D&D attributes, equal random generation, and skills

by robdalexander


Immergleich’s rules are based on Lamentations of the Flame Princess (LotFP), and so it inherits the following:

  • Character attributes (the D&D set of Str, Dex etc) are randomly generated — 3d6 in order, roll again if modifiers sum to less than zero, player may make one swap.
    • I like this — it makes character generation a “let’s see what I get to work with” rather than a pure act of design. It cuts through overthinking and it pushes players to try concepts they would instinctively avoid. It helps to cue up players that my game is about rolling with situations, not about grinding through fair challenges for a fixed-schedule reward. It probably discourages the mechanical-optimisation-oriented players who won’t do well in my game anyway.
  • There are no direct rolls of attributes or their modifiers — there are attack rolls, saves, and skills rolls, some of which are modified by attributes
  • It is implied that you use the common OSR approach of resolving most challenges through player creativity assessed by GM judgement. Outside of combat, most challenges don’t involve rolling.
  • There are skills, but they only cover near-impossible things e.g. climbing a sheer surface. Most characters are stuck with a 1 in 6 success chance in all skills; only thieves (“specialists”) can improve them.


  1. Although PCs are never very weak (the roll-again rule prevents it), some PCs are stronger than others (it’s not that rare to have your modifiers sum to +5)
  2. Even if PCs end up with a balance of overall scores, some attributes are much more useful than others (e.g. Int is almost irrelevant)
  3. I don’t like the pure creativity-and-judgement approach to problems. When failure is a possibility and would be interesting, I like to roll.
    • I particularly like to use rolls to skip over complicated interactions (with objects or with NPCs) and get straight to the result
  4. Attributes are underused.  They’re right there on the sheet, concisely describing characters in ways that make obvious sense to many players, yet most of the time they are only used indirectly. In some situations where they sound like they’d be relevant, they’re not used at all.

Initial solutions

I have houseruled thus:

  • We use 5e saves
    • e.g. “a Con save” means you roll d20+Con mod+(2 if your class is good at this save) versus a DC set by the power of the threat
  • By means of a computer program, all PCs start with attributes such that their modifiers sum to 1 or 2
    • This gives a mean of 1.5, the same as you get with LotFP RAW
  • I frequently call for Dungeon World (DW) –style attribute rolls — 2d6+(most relevant attribute modifier)+(1 if your character’s non-class background is relevant here) and interpret them thus:
    • 6- means you failed, and now a bad thing happens
    • 7-9 means you succeeded, but some bad thing also happened
    • 10+ you succeed

This gives characters that are closer to each other in overall ability, greater use of the attribute scores, and a mechanic for rolling in ordinarily-risky situations.

Remaining problems

  1. The raw attribute values are still almost never used — they mostly sit around on the sheet while the modifiers do all the work
    • Exception is when tracking attribute increase (e.g. some potion effects) or decrease (e.g. attribute damage). These exist in Immergleich but are not common.
  2. It’s not clear what the relationship of attribute rolls and skill rolls should be
  3. Small differences in attribute modifiers make very large differences to roll outcomes (see Dungeon World Dice Statistics):
    • A character rolling with -2 is going to do badly (72% fail, 3% pure success)
    • One rolling with +3 is going to do well (8% fail, 48% pure success)
  4. If you apply extra modifiers (e.g. my +1 for relevant background, DW’s frequent “take +1 forward” (i.e. +1 to the next roll you make), DW’s +1 or -2 for another PC’s aid or interference) then the results are even more extreme
  5. Now that they matter, it’s unsatisfying that there are only few, rare ways to increase attribute values
    1. I don’t want to just patch in the 5e model (2 attribute points to spend at 4th/8th/etc level because:
      1. PC levels in Immergleich are pretty much capped at 6
      2. With the Mentzer/BECMI bonus progression that LotFP uses, +2 can sometimes mean nothing (e.g. 10–>12), which is frustrating
    2. The DW model (1 point per level) does better with the level cap, but still has the dead zone problem
  6. I usually feel dissatisfied when a player takes the “swap two attributes” option — it takes the edge off “you play what you get”
  7. Having some characters start with +1 overall, while others get +2, feels petty to me
  8. Some attributes are less useful than others (Int remains underused)

Further, untested, solutions

  1. Discard the actual 3-18 values after initial rolling; keep only the the modifiers
    1. Resolves problem 1, takes six numbers off the character sheet
    2. For attribute change found in other materials (including traditional attribute damage), convert on a ~2.5:1 basis to modifier change
    3. The “attribute modifier” is now the “attribute value” (and I’ll use this term in the rest of the post)
  2. Define a way for specialist skills to enable impossible tasks on the difficulty scale used for attribute rolls
    1. Resolves problem 2
    2. See Combining Dungeon World attributes with LotFP skills, badly
  3. Be very strict about giving attribute roll modifiers — maximum +1 for background, +1/-1 for situation (e.g. help, equipment)
    1. Helps manage problem 3
  4. If their modifier total reaches +5 then don’t roll — they automatically succeed
    1. Helps have clarity as to when problem 3 has made rollingpointless
  5. Allow a +1 to an attribute at levels 3 and 6
    1. Resolves problem 5
    2. This is roughly equivalent to the DW progression (+1 to value at each level)
    3. Maybe make +2–>+3 cost double (e.g. your level 2 and 4 advances), because that change radically improves your success chances
  6. Change the generator program to
    1. Generate only values between -2 and +2
      1. Reduces problem 3 — there will be fewer +2’s in circulation
      2. (-3’s and +3’s are very rare with the “sum 1 or 2” rule, anyway. In the 15? characters who’ve seen play so far, none have one.)
    2. Discard the raw values and only print the modifiers
      1. Supports solution 1
    3. Only generate characters whose modifiers sum to +2
      1. Resolves problem 6
  7. Remove the “swap two attributes” option
    1. Now that the stats are balanced, players don’t need that to have viable characters
    2. Resolves problem 6
  8. Look for opportunities to add rules involving underused attributes
    1. E.g. when players ask “Do I know anything about…” I’ve started to call for Int checks in the vein of DW’s “Spout Lore”

Before I implement all of that…

Do you have thoughts? Tell me.

In particular, tell me if I have missed some implications of my proposals that will ruin everything.