer. Thanks, guys.

One last thing, a snippet of setting:

Among the warrior castes of the Tarag Thaani, there is a tradition that has preserved their skills through centuries of peace.
When a woman concieves, her husband, or her male next of kin if she has no husband, begins to forge a sword and build a bow. They are left uncompleted until the day the woman gives birth; then the man goes into seclusion with her and completes the weapons in her presence, and assists her with the birth. For three turnings of the moon the child is never separated from these, the sisterbow and brothersword. In a ceremony at the end of those months, the weapons are used to draw a tiny bit of the child's blood and anointed with it; this awakens their spirits and binds them mystically to their owner.
The child trains with these weapons only, for the rest of his or her life, and they are traditionally buried together. In the hands of a great warrior, the spirits will become stronger and change their shapes, eventually becoming perfect extensions of their master's will and representations of his soul.

It is a great honour to have an ancestor's weapons exhumed and reforged to create the weapons of one's child, or to have one's weapons completely destroyed at one's death. This comes from the habit of weapon spirits who are very close to their masters; when he dies, they will collapse into dust. Spirits that linger from the weapon of an ancestor are often wise and experienced; they advise the child as any close elder relation would do.

Some even reforge their own weapons for their grandchildren and become shamen or mendicants; these weapons often become heirlooms and are passed down through generations, becoming filled with different weapon spirits that each apply their own perspective to a situation.

Shreyas Sampat wrote: Use the cancelling mechanic as before, but each cancelled pair contributes to the magnitude of the result as though it were a Flame.

Jonathan Walton wrote: Sweetness. That's an option that I hadn't even considered, but a very insightful one. So then, pairs basically serve to determine "Contrast" don't they? The dice that get rolled determined the degree of success or failure (complete success, partial success, complete failure, partial failure, etc.) but the pairs, the dice that don't get rolled, decide the magnitude of what has occurred.

So you could roll a complete draw, but have Dragonball Z style special effects mowing down forests and cities. I like it. I think it will suit Torchbearer pretty well too.

Does the player who rolls (i.e. the player with the most dice) always get to narrate, or do they just narrate victories?