Thursday, June 2, 2011

The dominant strategy of stat raises

Stat raises are fun because they give us a tangible way to represent the character progressing and growing in his or her most basic form: attributes. This is a legitimate goal to try and represent through game design. Stats are so fundamental to the game, though, that stat raises have the potential to do more damage than good. I believe that they do just that and will attempt to demonstrate this through the lens of combat and skills.

There is a dominate strategy when allocating any resource to allocate it where it will provide the largest benefit as often as possible. Very obvious. This dominate strategy is also present during character creation, which leads to characters with a single very high stat that almost all actions key off of. As a result, stat raises are concentrated into that single stat, pushing it even higher.

This strategy is so obvious that the game had to account for it. Over the 30 levels of 4e, there is the potential for any given stat to receive eight stat raises for a +4 modifier. That is 20% the entire range of possibilities represented on a d20 and too big a variance to not build into the math. So what we have is a wash: everyone gets higher attacks and higher defenses to keep the chance to succeed in that fun-tense zone.

The problem arises if someone decides not to follow the dominant strategy. If someone decides to increase non-primary abilities, they get some benefit, but not necessarily enough to keep pace with the game’s expectations. Over the course of the campaign they could fall 5 – 20% behind their party, hitting less often and being hit more often. This is punitive enough that most people take the path of least resistance (and highest reward) and follow the dominant strategy.

The end result is increasingly focused characters who become intensely focused on a single ability. Not necessarily bad, after all, people specialize all the time, but it doesn’t really feel like it achieves the spirit originally set out for: to represent the growth of the character.

Again, the game must take into account the possibility of a stat increasing by eight over the course of the game and so DCs reflect the dominant strategy.  It makes sense that the super strong fighter (str 20) can jump farther than the rogue (str 10) despite the fact that they are both trained in athletics. When the DCs try to reasonably challenge the fighter with a 65% success rate, it feels fair that the rogue only has a 40% success rate; he invested in other places. But then the game coerces the fighter to invest in strength and the rogue to invest in dexterity and updates DCs. By the end, the fighter maintains a 65% success rate but the rogue has fallen to 20%.

Throughout the campaign, as the dominant strategy encourages characters to increasingly specialize, the net impact is that we focus less on what characters do well and more on what they can no longer do. What you do well is still reasonably challenging because we artificially set the DCs to ensure this fact, but what you don’t do well is now almost out of reach.

The change
A simple change would get us back to the spirit of the original rule: stat raises cannot be assigned to your highest stat. Stat raises would still demonstrate character growth and actually reflect the character shoring up weaknesses instead. They would also still be really powerful but instead of one dominant strategy, there would be a lot of good strategies. During the campaign, it would actually tend to bring characters closer together with regards to skills. More characters would have a reasonable chance to jump over the pit. The fighter will still be the best, but at least his party can come along.

The epically bad downside is that it requires a reworking of the math because the dominant strategy is no more. This is relatively simple for things like DCs, but combat proves quite a bit more complicated. In the end it probably isn’t something that could be adopted into an existing game, but it is useful to consider in looking forward.


  1. Interesting post. I came to the same conclusion when I was looking at the 4e math.

    I don't like the gaps that appear at higher levels between optimized and non-optimized characters. This is evident both in combat and skill use.
    I especially don't like how large a gap in the skill modifiers can be created at higher levels. The bonuses are just too many and too large.

    I am working on some tweaks that hopefully, eventually, would fix this. My current solution is to add two new "stats", Primary and Secondary (I need better names). They automatically start at 18 (Pri) and 14 (Sec). When creating a character and buying stats, Primary and Secondary can be improved as well (to a max of 20, since they have no racial modifiers).
    If you hadn't guessed already, Primary and Secondary are used for all powers instead of the regular stats.
    I had originally eliminated ability increases, but with this system they could be introduced again.
    This makes any race able to be fully optimized in any class and would allow "dexterous" fighters (swashbucklers).
    Of course, it is not without issues. Some skills are better than others, and some stats have more skills tied to it. And you could theoretically play a fighter with low str, dex and con.

    I have changed some more stuff, which made defenses and attacks able to increase by 1 per level instead of 1/2 level (removing reliance on magical items, expertise feats/defense feats), which makes the numbers increase at exactly the same rate as the monsters.

    I am still working on damage (it is just shy of +1 per level when ability score increases and magical weapons are removed from the equation) and the skill section. The skills are the biggest headache as they can have such great variance (trained, stats, armor check penalty, item bonus, racial bonus, power bonus).

    I like your suggested change that stat raises can't be assigned to the highest stat. It would fix (or at least limit) a lot of the problems that comes with having a one-sided character.
    I don't think it requires much reworking, if you simply compensate for the lost benefit. However, in would be much more elegant in a new system.

  2. Interesting system. I like the general direction but, like you say, I imagine it will just shift which abilities people focus on *after* maxing Prime and Secondary. In 4e I imagine you'll see a lot of Con.

  3. I also spent a fair amount of time thinking about removing items/etc to make the math more consistent. The problem I kept stumbling on, though, is that most people need *stuff*. My players fit in the *most people* group so I ultimately can't cut it.

    Any solutions? or is your group immune from the need?

  4. In response to your comments.

    Removing magical items:

    I didn't remove magical items, I simply removed the static bonus (which can be considered rather boring). I can choose if I want a low magic or high magic setting, as the required bonuses are no longer tied to the items. Also, I can surprise the players while they are without their weapons, or even steal them, and they would still have a fighting chance.
    You can still have a weapon that has a +1 to hit (but no higher!), either as a property or encounter/daily power (depending on how rare you want it to be). In my mind, this weapon would be rare, perhaps even an artifact (and it stays relevant from level 1 to 30), depending if the effect is a property or one per encounter/day.
    But besides the above weapon, all the items in the current books still work (except the "magic" version that does nothing but give a static bonus). Of course, the players may balk at magical weapons being less powerful than they are used to (and will not feel "magical").

    Of course, there are also the additional crit dice. I haven't decided if I want those integrated on a magical weapon or character level basis. I am inclined to go with character levels, as this will work for both high and low magic settings.

    Changing the stats system:

    True, some stats become purely better than others since the primary effects of stats become increasing skills and defenses.
    This change either requires a group that can choose the stats based on the character and not make the character based on the stats or finding a way to boost some stats (either by reworking the skills or adding/removing something from certain stats).
    High Constitution (besides the very important health bonus) only adds to a single skill, Endurance. I am more worried about Strength, since only a single skill are dependant on it and the Melee Basic Attack. (As an aside, I am currently leaning towards letting all characters use the "primary" stat for basic attacks, since there is a feat(tax?) for this already).
    The breakdown of the number of skills a stat affects are like this:
    Str/Con affect 1 skill, Dex/Int 3 skills, Cha 4 skills, Wis 5 skills.

    I haven't started attacking this problem yet. Any suggestions? :)

    There are many facets to this change. If Primary and Secondary does not scale, then effects and damage from Powers will not scale. In some cases I am okay with this, in some I am not (you do not need a scaling to-hit bonus, but you do need a scaling damage bonus).

    Last, but not least, some races get better racials because the placement of their bonuses isn't optimal (Eladrin getting +2 dex/int for instance). Because defenses are still tied to the stats, this means that it is still a small disadvantage, but I haven't yet given much thought to how much it affects balance.

    In my earlier comment, I mentioned that it doesn't require much reworking to perform a change. I guess what I am saying with this post is that it can quickly become large if you aren't careful. :)

  5. I had written a long reply earlier and I could have sworn I posted it, but it isn't showing up now. Is it gone or do you have to approve comments before they are published?

  6. It put it in a spam folder and, hopefully, shouldn't be doing that any more.

  7. No worries, it just made me slightly grumpy. :)
    I actually jotted down all the points I remembered, in case I had to retype it.

  8. I think we are headed in the same direction with magic items. My current thoughts are to have inherent bonuses like Dark Sun and then retain the bonus aspect of items along with more flavorful powers. Then you can apply one of the numerical bonuses (either the inherent or the bonus from the magic item itself).

    But here's the rub--the flavorful power is a rider on the item's bonus. So this introduces a narrow set of scenarios where a high level character foregoes his +3 inherent bonus to accept the +2 weapon bonus because it brings with it some other interesting power.

    When the item is first received the bonus will probably be higher than the inherent bonus and so the decision is a no brainer. After a while, the inherent and item will be the same, but because the item has a rider power you'll, again, go with the item. So it is a narrow window when this is likely to arise, but I think it might be interesting enough to warrant it.

  9. I think this is a great criticism of 3e D&D, but 4e gives you a raise in 2 different stats, which I think really addresses this problem effectively; the second stat raise gives the player a lot of leeway to shore up a weak stat or improve an area of secondary strength. For instance a Rogue who wants to be decent at Athletics etc may boost STR as well as DEX.
    4e's auto +1/2 per level to all rolls also gives the DM leeway to set tasks that are easy for specialised PCs while being challenging but not impossible for untrained PCs.

    As a general principle I'm in favour of setting DCs low enough that maxed-out PCs are allowed to auto-succeed; the important thing is not to let any amount of success have an unbalancing effect. The poster boy for the wrong approach is 3e Diplomacy, which by the RAW could turn any foe from Hostile to Friendly. Better to set low DCs for many tasks, but not allow outrageously high rolls to have open-ended effects - eg the merchant won't sell below cost no matter how high your Diplomacy or Bluff roll; you can't balance on a cloud no matter how high your Acrobatics roll (use a Utility power!), and so on.

  10. S'mon-
    With regards to 4e, my gripe is with the underlying logic of easy and difficult task DCs. Easy presumes no bonus at all that never improves (except for the two sets of +1 to all stats). Hard, on the other hand, presumes an 18+ to begin, a racial bonus, all stat raises being committed to it, and the occasional feat or magic item.

    So against a character that is 18/10/10/10/10/10, the DCs might work out okay. Against anyone else, they are too easy or too hard. Moreover, I believe the DCs actually push people into that format (the 18/10/10/etc) because they are naturally incented to want to succeed at things and the way you succeed in 4e is to super specialize. As a result, anything outside your domain of super specialization probably fails.

    Now, I also am willing to let some people auto-succeed to ensure that other have a fun chance, but we ultimately shouldn't have to make that tradeoff. Not allowing stat raises on the highest stat helps remove the need to make that tradeoff.

  11. I think there is a lot more to this than just shuffling around where the stat increases go; particularly when such shuffling sees inappropriate/unjustified/unrealistic development for a character. Essentially the problem is stat modifiers and the overly dramatic effect they have on the game. The reward for development is too high seeing over-focusing to develop an optimized/unrealistic/boring strategy of character building. To address this, I think you need to do a handful of things:

    - Go back to ability scores and get rid of the modifier.

    - The ability score then becomes a number used as a prerequisite to attain certain abilities/feats/powers. In this way, you can engineer attainment of these abilities/feats/powers such that a 14 strength fighter can do the bulk of what an 18 strength fighter can do. The 18 strength fighter might be able to pick a couple of special options unavailable to the 14 strength fighter but the 14th strength fighter is compensated by having better all-round stats and alternative ability options.

    - Stats are increased based upon abilities/powers chosen. You might have particular-level powers also have a stat increase associated with them so that with the acquisition of that power is also the associated stat increase. In this way, what you practice is what you get good at.

    - Combined with this is a hard limit on how high each stat can go. There are a variety of ways of determining these hard limits but for me, the most interesting way is to make it a part of ability score determination during character creation.

    The above does not quite gel with a 4e ethos with numerous changes required to make this work. However, I think that solving the "modifier disease" spawned by 3e/4e needs to be something addressed if you truly wish to deal with ridiculous optimization as highlighted in your post above.

    Best Regards
    Herremann the Wise

  12. That's a neat comment, Herremann. I had never thought of that route but it is giving me a lot to think about. Thanks.

  13. Runeward,

    Doesn't your approach lead to party members having more homogeneous abilities, and thus remove the focus of each character on a particular role?
    Or put another way, aren't your characters _meant_ to diverge in ability, so that certain characters shine in situations others couldn't handle?

    That may be a good or bad thing depending on whether you like the idea of characters having particular roles in a party. I think you should take it into careful account before making the decision to eliminate a dominant strategy that also fits the character's role in a party.


  14. I like the idea of roles but I dislike (and disagree) that roles become dependent on stats. A cleric heals through his powers, not his wisdom. A defender ties opponents down through mark, not stats. Sure, stats play a part in effectiveness, but the essence of the role comes from somewhere else. This change wouldn't impact that.

    To address your second paragraph, I think reducing the role of stats is a good thing. If someone wants to play a swashbuckler, there shouldn't be so many demands of their stats that they can't achieve other goals. If Dex/Cha become too critical that our guy cannot put points into Wis (or whatever), I think the game lost something.