I'm working on resolution rules for my first rpg design. I haven't even decided how to call it yet. To a considerable extent, it is inspired by the "Black Company" books by Glen Cook.

There are players playing characters and there is a GM.
Each player (including GM) has a pool of dice, they spend dice for conflicts. Each gets a number of dice in the beginning of a reward cycle - "Mission".

Reward cycle consists of several scenes.

Scene: It's decided what characters are going to do, GM setups a scene (NPCs, surroundings etc) with some basic conflict embedded, everybody generally knows what this conflict is.

To decide outcome of a conflict sides choose a number of dice they want to spend (they lose them regardless of outcome), they roll them and compare highest, then second highest and so on - they only decide who won. Potential Effects (from IIEE) are decided before the roll, winners Effect happens.

Dice are six-sided.

Players-with-characters can have additional dice if they risk their character lives. Each of them can get up to six of them - for that, before the main roll, player rolls the Risk die - to win it player should get a number greater then number of additional dice she got. Like if she got 2 additional dice, 1 and 2 are failures.
If player fails risk roll, her character suffers a fatal wound.

So, the part of the rules in question is procedure of choosing a number of dice.

The idea is that scene is played like "freeplay" and when something happens that influences scene conflict outcome, player that initiated these happenings puts a die or dice (from her pool) on the table.  These are limited to player's character actions only when events involve Risk, otherwise it can be as well friendly NPCs and "good luck".
To block the event, opposition puts on the table equal amount of dice, to say "instead" opposition puts more dice. This is all about Initiation and Execution. Each player can put dice for single such "task resolution" only once.

Such "task resolutions" happen once or more times. Risky parts involve a Risk roll. When there is no sense or possibilities to continue and it's time to resolve the conflict (GM decides), conflict roll is made, with all those dice opposing sides put on the table.

At last, a question. To spend dice for a conflict player (especially player-with-character) should make up something. Players wants to win the conflict and player (as everybody at the table) wants things happening in fiction to be credible and good.

In practice I get struggle inside the player - "I let some bullshit in the game or my character is more likely to lose". Player knows that game should be good whenever character loses or wins, but this way there will always be a considerable difference between players emotional state and characters emotional state.
And such difference is against my design goals.

What do you think about such "struggle"?

Have I done something wrong in rules or in play or maybe it's my and my friends "old-school RPG training" getting in the way?

I can tell you more about the game - I've just tried to keep the post short (failed though).

Thank you.

PS. If you think that something here is just stupid, I beg you say it straight :)
(maybe PM, to keep forums clean)

I love your idea and would love to see whatever more you might have written down.

Here's some questions and ideas to set up some brainstorming:

Should everyone get the same number of dice during each reward cycle?  I can see some groups and some genres giving more or less dice to the GM (or perhaps larger sided dice?).  Even in a not-very-gritty game, the second act should be difficult enough that the players feel all is lost. 

In Star Wars terms, what would be considered the reward cycles and missions?  For Luke, I am guessing it would be "convince uncle to let him to go to academy early," "recover the missing droid, "find a pilot to go to Alderaan", "escape the deathstar," "blow up the deathstar."  Am I getting the gist of the reward cycle here?

I'm assuming that multiple conflicts exist with the course of a single reward cycle (otherwise everyone would choose "all my dice" all the time!).  How does a conflict "come up."  You said the GM gives people an idea at the beginning of a new mission what the general conflict will be - does the GM establish certain conflicts initiated by circumstance and NPCs while the players establish certain conflicts established by their characters or character's reactions?

i.e. (more star wars) in the attack of the death star mission, would the GM say "the first conflict is piloting in through the enemy weapons fire, let's put out dice" or would the GM say "you are all in tight formation preparing for your attack run, what should our first conflict be?" - how much power does the GM have to establish conflicts, how much power does the player have to establish conflicts, in what order do conflicts get resolved, how brought or specific does a conflict need to be?

So, somehow everyone knows the conflict upon which they are bidding is "piloting in through the enemy weapons fire."  Who describes what they'd like to have happen first (the effect)?  Assume all players want the same thing, can they pool their dice (I hope not...yeck).  Who bids first?  Are the bids secret? 

When players use their risk dice, what do you mean by "failure."  If someone uses 3 regular dice and 3 risk dice, does that mean if any of their risk dice comes up 1,2, or 3 they suffer a fatal wound?  And what about multiple fatal wounds?

Coming up with who puts out their outcome first is probably important.  It sounds to me like you are thinking in order to put an outcome on the table, a player needs to start the bidding... Is that what you mean, or might you just have them put an "ante" on the table to take some responsibility for their outcome.  So if no one opposes that outcome, then that is the outcome that occurs right?  If a player doesn't like the outcome, though, they have to put out the same or more dice on the table and describe their alternative outcome, right? 

Your question about players playing it safe is a good one.  The big problem with sharing the GM and/or rules power of resolution is that players may well take advantage of their colleagues when they run out of veto power.  From my reading through the Capes goal and conflict resolution discussions, sometimes you have to just call your friend an asshat and move on.  On the other hand, I think we could devise a reward system (perhaps in terms of character development or the bigger plot development?) whereby players who "spend out" are rewarded - in that way it should encourage people to spend it all early.  Further, the purpose of the GMs dice should be to look around the table and see when the group really doesn't like what a player is doing..