AD&D 1E What is the best way to generate ability scores for 1e AD&D?

[Mannahnin]

Back in the day my AD&D groups used Method I (4d6 drop low, arrange), but we gave each character three sets, pick your favorite.

One set was not going to reliably get you decent stats by AD&D standards. You'd almost always run into the "1-2 people get lucky*, 1 person gets screwed, everyone else has mediocre stats" pattern.

Giving each person 3 sets had SOME of that, but less extreme variance between the luckiest and the least lucky.

*(Although "lucky" in this case still usually only means getting two stats high enough for "real" bonuses. The average score of 13.27 Method 1 gives you is nice by B/X standards but junk when you're using the AD&D tables for bonuses).

 

[Celebrim]

Definitely partial to Method III. Like you said, not a super high chance of an 18 (about 15%) in the array, and rolling in order generates the “Huh, I didn’t think I would play that” effect which is more desirable in a strongly random game like AD&D.

If you want to give players more control over their class choices, use method III but one stat of their choice becomes an 18 is interesting.

I think method III is definitely my favorite, though it might not be great for optimizers because fighter classes really shine with an 18 to play with, and can almost ignore anything outside of Strength and Constitution as long as they have at least 9's or so elsewhere. The bad thing is that attribute dependency is both high and low at the same time. You are two or three times more powerful if you have 18's in your key attributes, but you hardly don't care what is in your other scores. So if you are power gaming, the fixed array and or the 5d6 take 3 and arrange to taste method will probably work better for you.

 

[Celebrim]

Back in the day we used 4d6 drop the lowest, arrange as you see fit. I never used it, but I kinda like BECMI's option to drop 2 points from another stat to raise your prime requisite by 1.

The problem with this is that AD&D (and I'd presume BECMI) has so much single attribute dependency that it's pretty much always a good idea. For most classes only your prime requisite and your constitution matter, and intelligence is basically a dump stat for all characters that aren't M-Us. There was a period where I was allowing this multiple times, and the result was always several 18s and several 6s and 8s, and the 8s never mattered.

 

[PHATsakk43]

If you require rolls before anything, 1d12+6 straight down the line. Its swingy, but swingy around a moderate number. Possibly can get sub par scores (7 or 8) but nothing less and roughly the same chance as a 17 or 18.

Pretty reasonable success rate to get any class/race combo except paladins.

Seriously, give it a shot and roll up a few characters and see how it works.

 

[cbwjm]

I'm normally a 4d6k3 arrange as desired kind of guy. Back when I ran 2e in high school I allowed 8x 4d6k3 and discard the two lowest scores.

 

[Celebrim]

If you require rolls before anything, 1d12+6 straight down the line. Its swingy, but swingy around a moderate number. Possibly can get sub par scores (7 or 8) but nothing less and roughly the same chance as a 17 or 18.

Pretty reasonable success rate to get any class/race combo except paladins.

Seriously, give it a shot and roll up a few characters and see how it works.

If you were reading in the other thread you'll see that I recommended something like this as a possible "solution" to what the system was demanding of the attributes, except I recommended 1d6+12.

However, you'll see I don't put that among my final recommendations and the reason for that is that the most functional strategies all share that they roll at least 24 D6's in order to narrow the standard deviation. If you roll just 6d6 you will get the one rolled a Yahtzee of 6's and another guy a Yahtzee of 1's and everything in between. You still have that problem rolling 24D6, or 36D6, in some cases 108D6, but they are much less likely when you roll that many dice. You end up with pretty tight grouping, or at least tighter grouping.

And you'll see the other recommended versions people are offering have that in common. Someone combines method 1 and method 4, and rolls 72D6, and again that gets a really low standard deviation around some average. And someone extends method 1 and rolls 32D6 with a slightly different variation, and again that's going to have a slightly different average and standard deviation based on what they throw out, but the idea is fundamentally the same.

 

[Eyes of Nine]

Adjacent to this topic, was there ever a Dragon or other third party publication that provided rules for "18/00" for the other abilities besides Strength?

 

[Eyes of Nine]

Also, here's a dumb idea:

1. Roll 7d6, take 3 of them and apply them to any ability. Set aside the other 4 dice .
2. Then roll 3d6 down the line (and back up to the top again if necessary).
3. Now, take the results of those other 4 dice and you can each die result as wish to your stats, including the original stat you applied the first roll to; and you can add more than 1 die to your stats. But once the die is used, now its gone.

Example:

• Roll 7 dice: 1, 2, 3, 3, 4, 6, 6
• Take the 4, 6, and 6 and assign those to say... Dex
• Then roll 3d6 down the line. I just did it in Roll20:
• 
• Crazy above average roll. So now my stats are:
  • STR - 11
  • INT - 11
  • WIS - 17
  • DEX - 16 (my original roll)
  • CON - 14
  • CHA - 13
• I've got a 1, 2, 3, and another 3 to apply from those set-aside dice. I add the 3 to my CON, the 2 to my DEX, a 1 to my WIS, and the other 3 to my... CHA.
• For a Monty Haul character who's got:
  • STR - 11
  • INT - 11
  • WIS - 18
  • DEX - 18
  • CON - 17
  • CHA - 16

Maybe only roll 6d6 to start and reserve 3 of them or something.

 

[GreyLord]

Any method in the DMG. (Added Note: UA methods not included in this, if one used UA, they still had to roll in front of me, and that was what they got and no one else could use it...).

After players have rolled the numbers, any player can use that set for their characters. Thus, if someone got lucky and rolled an 18, everyone could use that rolled set of numbers if they wanted to.

Makes it so no one has the best ability scores because of their rolls, no one is unhappy, and no one is complaining.

Oh...and yes...everyone has to roll their ability scores in front of me and I record it down. Makes it easier to verify they are legit, and makes it easier for everyone else so that they can see what those scores are as they happen.

Edit PS: Almost always 4d6 drop the lowest was the most popular method used.

 

[Celebrim]

[*]I've got a 1, 2, 3, and another 3 to apply from those set-aside dice. I add the 3 to my CON, the 2 to my DEX, a 1 to my WIS, and the other 3 to my... CHA.

The idea is creative and it deserves credit for that, but it has the same general problem that the BECMI idea of subtracting from your stats to add to a different stat has. The totals aren't generated linearly so if you modify them linearly, it's a very large modification. An 18 is 1/3rd as common as a 17. A 16 is twice as common as a 17. So even a +1 bonus is a large bonus. Whatever leftover dice you have will produce a very large number of extreme results almost every time.

I wouldn't even be surprised if the correct strategy here is to counter intuitively save the large dice. If you instead of making a 16 make a 10, saving the 6, 6, 2, 1, then your final results are 18, 17, 17, 16, 13, 10. That is perhaps slightly worse than the results you had, but only because I'm assuming no overflow. If I can overflow the stack, then I could do 18, 18, 18, 13, 11, 10

And of course, because you are really only rolling 22 dice, and because it matters so much what the first 7 are since they matter more than the rest, you'll get pretty large deviations.

Maybe only roll 6d6 to start and reserve 3 of them or something.

All humans are inherently bad at probabilities. It's a blind spot in our supposedly "general intelligence".