D&D 5E How do you avoid overshadowing while applying ability scores as rolled?

[cbwjm]

I tend not to worry about trying to avoid overshadowing, just roll the dice and start playing, I follow the old Gygax method (which I was reminded about by Matt Colville) for my current game and just let everyone reroll until they had at least 2 15s.

Admittedly, I have thought about requiring that they have the 2 15s and that total modifiers before racials add up to between +5 and +8. Still gives a wide range of abilities and after racials you still end up with a range of power but it tends to truncate it a little.

 

[Hriston]

This follows up my earlier thread. For my next campaign I want my players (who have high system mastery) to allocate their scores as rolled. I also want a lower range than in the PHB i.e. nearer 3d6 than 4d6k3.

People pointed out the undesirable possibility of overshadowing. Where a player with less lucky rolls is in the shadow of another's very lucky rolls.

How do you mitigate or avoid that?

[EDIT] I like the redrick roller mentioned in my previous thread, and want to consider also 3d4+4, or bounding rules like "must nett positive" or "must sum to 60" as others suggested.

The issue this thread is asking to address, if I understand correctly, is the “swinginess” of 3d6 that some of the posters brought up in the other thread, the problem being that the distribution curve of 3d6 is somewhat “flatter” than that of 4d6 drop lowest, thus producing proportionately more high and low results relative to 4d6 drop lowest and proportionately fewer middling results, again relative to 4d6 drop lowest. This is what I understand those posters to have been saying, and I agree that this is a property of using 3d6, and since I don’t have any experience using 3d6 score generation in an actual game that I can remember, I have no reason to doubt that the swinginess was a noticeable issue at their tables.

You can remove the relative swinginess but get the same average by using “averaging dice” which are marked 2, 3, 3, 4, 4, 5, or use the “averaging method” that Gygax gives for “general characters” in the 1E DMG, which is to roll 3d6 and count 1’s as 3’s, and 6’s as 4’s. This will produce an average result of 10.5 and a range of 6 to 15. This may make extreme results a bit too rare for adventurers, however.

If you like the high proportion of middling results but want the full 3 to 18 range, you can use 15d2 - 12. It produces a result very close to a normal distribution.

 

[Bawylie]

This follows up my earlier thread. For my next campaign I want my players (who have high system mastery) to allocate their scores as rolled. I also want a lower range than in the PHB i.e. nearer 3d6 than 4d6k3.

People pointed out the undesirable possibility of overshadowing. Where a player with less lucky rolls is in the shadow of another's very lucky rolls.

How do you mitigate or avoid that?

[EDIT] I like the redrick roller mentioned in my previous thread, and want to consider also 3d4+4, or bounding rules like "must nett positive" or "must sum to 60" as others suggested.

Don’t know if this got brought up, but if you want a 3d6 distribution, I find my players prefer some control over the outcome. So when I do 3d6 down-the-line, I also do this:

Any array of abilities that comes out less than +2 net should be re-rolled.
A player can switch two any two numbers - twice.
(Meaning I might want to play a fighter but I rolled an 8 Str an 11 Con and a 15 Cha. So I can switch the 8 and 15. And if I want to, I can make one more switch, too).

The floor of “no less than +2” and the customization offered by allowing 2 swaps is generally sufficient.

I’ve done two campaigns like this. Both in 3.5. I think one was in Ravenloft and the PCs were normal people discovering the horror of the land they were in. Anyway, it worked.

Probably some judgment might help too. If one person is notably behind the rest of the party, tossing them a +2 isn’t going to hurt.

 

[clearstream]

You can remove the relative swinginess but get the same average by using “averaging dice” which are marked 2, 3, 3, 4, 4, 5, or use the “averaging method” that Gygax gives for “general characters” in the 1E DMG, which is to roll 3d6 and count 1’s as 3’s, and 6’s as 4’s. This will produce an average result of 10.5 and a range of 6 to 15. This may make extreme results a bit too rare for adventurers, however.

For some reason I feel drawn in this direction. Maybe 3d4+3.

A few asked the reasonable question of motivation @Umbran I'd like to push my players off comfort picks, and away from party balance or high optimisation. I'll be happy if that leads to four wizards or whatever! Class choice will come after roll so best stat will presumably inform that.

After reading all responses so far I feel less worried about overshadowing. I'd like to see a bit more RP than in my previous campaign, which was very crunchy. Stats will still matter as I also consciously want the base material to set a higher difficulty level at every tier. I'm expecting some deaths and a few characters to reach 15th.

 

[Harzel]

For some reason I feel drawn in this direction. Maybe 3d4+3.

You will end up with a lot of the stats being near the mean, which somewhat diminishes the impact of assigning them in order. Given your goals, I think you should consider a card-based method. It retains some of the random feel of dice, while putting a pretty tight constraint on the overall goodness of the set of stats, which addresses* your overshadowing (due to stats) problem.

In my current campaign I used playing cards to have players generate their PC stats, with the numbers acquired being assigned in order. Due to the set types of cards, while there is slight variance between PCs, they all maintain a pretty good balance to each other. The method was also really useful in getting some players to actually consider the Standard Human, as the method generates 4 "odd" scores fairly often, and the "get six +1s" bumps most of their stats to the next modifier.

Make a deck of 12 cards:
two 4's
two 5's
two 6's
three 7's
two 8's
one 9

The player shuffles them up and then deals six piles of two cards each. As they flip each pile, they add the numbers together and assign them to the six ability scores in order. Once they have their six stats, they can then choose their race and class based off of them.

Most PCs will end up with at least one (if not 2) stats of 15+. A few players will luck out and get a 16 or 17 (which will let them jump up to a +4 mod with the appropriate race selection.) And you may have the unluckiest individual who might end up with 12s and 13s across the board, but that is exceedingly rare. So you might decide to put in a rule that says if a PC has no +3 to at least one ability score after racial bonuses, they can re-draw.

That particular deck is a little rich for my taste - it adds to 76 - but by modifying the composition of the deck it's fairly easy to adjust the total, the min and max scores and the expected amount of variance in the scores. Also, you can get somewhat different effects using a deck of 18 cards (with values, e.g. 1-6 or 2-6 or 2-5) instead of 12 cards.

* It doesn't solve it completely, of course. The only way to do that is to give all PCs the same stats.

 

[iserith]

For some reason I feel drawn in this direction. Maybe 3d4+3.

A few asked the reasonable question of motivation @Umbran I'd like to push my players off comfort picks, and away from party balance or high optimisation. I'll be happy if that leads to four wizards or whatever! Class choice will come after roll so best stat will presumably inform that.

After reading all responses so far I feel less worried about overshadowing. I'd like to see a bit more RP than in my previous campaign, which was very crunchy. Stats will still matter as I also consciously want the base material to set a higher difficulty level at every tier. I'm expecting some deaths and a few characters to reach 15th.

It's not entirely clear to me what you mean by "a bit more RP" here, but if you mean "striking a balance between relying on their bonuses and abilities and paying attention to the game and immersing itself in its world," then I'm not sure this approach with ability scores is going to achieve that. That's better achieved in my view by balancing the use of dice against deciding on success. That's on the DM side by ruling accordingly based on the players' description of what they want to do. See DMG pg. 236-237. You'd need to get your players to play differently, too, by stating what they hope to achieve and how.

If their ability scores are by some measure lower than the 27 point buy or standard array, combat challenges may be slightly more difficult is all which may lead to more of the sort of "crunchy talk" that you seek to avoid as the players work harder to survive under the new paradigm.

 

[Guest 6801328]

It wasn't an attack. My apologies if the prhasing left you feeling that way.

I blame the self-doubt and maybe even PTSD you must be experiencing from having kicked me out of that other thread. It's ok, man, really. I'm fine. You can forgive yourself.

Don’t know if this got brought up, but if you want a 3d6 distribution, I find my players prefer some control over the outcome. So when I do 3d6 down-the-line, I also do this:

Any array of abilities that comes out less than +2 net should be re-rolled.
A player can switch two any two numbers - twice.
(Meaning I might want to play a fighter but I rolled an 8 Str an 11 Con and a 15 Cha. So I can switch the 8 and 15. And if I want to, I can make one more switch, too).

The floor of “no less than +2” and the customization offered by allowing 2 swaps is generally sufficient.

I’ve done two campaigns like this. Both in 3.5. I think one was in Ravenloft and the PCs were normal people discovering the horror of the land they were in. Anyway, it worked.

Probably some judgment might help too. If one person is notably behind the rest of the party, tossing them a +2 isn’t going to hurt.

I like this.

I assume by "less than +2" you mean the sum of all the modifiers?

EDIT: I just wrote up a quick monte carlo sim to test this, and it took an average of 3 tries, using straight 3d6, to get a set of scores with a net modifier of +2 or better.

from random import *

def roll():
  return (int)(random() * 6) + 1

def rollAbility():
  return sum(roll() for i in range(3))

def mod( score ):
  return (score - 10) // 2

def modSum( scores ):
  return sum(mod(x) for x in scores)

def rollAbilities():
  return [rollAbility() for i in range(6)]

total = 0
iterations = 1000
for i in range(iterations):
  count = 0
  mods = 0
  while mods < 2:
    mods = modSum( rollAbilities() )
    count += 1
  total += count
total / iterations

 

[5ekyu]

For some reason I feel drawn in this direction. Maybe 3d4+3.

A few asked the reasonable question of motivation @Umbran I'd like to push my players off comfort picks, and away from party balance or high optimisation. I'll be happy if that leads to four wizards or whatever! Class choice will come after roll so best stat will presumably inform that.

After reading all responses so far I feel less worried about overshadowing. I'd like to see a bit more RP than in my previous campaign, which was very crunchy. Stats will still matter as I also consciously want the base material to set a higher difficulty level at every tier. I'm expecting some deaths and a few characters to reach 15th.

" I'd like to push my players off comfort picks, and away from party balance or high optimisation. "

Ok just wanting to add that in my experience, the lower stats overall combined with the other aspects of choice etc... this will expand and exacerbate any issues that optimization and differences in system-fu will create. It increases the impact of those actually seen in play, at the table.

I d9nt know why you would want to push away from "party balance" or what that term means to you, but 3d6 in order etc would make optimization and system-fu much more a benefit than it would if you just had say standard array.

So if your goal includes moving the play away from those being important, this, to me, is the opposite direction I would take.

 

[DEFCON 1]

You will end up with a lot of the stats being near the mean, which somewhat diminishes the impact of assigning them in order. Given your goals, I think you should consider a card-based method. It retains some of the random feel of dice, while putting a pretty tight constraint on the overall goodness of the set of stats, which addresses* your overshadowing (due to stats) problem.



That particular deck is a little rich for my taste - it adds to 76 - but by modifying the composition of the deck it's fairly easy to adjust the total, the min and max scores and the expected amount of variance in the scores. Also, you can get somewhat different effects using a deck of 18 cards (with values, e.g. 1-6 or 2-6 or 2-5) instead of 12 cards.

* It doesn't solve it completely, of course. The only way to do that is to give all PCs the same stats.

Yeah, you have to okay with the slight possibility of someone starting with an 18 or 19 if they select a race that takes advantage of where their cards landed. I myself did not have any issues in that I think for my 14 players in both campaigns, only two of them drew the 8-9 combo for a 17, and the one who took advantage of it was a new player who got the 17 in CON and then selected Dwarf, giving them a total of 19. And in that case... having this new player playing a Dwarf Fighter with a +4 CON modifier helped offset his inexperience and lack of really optimized combat tactics. So it hasn't been an isuse at all. The other player also drew the 17 in CON but really wanted to play a half-elf, so they kept the +3 CON as-is, which matches many other players who started with +3s in their primary stat.

I myself probably wouldn't want to do three-card decks with numbers from 1-6 or 2-6, because I would want to eliminate the possible 5s and 6s that could come up as someone's ability scores. My method made the single minimum as 8, and only a couple chances of 9s, which is what I preferred. Also, if I was doing three-card piles... at that point I might as well just use the dice as normal.

To reduce the possible high end numbers in a 12-card deck you could go something like: two 4s; three 5s & 6s, two 7s & 8s, which would still give the possibility of that illusive 16 to start with (allowing for the rare +4 modifier at the start of the campaign), but usually will result in lots of 13s and 14s and occasion 15s for those who are lucky (and the ones who get the 15 will usually end up with an 8 or 9 to balance it.)

 

[Oofta]

I know we have a barbarian with INT 16 and a female wizard with STR 18! Neither would happen with Point Buy, & it's certainly made them more interesting characters.

Are primary stats are also high, even if not spectacular? Because sure. If I can have a decent int and a high strength, why not? Might be fun.

But that's the exception to the rule unless you use a system that skews to high numbers. I was talking about any game I've ever played that used point buy. I've never known, nor cared, what my player's scores were.

In any case, people can do whatever they find works for them and their group. Just put me into the "if you roll for stats I don't want to participate" crowd.