Here are the basics of my system, as is. I'm looking for advice on fine tuning some elements.

The Mechanic:

1d20 blackjack roll-under

The Attributes (range 1-13):

Build - overall strength and health
Agility - speed
Perception - the five senses
Intellect - pretty obvious
Bearing - "spiritual" presence

The Skills:

Skills function basically the same as GURPS. Every skill is linked to an attribute. So, something like Brawling might be Agility -3 +ranks. A character with an Agility of 8 and 4 ranks in Brawling would have to roll a 9 or under.

The Adjustments:

1) I'm thinking it might be better to have attributes range from 1-20. This way I can do attribute tests and have them be on the same scale as skill tests. Otherwise, I would probably have to add a constant to an attribute for a check. However, a constant seems artificial. In terms of the attributes relation with skills, if I increased the range from 1-20, I would then have to do something like halve the attributes when determining base skill values. Any thoughts?

2) I'm tempted to split Build into Strength and Health. The idea is that you don't have to be strong to be healthy. However, this would make Strength almost useless in a contemporary setting I have planned as physical strength is valued less (like in most industrial/post-industrial societies), and there really aren't many skills I plan to link with it. I could use suggestions on this one.

3) Also, I want to rename Bearing. It's used for the "magical" elements and social skills. These may seem like oddly different categories to lump together, but it actually works with the metaphysics of the campaign. Charisma sounds too social and Spirit/Soul sounds to metaphysical. I can't think of a good word to cover both elements.

4) Skill ranges. As you can see, it's possible to have skills that exceed 20. Now, this isn't a problem is penalties stack up enough to drop it down again. However, a player can still develop a character's skill to a near flawless level. Would putting a cap on skills be a reasonable solution, or would that just be too artificial as well. I'm open to other ideas on this one.

5) Automatic/Critical Success. I don't just want to go with a "1 is always a success". Setting the "luck" factor at a solid 5% is a little too high for me. What I was thinking is that if the penalties stack up so much that the task is "impossible", then only impose enough penalties to drop it to a 1. If that roll is successful, then roll-under again (possibly carrying over the remaining penalties. Critical success (which would only apply to "possible" tasks) could work the same way. If you roll a 1, roll under the skill again. If successful, you have a critical success. I'm open to hearing other options, though.

6) Automatic/Critical Failure. Likewise, I don't want to go with "20 is always a failure". Also, I want something smaller than the 5% margin. Actually, I had something different in mind. Instead of having a player roll, coming up with a critical failure and acting accordingly, I was thinking of having critical failures secretly determined by the GM. Why? How about this? The bookworm character is deciphering some text on an ancient tablet. The player rolls, thinks he succeeds, and the party heads off in the wrong direction because of bad info. Sure, the player could roll, crit fail, and RP it out. However, if I was a player, I would actually think it would more fun not to know that I tanked before my character could figure it out. For the game(s) I plan to run, something like this provides more of an opportunity to throw in some more sub-plots and side-quests. Anyway, I'm still working on a mechanic for this. It might be a burden to have the GM (likely me) make a secret roll or two every time a player makes a check, but that's the only solution I've come up with so far. Any other ideas?

Thanks for any advice.

OK, here are some questions/thoughts:

0) What's a 1d20 blackjack roll-under? What does a failed roll mean? What does a success mean?

1) Why do you want attribute tests? Depending on what failure/success means (see #0, above,) why is it important for attribute tests to have the same chances of success as skill tests?

2) Why split Build? Is it important in your setting/game that Strong people not necessarily be Healthy as well?

3) How about Presence?

4) If a skill is too high to fail, why not just skip the roll?

5 and 6) I can't understand the critical/automatic success and failure until I understand the part about "blackjack roll-under" and what it means to succeed or fail, but again, before answering whether to have a particular auto/crit system, I have to ask: is an auto/critical success/failure important? Why?

First, I should mention that I talked over the general system idea with my group and they like the idea of a system that has the elements I mentioned and isn't a rules-lite storytelling system. What I'm looking for is advice on how to fine-tune the mechanics. Anyway, to answer the questions.

0) Blackjack roll-under means you want to roll as high as you can while not exceeding the test value. For example, if I'm rolling against a 13, a 3 is ok, a 8 is better, and a 12 is even better than that. Statistically it's the same as a regular roll-under, but you don't have to bother subtracting what you roll from your skill to determine the margin of success. As for success/failure, what exactly are you asking? A failed roll means you failed at the task, and a successful roll means you succeeded. Though I would think that would be obvious.

1) I want attribute tests because they could be necessary. Two characters are lunging for the same item. I'm not just going to say the highest attribute wins, nor do I think I should have a "lunging" skill. Instead, it's a simple Agility test. And it's important for attributes to have the same chances of success as skill tests because it just makes sense that way. A test is a test. No matter what one is testing, for this system, it needs to be on the same probability scale.

2) It is somewhat important that healthy people are not necessarily strong. This is another one of the it-makes-perfect-sense-to-me things that I'm hung up on.

3) Good idea. I think that might work.

4) Because I want the player to feel that s/he can try anything s/he wants, even if there is only a sliver of a chance for success.

5 and 6) Yes it's important. Regarding auto success/failure, I don't want any guarantees in this system. No matter how small, there's always a chance things can go south. Crits are also important because sometimes people are able to perform beyond their normal ability, and sometimes people just screw up. That's life. Sh*t happens, sometimes for the better and sometimes for the worse.

Hmm, it sounds like you're pursuing a goal with your game design that's not really popular around here.  Which is not to say that it's not a valid goal, just that if you don't get the response you want, that's probably the reason.  It sounds like you're trying to create a "realistic" set of rules, which for many reasons is a bit of a minefield.  I'm going to ignore these issues though, becasue there are a hundred threads that explain why this is a problematic goal, and a hundred articles explaining why you can't create the game you're trying to. I'm going to assume you've read all these, and are still interested in doing this.

It's hard for me to give you answers to your questions, because ultimately they're design choices that only you can make.  I have no idea what kind of "reality" you're trying to model with your rules, and I don't know what effect you're trying to create, so I can only suggest questions you might want to ask yourself.

Here are a few questions which commonly cause complications for this kind of system:

What happens when two characters try to help each other?

What happens when two characters are opposing each other?

How does the GM (if there is one) decide what the difficulty of a task is?

How do you set stakes? (This means "how do you decide what the consequences for sucess and the consequences for failure are?" Most systems have a very detailed system for doing this with combat, but almost no system for any other kind of roll.  Also, they tend to foucs on what you can do if you succeed, but don't discuss what happens if you fail) An interesting answer to this question could be the difference between a great system and an average one.

What are the limits of a skill? Who decides what actions are covered by what skills? How many skills are there in your game, and how broad are they?  The number and type of skills you choose will drastically effect your game.  Consider the difference between a game with a "fighting" skill, and a game with a "bladed weapons - one handed" skill.  Please, please, resist the temptation to have a huge shopping list of skills.  It is in every respect a bad idea.  The skills you choose to detail will determine what is important in your game.

The last, and (in my opinion) most important question, is "why this game?" Why play this game, and not the myriad of other, better playtested, better supported, more well known systems out there.  GURPS, d20, Savage Worlds, Fate, Fudge, Rolemaster... There's an endless list of games, from "realistic" to pulpy, with great fan support, years of playtesting, and often free rulebooks.  What's your game got that these ones don't? This doesn't mean that you should give up and play those games, just that you need to identify what your game does that those ones dont.  Think about why your game will be the best for your group, and focus on that.  Make it the centre of your system.  If you think there's no other game that makes the "effort" a character puts into an action important enough, build your game around an "effort" system.  If you think every other game makes it too easy to get killed, make an innovative system for keeping characters alive.  Above all, play, read, and understand those games.  Learn from them, steal from them, and you'll make a better game.

Good luck!

Hmm, it sounds like you're pursuing a goal with your game design that's not really popular around here.

It sounds like you're trying to create a "realistic" set of rules, which for many reasons is a bit of a minefield.

Why play this game, and not the myriad of other, better playtested, better supported, more well known systems out there.

Ok, cool.  What I'm hearing is that you've got a very strong "simulationist" agenda (This means, roughly, that play is about exploring the setting and experiencing "being there" in the fictional world). That's not very popular here as a primary goal, but it is a popular and functional way to play, and I'm sure you can make a game that works well for your group.  That said, all of the questions I posed above still stand.  Of particular interest to me is this:

Quite simply, none of them fit the setting I have in mind.

Can you tell me more about how your setting is different so that it doesn't fit with any other published set of rules?

I mean, what is it about the "reality" of your setting that makes it work differently from how we understand reality?

What's problematic about this is that there's no clear rules for saying what happens when you succeed or fail.

Consider this alternative...

"before the dice are rolled, the GM must say what the consequences are of each possible result"

I think we're coming from pretty different philosophies about gaming, but hopefully I've given you some ideas that are germane to the way you like to game.  Seriously, I'm not trying to tell you that the way you play is wrong, I'm just asking tough questions that will (hopefully) help you make a game that works better for you.