Tuesday, June 29, 2010

AD&D's Most Enduring Rule

One of the strange little oddities of AD&D is that the 1978 Players Handbook does not offer any explanation of how to generate ability scores for a character, except to note that "Each ability score is determined by random number generation" and that the Dungeon Masters Guide provides "several methods of how this random number generation should be accomplished."

Now anyone who'd played OD&D might reasonably assume that one of those "several methods" would include a straight 3D6 roll, but such an assumption would be misplaced. The DMG explains that
While it is possible to generate some fairly playable characters by 3d6, there is often an extended period of attempts at finding a suitable one due to quirks of the dice. Furthermore, these rather marginal characters tend to have short life expectancy -- which tends to discourage new players, as does have to make do with some character of a race and/or class which he or she really can't or won't identify with. Character generation, then, is a serious matter, and it is recommended that the following systems be used. Four alternatives are offered for player characters.
These four methods are:
  • Roll 4D6, drop the lowest (or "one of the lowest," interestingly) die from each roll, and arrange as desired.
  • Roll 3D6 twelve times, pick the six highest, and arrange as desired.
  • Roll 3D6 six times for each ability and retain the highest score for each ability.
  • Roll 3D6 in order a number of times sufficient to generate 12 characters and choose the most desirable set of ability scores.
As you can see, 3D6 in order isn't among the methods offered in the DMG to generate ability scores. Now, obviously, there's nothing to prevent one from using such a method, but it's clear, as per the text quoted above, that such a method is deemed a source of potential discouragement and a creator of "marginal" characters. If history is any guide, it would seem that most D&D players agreed with the DMG's advice, to such an extent that 4D6-drop-the-lowest became the implicit standard way to generate ability scores, a position it still seems to enjoy today.


  1. 4d6 and drop the lowest is the method I used for running my AD&D games. I consider it harmless: the stats are what the players sees first that tells them what their characters "are," and they usually want to see that they're a cut above the herd. This gives them a better chance to experience that, without going overboard. Fine by me. (4d6 & drop also seems to be in line with the increased overall power of AD&D compared to OD&D.)

  2. In an AD&D campaign, Gary is right. The stats are too important to the character's effectiveness to use less than the methods offered in the DMG. But 3d6 in order is fine for OD&D or Holmes Basic.

  3. What Anthony said. And like others, I often put in some wiggle room. I allow an elimination roll for any score below 9 (although in my next campaign I might reserve that for human characters). I also allow a point or two be moved around for good reason, but never to make something an 18.

    For some of my reasons for 4D6, check my post on "Elmer Fuddism in D&D"


    (yeah, I created a term for 3D6 character play) from a couple of weeks ago. Yep, lots more character self-esteem with AD&D over OD&D.

  4. If I want to get the highest scores possible, I go with:

    Roll 3D6 six times for each ability and retain the highest score for each ability.

  5. The default these days with 4e actually involves no dice rolling. The point buy system is strongly encouraged, actually one of the things I dislike the most about 4e.

  6. The thing is, AD&D rewards high scores, which in and of itself is ok, but you really need them to have a solid character. Most benefits don't start until 15 or more. So to get characters with some variety you need really high scores. Whereas with B/X and BECMI the ability score bonuses are mapped to the bell curve, and 4d6 drop lowest actually gives too much away IMO.

  7. I always liked this rule - particularly as my first ever character generated under Basic had fairly low ability scores. I think it was good idea from a 'realism' perspective - your characters were heroes and therefore would be a cut above the rest of the population.

    I used 4D6 like most people - but the third alternative - 3D6 six times for each score - seems remarkably generous. Maybe someone more mathematically inclined than I am could figure out which method actually gave the most powerful characters.

  8. I was going to go lay down early tonight, but the notion of removing "one of the lowest" intrigued me enough to write a script to get the numbers. Sorry for the length, but the size is the size, what can be done?

    Straight 3d6 (score -- count -- percent):

    3 -- 1 -- 0.4630%
    4 -- 3 -- 1.3889%
    5 -- 6 -- 2.7778%
    6 -- 10 -- 4.6296%
    7 -- 15 -- 6.9444%
    8 -- 21 -- 9.7222%
    9 -- 25 -- 11.5741%
    10 -- 27 -- 12.5000%
    11 -- 27 -- 12.5000%
    12 -- 25 -- 11.5741%
    13 -- 21 -- 9.7222%
    14 -- 15 -- 6.9444%
    15 -- 10 -- 4.6296%
    16 -- 6 -- 2.7778%
    17 -- 3 -- 1.3889%
    18 -- 1 -- 0.4630%

    Remove lowest of 4d6 (score -- count -- percent):

    3 -- 1 -- 0.0772%
    4 -- 4 -- 0.3086%
    5 -- 10 -- 0.7716%
    6 -- 21 -- 1.6204%
    7 -- 38 -- 2.9321%
    8 -- 62 -- 4.7840%
    9 -- 91 -- 7.0216%
    10 -- 122 -- 9.4136%
    11 -- 148 -- 11.4198%
    12 -- 167 -- 12.8858%
    13 -- 172 -- 13.2716%
    14 -- 160 -- 12.3457%
    15 -- 131 -- 10.1080%
    16 -- 94 -- 7.2531%
    17 -- 54 -- 4.1667%
    18 -- 21 -- 1.6204%

    Remove second lowest of 4d6 (score -- count -- percent):

    3 -- 1 -- 0.0772%
    4 -- 4 -- 0.3086%
    5 -- 14 -- 1.0802%
    6 -- 29 -- 2.2377%
    7 -- 54 -- 4.1667%
    8 -- 90 -- 6.9444%
    9 -- 135 -- 10.4167%
    10 -- 174 -- 13.4259%
    11 -- 204 -- 15.7407%
    12 -- 207 -- 15.9722%
    13 -- 180 -- 13.8889%
    14 -- 108 -- 8.3333%
    15 -- 59 -- 4.5525%
    16 -- 26 -- 2.0062%
    17 -- 10 -- 0.7716%
    18 -- 1 -- 0.0772%

    As you can see, removing the lowest of 4d6 gives a significant skew towards the high end of the scale, while removing the second lowest causes a skew upward, but more towards the higher range of average or just above average while keeping 18 as a very rare, hugely talented individual. In fact, it makes those individuals even more rare than straight 3d6, which may or may not be of interest to you and also is a bit counterintuitive at first glance.

    It lends itself to a feel of adventurers being ordinary folks, although perhaps a little more capable, meeting extraordinary circumstances. I think I have to cogitate on this, because I sort of like it.

  9. I'm a 3d6 straight down man myself, but as far as being generous with character generation I'd go with the fourth option...lets the player create a stable of potential PCs.

  10. Because the "Dungeon Master will inform you as to which method you may use to determine your character's abilities" and the beat goes on . . .

    "One of the lowest" heh heh :-D

  11. 3d6 six times is indeed super generous. Comparing it with 3d6 or 4d6 choose 3:

    Straight 3d6:
    Average - 10.5
    Chance of 18 - 1 in 216
    Better than 15 - 1 in 21
    Less than 9 - 1 in 4
    Chance of 3 - 1 in 216

    4d6 choose 3:
    Average - 12.25
    Chance of 18 - 1 in 60
    Better than 15 - 1 in 8
    Less than 9 - 1 in 10
    Chance of 3 - 1 in 1,300

    6 x 3d6, choose best:
    Average - 14.25
    Chance of 18 - 1 in 36
    Better than 15 - 1 in 4
    Less than 9 - 1 in 3,300
    Chance of 3 - 1 in 100 trillion

    So you can expect all of your stats to be above average, with at least one being far above average, and almost never see a really bad score. Rolling each stat three times would make a much more reasonable method:

    3 x 3d6, choose best:
    Average - 13
    Chance of 18 - 1 in 72
    Better than 15 - 1 in 8
    Less than 9 - 1 in 60
    Chance of 3 - 1 in 10 million

  12. This comment has been removed by the author.

  13. Using the incredible power of my laptop, I generated 100,000 stats using the "6 x 3d6" method and get a spread of results like this:

    8 - 0.033 %
    9 - 0.283 %
    10 - 1.264 %
    11 - 4.257 %
    12 - 10.484 %
    13 - 18.241 %
    14 - 21.366 %
    15 - 19.487 %
    16 - 14.205 %
    17 - 7.64 %
    18 - 2.74 %

  14. 3d6 in order can work for AD&D-- it's just a matter of which rules you choose to apply. or, as James suggested a few months ago, no one really played ad&d, they played d&d with added spells, classes, monsters, etc.

    as far as the "realistic heroes" argument, I think it was someone from the on-line dwimmermount game who suggested that it would be the types who weren't cut out for an honest trade that would risk their lives fighting kobolds for their coppers.

    in other words, I find that 3d6 in order really helps in generating a real-seeming character. "why is his charisma *so* low?"

  15. >as far as the "realistic heroes" argument, I think it was someone from the on-line dwimmermount game who suggested that it would be the types who weren't cut out for an honest trade that would risk their lives fighting kobolds for their coppers<

    No offense Bri, but this theory is such BS (hear it a lot on DF). The dudes who go into sports, cave diving, mountain climbing, or into combat tend to be above average athletes, not Elmer Fudds.

    And there are only one kind of adveturers who fight kobolds for coppers - 1st level ones.

  16. According to anydice.com, highest of 6 x 3d6 comes out to roughly:

    Under 9: Negligible chance
    9: 0.3%
    10: 1.3%
    11: 4.4%
    12: 10.6%
    13: 18.1%
    14: 21.2%
    15: 19.4%
    16: 14.15%
    17: 7.9%
    18: 2.8%

    So, pretty much impossible to not get a stat that's above average, with a strong curve basically centered on 14.

  17. Extreme sports people fail to compare anything approaching a medieval template, as far as I can see. Perhaps an adventurous aristocrat in the renaissance, but that doesn't align with my image of D&D adventurers from the early period of the game.

    Of course, post 3rd edition, it seems that most parties were composed of heroes with comic-book style power-levels (think Batman, not Superman). Fredrik, Black Dougal and Morgan Iron Wolf seemed to reflect a different expectation of what players thought constituted a cool set of rascals to roleplay. But then, when the Moldvay book came out, our collective heroic image-template had not been saturated with quite so many muscle-bound Arnie and Stallone tropes.

    I'll always remember with some amusement, however, one of players complaining after rolling four sixes and being told that, no, it was NOT allright if he took the extra six and applied it to another score...

  18. It's 3d6 x 6 in order for me.

    If you're such a fool as to conflate character "stats" with character "survivability", you were not long to live in one of my games under any circumstances.

    It's amazing how few people realize how loosely ability scores map to success in actual play in classic D&D.

  19. The push to abandon 3D6 in order follows naturally from the tinkering Gary did after OD&D to increase the effect stats have on the game. If he's going to make stats so important, then players are going to want high stats.

    This is one of those things you can imagine him adding to increase Gygaxian naturalism, to better represent the real differences among people, but one wonders if he realized when he started down this road how much was going to change as a consequence. It led among other things to a stat arms race, to all characters resembling each other in their lopsidedly excellent qualities.

    In hindsight, I find myself persuaded to return to the older model in which stats don't have nearly as much of an effect on the game, which liberates players to accept a wider range of stat scores in their characters, and hence more real character diversity.

  20. @Brunomac

    My characters aren't so much on the mountain climbing. They might have more in common with mall cops, meth cookers, street preachers, and con men.

    And there are only one kind of adveturers who fight kobolds for coppers - 1st level ones.

    In other words, every adventurer once.

  21. @ Rick Marshall: I think you're exactly right.

    I posted my thoughts on the subject here.

  22. Well, for those interested, there's no need to create your own scripts for checking the odds of different chargen methods. Anydice is very easy to configure to give you what you need

    For example for 6x3d6 keep highest, the script is this one http://anydice.com/program/120

  23. It wasn't a big deal writing my script, but Anydice is sort of nice.

    Superimposing the three methods I talked about above in a graph at Anydice makes me like the 4d6 drop second lowest even more.

  24. I've just read TSR's Buck Rogers rpg, based on AD&D2, and alongside the familiar methods, there's also 10+1d6, which I've never seen before.

  25. I strongly believe that heroism is not a product of strength, intelligence, or other quantifiable characteristic. So, to me, it seems like a really, really good idea to minimize the effects of those things in game play. That doesn't make the characters "Elmer Fudds" at all, it means that the players have to exhibit, through their characters, the actual, unquantifiable, qualities that make a person heroic: courage, conviction, moral fortitude, kindness, and so on. The things that the naïve equate with being "better" than the run-of-the-mill of humanity, such as physical strength, book knowledge, and so on, are not the things that make heroes, not in the real world. See Nausicaä of the Valley of the Wind for an example of a character whose traditional gaming qualities are not particularly strong (she is a young girl, with no particular strength, amazing intellectual ability, combative skill, or much of anything except for an ability to pilot a light aircraft, an insatiable curiosity, and an absolute fearlessness and kindness), but whose heroic qualities are far in advance of most.

    Verification word: intgasm. I don't think I even need to comment on that one.

  26. I understand the "need" for high prime requisite scores in AD&D, but what I don't understand is the desire to completely eliminate low scores (with "low" often being defined as anything under 9!)

    Just as mechanical bonuses for high scores in AD&D don't kick in until 15 or 16, mechanical penalties for low scores don't strike unless a score is 7 or lower, and the two penalties that can really imperil survival (armor class and hit dice) don't occur unless the character has a Dex or Con score of 6 or lower. Is a fighter with high Str but with Int or Cha of 7 really a hopeless character? Does a 6 Str really cripple a (single-classed) magic-user that much?

    I guess the fear of low scores is a result of widespread use of the "roll your stat or less on d20" mechanic (starting in modules and later promiscuously adopted by DMs). I have always thought that this mechanic has no valid reason to exist at all in AD&D, with all the alternatives available.

    Strength check: The PHB offers three different probabilities for accomplishing a feat of strength: open doors, bend bars/lift gates, and open magically-locked doors, in order of difficulty. Also, the PHB gives a simple formula for the amount of weight a character with a given Strength score can lift, and the DMG extends that formula to percentile Strength scores.

    Dexterity check: This is massively abused as an ersatz saving throw (i.e. "roll under your Dex or fall off the rope bridge to your death") Why not just use an actual saving throw, adjusted by the character's defensive Dex bonus? Petrification is a good general one to use that doesn't favor any class. Use Dragon Breath for hazards that you think fighters should be better at escaping than squishy spellcasters.

    Constitution check: Like the Dex check, this is most often abused as an ersatz saving throw (though probably not nearly as often as Dex) Like the Str check, there are multiple alternatives. You've got saving throws, system shock, and the DMG has formulas for the probability of contracting a disease or parasite and for the severity thereof.

    Charisma check: It's called a reaction roll.

  27. kelvingreen said...

    I've just read TSR's Buck Rogers rpg, based on AD&D2, and alongside the familiar methods, there's also 10+1d6, which I've never seen before.

    "the dark eye", the german system i started playing with, used this method as well (in 1984 i think). the 5 stats were all d6+7.

  28. We use a variation method. We roll 4d6, re-roll ones or twos, drop the lowest, fit where you want them. We do this for two reasons, 1. Heroic characters are supposed to be above and beyond the norm, I like to think of my PCs as people like Achilles or Hiawatha, ie nearly legendary type of figures who will become stories for many generations. And 2, it helps people play the type of character they want to play instead of something they got stuck with.

  29. I'm currently working on an article about distribution of results from various multiple dice rolls. I don't really like mathematics but I think it's fascinating when you look at the spreads. I'm also going to include some monster attacks that do multiple damage dice.

  30. "In hindsight, I find myself persuaded to return to the older model in which stats don't have nearly as much of an effect on the game, which liberates players to accept a wider range of stat scores in their characters, and hence more real character diversity. "

    This makes me laugh. Instead of everyone being the same with high stats, people are more diverse if they have different stats that just don't matter.

    Either way they are exactly the same, if you give everyone straight 18's or give everyone a random stat from 1 to 20..but they don't do anything, then all of the characters will be the same mechanically either way.

  31. Am I the only one that used a d20 (moving all 1's and 2's to a 3, and all 19's and 20's to an 18)?

  32. Instead of everyone being the same with high stats, people are more diverse if they have different stats that just don't matter.

    This is true if stats are used as role-playing hooks rather than as game mechanics. For example if Intelligence doesn't have much of a game impact beyond magic-users, then the range of Intelligence available for non-magic-users gives you more possibilities to play with - dumb clerics, smart fighters, mediocre thieves - you get the whole range. But if Intelligence is also governing things like how many skill points you have available, you're going to be much more leery about playing a "dumb" character. Dexterity is another good example - no one wants to be an "oaf" in a game where Dexterity has a major impact on AC and saving throws. If the majority of the game mechanics you use on a round-to-round basis come from class abilities (as in OD&D), that leaves the field open for your stats to be basically role-playing suggestions.

    Stats that have game-dependent effects tend towards homogeneity in characters mechanically, which in an RPG tends to lead to characters that are role-played the same way. I saw this as a real problem for 2nd edition AD&D and it was made even worse in 3rd edition D&D. 4e tries to mitigate it by allowing for pairs of abilities to determine mechanical bonuses (but they do different things to encourage homogenous characters, so its a trade-off).

  33. I seem to remember playing at one point using 3d6 in order, then allowing players to increase a stat 1 point by dropping another 2 points. (OK, your cleric can be wiser, but he's going to be REALLY clumsy.)Was that ever an official suggested method, or something we just houseruled?

  34. @Jonathon:

    I seem to recall a suggestion to that effect, but that kind of point trading is something I remember more from Chaosium games.

  35. I remember that the Unearthed Arcana had methods included that practically guaranteed you would have an 18 in the primary stat for your class. I remember wondering what the point was. You may as well assign stats then (which for certain types of games is fine, but not something I care for as a rule).

    We usually used the 4d6 and drop the lowest, but sometimes we did 3d6 once for each stat, but allowed you to pick which stat to assign them to.

    Sometimes we tried adjustments, taking two points off of one stat to raise another by a single point. A lot of low charisma scores came out of that idea and we got back to almost always having 18s in the primary stat.

    Did anyone else try allowing adjustments like that in their games? Did it work out or not?

  36. Jonathon:
    2-for-1 swapping is the rule in B/X (Moldvay) and BECMI (Mentzer) Although, IIRC, it was accidentally (?) omitted from the Rules Compendium. I don't know if it's in Holmes or not.

    In Moldvay/Mentzer you can only raise your Prime Requisite, you can only lower Strength, Intelligence or Wisdom, and you can't lower a score below 9 (a rule which many players seem to have internalized as "PCs aren't supposed to have any scores below 9")

  37. >It's amazing how few people realize how loosely ability scores map to success in actual play in classic D&D<

    Will, I think you often miss the point on this. It isn't just about the bonuses. It's about concept of character. How strong or how fast or how charming are all intuitive things that help you define a make believe character. At least for me, but that might just be my old sports background where you are often perceived by how far you could throw a football or how fast you can hit a heavy bag.

  38. In my AD&D days, we used the roll 3d6 six times for each ability method. It wasn’t for survivability or any need. We just observed how big the flat spot in the middle of the tables was (on top of the wide spot inherent in 3d6) and felt it was more fun if all PCs had at least one bonus from an ability score.

    (This is also one of the reasons I tend to prefer ability checks against the actual score instead of a modifier. Why have a 15 point range if you aren’t going to use it?)

    Not to mention the prereqs some classes had.

    Playing classic D&D these days, we’ve found 3d6 in order with the 2:1 trade or swap a single pair to be fine.

  39. "Will, I think you often miss the point on this. It isn't just about the bonuses. It's about concept of character."

    To be fair, you *did* invoke what could be termed the survivability fallacy in one of your own recent posts on the topic.

    "So 4D6 is the way for me, and I think it is the most popular. I’ll even allow an elimination roll for anything under 9. Whatever it takes to get you a decent, survivable character you can be happy with and get the game under way."

    It's not the concept of character notion I'd object to.

  40. Will: Well, I think I was pretty much equating 3D6 with that original style of DM'ing. Sub-par human specimens walking into a humliating death trap. Those things went hand in hand and 3d6 might as well be used for it, because for me the trap laden dungeon crawl gets old anyway. I think allowing 4d6 or similar methods are more suited to heroic play, which is what I like these days (since the 80's, really).

  41. This discussion highlights for me the weird fact that players' ability to invest in a character depends on how "attractive" we find that character's scores. Scores with little or no mechanical meaning (e.g. Charisma, Intelligence) can be the most important. That is, I don't think this has to do with mechanics at all.

    Ironically, in a 3d6 system, getting an 18 or multiple 17s was a kind or curse-- you didn't want to play that character too aggressively for fear of losing those treasured scores.

  42. Many seem to ignore the inherent gaming glory involved in dealing with crap rolls without all of those plaintive moans- isn't anyone up for a challenge anymore? I love this thread over at Knights & Knaves about one of Trent Foster's games:
    It starts with a joking discussion about how long his new character could survive considering his strength score of 3. The character went on to be something of a legend in that campaign. The reason? T. Foster's good attitude. He seemed to be aware that he was playing a game, and the disadvantage just whetted his stoic competitive spirit. You know you have EARNED the fun you have with a character like that, IMHO.

  43. I think "one of the lowest" simply means that if two of the dice come up "1," for example, you only discard one of them.

  44. As the different editions of D&D progressed, ability scores became more and more important. Especially when skills and saves are tied to them, then they become more important than levels. In some ways I prefer higher than average stats, with bonuses starting at 12, because the country bumpkin who starts out wimpy and ends up champion of the universe is so cliche as to be goofy.

    On the other hand, being disgusted with the complexity of it all, I stripped out skills and saves and folded them both into a Ability Check system where your stat bonuses help determine your target number, so YMMV.

    I usually use an extremely simplified system of point-buy, alternating with the roll-4d6-drop lowest mechanic. Absolute randomness can be silly: when you go to a restaurant do you roll for what you're going to order?

    Word verification: pronc (a piece armor usually worn around the arm that plugs in to your shoulder).

  45. I was in a straight 3d6 game- we ended up with a fighter with 18:76 str, and 4 cha with hte full helm on; a wizard who took two aging categories to get is int up to 18 (he learned magic from the dwarves...); and me a dwarven cleric with an 18 wis, but an int of 7. the stories from that game are some of my favorite.

  46. I prefer 3d6 in order. Then figure out who that is, and play it.

  47. I prefer to just roll 3D6 enough times over and over again, until you reach 18, 6 times.
    What, yet another 18 ? I just can't believe it !