Back to Top


Gnomish Workshop: People Building

This article was first broadcast in Episode One Hundred Eighty Two on 22nd September 2021.

(Sound of grinding gears)
Ryu: What is he doing in there? 
Lennon: Maybe he’s upgrading ROSTRO? 
ROSTRO: If you are interested, my processing faculties were largely taken up with an analysis of persona synthesis and optimal configuration techniques of same. 
Ryu: If we don’t take an interest, is Ostron going to wake up? 
ROSTRO: Probability suggests it is not a likely outcome.
Lennon: Ugh, fine, what are you on about this time?
(The screen powers on)


When creating new characters for a campaign or a one-shot, a frequent point that gives both players and DMs pause is the issue of ability scores. There is often a lot of thinking and wrangling about what statistics to put scores in, but before people can even get there everyone has to decide what scores are being used in the first place. 

Now for a lot of longstanding groups this is a non-issue; the group figured out how to generate ability scores sometime back when they could verify the stats of a dinosaur in the monster manual by going outside and looking for one, but for new groups, players, and especially DMs it’s worth reviewing where these numbers can and do come from. 

Conceptually there are three primary methods for generating the numbers required to fill out the ability scores: use of a prescribed list provided by Wizards of the Coast, utilizing the so-called “point buy” system wherein scores can be increased through the expenditure of a predetermined pool of intangible resources, and rolling dice to generate the desired totals. 

It should be noted that with the exception of the prescribed list there are subtle variations in implementation for the other systems that can generate statistically varying results. Legacy methods of ability score generation also produce some interesting options and variety that some people may not be aware of. Each major variation shall be discussed here along with an analysis of the primary impact to eventual character profiles. 

We’ll start with the easy one. With the standard array everyone is using the same numbers for their characters, specifically: 15, 14, 13, 12, 10, 8. The six scores can be assigned to any abilities the player wants. Now everyone knows that after character creation you almost never look at the actual score anymore; all anyone cares about is the modifiers. In this array you end up with two “+2” modifiers, two “+1” modifiers, one zero, and one “-1” modifier. That produces a total ability modifier total of +5 (2 + 2 + 1 + 1 + 0 – 1. We’re noting the ability modifier total for two reasons. First, it’s a measure that’s often used to determine if an array is “valid” or not, particularly with ability score generation that involves rolling. Second, it’s a decent measure of how competent the character is going to be overall, although the specifics of that can vary a lot; a character with an ability score total of +6 could have a decent spread of +2s, +1s and zeroes or they could be rocking two +4s and a -2 somewhere. Either way, though, as a whole those two characters are going to be equally competent, it’s just that one character is going to be decently middle of the road with a few specialties while the other one is going to be the obvious person to solve some problems and will not even be told about certain others.

The standard array allows characters to have two areas of specialty, two slight competencies, one completely average score, and one “dump stat”. The presence of a 15 and a 13 means they either immediately have the ability to take two of their scores and improve them to bump up their modifiers, or they have one they can target with their racial +1 score to improve it immediately. 

The concept of a standard array has been around since 3rd edition, but as an optional rule; it was only present in the Dungeon Masters’ Guides for both 3 and 3.5, and it used the same array as 5th. It was first presented as a default option in the player’s handbook in 4th edition. That edition’s standard array was more impressive though; players got a 16, 14, 13, 12, 11, and 10 to assign. Apart from moving the ability modifier total to +10, it also introduces the possibility of someone using their racial bonus to start the game with an 18, or +4 modifier, while still providing two scores that will immediately improve the modifier with only 1 point added. Some would say that makes sense because 4th edition characters were ridiculously powerful and this just proves that.  

The primary advantage to implementing a standard array is that it saves time and reduces confusion among players trying to figure out a mechanical system of generating ability scores and then subsequently requiring them to determine the distribution of said scores based on their desired character profile. With the standard array the former concern is eliminated, though there remains the issue of the players determining an optimal distribution for the ability scores. 

As mentioned, the standard array from fifth edition provides a character with a wide range of competencies and only forces one instance of a sub-par area of performance. This will produce a group of characters able to generally tackle any given obstacle with a similar average competency; any experts in fields will be dictated by class competencies more than ability scores. The characters will also be at an objectively average level of power. 

Using the fourth edition array may be desirable if the players or DM feel that characters produced by the default array are not impressive enough or the campaign is intended to be a higher difficulty, as the power levels both by default and after applying ability scores will produce a group of characters showing above-average competencies. 

Moving on from standard, the next closest thing is the point buy system. This is another advent of 3rd edition, though in both that and 3.5 the system was hidden in the Dungeon Master’s Guide again, and only made it’s way to the Player’s handbook for 4th. 

With this system, every ability score starts as an 8. The player is then given 27 points to spend to improve the scores based on a table in the rules. For 5e, it costs one point for each score increase until you get to 13. Then it costs two points to move from 13 to 14, and two again to move from 14 to 15. Players are not allowed to increase scores above 15. With this system, the most lopsided array a character can generate has three 15s and three 8s. It’s also possible to generate the scores used in the standard array exactly. 

The point buy system for 5e caps the possible ability modifiers at +3 and the total at a +6, just like the standard array, but like we just mentioned it allows for more variety and extremes in application. It also provides the players with a way to ensure none of their scores are negative if that’s something that bothers them. 

Now the system from 3.5 was a little different. It still started with six 8s, but the scale was different; improving the scores cost only one point up to 14. Then it cost two more to get to 15, and two more to get to 16. Three more points would get you to 17, and three on top of that would give you an 18. That system only gives out 25 points by default so no one can have more than one 18 in their array but even being able to buy up that high meant there was a lot more variety in the arrays that could be produced. It still wasn’t possible to get an ability score array with modifiers totalling above +7, and most of them would end up lower, but you would have the option of making characters who were total experts at one or two abilities. 

4th edition also had a point buy system but like the standard array it was more powerful overall. We won’t rehash the details of it but suffice it to say it was possible with that system to generate arrays where a character would start with an 18, multiple +1 modifiers, and no negative modifiers. 

A few modifications to the point buy system have been suggested or implemented in D&D’s history beyond the differences in cost scaling and score ceilings. In third edition and 3.5, alternate point totals were suggested for DMs or players who wanted their characters’s abilities to reflect different backgrounds or challenge levels for the campaign. A total of 15 points was recommended for a so-called “low-powered” campaign, while high-powered ones were allotted 32. 

Second edition provides a unique hybrid system similar to point buy but providing far more variety. In this, each score is initially an 8. Then seven d6s are rolled. The values are maintained separately and added to the scores as the player desires, with the caveat that no score can exceed 18. The totals of the dice can be added together to provide greater increases (for example, a roll of a 2 and a 3 can both be added to the same score to provide 5 extra points), however the sum total cannot exceed the limits imposed; if a character has rolled two sixes, they cannot apply both to one score and discard the extra two points to remain at 18; they must locate a rolled four among their dice pool or make do with an alternate score. 

This method would make the resulting character profiles wildly different from one another as not only would the players be able to allocate scores differently, they would not all be working from the same available pool of points. Also there is no mechanism for recourse if a series of substandard rolls results in the official ruling, so it is possible a player could be left with a lower power character simply by virtue of bad rolls. A particularly fortunate player may also evoke envy from other players who were unable to generate a beneficial set of die rolls. It is therefore recommended that this method be used sparingly, such as with very advanced groups who desire greater volatility in ability scores or in characters for limited use situations, such as a single session adventure. 

If that all sounded complicated, buckle up, because it’s only downhill from here. 

Most people who’ve been around for D&D since third edition will know about the “roll 4d6 and drop the lowest” formula. You do that six times to get a set of scores that total within the normal ranges for ability scores…usually. There’s always some poor sap who ends up with two 10s, a 14, an 8, a 6, and a 3. This was the proscribed method for generating ability scores in third edition and 3.5,  was one of the three equal options in 4th edition, and is the first one mentioned in 5th just before the idea of a standard array is presented. 

The thing that makes this method confusing for some people is the fact that there’s what ROSTRO calls a “reroll threshold”; after generating the six scores, you’re supposed to take the *modifiers*, add them up, and see what the total is. If the sum of the modifiers is below +4 or above +8, you’re supposed to reroll all the scores. This is where a lot of new players get lost, because first they have to do the die rolling and the math to get their six numbers, then they have to convert the scores to modifiers, then they have to add them all together and include the negative numbers. 

This method on average tends to produce slightly more powerful characters. First of all the most likely results for the rolling are 12, 13, and 14, and it’s statistically unlikely you’ll get rolls for negative modifiers a lot. Also, with totals of +7 and +8 being valid, you can walk away with arrays that are better on average than the standard array or anything you can build with basic point buy. Also, while the power levels of the various characters will be more in line than with that random hybrid system, you’re going to see a lot of variation in the specific scores and capabilities if everyone rolled their own stats. 

There are actually a lot of different ways rolling dice were used to generate ability scores, but to find most of them you have to go back to 2nd edition or earlier. As Stephen Colbert told Matt Mercer when they sat down for a game, back then not only did you only roll 3d6 for the scores and just had to deal with whatever was rolled, you did the rolls in order. So roll one went to the strength score, roll two went to dexterity, and so on. A more “permissive” system suggested in the Dungeon Masters Guide at the time suggested allowing one reroll and one opportunity for characters to swap scores between two attributes. For reference, in 2nd edition it was common to roll the ability scores first, then choose a race and class based on those, rather than the other way around that is more common in 5th edition. 

The 2nd edition DMG did mention letting players assign ability scores as they wanted, but included a warning that this would lead to more time spent creating characters and attempts at min/maxing rather than simply “creating the character of their imaginations.” Apart from the limitations of having the ability score rolls locked to specific attributes, this method also produces less powerful characters than the “4d6 drop lowest” method because rolls of 3d6 average around 12, making the numbers 10 – 14 more likely. 

Another unique system suggested by 2nd edition involves even more rolling. The first starts with rolling for the ability scores in order, but allowing the players to roll twice and choose one of the two to use. That limits the chances of generating wildly powerful or skewed characters but it still can leave the player unsatisfied if they were forced to choose between two poor options for an ability they wanted more power in. 

The final dice rolling method used in official D&D is similar to the previous one but allows greater variety and freedom than other 2nd edition methods. In this, a player rolls 12 sets of 3d6, generating 12 scores. They choose six of the twelve and assign them to scores as desired. This alleviates or eliminates the scenario of a key score ending up with a substandard value as it is statistically unlikely for 12 rolls of 3d6 to generate a set of scores that are universally substandard. However, outliers exist in all systems governed by chance. 

It should be noted that for the systems mentioned apart from the four d6 drop lowest method, there was no prescribed limit on the scores generated. Mathematically it is impossible for any one score to exceed 18 just based on the die roll, however a subsequent ability score increase could push the total to 20. This is a consequence of the generation method originating in a system without ability score bonuses attributed to races; the scores dictated the species, rather than the scores being applied after the fact. If this method is used without limitations such as those imposed in the 4d6 method, it is possible to generate abnormally powerful or abnormally substandard characters. 

As the garbage processor tried to say, in 2nd edition they didn’t have balancing limits on the characters; you got what you rolled and had to deal with it as both a player and a DM. If you’re using one of those systems for a 5e game, putting in the +4 to +8 limit is probably a good move, although the Lord of the Rings showed us an adventuring party with a mixture of highly competent adventurers and less impressive novices can work if the story and roleplaying are done right. 

Some of those methods are also complicated for people to figure out. As I said before, despite 4d6 drop the lowest being the standard for D&D in a lot of peoples’ minds, it’s not the easiest formula for some people to follow. If you start looking at some of these other systems, especially something like the hybrid point buy, it can be even worse. So if a DM really does want to use one of the alternate methods for some reason but players are having trouble understanding it, it might be better for the DM to do the actual rolling and math, either rolling for each person or just generating an array they’re happy with and then telling everyone to use that same one. 

Generating a different array or a few of them and having each player use that one can speed up character generation and be less work for the DM while being an answer to players who for whatever reason have a problem just using the standard array. 

So if you’re doing character generation and want more powerful characters, more control over how the scores are generated, or just want to go crazy, try looking at some of the alternate formulas. 

Lennon: Okay all that information is nice and all but what was all of that noise? Were you grinding out your own dice or something? 
ROSTRO: When my creator’s body succumbed to gravity it knocked loose a weapon which lodged in my chassis. Some manipulation was required to overcome the internal obstruction. 
Ryu: Oh yeah I see…um..Lennon… That’s a vorpal sword. A vorpal sword fell into it and it just had to wait a bit to work it loose!? Shouldn’t it be all in pieces? Is it trying to say it’s immune to *VORPAL SWORDS* The more I know about this machine the less I like it. 
Lennon: Well let’s head over to the scrying pool and see if there are any uplifting listener messages in there to take your mind off of things. 

No Comments

Add Comment