Quote from: Aglondir on August 09, 2024, 01:27:44 PMGenerally:

Untrained: 0 to 30%
Basic trained: Start at 60%
Best trained: 95%

Assumptions:

1. There's always a 5% chance to fail on any check, no matter how good you are.

2. There's no need to make checks for tasks that automatically succeed or fail.

3. There's also difficulty modifiers and degree of success to consider, but that's outside the scope of the question.

I think you're close to being spot on.

Quote from: Mishihari on August 07, 2024, 03:50:30 AMI remember something about actual research done, prolly by WotC, that concluded that a 70% hit rate was the most fun in combat.  That probably generalizes.

This.  Game theory says people consider a 70% success rate as, "fair."

I know it makes no sense, but people like to win. 

I'm not just talking tabletop RPG's.  Video game programmers try to make the medium difficulty level be 70%.  They collect data about how many people die and need to restart a level/section of a game, or kill the level's boss.  Then they tune their game so the majority of players win 70% of the time. 

Quote from: weirdguy564 on August 09, 2024, 07:04:21 PM

Quote from: Mishihari on August 07, 2024, 03:50:30 AMI remember something about actual research done, prolly by WotC, that concluded that a 70% hit rate was the most fun in combat.  That probably generalizes.

This.  Game theory says people consider a 70% success rate as, "fair."

I know it makes no sense, but people like to win. 

I'm not just talking tabletop RPG's.  Video game programmers try to make the medium difficulty level be 70%.  They collect data about how many people die and need to restart a level/section of a game, or kill the level's boss.  Then they tune their game so the majority of players win 70% of the time. 

I suspect 70% is rounded up and the actual threshold number to be 66.667% because humans have an odd affinity for threes and as the old saying and the Meatloaf song goes; "two-out-three ain't bad."

Quote from: Man at Arms on August 09, 2024, 02:14:18 PM

Quote from: Aglondir on August 09, 2024, 01:27:44 PMGenerally:

Untrained: 0 to 30%
Basic trained: Start at 60%
Best trained: 95%

Assumptions:

1. There's always a 5% chance to fail on any check, no matter how good you are.

2. There's no need to make checks for tasks that automatically succeed or fail.

3. There's also difficulty modifiers and degree of success to consider, but that's outside the scope of the question.

I think you're close to being spot on.

Thanks! Here's what a simple D20 roll-over system could look like, using those parameters:

Attributes: Range 1 to 5. Average is 3.
Skills: Range 1 to 10.
Checks: Roll D20 + Att + Skill >= 19. If you are trained, add 6 to your skill.
Crit/Auto-fail: 20/1

Example 1: Alan has Dex 4 and Stealth 1. He rolls D20 + 4 + 1 + 6 >= 19 to succeed (60% chance)
Example 2: Bob has Dex 4 and no Stealth. He rolls D20 + 4 >= 19 to succeed (30% chance)
Example 3: Curt has Dex 4 and Stealth 10. He rolls D20 + 4 + 10 + 5 >= 19 to succeed (105% chance, but a 1 always fails)

Conclusion: The DF 19 looks high, but the math is solid.  Another option is to reduce the att range to -2 to 2 and adjust the TN to 16. Actually I like the look of that better.

Here's (almost) the same thing, but with a simple D100 roll-under system:

Atts range 15 to 35, average 25
Skills range 1 to 50
Roll D100. You succeed if the result <= Att + Skill.
If you are trained, add 25% to your skill rank.
CS/CF: doubles under/doubles over

Example 1: Alan has Dex 30 and Stealth 05. He must roll D100 <= 30 + 05 + 25 to succeed (60% chance)
Example 2: Bob has Dex 30 and no Stealth. He must roll D100 <= 30 to succeed (30% chance)
Example 3: Curt has Dex 30 and Stealth 50. He must roll D100 <= 30 + 50 + 25 to succeed (105% chance, but a 00 always fails)

Nothing really new there, as there are tons of D100 systems that do something like that. But it is lightning fast, and easy to see the odds.

I am stumped as to why so many people in this thread are assuming 50-50 is the gold standard for fair and that players are wrong for preferring otherwise. Fairness is not determined by what your personal odds of success are; they're determined by how your odds of success compare to the other player characters' and to the enemies you fight.

Quote from: Chris24601 on August 10, 2024, 09:09:22 AM

Quote from: weirdguy564 on August 09, 2024, 07:04:21 PM

Quote from: Mishihari on August 07, 2024, 03:50:30 AMI remember something about actual research done, prolly by WotC, that concluded that a 70% hit rate was the most fun in combat.  That probably generalizes.

This.  Game theory says people consider a 70% success rate as, "fair."

I know it makes no sense, but people like to win. 

I'm not just talking tabletop RPG's.  Video game programmers try to make the medium difficulty level be 70%.  They collect data about how many people die and need to restart a level/section of a game, or kill the level's boss.  Then they tune their game so the majority of players win 70% of the time. 

I suspect 70% is rounded up and the actual threshold number to be 66.667% because humans have an odd affinity for threes and as the old saying and the Meatloaf song goes; "two-out-three ain't bad."

I would actually say this is almost certainly about breaking bad luck streaks. 50-50 odds are infamous for producing long streaks and if one of those streaks goes against the players it could break the game balance. A 70% success rate will almost certainly not produce a long bad luck streak the way a 50-50 success rate will.

There's also the matter of lost turn economy. A 70% success rate means that you are making efficient use of 20% more action economy than you would with a 50% success rate. Obviously this logic has diminishing returns because you do need to miss eventually, but systems with higher hit rates are faster and feel more responsive than systems with lower hit rates, all other things being equal.

The disconnect is not in that players prefer higher success rates.  They do, in all kinds of games. The 2 out of 3 center comes up again and again.  Though note that this is the rate where players will settle.  They'd prefer 99.99% if you'd give it to them. 

What is less clear is what rate is good for the game.  I submit that it varies depending on the feel, style, and a lot of other intangibles. I've got a lot of players dealing with 40% to 60% chances, frequently.  They don't like the individual failures, but they do enjoy the overall experience.  Part is the challenge.  Part is removing the treadmill.  That is, a person will put up with 50% or even lower for some time, if improving their skills is a thing they can do, and the world doesn't shift under them to keep them at the same point. 

Quote from: Steven Mitchell on August 11, 2024, 07:35:57 PMThe disconnect is not in that players prefer higher success rates.  They do, in all kinds of games. The 2 out of 3 center comes up again and again.  Though note that this is the rate where players will settle.  They'd prefer 99.99% if you'd give it to them. 

What is less clear is what rate is good for the game.  I submit that it varies depending on the feel, style, and a lot of other intangibles. I've got a lot of players dealing with 40% to 60% chances, frequently.  They don't like the individual failures, but they do enjoy the overall experience.

I don't think there's anything wrong with a 100% success rate. In an RPG, many actions have a 100% success rate - they're just not usually described as skill rolls. Many things like running, jumping, speaking languages, casting spells, and others are automatic. And I think it's a reasonable expectation that, say, an experienced pilot PC shouldn't have to roll a 60% chance to have to land an airplane -- but someone who has never flown should have a very difficult roll.

The problem is that automatic actions are common in games - but 95% success rates are boring to roll on, because the outcome is almost always the same. (The same is true of 5% success.) In terms of game play, you want rolls to be unpredictable to keep up tension.

The simplest way to keep uncertainty is degree of success systems. So even if a PC has a 100% chance of minimal success, they can still have uncertainty over getting a higher degree of success. I think there are other system principles that can help as well - I'll have to think on that.