If this is the wrong thread area for this, let me know where it belongs, please.

i recently picked up Sorcerer at my local store and i'm enjoying the read so far.  Just before bed i was reading Verge and this mechanic (mx) came to me.  i'ven't had much time to work on this, so be gentle.

Pertinent Info

- Abilities are measured in Experience and Training.  Experience comes from DOing, Training comes from STUDYing.
- Training determines the size(d4, d6) of the dice to roll.  Experience dictates the number of dice to roll.
- Experience gives consistent results, Training exposes character to more information, and is more adaptable

Process
1. GM sets a number as a Penalty/Difficulty/TN
2. Player rolls the appropriate number and type of dice
3. Player counts the number of matching dice
4. Subtract the penalty
5. Remainder indicates the level of success

Sample
Jeff says the Penalty is 1.  Bob rolls 4d6 and gets 2, 3, 3, 4, making 2 matches.  Remove the Penalty to make for 1 Success.

Problems
1. Requires having lots of dice and of different types
2. "Smaller dice are better" is counter intuitive (to me)
3. Is much like many other systems
4. Might not scale well, but isn't appropriate for game that require massive scaling anyway, system matters

Benefits
1. Good for narrative games
2. Simple
3. Fast
4. Represents broad training vs. real world experience well
5. Can use a wide variety of meta mx

Help Needed
What are the pitfalls with this system and how might i remove them?
What systems do the same thing?
What are the probabilities involved?
What does it do well?

Thanks in advance!

Peripheral Info

Virtual Dice
As presented, Attributes are missing.  i would add them as Virtual Dice, a non-existent die that you can use to make a match. For instance if Bob had a Virtual die of 3, he'd have 3 matches in the above example. To be useful in conjunction with a smaller die, Virtuals should be 4 or lower.

Or instead of changing the number on the virtual die, you could add additional virtual dice.  A weak attribute would have a one virtual at 5, then improve to be 4 and 5, then 3, 4 and 5 and so on.  Perhaps there is only one virtual die, but it could be a 3, 4, or 5 as needed. If d8 is the biggest die in the game, i'd right that as v3, meaning that die is showing all numbers from 3 to 8, inclusive.

SDK
1. TN/Penalty scales with the availability of dice and small dice, lower number might work best
2. Likelihood of a match is really low if the Penalty is more than half the number of the die
3. It doesn't have to represent training and experience, it could also be (Skill)d(Attribute)
4. Straights could also count, 1, 2, 3, 4 would be equivalent to 4 matches.  Straights are much more likely, perhaps that could represent a special trait, or you could throw out the match thing and just use "in a row".
5. Instead of making all the dice the same, it could be... add a d4 for this, 2d6s for that and a d8 for the other thing. Mixing die types might mean fewer matches, but could represent something specific in your game

I love games that separate training vs experience. Its very good for "realism". That being said, does your game seek to simulate accurately the practice and use of skills, or does it seek something else?  As a system, I think this is supportive of some things, it strikes me as making opposed rolls very simple to arbiter (did I get more matches than him) which strikes me as great for CR. The problem with scaling is really that you quickly ascend toward perfect successes as you roll more and more dice.  Further, does having more than 1 success after the penalty is assessed mean anything?

This isn't a game yet, just a mechanic.  But i would like for it to be a part of a system that, yes, expresses that experience and training are different.  Realism in the simulationist meaning is not the goal, so much as a feeling of "yeah, that feels right".

CR?

Yes, the number of remaining matched dice gives you the degree of success.

Welcome to the forge, Justin.  And thanks for posting on my otherwise ignored thread.

sorry CR is conflict resolution:

Conflict resolution (from the glossary)
A Technique in which the mechanisms of play focus on conflicts of interest, rather than on the component tasks within that conflict. When using this Technique, inanimate objects are conceived to have "interests" at odds with the character, if necessary. Contrast with Task resolution.

The questions I now have are:
How many successes are necessary to accomplish tasks?
(1 for basic success, 5 for perfect?)
How many dice are present for normal rolls?
What is a "normal" penalty (in your mind) and for what types of situations?
(1 for annoyances, 3 for having an arm off and trying to drive a stick shift in a torrential downpour?)

i like to leave those things open for whoever will make the System, this is just the process the system will use.  Keeping the numbers nice and low seems to be the best idea to me.  The penalties can be higher if "straights" count. likewise, pools could be smaller and dice bigger.  Without straights, a penalty of 2 would be really tough, one success should be a nominal result, 2 would be pretty good and 3 would be excellent.