Abstract System in Application

I'm considering using the Abstract system in application towards my earlier project. However, it seems lacking in a couple areas. Foremost in advancement to the level 100s.

Once the characters reach level 100, their specializations would likely be around 102 and their average and upper limit for even the highest attributes would be around 17.5 to 30. Is this too little? Is it too much? Do attributes then become meaningless? What does a 102 mean in the long run? If I had smithing 102 and the average NPC level was 5 that makes their smithing at 7. What does that mean?

Should I scale the skills slower? Should I increase the effect of attributes to make an average attribute have more dramatic effect? Should I implement the pay to succeed? Or maybe I should consider adding a scaling attribute system for my game. Perhaps every ten levels an attribute can be increased?

Of coarse then we would have rank 13 attributes which would be a problem, every twenty could result in at most a rank 10, and furthermore if I were to limit each attribute to one increase before 50 and one after then it would max out at 7. I think 7 is tolerable. With an average range of 24.5 and an upper limit of 42 it has up to a 40ish% boost to your rank 102 specialization. Not too shabby. Furthermore it can be another point of separation from non-heroic NPCs by not granting them statistic bonuses. This regular numerical increase, does however bring the game a few more steps away from a role playing game and towards a roll playing game, however I wonder if this would be another reward players could be granted to help encourage them to grow and develop.

So the highest roll possible without 'other bonuses' would be 144. I'm fairly certain that there will be other bonuses. Environment probably ranging from 5-30, tools ranging 5-30, perhaps bonuses bringing it up 10-30 and finally a 5-20 oversight bonus. All told, the upper limit becomes 254. With the only random numbers being attributes, setting the range as 7 to 42 making the range 219 to 252. Just short of 1/5 variance.

Not shabby I think. Not shabby at all.

The question becomes, what will the target numbers be? Should we be looking at two master fighters, both specializing in their respective fields to the maximum, (striking and dodging) The dodger would be looking at armor 30, bonuses 30, skill 102 and attribute 21, for a total of 183 and the master striker would be getting 162 + 7d6 with an average of 186.5. This means at the highest levels of combat we notice the attacker has a 3.5 edge against the defender. This edge is the .5 edge gained per die when using a +3 per die for static numbers while there is a 3.5 average per die.

Using http://anydice.com/ this means the attacker has about a 75% chance to hit the defender. Technically I could lean either way in this matter. I could bring the bonus up to a more average number, granting +3 OR 4 per rank of attribute, alternating between and narrowing the advantage, I could increase it to 4 per rank of attribute, and turning the advantage, or leaving it as is. It all depends on my personal agenda.

Alternating between 3 and 4 would make it 'equal' and narrow the odds closer to 50%. It would however complicate setting numbers for each attribute and balance combat times.

Raising the bonus to 4 would make it biased strongly to the defender. This would extend combat times and drop hit rate. It also would make getting the first strike less valuable.

Leaving it as is would make combats faster, give the first strike an advantage, and have generally a higher hit rate.

Having these high numbers would require figuring out the high results anyhow, so I think perhaps that would make a good topic to consider. When the players have received an indication of notable advancement every level. Does this, and SHOULD this still allow lower level application? What variance should even be considered, and can it even be considered? Once again I think that this should be examined with cross sections using extremes and averages. Perhaps it will uncover some form of major flaw.

First, lets look at starting levels, 1, 3, 5. Then I think 40, 43, 45 should be examined, followed by a stretch range of 20 vs 40. Or even 20 vs 100.

Lows will be untrained and minimum attribute at start. Likely a constant number will be used for all levels.
Ave will be a point every other level and two at start for specializations. It will also have average starting stats.
High will be a specialized starting stat and max points every level.

1
Low 2d6, 6
Ave 3d6 +2, 11
High 4d6 +3, 15

3
Low 2d6, 6
Ave 3d6, +3, 12
High 4d6 +5, 17

5
Low 2d6, 6
Ave 3d6 +4, 13
High 4d6 +7, 19

Looking at this, we can go ahead and start comparing well go with the notation of instigator, level and specialization rate, vs reacting level and specialization rate.

3L vs 3A: Unable to win as defender wins and the defender has 1 rating higher than attacker. Had they invested at least two points, they would have had a chance.
3A vs 3L: Cant lose, their bonus and minimum rolls are 1 higher than the defender.
3A vs 3H: Small chance of attacker winning. There are three possible rolls for success.
3H vs 3A: Small chance of failing. There are three possible rolls for failure.

Alright with this quick comparison it appears that not taking an applicable specialization can put you in hot water very quickly. Looking at this I wonder if maybe a mitigating factor is necessary. But this may be a bit of a quick knee jerk reaction. If players recognize this on their own, they may be mitigating the problem quickly, or accentuate it with some very situational glass cannons. Of coarse this will result in some of the above pairings. I personally am thinking the problem will continue on to accentuate itself in the future....

40
Ave 3d6 +21, 30
High 4d6 +42, 52

43
Ave 3d6 +22, 31
High 4d6 +44, 54

45
Ave 3d6 +24, 32
High 4d6 +46, 56

And here it is. The problem has become evident. Now anyone who hasn't placed full points into the ability has become unable to compete in a direct trait check. Now this will only become worse in the future and assuming we didn't want to gear the game towards specialists, or allow the strong players to always dominate should we gear the game towards the average players we need to seek to mitigate these numbers.

My first thought is to compare how some of these numbers are mitigated in other games.

DnD 3rd either uses a asymmetrical target number or keeps the numbers far from reaching the large scaling problem we just encountered. In the skill system the characters are generally limited to level 20 (we'll ignore epic levels for now) which means additive skill points will reach an upper limit of 23. Assuming average and high once again the expert would have a bonus of 27 (ability bonus of +4) and the average would have a bonus of 11. Given that 3rd edition uses a d20 for skill that means there is an overlap of three possible results (18-20) assuming ties go to defender and defender is the high number.

Their asymmetrical system is their skill vs DC, attack vs AC, save vs Spell DC system. The ability to boost spell saves is limited to only the base ability and level of spell (A couple feats altering this). The AC is basically based on only money and base statistic. The DC for a skill being a static number based on whatever the hell the GM wants. Furthermore these checks use an always fail or always succeed condition in the upper and lower 5% ranges. Meaning only 90% of the rolls even need the bonuses.

DnD 4th implements a similar skill system, but they widthdrew points and ranks. In fact, you are simply trained in a skill or not, meaning you get +4 in that skill or not. Furthermore the remainder of the bonus is gained simply by your level (half in this case) meaning a direct trait comparison between trained and specialized and untrained and unspecialized will at best have a 8 difference at first level. (This expands at later levels with their increasing base attributes later. Without items probably peaking near 11 or 12)

This also uses the asymmetrical system of AC/DC and so on. (I had to, don't hate me.)

White Wolf systems incorporate always having a chance to fail. The range is ALWAYS zero to rating+ successes. If you include rerolls or double counts (depend on which system you use when you roll a 10 on the d10) you can also get higher than skill results. With ratings varying from at least 1 to 10, there isn't even that much of a difference. A person with one die can succeed against someone with 10.

Savage Worlds is much like the white wolf system, except they also use the asymmetrical system of target numbers like DnD's AC, and have small maximums like White Wolf's ten dice max (I know it can go higher, but just going of base stuff here.)

Heroes Unlimited (based on the Palladium system which I likely may not be completely accurate in the following description due to some variances between games) appears to follow a comparative system, but the bonuses crawl forwards slowly. Possibly gaining one every few levels and with only small bonuses for specializing that MAY add up in the end to be a large number, but probably will not breach 20 in combat very far. This system rolls a d20 just like DnD skills and therefore is looking a lot like comparative DnD skills.

The reason I bring this system up however is that it uses percentile skills. It has the player start out with 20-60% ability within a skill then slowly crawl upwards at a few percent per level. Comparative skill checks would be about distance from your percent, trying to succeed at an action alone would just be aiming below said percent. There's also a hard cap for all skills at 98% for most classes. Any penalties to the skill are cut down from there. The mitigations implemented are a hard cap, competing against your own score and static increase.

So from my experience, the mitigation implemented by systems is generally broken down into some of the following.

Asymmetrical Competitions: Creating a non-specialization based target for rolls that merely grows with the character and may be increased or decreased somewhat based on wealth, preference or other method that can allow boosting past creation. I feel like this will draw away from some of the simplicity of the base system, forcing charts and a number of other advancements.

Soft Difference Limits: Indirect caps such as level that prevent specialization from going too far and creating no-win situations. Apparently such soft difference limits seem to hit around the variance of the die with a 10% chance of success. This means I would have to calculate the variance of the average die to calculate my max level. Somewhat restrictive, it would prevent the broad scale I had in mind.

Hard Difference Limits: I've already implemented one, but going further I could implement a hard limit like the maximum of 98%. For instance, the bonus can never exceed 50, or 100. This makes for an abrupt and artificial feeling cap for additive bonuses, but if the system were a percent one it would naturally flow in. This could be achieved using the current system by ensuring that the leveling process is simply slowed to ensure this cap is reached late game and not half way through. Currently I have it reaching 102 with my current hard difference limit, but it appears that it is necessary to either reduce the hard level cap or the points increased per level. This will require finding the breaking point where specialization creates no-win situations. Which appears to occur somewhere likely in the level 10 region.

Always Conditions: Implementing an always condition like the DnD Asymmetrical system or the bonuses and botching in White Wolf and Savage Worlds would allow for comparisons to become more likely to suddenly fail or succeed. The question becomes how to implement them. With our random range changing per rating of Attribute, the odds would be difficult to set as static results. However, since the dice are added not as success per die, they become more difficult to implement using the White Wolf style.

Forced Advancement: Looking at DnD in 3rd edition saves were forced progression. Everyone got them, and they always increased eventually at a steady rate. They could be increased or decreased by other modifiers, but onwards they trudged. More obviously as brought up the skill system in 4th edition. These ensure the character remains relevant. This system could implement it, a forced advancement in certain relevant skills, possibly as a class system. It will not solve the problem for other skills, but could be a way to force the player to at least be average in some key categories. I don't like the idea as it goes against my initial concept of customization, but seeing some of the mess encountered in GURPS skills, it might be useful.

My first impression is to consider a system of forced advancement, always conditions, the current hard limits and asymmetrical comparison and considering some soft limits.

Forced Advancement would be implemented with classes and possibly races. Combined with the general and specific specializations mentioned in the specializations post. A class would likely have major and minor skills. Major would increase constantly at max or two at start and one every other (not sure which yet.) Minor would have one at start and one every other or third. Similarly with races. Players could increase the specializations on their own above this (as long as it is not a generic one, then they would probably grab a sub category and increase that.)

Always Conditions I am considering implementing as follows. If half or more dice are sixes, then reroll and add the second total as well. If half or more are ones, then no dice are added. Should this be 50/50, bias towards the sixes. Only reroll once. This means if you have one die, you can get from 0 to 12, if you have three 0 to 32, five 0 to 60. It also means the odds of getting an always condition changes as the dice change, but not so impossibly as to require all sixes or all ones.

I didn't cover asymmetrical comparisons previously. That is because asymmetrical comparisons will only apply to player vs task. Such as crafting a sword. Your target number will change depending on the qualities of said sword.

As for the soft limits, I want to see what the results of the above do before considering how that applies to the soft limit balance.

So lets review the earlier comparative checks with the idea of Always Conditions.

1
Low 2d6, 6
Ave 3d6 +2, 11
High 4d6 +3, 15

3
Low 2d6, 6
Ave 3d6, +3, 12
High 4d6 +5, 17

5
Low 2d6, 6
Ave 3d6 +4, 13
High 4d6 +7, 19

3L vs 3A: Low is able to win assuming they get at least one six.
3A vs 3L: Low is able to fail if they get 2 ones.
3L vs 3H: Low is able to succeed if they get some good rolls.
3H vs 3L: High cannot fail against low.
1A vs 3A: Definite chance for success.
1A vs 5A: Still fair chance for success.
1A vs 5H: Continuing chance for success.

I like the look of this much better than previously. There is benefit to getting high specializations, but a low specialization does not mean it is impossible to succeed. Let's see how this plays out in later level pairings.

40
Ave 3d6 +21, 30
High 4d6 +42, 52

43
Ave 3d6 +22, 31
High 4d6 +44, 54

45
Ave 3d6 +24, 32
High 4d6 +46, 56

40A vs 40H: There is a chance to succeed. Exactly one. This doesn't appear to work too well still, especially considering the odds of getting two 18s in a row is pretty low to put it lightly.
40A vs 43A: This looks almost identical to the low level similar investment scenarios. Probably doesn't need further investigation.

So we've managed to reduce the difference greatly. But probably not enough. Not enough if I want those who have invested only some to still have a chance to compete with those who invested a lot. Some of the other solutions that could be implemented I'm considering include...

Increasing the variance: Every 20 levels giving the player the option to increase an attribute one die, with a maximum of one increase per stat up to 60 and 80 and 100 allowing a second. That would add a potential 12 to the variance for both, and make the game a bit more high powered. I was also considering the potential of increasing attributes by class and race. This means that should the player be a mage, human let's say, they will get two to INT bringing their variance up to rank 6 dice, at 100 this could be 8 dice. For D6 this isn't that bad as far as number of dice, and at 100 assuming the player assigned one rank and chose a class somewhat relevant brings the variance up a total of 24 for the average rated stat. The specialist gets a total of 48. However, since both can be 0, this makes for a wider potential range of results.

Cutting the Levels: I really don't like this idea. The main reason is that the 100 level system allows for a slow progression of abilities and bonuses in the other half that I wish to develop. With all 100 levels divided for 5 scales of play, and 20 levels within a scale to develop and gain abilities, it grants a decent progression rate allowing players to learn about the abilities they took, then grow.

Alright, let us take a look at the problematic 40 range adapting for attribute increases.

40
Ave 5d6 +21, 36
High 7d6 +42, 63

40A vs 40H: peaks at 81, target 63. That gives us a bit of variance.
40H vs 40A: Still no chance of failure

This is a bit better, it shows the attacker advantage that has been playing into previous concepts, and the higher skill still benefits. The question I'm now asking, is how often should the static number be used? Personally I like having the static number as an option to accelerate tests, but we'll see. For giggles, I'd like to see what happens at 100.

100
Ave 5d6 +51, 66
High 7d6 +102, 123

100A vs 100H: With a peak of 111, there's no chance of success. Odds are there hasn't been for what appears to be about 30 levels or so. Of coarse, after level 70 we've probably reached a high powered point where the characters have probably gained enough tricks to work around such walls. I'll have to be sure to build them in.

I think now I've finally reached the point where I'm willing to settle and move on. So in summary here's what we've got.

Attributes can be increased by class and race by one rating.
An attribute can be increased once every 20 levels, but a specific one cannot be raised more than once before 80 and no more than twice after that.

Critical fails and successes are when the dice rolled have at least half ones, or at least half sixes. With a bias towards successes.

At first level specializations can be bought up to rank 3, after that one point can be allocated per level.

Some specializations will be automatically granted, and can be raised beyond the base allotment up to the same cap as if bought manually.

I need to build work arounds to prevent unbeatable opponents.