Reconsidering Skills

I have been considering skills in Echelon, and I’m not entirely sure they will work the way I want.

One of the difficulties found in D&D 3.x is ‘falling off the Random Number Generator’ (RNG).  It is trivial for two characters of the same level to be so far apart in bonuses that one of them literally cannot roll as well as the worst roll of the other.  Spot and Hide checks are a commonly-used example — at 20th level, a Rogue with maximized ranks in Hide (+23 to checks) cannot be Spotted by a Fighter with average Wisdom and no ranks in Hide (+0 to checks).  The best the sentry (even at 20th level) can do is still worse than the Rogue can possibly roll, even without taking into account the likely high Dexterity and possibly equipment bonuses the Rogue has.

To a certain extent this may be desirable.  The Rogue is a Hiding God, while the Fighter really shouldn’t be on sentry duty at all.  However, this applies not only to direct challenges, but also cooperative situations such as Balance checks, such as when the party needs to make their way across a narrow, rickety bridge.  The Fighter (0 ranks in Balance and Dex 12 — full plate limits Dex bonus to AC, so he doesn’t have more) has a +1 to his Balance check, the Rogue (23 ranks in Balance and Dex 29) has +32… before equipment bonuses.  Something that challenges the Fighter is trivial for the Rogue, something that challenges the Rogue (i.e. he has to roll at all) cannot be practically attempted by the Fighter because the DC is off the RNG.

In fact, an ‘even challenge’ (where the Rogue needs to roll 11 or more on 20, 50% success) has as DC of 43two full RNGs away from what the Fighter gets.  If the Fighter rolls a 20, and adds another 20, he still can’t succeed.

There isn’t even a middle ground where the Fighter has a chance to succeed and the Rogue has a chance to fail.  For characters at the same level, this is not acceptable.

I have no problem with falling off the RNG as levels increase.  A high-level character who has mastered the art of not being seen can quite reasonably be able to hide in ways that low-level characters can’t.  The game changes enough as levels increase that, despite having similar core mechanics, they are not the same game.  I have no problem whatsoever with 20th-level characters being able to do things 1st-level character cannot — Echelon is predicated on high-level characters being much better at what they do, whatever it is, than low-level characters.

Goals and Desirable Characteristics of a Skill System

The design of skills in Echelon is based on the following assumptions and expectations.

  • High-level characters are better at pretty much everything than lower-level characters, given the same training.  Experience counts for a lot.
  • For any particular ability, trained characters of a certain level will be better at what they are trained in than an untrained character of the same level.
  • Training can overcome experience, to a point.  Lower-level trained characters can expect better results than higher-level untrained characters (either based on check result or trained capability).
  • Experience can overcome training in limited ways.  Higher-level untrained characters can expect success more often than lower-level trained characters, if the difference in level is large enough.
  • Natural ability (high ability scores) can provide some benefit but should be possible to overcome with a combination of training and experience.
  • At any given level, in straight checks, trained and untrained characters should remain on the random number generator.

Echelon is largely based around such assumptions.

Basics of Echelon Skills

Echelon models skills somewhat differently from D&D 3.x.  The same basic skill definitions (name, key ability score, what happens on a successful check) are much the same.  They want they are trained up is rather different, though.

First, skill ranks are gone.  They were a pain to manage and almost always amounted to “pick a number of skills equal to your base skill points per level plus Intelligence modifier and you get level+3 to checks with those skills.  Unless they’re cross-class skills in which case you halve that number.”  Unless you’re dealing with a multiclass character, in which case you might need to decide which skills to stop maintaining (or what skills to add at a lower value)… I’ll be honest, I usually just make it up because it’s too much hassle.

Skill use is improved through the Level Bonus (increases every even level).  This addresses the first goal, higher-level characters being more reliable with their skill use than lower-level characters.

A character that wants to be better than that with a skill can take a talent to improve his use of that skill.  This talent grants a competence bonus to checks with the skill (they don’t stack with multiple talents affecting the same skill) and at each tier the talent provides specific benefits or abilities not available to those without the talent at that tier (which takes care of the second goal).  This trivially satisfies the second goal.

Having talents at each tier grant specific abilities not available to untrained (or to less-trained) characters means that trained characters can achieve results not possible to untrained but experienced characters.  This satisfies the third goal, though some consideration should be given to relative reliability between low-level trained and high-level untrained characters.

A large competence bonus from training means that low-level characters will have more reliable and consistent results than higher-level untrained characters for large differences in level.  Too small means untrained characters can get more reliable results fairly easily.  The size of the bonus will need to be carefully chosen so the level difference is reasonable.

The difference in natural ability is four points between “best’ (9) and “average” (5), and “average” and “worst” (1 ).  As long as training and experience can overcome this natural talent fairly early it will become a small component of the total check modifier.  Talent plus training plus experience can still be better yet, of course.  Remembering this constraint will allow the fifth goal to be met.

The sixth point has not been carefully examined before this post.

The first three goals are fairly easily met with the current design.  It looks like the only part requiring consideration at this point is the size of the check modifiers for trained characters.  Chosen well they can satisfy the fourth through sixth goals.

Current Design

The current design has training provide a single, +4 competence bonus to the checks it applies to.  Normally this is to all checks made with a single skill (such as Spot), but I can see a case for having the bonus apply to limited applications of multiple skills.  For instance, a ‘Forester’ (talent modeling a “Favored Terrain” Ranger class ability) might apply to all Survival and Knowledge(Nature) checks relating to forests (but not desert or mountains).  This would be specified for each talent.

Check modifiers for Current Design, by Level and Ability Score
Untrained Trained
Level Bonus 1 5 9 1 5 9
1 +0 +1 +5 +9 +5 +9 +13
2 +1 +2 +6 +10 +6 +10 +14
3 +1 +2 +6 +10 +6 +10 +14
4 +2 +3 +7 +11 +7 +11 +15
5 +2 +3 +7 +11 +7 +11 +15
6 +3 +4 +8 +12 +8 +12 +16
7 +3 +4 +8 +12 +8 +12 +16
8 +4 +5 +9 +13 +9 +13 +17
9 +4 +5 +9 +13 +9 +13 +17
10 +5 +6 +10 +14 +10 +14 +18
11 +5 +6 +10 +14 +10 +14 +18
12 +6 +7 +11 +15 +11 +15 +19
13 +6 +7 +11 +15 +11 +15 +19
14 +7 +8 +12 +16 +12 +16 +20
15 +7 +8 +12 +16 +12 +16 +20
16 +8 +9 +13 +17 +13 +17 +21
17 +8 +9 +13 +17 +13 +17 +21
18 +9 +10 +14 +18 +14 +18 +22
19 +9 +10 +14 +18 +14 +18 +22
20 +10 +11 +15 +19 +15 +19 +23
21 +10 +11 +15 +19 +15 +19 +23
22 +11 +12 +16 +20 +16 +20 +24
23 +11 +12 +16 +20 +16 +20 +24
24 +12 +13 +17 +21 +17 +21 +25

The first three goals are already met.

The fourth goal (experience can allow more reliable results than lower-level training) is fairly well satisfied.  It takes eight levels (two full tiers) to be as reliable as someone trained (so training counts for a long time), but experience does eventually make someone more reliable at checks.

The difference between average and good or average and bad is 4 in both cases.  Training alone or two tiers experience can match this, training and two levels (+1 Level Bonus) or ten levels experience can exceed this.  That eight or ten levels seems pretty big, but it assumes general experience rather than specific training.  It looks like this is workable.  The difference between best and worst is twice as big; it would take someone with Dex 1 training and eight levels experience (or sixteen levels of experience) to match the untrained character of the same starting level with Dex 9.  This seems a little long, but it’s a degenerate case.  It looks like the fifth goal (training and experience can overcome natural ability) is met.

The sixth goal (trained an untrained stay on the RNG) is clearly met at all levels.  The best (Dex 9 trained) and the worst (Dex 1 untrained) are always twelve points apart.  This is pretty comfortably on the RNG.  In fact, this may seem too comfortably on the RNG (especially at a particular ability score value — two characters of the same level and with the same Dex score will be only four points apart), but specific abilities gained by higher tiers of training are expected to make up for this.

So, it seems a +4 bonus for Basic training, that does not get increased at higher tiers, looks workable.

Alternate Design

This entire post was prompted by consideration that the +4 bonus from Basic training might not be large enough as levels increase.  I decided to examine a simple option where the check was equal to +2 per tier.

Check Modifiers for Alternate Design, by Level and Ability Score
Untrained Trained
Level Bonus 1 5 9 1 5 9
1 +0 +1 +5 +9 +3 +7 +11
2 +1 +2 +6 +10 +4 +8 +12
3 +1 +2 +6 +10 +4 +8 +12
4 +2 +3 +7 +11 +5 +9 +13
5 +2 +3 +7 +11 +7 +11 +15
6 +3 +4 +8 +12 +8 +12 +16
7 +3 +4 +8 +12 +8 +12 +16
8 +4 +5 +9 +13 +9 +13 +17
9 +4 +5 +9 +13 +11 +15 +19
10 +5 +6 +10 +14 +12 +16 +20
11 +5 +6 +10 +14 +12 +16 +20
12 +6 +7 +11 +15 +13 +17 +21
13 +6 +7 +11 +15 +15 +19 +23
14 +7 +8 +12 +16 +16 +20 +24
15 +7 +8 +12 +16 +16 +20 +24
16 +8 +9 +13 +17 +17 +21 +25
17 +8 +9 +13 +17 +19 +23 +27
18 +9 +10 +14 +18 +20 +24 +28
19 +9 +10 +14 +18 +20 +24 +28
20 +10 +11 +15 +19 +21 +25 +29
21 +10 +11 +15 +19 +23 +27 +31
22 +11 +12 +16 +20 +24 +28 +32
23 +11 +12 +16 +20 +24 +28 +32
24 +12 +13 +17 +21 +25 +29 +33

As above, the first three goals are already met.

The fourth goal (experience can allow more reliable results than training) is partly satisfied.  It takes four levels of experience to match the competence bonus from each tier of training.  This means it takes more tiers to match training as more training is gained (four levels to match Basic training, eight levels to match Expert training, twelve levels to match Heroic training, and so on).  I’m not certain I’m happy about this, since it is on top of the levels needed to get the training in the first place — an Expert untrained character could match the check modifier of a Basic trained character (with the same ability score), but it takes a Master to match the check modifier of an Expert-trained Expert character (and a Legendary untrained character to match the check modifier of an Heroic-trained Heroic character).  I’m not happy with this, but technically the condition is met for part of the time.

The fifth goal is certainly satisfied.  The competence bonus climbs to +12 (which is alone bigger than the ability score gets, let alone the difference between best and average or average and worst), as does the Level Bonus.  An average person with training and experience meets the difference in no more than four levels (+2 for competence, +2 for Level Bonus) and exceeds it within two more (+1 more Level Bonus, or +2 more competence if he upgrades the talent).  In the worst case it will be eight levels (competence bonus and Level Bonus).

The sixth goal is almost satisfied.  At the top end, the difference between trained and untrained gets too close to  the edge of the RNG.  At any given ability score the competence bonus is +12 (which is workable), but the difference between best trained and average untrained is +16 (barely on the RNG… but still is).  Best trained and worst untrained?  Off the RNG.  I’ll call it satisfied, since it barely falls off, at the top end, and only for an edge case.  As with the fourth goal, while I’ll consider it technically met I’m not very happy with it.

Composite Design

I briefly considered a composite design using the scaling bonuses to govern access to tier abilities but discarded it pretty quickly.  If only the competence bonus is considered in gaining access to the tier abilities there is no difference between that and the Current Design.  If the Level Bonus applies, you can end up with high-level characters essentially having access to all trained skill abilities even untrained (which I don’t want)… unless you put a requirement of having the relevant talents at the suitable tier, in which case it’s back to the Alternate Design.


It seems my first guess at it — and +4 was more or less a guess — is a better way to go.  The way skills work in Echelon I can meet my goals more simply with a flat bonus than I can with a bonus that scales with level.

No related content found.

One comment

  1. hadsil

    If a Fighter puts no ranks in Spot while a Rogue maxes Hide, it’s not fair to blame the skill system for the fact the Fighter can never find the Rogue. It is fair to fault the skill system if the Fighter had maxed his Spot and still can’t find the Rogue. Given the cross-class rules, that is 3E’s problem.

    Pathfinder fixed that by getting rid of cross-class. Now a 20th level Fighter can have 20 ranks in Perception vs a Rogue’s 20 ranks in Hide. At 20th level, say Fighter has 14 Wisdom while Rogue as 22 Dex. Fighter’s total Perception is +22. Rogue’s total Hide is +29, including class skill bonus. The Rogue is better at hiding, but he’s supposed to be. However, the Fighter still has a chance. If the Fighter cared to, he could have spent one Feat on Skill Focus (Perception). Now his total Perception is +28 vs Rogue’s +29 Hide. The Rogue will be spotted 50% of the time. The skill system isn’t failing.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Back to Top