On the Virtue of Ability Scores

We've learned a lot about ability scores in the last 20 years. Some of the first points are trivial, obvious, or truistic, but are covered for completion.

Mostly, they are the way they are in Dungeons & Dragons due to legacy compatibility. You can see this in more modern systems that go to a much simpler modifier system. Instead of 18, your score is +4 (or +3 or whatever).

The value in the 3-18 system is that the abilities they produce are a bell curve of scores. This means that most people are average. If you have a range of -3 to +4 and you roll a single D8 to determine your score, you're going to have a huge variation in natural ability.

Ability scores can be very useful for things like saving throws or as an ad hoc measure to determine the success of actions. But how you use the ability score to check things is kind of related to your intent.

That juicy design nut

You can check with a single die (usually a d20) or multiple dice, added together (usually 3d6) There are several variations of these methods. Personally, I usually use a 1d6 roll with the Moldvay statistical bell curve bonus added on. The key here is determining how you want the result to be calculated. There are two results: interesting and consistent.

Consistent results are achieved with a bell curve. The intent here, is that people who have low statistics are bad at the check and people with good statistics are good at the check.

This may sound awesome, but really, often consistent play is boring play.

Interesting results, generally give a 60/40 to 70/30 chance of success/failure. There is a reason most older games used 1 in 6 or 2 in 6 as chances of success, because failures (or successes, dependent on the system) happen frequently enough to create interesting situations in play.

Damn it Jim! You're not a computer!

Some Options

There are several different methods of addressing the above, but they are highly dependent on the weight you want to place on ability scores. The more weight that is placed on ability scores, the more the design and skill of the character is a factor.

Some common options are rolling multiple D20 dice, and using the higher or lower one to represent a bonus or disadvantage. This is the D&D next mechanic, but it's been around for a long time. It shifts the average result to a higher or lower value, but produces an s-curve, not a bell curve.

A way to modify bell curve rolls is to add or subtract a d6 to indicate additional difficulty. This is effectively the same as applying a rolling +2 or -2 to the check, and setting what you think the value should be for a success at the task.


  1. I need to write posts that are more like yours, dude. This is like, the opposite of rambling.

    And this is what Roll 2d20-keep-the-highest looks like.

  2. I like ability scores they tell you a lot about a character but how they are applied can vary a lot if the rules or DM (even players) view an ability score as a capability or limit.

    If the game allows the ability score to be damaged with regularity I want the modifiers associated with ability scores to be few and far between.

    If character level is meant to be a significant factor in the game ability checks (tied exclusively to ability scores) can be bothersome as well.

  3. I don't remember where I saw this, but someone recently pointed out that one of the great things about Pokemon is that your attacks usually hit. There are very high success rates for most things. I like to run my games that way, where success is typical but failure is interesting and unexpected.

  4. Minor correction: +1d6 is closer to +3 than to +2 (expected value of 1d6 being 3.5).

  5. I'm not sure how you reached the conclusion that consistent results are uninteresting. Simply because results over a long period of time are consistent, it does not follow that they will be uninteresting on any given roll. Further, simply generating a flat rate of success/failure (like 30/70) is JUST as consistent and probably less interesting.

    Further, statistically speaking, rolling 2d20 and taking the highest/lowest is the equivalent of about a +/-3 modifier. So, lots of jumping and conclusions here.

    1. I don't think the problem was with probabilities for events being consistent, but that they're too small. If a "critical failure" only has a 0.01% chance of occurring, it's not going to see much use. While it's desirable for expected outcomes - like a strongman busting down a door with a single smash or a weakling not even being able to climb a rope - to be more common than unexpected ones, it's kinda boring if the latter almost never occur in the game

  6. Success all the time isn't too bad. The point is that failure can create cool results as well and become a story in and of itself. Players are afraid of failure in part because GMs don't know how to "fail forward" and make rolling that 1 or 2 as suitably epic as rolling a19 or 20

  7. One other advantage of the 3-18 system against Wizards-era modifiers is granularity. You simply have more points on the distribution to show how different characters stack up in terms of strength, etc.


Related Posts Plugin for WordPress, Blogger...