What I Like About Orc Babies

This is a touchy subject, and I'll try my best to discuss it with care, but I want to talk about children in D&D - human children, halfling children, orc children, giant children, beaver children, and everything in between. 

This has been kicking around in my mind since reading Lonely Star's post on the absence of pregnant women in TTRPGs, which sparked a related conversation about children in TTRPGs. This of course brought to mind the AD&D Monster Manual and its very thorough inclusion of children of all kinds. The implications are grim, and this is infamously brought to the forefront in Gygax's own Dungeon Module B2: The Keep on the Borderlands and its use of "monstrous" humanoid children.

I will also be talking about noncombatant adults (in the AD&D Monster Manual these are referred to universally as "females", but I will be using my preferred nomenclature because I believe that anyone can fight or not fight regardless of their gender). These adults go hand in hand with children, as they are often found in lairs and either don't have combat statistics or fight as lesser combatants compared to their "male" counterparts. In the case of those that do not fight at all, they pose the same thorny questions as do children. That is, once you've slashed and burned your way through the warriors guarding the Caves of Chaos, what do you do with the humanoid civilians?

So, let's talk about children and noncombatants in the fantasy colonial violence game.

Interestingly, there are no demihuman or humanoid children in OD&D's Monsters and Treasure. There are young dragons, lycanthropes, centaurs, and rocs, but no orc babies. The closest thing to children of people in the booklet are the lycanthropes and centaurs (lycanthropes being basically people most of the time, and centaurs being people from the waist up). Likewise, dragons, lycanthropes, and centaurs are the only entries in which female members of the species are mentioned (and only the female centaurs are noncombatants). 

AD&D's Monster Manual is vastly different. Here is a list of every (?) Monster Manual entry which mentions gender or children:

• A bugbear lair will have female and young bugbears each equal to 50% the number of male bugbears. In a life or death situation, the former fight as hobgoblins and the latter as kobolds.
• A centaur lair will have females equal to 200% the number of male centaurs and 5-30 young. In a life and death situation, the former will fight with hooves only (unlike the male centaurs, they do not wield weapons). Centaurs are 90% likely to ransom their threatened women and young using their treasure hoard - an even more grim implication than usual.
• A dwarf lair will have female and young dwarves equal to 50% and 25% the number of male dwarves, respectively. Presumably, neither fights (from here on, I will presume female and young monsters do not fight unless mentioned otherwise).
• An elf lair will have female and young elves equal to 100% and 5% the number of male elves, respectively. 5% of the time there will be 10-30 warrior elfmaids mounted on unicorns (cool).
• If more than one cloud giant is encountered in its lair there is a 75% chance that the second is a giantess (which fights as a fire giant), while the rest are young giants. Percentile dice are rolled to determine the young giants' maturity, and they have hit points and inflict damage equal to that percentage of an adult giant's hit points and damage. This same rule of thumb is more or less repeated with all giant types, except the storm giant for some reason.
• A gnoll lair will have female and young gnolls equal to 50% and 200% the number of male gnolls, respectively.
• A gnome lair will have female and young gnomes equal to 50% and 25% the number of male gnomes, respectively.
• A goblin lair will have female and young goblins equal to 60% and 100% the number of male gnomes, respectively. Here the Monster Manual explicitly states, "As is usual with creatures of this sort, the females and young do not fight," which is why I am presuming that female and young monsters listed in other entries do not fight unless mentioned otherwise.
• A halfling lair will have female and young halflings equal to 100% and 60% the number of male halflings, respectively.
• A hobgoblin lair will have female and young hobgoblins equal to 150% and 300% the number of male hobgoblins, respectively.
• A kobold lair will have female and young kobolds equal to 50% and 10% the number of male kobolds, respectively, as well as 30-300 eggs (worth mentioning as deciding what to do about the eggs is similar enough to the decision made about the females and young).
• Werewolf packs are typically a male and female werewolf and 3 to 6 young of 60% to 90% growth. Male and female werewolves fight the same, except the male werewolf gets +2 to hit and does full damage if the female werewolf is attacked, and the female werewolf gets +3 to hit and does full damage if the young werewolves are attacked. The young fight at -4 to -1 to hit depending on their maturity and do 2-5 points of damage (versus the adult werewolf's 2-8).
• A caveman lair will have female and young cavepeople equal to 100% and 50% the number of male cavemen, respectively.
• A tribesman village will have female and young tribespeople each equal to 100% the number of male tribesmen.
• A nomad lair will have female and young nomads equal to 200% and 100% the number of male nomads, respectively.
• A mermen lair will have merwomen and merchildren each equal to the number of mermen.
• An ogre lair will have 2-12 female ogres and 2-8 young ogres. The female ogres do 2-8 damage with a weapon (compared to the male ogre's 2-12 damage) and have a maximum of 6 hit points per hit die. The young ogres fight as goblins.
• An orc lair will have female and young orcs equal to 50% and 100% the number of male orcs, respectively.
• A sahuagin lair will have 30-120 female sahuagin and 10-40 hatchlings. Female sahuagin have 2 hit dice and hatchlings have 1 (compared to the standard 2+2 hit dice). It is worth noting that "the religious life of [sahuagin] is dominated by the females." There is a 10% chance for every 10 male sahuagin that there will be a 5th to 8th level priestess with 1-4 3rd or 4th level assistant priestesses. Female sahuagin are only slightly smaller than male sahuagin, and otherwise indistinguishable.
• A triton lair will have female and young tritons each equal to the number of male tritons.
• A troglodyte lair will have female troglodytes equal to the number of male troglodytes. Female troglodytes have 1+1 hit dice (versus the male troglodyte's 2 hit dice). "There will also be hatchlings and eggs, but these are of no importance."
• A yeti lair will have a 30% chance to have 1-3 female yetis and, if female yetis are present, a 15% chance to have 2-5 young.

This was compiled with a PDF of the Monster Manual and Ctrl + F (search terms "female" and "young"), not an eagle-eyed readthrough of the book in its entirety. Sometimes words like "hatchling" or "fledgling" are used in place of "young", so there may be some occurrences I've missed. I think you get the idea, though.

For the sake of brevity, I included only human-shaped monsters like demihumans, humanoids, giants, and the like - no animals or animal-like monsters such as giant beavers, pegasi, or owlbears. I already went into a great deal of detail as to how these other monsters' young are handled in my post on AD&D's monster economy, so it felt redundant to retread the same ground here. I've excluded dragons for similar reasons.

I also did not include "female-coded" monsters for whom biology is not really a thing - mostly demons and devils like the succubus, marilith, and erinyes, but also the banshee, which is always the spirit of a female elf, or the lamia, which has the upper body of a human woman but no explicit biology.

AD&D's presentation of "demihumans" (dwarves, elves, halflings, etc.) versus "humanoids" (goblins, kobolds, orcs, etc.) is very similar. That is, the Monster Manual's description of these different types of people (and they are all people in my eyes) is relatively consistent. They all have "lairs", treasure, combat statistics, leader-types, pets, an implied society, and, of course, noncombatants. They are all treated the same, which is as "monsters".

From high enough up, looking down, they are all simply "game pieces" with which to populate your world. They all show up on the encounter tables. They are in the Monster Manual because they exist in the milieu, not necessarily because they exist to be killed. The players will encounter elves and they will encounter orcs. Sure, the elves are Chaotic Good and the orcs are Lawful Evil, and one can imagine that you will fight the orcs more often than you fight the elves, but that doesn't mean you can't fight the elves and befriend the orcs.

This stands in contrast with modern D&D's approach to the issue of personhood. In the original 5e Monster Manual, you do not find any statblocks for halflings. There are technically statblocks for dwarves and elves, but these are Drow and Duergar, the dwarves and elves who live underground, who it is okay to kill because they are Evil. Dwarves, elves, halflings, dragonborn, tieflings, and the like have societies (described in the Player's Handbook), but they are not monsters. You can fight a bandit who the DM decides is an elf, but you cannot fight a regular elf. These races are broadly, as a people, "off-limits".

Compare that to goblins, kobolds, and orcs. All are in the Monster Manual and have generic statblocks. You could decide that your bandit is an orc, but you can also fight a regular orc using the orc statblock. Unlike the dwarves and elves and halflings, they're combat fodder. However, they also have the "humanoid" creature type - the same type used by player characters (before supplements introduced playable races with different creature types), suggesting that while they are low level combat fodder, they're also people.

In the new version of the 5e Monster Manual, orcs were elevated from the rank of monster fodder to off-limits personhood alongside dwarves, elves, and halflings. This is presumably because they became one of the core player character races (half-orcs were understandably removed as a player option and replaced with regular orcs). 

On the other hand, goblins are now fey type creatures. Kobolds are now dragon type creatures. You can see this same pattern with other formerly humanoid monsters. Aarakocra are elementals. Bullywugs are fey. Gnolls are fiends. While orcs were made "off-limits", the remaining humanoids were not, and in fact are no longer humanoid type monsters at all. The game seemed to double down on the distinction between human-shaped creatures that are people and human-shaped creatures that are not, right down to their game categorization.

Of course, you can just do what you want with the game and say your goblins are humanoids or your dwarves are elementals and your elves are fey, but it troubles me. In 5e, you can kill a goblin, because it's in the Monster Manual. You can kill a bandit who your DM has decided is an elf, but you can't kill a plain old elf. Dwarves and elves and halflings and orcs are off-limits. Goblins and kobolds and gnolls are not.

Returning to AD&D, you could call out the demihuman/humanoid distinction as being similar to the one I've pointed out in modern D&D (one is a Good not-human and one is an Evil not-human), but in terms of game mechanics, they're presented exactly the same. They are all there in the Monster Manual. One is presented as being more villainous, but they're all creatures living in the world. Here are their game statistics in case you have to kill them, dwarves and elves and halflings included. Everyone is on the encounter table, and they all bleed if you stab them. And they all have children.

Of course, you don't have to kill anyone. Just as the DM can use the entries on dwarves, elves, and halflings to create a "lifelike" demihuman enclave with which the players can interact, the same can be said of the entries on goblins, kobolds, orcs, and other humanoids. When I see how similarly AD&D presents the societies of demihumans and humanoids, my impression is that this is a game where all of these creatures are people, even if some are vilified while others are not.

This is, of course, not the default assumption in D&D - especially not the D&D of Gygax, who once famously asserted that it is both Lawful and Good to kill orc babies, going so far as to cite an infamous perpetrator of Native American genocide. AD&D's domain game requires that a player character who wishes to build a stronghold must first clear the land of any monster lairs present - i.e., those very same goblin, kobold, and orc lairs which contain so many noncombatants and children. Gygax's Keep on the Borderlands is one of the most foundational of D&D's early adventure modules, with the objective of clearing out the Caves of Chaos and their humanoid residents - including their noncombatants - so that the march of human civilization can push ever onward.

It is also worth mentioning that AD&D's section on CONSTRUCTION & SIEGE details the time it takes to excavate rock for the construction of dungeons beneath the player character's stronghold. This section includes rules for the cubic volume of rock excavated by different types of human-like creatures, including humans, dwarves, gnomes, and halflings, but also gnolls, goblins, hobgoblins, kobolds, orcs, ogres, and several types of giants:

The use of humanoid and giant laborers is somewhat discouraged by this section, which describes these workers as generally unwilling and difficult, but here any notion of subtext becomes text. The player character can use these people to construct their stronghold dungeons, but they will be slaves, and will have to be heavily guarded to ensure peak efficiency:

When you consider this in combination with the prevalence of noncombatants in these monsters' lairs, it paints a bleak picture.

It's obvious that Gygax's idea of Law and Good is deeply flawed. It is certainly not the definition of Lawful Good I use in my games. Alignment in my games is more an ideological or cultural predilection or bent, not a cosmic or ontological truth. Orcs are not naturally Evil creatures born of shadow and sorcery. They might culturally tend towards Evil because their ideology espouses strength and mercilessness (likely because they have been done wrong time and again by their more powerful neighbors), but they are not universally or irredeemably Evil in any way. 

Even if the particular group of orcs in a scenario is Evil, that does not mean they are actively doing Evil. I use reaction rolls to determine how such people might interact with the world around them (generally, I find these older mechanics super helpful in making these scenarios more complex and engaging). The classically Chaotic or Evil orcs might be more likely to be unfriendly or hostile to their neighbors, but there is also a chance they are simply keeping to themselves, or even attempting to establish friendly relations or trade. Any character whose first course of action when dealing with such a people (even those who are Chaotic or Evil and hostile towards their neighbors) is to go to their home and kill them and their children is not Lawful Good in my eyes.

While AD&D's domain game may insist that player characters use violence to colonize their sphere of influence, I see no reason why this should be required. A detailed reading of the domain game rules suggests that player characters can eventually assimilate monster populations into their realm. This is dependent on the player character's and the monsters' respective alignments, but in a game where alignment is less prescriptive, why shouldn't the Lawful Good paladin be allowed to negotiate terms with the orcs, make them vassals, and attempt to change their ways? It is still colonialism, but with less bloodshed, and I don't see much of a way around it without excising the domain game entirely (or building your stronghold someplace where there is somehow both good land and also no one already living there).

Similarly, while AD&D requires that humanoid and giant workers be enslaved and heavily guarded in order to construct stronghold dungeons, there is no need for the individual DM to cling fiercely to this rule of thumb. Why must all goblins, ogres, and giants be loathe to do labor (any more than the average person loathes work)? Why can't they be recruited in good faith and compensated for their time the same as any other laborer? 

The Keep on the Borderlands is also an interesting case. Just as in the Monster Manual elves are detailed in much the same way as orcs, in B2 the Keep is given nearly as much description as the Caves of Chaos. It is not just a nondescript bastion of human civilization and home base for forays into the untamed wilderness, but a richly detailed play environment in which every defender, civilian, and treasure is placed with care. Do with it what you will.

The inhabitants of the Caves of Chaos may be Evil (or Chaotic, in this case), but they aren't actually doing anything besides fighting amongst each other. The rich detail of the Keep itself suggests that it is intended as another play environment which the players might need to "solve". The players might be able to unite the factions of the Caves against their common enemy. Better yet, why not lead them against the Cult of Chaos and negotiate terms for them with the Keep's Castellan? It's a bit of a stretch, certainly not Gygax's intention, and maybe not the most adventurous way to play out the module, but TTRPGs are all about tactical infinity. It's an option, is all I'm saying.

When human-like monsters are portrayed as inherently, irredeemably Evil or adversarial, the players' agency in interacting with them becomes extremely limited, lest they be forced to engage in "Lawful Stupid" behaviors. That is, when the orcs are Evil no matter what, the players have no choice but to thwart them in some way. If the players allow them retreat or surrender, they are presumably only leaving human civilization vulnerable to renewed offensives in the future.

The big problem with orc babies (here I'm using this as a catchall term for noncombatant monsters) is that if these monsters are irredeemably Evil or adversarial, it is easy to dismiss the wanton slaughter of their civilians as an act of "greater Good" - after all, the young will grow up to be Evil warriors, and the adult noncombatants will only exist to facilitate their Evil and to bring more Evil into the world. Genocide becomes permissible - even optimal - for a "Good" character to partake in.

Another problem with orc babies is when they're wielded against the players as a gotcha. "Aha! I, the DM, have created/am running a scenario in which the default mode of play is to go to the monsters' houses and kill them. You've done what you were intended to do, but you did not consider that at the end of the line you would encounter these helpless civilians. Now you have a moral dilemma, which I have given you no means of avoiding!"

For some groups, this dilemma might be fun to engage with. I think it sucks, at least when it's setup this way. Even if we decide that human-like monsters are not ontologically Evil, but must nonetheless be thwarted, and violence is somehow the only way, let's say the players decide that the threat has been neutralized and they allow the noncombatants to live. They are probably condemning the remainder to starvation or predation by some stronger foe, but they get to feel good about doing the "right thing", I guess.

The Keep on the Borderlands is such an infamous example because - while other approaches do exist - the default objective of play is to clear out the Caves of Chaos. Many DMs have and will run this module assuming that the player characters will go kill every humanoid combatant in those Caves, and many players have gone/will go along with the path of least resistance, only to encounter the gotcha once their grisly task is done. They are mostly not at fault because they were just going along with the quest hook, which is often considered the courteous thing to do when the DM is running a module.

A proper sandbox needn't have any assumed objectives, however. Sure, when I design a sandbox, I assume that the players will try to establish themselves in town, that they will explore the nearby dungeon, hunt down the owlbears, roust the brigands from the castle, clear the ogres from their lair, and overthrow the town's Evil overlords, more or less in that order. But they might also make common cause with the town's rulers, subdue the owlbears for the wizard's menagerie, become brigands themselves, and recruit the ogres to their side rather than run them off or slaughter them.

If there is an orc lair in my sandbox, there will probably be propaganda about how the orcs are uncouth, inimical to civilization and impossible to reason with. The only good orc is a dead orc and all that. Again, because of the way alignment works in my games, the orcs may not even be Chaotic or Evil. Even if they are, they may not prey upon their neighbors. They may be isolationist, or even cautiously friendly. Perhaps their mere presence is an inconvenience to the powers that be, who want to run them off their land, put down a parking lot for their dragons, and build a magic Walmart in place of the orcs' lair.

If the players dig into it, they may uncover the truth. They may in fact find a straightforward den of orcs with sinister intentions. Even this scenario is not exclusionary of less violent means of settling the dispute, however. Perhaps the players will find that the orcs do have some sort of ancestral claim to that territory, or they learn that the orcs can be appeased in some way which does not involve plundering all of human civilization. Perhaps there are factions of orcs, some less warlike than others, and negotiations can take place if the orcs have a change of leadership. Perhaps the human ruler is Good and is willing to make certain concessions, or perhaps the human ruler is the greater Evil and it is they who must be overthrown to ensure a lasting peace.

What I like about orc babies (again using this as a general term to encapsulate noncombatant monsters) is that they signal to the players that these monsters are people. When the players find out that there are hundreds of noncombatants in the orc lair, it should give them pause about how exactly to handle whatever conflict exists. Their first course of action shouldn't be to find a way to kill everyone inside. If they wrote "Good" on their character sheet, they have to find another way.

It complicates the scenario, but only if the orcs themselves have a level of complexity. In a proper sandbox, the presence of humanoid civilians shouldn't be a "gotcha". The players have the means to acquire whatever information they need, to make a choice based on that info, and to have their choice make an impact.

This all assumes that the player characters are basically Good and care about the wellbeing of civilians. If they are instead Evil, the result of including civilians in your monster populations can again be quite grim. 

I've already discussed this in the aforementioned post on AD&D's monster economy and in my post on subduing dragons. Generally, I feel that the assumption that player characters will want to subdue dragons and auction them off in the nearest town or city (and that most such places will usually have buyers eager to acquire them) is a fun bit of worldbuilding. Similarly, there are all sorts of animal-like creatures which the players are incentivized to kill, butcher for their hides, and steal their eggs and young to sell to the highest bidder.

I am not shy about referring to these practices as slavery, which they very plainly are. Many of these animal-like monsters are intelligent, peaceful, and even lean towards Goodness. All of the dragons are intelligent, half of them are Good, and all of them explicitly loathe their imprisonment and seek to break free of their masters at the earliest opportunity. Skinning giant beavers and kidnapping the children of monsters to be enslaved are clearly Evil acts, even if the monsters themselves are designated as "Evil".

Still, I argue that players should be able to play Evil characters, and that there should be opportunities for them to do Evil. Even if they are not Evil, they must still have the opportunity (and, even better, an incentive) to do Evil for their choice to be Good to matter:

Being Good is hard. It should be. It should be easier (or at least more profitable) to be Evil. Players can write "Good" on their character sheet all they want, but they have to actually do Good for that to mean anything. And for doing Good to actually mean something, it has to be a choice. For doing Good to be a choice, the players must also have the choice to do Evil. 

If they spend the whole game doing little quests for the dirt farmers of the local thorp out of the Goodness of their hearts with little wealth to show for it, they may start rethinking what they wrote on their character sheet once they hear that their (Evil) rivals just earned 80,000 gold from skinning giant beavers and selling their children into slavery. Maybe it just reaffirms their commitment to Goodness, and the existence of such Evil in the world instead serves to distinguish them from those sorts, and they vow to put an end to such injustices at any cost. But the (potential for) Evil deeds must exist to create that contrast.

You might have players who choose to play Evil characters, and that's okay too! A campaign of poaching monsters and selling their children into slavery is definitely not for everyone, so you might restrict these options, but I personally enjoy portraying a fantasy world in all its ugliness. If there are giant beavers nearby, the village may be negotiating with them to construct a dam, while simultaneously, outlaw poachers may be conspiring to prey upon them. The pegasi nesting nearby may be valuable allies to Good characters, but if there's a Chaotic Evil wizard in the area, they're likely to desire the creatures' eggs for nefarious purposes, and are willing to pay good money for them. The players get to choose how to engage with those dynamics one way or the other.

This is exactly the kind of thing session 0s and safety tools are for. If someone at the table is not comfortable with being a part of Evil acts, or with poaching or slavery specifically, then players should not play Evil characters, or those acts should be forbidden, or the person who's uncomfortable should consider playing in a different game. Violence against children and the enslavement of people is obviously an even thornier issue and should be handled with even more care. I totally understand why many or even most people don't want orc babies in their games.

But I like a bit of thorniness in my games. If the players decide that their best course of action is to carry out a genocide, then they should have to deal with the moral and spiritual toll of that. I am not going to trick them into it or force them into a situation where it's their only option, then inflict the dilemma on them as some sort of inane "twist", but if they make no attempt to gather more information or pursue less absolute avenues then that is their choice, and they must deal with the consequences. 

At the same time, if they have chosen to play Evil characters and all involved are interested in playing out the implications of that choice, then it is up to me to decide how much opportunity I'm willing to give them to engage in that sort of play and explore its consequences. If I know that my players are interested in exploring Evil, I have a choice as to whether or not to include orc babies in my sandbox, knowing exactly what the outcome might be. If the players decide to slay the orc civilians, then that is their choice, and it's also my choice because I put them there, knowing that this might happen.

And even if the player characters are not themselves Evil, just as with the subdual of dragons and participation in the monster economy, anything the player characters can do, NPCs can do too. The player characters might treat the orcs as people, but the NPCs might not. The player characters might seek to treat with the orcs as equals while the local Evil ruler is carrying out a genocide and enslaving their victims' civilians to excavate their subterranean dungeons (this is even represented in the background section of Dungeon Module B1: In Search of the Unknown).

Again, whether you choose to include this sort of thing will depend entirely on your and your players' sensibilities. Just as players will have different levels of comfort with carrying out Evil acts, they will have different levels of comfort in dealing with NPCs doing the same. Thwarting an Evil wizard who is doing cartoonish wizard villainy is one thing - thwarting a tyrannical slaver is another. Some players may relish the opportunity, while others will become despondent when forced to confront real world parallels in their leisure time.

I'm focusing on humanoid civilians because there is already so much baggage there with parallels to real world racism, colonialism, slavery, genocide, and so on, but it's important to again remember that AD&D's Monster Manual presents demihuman noncombatants the same way. Dwarves, elves, and halflings can present conflict in your setting as well, and players of Good characters who wish to resolve these conflicts will have to reckon with the fact that they're also people, just as you as the DM will have to consider whether to include noncombatant members of these people if you know your players are interested in exploring Evil deeds. The default assumption in many cases may be that the player characters are Good and that the orcs will be potential enemies and the elves potential allies, but this isn't always the case.

This is not to say that human-like monsters can never be fodder. This is a game, after all, and going toe-to-toe with thinking enemies with capabilities similar to yours is a fun scenario to play out. Remember that noncombatants only appear in lairs, so human-like monsters encountered in dungeons and outdoors are fair game (though again, that is not to say they are inherently Chaotic or Evil or even doing anything nefarious). Just as it is "okay" to kill a human brigand (that has its own baggage, as brigands are people too and probably have a good reason for turning to brigandry), it is okay to kill an elf or a goblin in an imaginary violence game.

All that being said...I think there is something very funny about an Evil child. My Hexmas sandbox, the Jinglebell Barrens includes a Chaotic Evil frost giant toddler:

• Kolvi: Hrothvar's only son. He is the frost giant equivalent of a toddler - as large as an ogre, but doughy and badly proportioned. He is always accompanied and protected by his pet winter wolf, Hoarfang. He is lazy and cowardly, has a child's temper and petulance, and wants nothing more than to see adventurers slain, roasted, and served up at a feast. He hates both his father and grandmother, but adores Rimehilda. He will run to her and make up any lie about adventurers which might get them into trouble. One adventurer, before being killed and eaten, spoke of a talking wishing well. Kolvi begs Rimehilda to take him there so that he might wish for all adventurers to be killed and eaten by him.

My recent post on halflings and their dogs includes this Lawful Evil halfling child:

Corin Thistlethorn (LE): 3 hp. A male Hairfoot child (AC 9). Thin and sharp-featured, with dark brown curls and keen grey eyes. His clothes are surprisingly tidy for a child. Although young, he is more intelligent than the average halfling (Int 14). He quietly observes conversations before joining in, and adults sometimes find his questions uncomfortably perceptive. He is selfish and manipulative and enjoys bending rules, knowing secrets, keeping score of favors, and figuring out how to gain advantages over his peers. He is tight-lipped around strangers, evaluating them carefully before deciding whether they are allies or people to avoid (neutral). Treasure: A carefully organized collection of bird feathers, insects, and unusual stones, a slate tablet and chalk used for keeping lists, small wooden tokens representing debts and favors owed by other children, and a pocketknife (no value). Corin has 2 dogs:

1. Bracken: A tan rabbit hound. Curious about everything, he has a habit of bringing interesting trinkets to Corin. Terrified by storms.
2. Soot: A black terrier with mismatched eyes. He distrusts strangers and watches them with the same suspicious scrutiny as his master.

Now, I'm not saying these children should be murdered because they're Evil. Once again, alignment is a mutable cultural tendency in my games, not an ontological truth about who a person is (the halfling child above, for example, is Lawful Evil despite halfling culture leaning towards Lawful Good). These are children, meaning they are even more malleable than most people, if gotten to early. Murdering a child, regardless of their fantasy race, personality, alignment, or whatever, is one of the most Evil things you can do.

Just because the frost giant child wants to eat you and will hide behind the Evil witch's skirts and tell lies about you to achieve that end does not mean he can be killed in good conscience. However, if you're trying to infiltrate the frost giants' home without ending up on a skewer yourself, maybe lock the child in a closet or trick them into drinking a sleeping potion or something. You probably don't have to do anything about the off-putting halfling child hanging around the shire, but it gives the place a little flavor for one of the children to be a bit unnerving.

And that's what I really like about orc babies. The fact that the Monster Manual tells me that 60% of the orcs in the lair will be children and noncombatant adults primes me to include those orcs as NPCs as well. If the orc lair has a certain number of warriors, 50% as many noncombatant adults, and 100% as many young orcs, then I know that 40% of the orcs in the lair are warriors, 40% are young, and 20% are noncombatant adults, and I can use that to give some variety to the NPCs encountered there.

If the player characters are hanging out in the orc lair, they should have a chance to encounter an orc child. Not "encounter" as in "fight" of course, but like...I don't know. Roleplaying an interaction with a orc toddler sounds fun to me. Make the child annoying to deal with just like you would any NPC. That doesn't give the players an excuse to murder them, no more than it gives them an excuse to kill a corrupt city official or a wizard's meddlesome henchman. 

In some cases, violence against children will be advantageous (yes, this is a weird sentence to write, but bear with me). In Dungeon Module G1: Steading of the Hill Giant Chief, there's an opportunity to murder or subdue some hill giant children in order to take their clothes so that the player characters can pass as giant children themselves:

Again, the assumption of this module is that the giants are Evil and anything goes, which has its drawbacks, but in a more nuanced game I think the dilemma is somewhat interesting. Are the players willing to do this wretched thing for a slight upper hand, or is it too distasteful, so now they have to go about things the hard way? Is the moral calculus at all altered by the fact that those giant children are actually a significant threat with 13 to 26 hit points and armed with weapons with which they fight as ogres? For some groups it will be, for others not so much.

These types of NPCs make a place feel lifelike. I want to portray these as real places, not video game levels to be cleared. The elves and orcs are simply "game pieces" to the DM, but they should feel like people to the players. The presence of noncombatants reinforces that the lair can be a place where people live, not just where monsters wait to be slain. I don't want to exclude these NPCs out of fear for what the players might do to them, no more than I want my hamlets and thorps to be stuffed to the brim with guards and 1st level fighters, with nary a peasant farmer or meddlesome child in sight.

If you don't want to risk it, don't put noncombatants in your game. Or, simply tell your players up front that those NPCs are off-limits and you won't tolerate any violence against them. You could go the Bethesda RPG route and either have children be conspicuously absent from your world or else immune to harm. It's also okay to have your orcs or whatever just be monsters birthed from sorcery with no explicit biology, if you want violence against them to be totally free of consequence.

Personally, I like having orc babies in my games. I like weird, annoying, funny NPCs, and no person is weirder, more annoying, or funnier than a child. So much of the legacy of children and noncombatant monsters in D&D is tied up in the rancid worldview of Gygax and the baggage of classic modules, which assume that it is unambiguously okay to kill these creatures and which punish the players with unfair forced moral dilemmas for doing what the modules intend for them to do.

It's because of those same modules that we never talk about the children of dwarves, elves, and halflings, or the fact that every village, hamlet, and thorp is populated by many civilians and children. Not many modules feature demihuman lairs or detailed settlements. In the case of Dungeon Module N1: Against the Cult of the Reptile God, in which the village of Orlane is detailed, the proportion of detailed men, women, and children is way off.

But the player characters can raid an elf village just as much as they can an orc lair (the elves have treasure too!), and the elves will have noncombatants and children just as the orcs do. Every time a village is attacked by a monster, we must imagine that noncombatants and children are being killed, eaten, burned by dragon fire, carried off into slavery, or whatever. It's not super fun to imagine, which is part of the reason it isn't talked about more, but it's there, happening in the background.

Fortunately, there isn't much (not really any that I can recall) violence against children or noncombatants happening "on screen" in my D&D games, but I've thought about running modules like B2 and G1. My villages, hamlets, and thorps have children in them. My sandboxes have orc lairs and elf lairs with all the usual noncombatants. There are petulant frost giant toddlers and unnerving halfling children. There are dragons that can be subdued. There are giant beavers that can be poached and enslaved. There is opportunity to do Evil. There are Evil NPCs pursuing their own goals in the sandbox, many of whom are willing to do these wretched things even if the player characters are not.

So, the potential is always there. It's something I'm comfortable with. I trust my players to engage with that material thoughtfully, and I trust that we can have a conversation about it if it becomes a problem. Not every table has the same luxury, so you will have to adjust to taste. 

D&D is a game where violence happens...a lot, whether the player characters initiate it or not. If you're trying to portray a lifelike world, it might also be a game where children are prevalent. You have to calibrate how much those two facets interface with one another, what happens on screen versus off, and how much opportunity to give your players to interact with those elements.

I don't begrudge anyone for preferring not to engage with it, but it's worth considering. Don't throw the orc babies out with the bathwater!

Who Can Rule a Stronghold in AD&D?

This is something interesting that came up when I was messing around with the AD&D DMG's Appendix B to create a sandbox (okay, many sandboxes - I have a problem).

As I've described many times on this blog, when using Appendix B to stock a sandbox, I roll on the Inhabitation table to determine the placement of settlements, castles (which I will call strongholds), and ruins:

Generally, three out of 100 "spaces" (6-mile hexes, in my case) will contain a stronghold. To determine the type of stronghold and who rules it, we are directed to the Castle Tables in Appendix C:

Strongholds can be ruled by bandits, brigands, berserkers, dervishes, or characters/"character-type NPCs". Deserted strongholds (which I generally rule to be ruins, not proper strongholds of good construction) can also be inhabited by monsters (or totally deserted).

Bandits, brigands, berserkers, and dervishes are all "men" - that is, humans. Character-types, I figure, can be of any race which is available to the players. So if I determine that a stronghold is ruled by such a character, before I determine that character's class, I roll on this table (found earlier in Appendix C) to determine the character's race:

80% of characters will be human, 5% will be dwarves, 5% will be elves, 2% will be gnomes, 5% will be half-elves, 2% will be halflings, and 1% will be half-orcs. You can actually simplify this by making it one roll, like so:

    d100      Race of Individual
    01-80     Human
    81-85     Dwarf
    86-90     Elf
    91-95     Half-elf
    96-97     Gnome
    98-99     Halfling
    100        Half-orc

I'll then determine the class of the character by rolling on Castle Sub-Table II.B., limiting myself to those classes which are available to that race.

But there's a problem - characters can only build strongholds beginning at a certain level, and demihuman races have level limits, as outlined in the PHB:

So the question is - which demihuman characters can actually rule strongholds in AD&D? Let's find out.

The Cleric can establish a "place of worship" and attract followers at 8th level, and at 9th level can construct a religious stronghold (presumably, they would build out the existing place of worship into a stronghold, not start an entirely new construction, but this isn't made explicitly clear):

While Castle Sub-Table II.B. states that clerics who rule wilderness strongholds will be from 9th to 12th level, I would allow them to be 8th level as well, with the caveat that these are not proper religious strongholds but the lesser "places of worship".

Aside from humans, player character half-elves and half-orcs can be clerics, but they are limited to 5th and 4th level, respectively, so they can build neither places of worship nor religious strongholds. Dwarf, elf, and gnome NPCs can also be clerics, but player characters cannot - I would personally ignore this as I don't really see the logic. Dwarves are limited to 8th level and elves and gnomes are both limited to 7th level. Dwarf clerics can build places of worship but not religious strongholds. Elf and gnome clerics can build neither. Only human clerics can build religious strongholds.

The Druid does not dwell permanently in castles, but at 11th level they can "inhabit building complexes set in woodlands and similar natural surroundings":

This is not a castle per se, but I would still count it as a stronghold. I imagine it being like the druid grove in Baldur's Gate 3, which is very much a stronghold.

Castle Sub-Table II.B. states that druids who rule strongholds will be 12th to 13th level, but I would allow them to be 11th level as well.

Only human and half-elf player characters can be druids, and half-elf druids have no level limit, so they can construct and live in such building complexes just as human druids can. Halfling NPCs can be druids, but are limited to 6th level, so they cannot.

The Fighter can establish a freehold ("some type of castle") at 9th level:

A player character of any race can become a fighter, but gnomes and halflings are limited to 6th level, elves to 7th level, and half-elves to 8th level - none of them can build a stronghold as a fighter. Dwarf fighters are limited to 9th level (but they must have Strength higher than 17), and half-orc fighters are limited to 10th - they can both build strongholds.

The Paladin can only be human. It isn't clear when exactly they can build a stronghold, but they can at some point. However, it can only be of the small type (a small shell keep, tower, moat house, or friary):

I assume that, as a sub-class of fighter, they construct their strongholds at 9th level (this is consistent with Castle Sub-Table II.B., which lists the lowest-level paladin with a stronghold as being 9th level, though as I've already demonstrated - and will continue to demonstrate - that table is not entirely reliable).

The Ranger can construct strongholds much the same as fighters:

That is, rangers can build strongholds at 9th level. This makes me more confident in my assumptions about the paladin, since both are fighter sub-classes. Again, Castle Sub-Table II.B. omits 9th level rangers (it includes only rangers of 10th to 13th level).

Aside from humans, only half-elves can be rangers, and they are limited to 8th level, so they cannot build strongholds.

The Magic-User can construct a stronghold at 12th level:

Castle Sub-Table II.B. incorrectly lists 11th level magic-users as potential stronghold rulers (it includes levels 11 to 14, but should probably be levels 12 to 15).

Aside from humans, only elves and half-elves can be magic-users, and they are limited to 11th and 8th level, respectively, so they cannot build strongholds.

It is not clear from the PHB whether the Illusionist can build a stronghold, but since they are a sub-class of the magic-user and are included on Castle Sub-Table II.B., I assume that they can, and that they follow the same rule as the magic-user (that is, they can build a stronghold at 12th level).

Castle Sub-Table II.B. includes illusionists of 10th to 13th level, but like the magic-user, it should probably be 12th to 15th level instead. It's also worth noting that illusionists' strongholds "will often be covered by an illusion to appear as a mound of rock, a ruined place, or a huge castle". Neat!

Only humans and gnomes can be illusionists. Gnome illusionists are limited to 7th level, so they cannot build strongholds.

The Thief can build a stronghold, but it must be "a tower or fortified building of the small castle type" and "within, or not more than a mile distant from, a town or city":

This means that you will never encounter a thief's stronghold in the wilderness.

It sounds like the thief can build the stronghold at any level, but can only establish a gang (i.e., attract followers) at 10th level, so for the purposes of this exercise I'll say they can build the stronghold at that level as well. 

This is consistent with Castle Sub-Table II.B., which lists the lowest-level stronghold-ruling thief as 10th level (although it is worth pointing out that since I use 6-mile hexes and a thief's stronghold can only be 1 mile distant from a town or city, I will never have a hex containing a thief's stronghold - if you instead use Gygax's scale of 1-mile hexes, the table works just fine, but the thief's stronghold must be in a hex adjacent to a town or city).

Anyone can be a thief, but half-orc thieves are limited to 8th level, so they cannot construct strongholds. Dwarf, elf, gnome, half-elf, and halfling thieves have no level limit and so can construct strongholds.

The Assassin is tricky, but essentially, they have two types of "stronghold" - a guild headquarters in a large town or big city, "typically a warehouse or other nondescript structure", and a second headquarters "of any form - cavern, castle, monastery, palace, temple" which (presumably) may also be in a large town or big city, but "if it is a large and obvious place...must be located well away from all communities":

The assassin must be 14th level to rule the former and 15th level to rule the latter (15th is the level cap for assassins). 

Castle Sub-Table II.B. lists the assassin stronghold ruler's level as 14th (and only 14th) level. As with the thief, this works with Gygax's scale of 1-mile hexes, but the assassin's stronghold must be in a hex adjacent to a town or city. 

Since I again use 6-mile hexes, 14th level assassins can only rule guildhalls which are "always within a large town or big city", and only 15th level assassins can rule wilderness strongholds, my version of the table should include only 15th level assassins.

Dwarves, elves, gnomes, half-elves, half-orcs, and humans can be assassins. Dwarves, elves, gnomes, and half-elves are limited to 9th, 10th, 8th, and 11th level respectively, and cannot rule strongholds. Half-orcs and humans have no level limit and can rule strongholds.

The Monk can build a stronghold at 8th level, but it must be a monastery or monastery-like headquarters:

Castle Sub-Table II.B. states that "Monks' strongholds will usually be monasteries, resembling a type of enlarged moat house, having fewer of the defensive constructions of a typical castle - but being nonetheless formidable." I take this to mean that monks can only build castles of the small variety, specifically moat houses or friaries.

Castle Sub-Table II.B. includes monks of 9th to 12th level, omitting 8th level monks just as it omits 11th level druids and 9th level paladins.

Only humans can be monks.

The Bard is of course a special case. The description of the class does not explicitly state that they can build strongholds, but they are included in Castle Sub-Table II.B., so we must assume that they can.

Bards begin as fighters, then become thieves between 5th and 7th level - they must make the switch before attaining 8th level as a fighter, so they cannot build a stronghold as a fighter does at 9th level. They can attain 5th to 9th level as thieves, then must become druids - again, they cannot build a stronghold as a thief does at 10th level. Once they become a druid, they begin progression as a bard. They have their own experience table but are treated as druids of the same level (kind of - after 12th level they remain a 12th level druid until 23rd level when they finally become a 13th level druid).

Thus, I would assume that bards abide by the same rule as druids when it comes to building strongholds - that is, they can do so at 11th level, but it must be a building complex in a wilderness or natural setting.

UPDATE (January 12th, 2026): Bards will only construct a stronghold at 23rd level:

Only humans and half-elves can be bards.

Here is a summary of my findings - Who Can Rule a Stronghold in AD&D:

• Clerics: Dwarf clerics of 8th level only and human clerics of 8th level and above can rule places of worship, and human clerics of 9th level and above can rule religious strongholds.
• Druids: Half-elf and human druids of 11th level and above can rule building complexes set in woodlands and similar natural surroundings.
• Fighters: Dwarf fighters of 9th level only, half-orc fighters of 9th or 10th level, and human fighters of 9th level and above can rule strongholds of any type.
• Paladins: Human paladins of 9th level and above can rule small strongholds only.
• Rangers: Human rangers of 9th level and above can rule strongholds of any type.
• Magic-Users: Human magic-users of 12th level and above can rule strongholds of any type.
• Illusionists: Human illusionists of 12th level and above can rule strongholds of any type. They are usually covered by an illusion which makes them appear to be something else.
• Thieves: Dwarf, elf, gnome, half-elf, halfling, and human thieves of 10th level and above can rule small strongholds only, and they must be within or no more than a mile distant from a town or city.
• Assassins: Half-orc and human assassins of 14th level can rule a guild headquarters (a nondescript location, not a castle of any type) which must be within a large town or big city, and half-orc and human assassins of 15th level can rule a headquarters of any type (cavern, castle, monastery, palace, temple, etc.). It can be in a town or city but must be located away from all communities if it is a large and obvious place.
• Monks: Human monks of 8th level and above can rule a moat house or friary.
• Bards: Half-elf and human bards of 23rd level can rule a stronghold of any sort.

Here is a different summary sorted instead by race:

• Dwarves: Can rule as clerics (places of worship only), fighters, and thieves.
• Elves: Can rule as thieves only.
• Gnomes: Can rule as thieves only.
• Half-elves: Can rule as druids, thieves, and bards.
• Halflings: Can rule as thieves only.
• Half-orcs: Can rule as fighters, thieves, and assassins.
• Humans: Can rules as any class.

You could determine who rules a stronghold at least two ways: Do you roll to determine the ruler's race or class first? If I roll first for race and determine that the ruler of a wilderness stronghold is a human, for example, they could be of any class. If I instead determine they're a half-orc, they can only be a fighter or Grandfather/Grandmother assassin. If I roll first for class and determine that the ruler is a paladin or monk, for example, they can only be a human. If they're instead a cleric, they could be a human or a dwarf.

I'm not sure which is better, but I'm leaning towards determining class first. Just remember that thieves and most assassins don't live in the wilderness and druids and assassin Grandfathers/Grandmothers don't usually live in cities.

It is worth noting that deserted strongholds (which I treat as stronghold ruins) can be inhabited (like a monster lair) by dwarves, elves, gnomes, and halflings, even if these types can't rule strongholds as most classes. The wilderness encounter tables in Appendix C include these demihumans under "rough" terrain, which I believe is where you're meant to roll when populating a "castle" of the "deserted (monster therein)" type (the footnote to the tables describes "rough" as including "ruins within up to five miles of the party").

What this suggests is that individual elves, gnomes, and halflings, for whatever reason, don't build wilderness strongholds, but enclaves of them may move into such places once they're abandoned.

Interestingly, all four of these demihuman types can lair in such castles in uninhabited/wilderness areas, but only dwarves and gnomes lair in them in inhabited/patrolled area. In other words, if there's a road to the dungeon, elves and halflings won't hang out there - not on the surface, anyway. I wonder why that is.

On Character Creation

It behooves any Dungeon Master to write out their character creation rules for their players. I include these rules in the same document as any house rules I might be using. Obviously, if character creation in the game adheres to the rules laid out in the Player's Handbook, this isn't strictly necessary. 

While character options (races, classes and subclasses, backgrounds, feats, spells, etc.) presented in later supplements (and in the PHB, page 163, in the case of feats and multiclassing) often include a disclaimer that they are options only at the DM's discretion, in my experience, players will often assume that every published option is available to them regardless of the campaign unless the DM makes the (un)available options explicit. Thus it's important to have at least a bulleted list with general guidelines (or, a short conversation between DM and players). 

This post summarizes the character creation rules I use in 5th edition Dungeons & Dragons (and some thoughts on how to use them for other versions of the game).

Choose a Race

The player can choose what race their character will be, or they can roll on a table to randomly determine this. The table is as follows:

d20

1-13      Human

14-17    Roll d6 (1-2: Elf; 3-4: Dwarf; 5-6: Halfling)

18-19    Roll d10 (1-2: Dragonborn; 3-4: Gnome; 5-6: Half-elf; 7-8: Half-orc; 9-10 Tiefling)

20         Anything else

If the player chooses to randomly roll, they can add +1 to an ability score of their choice after determining ability scores.

The conceit of this table is thus: 

• I want the races in the Player's Handbook to be most common (thus why "Anything else" makes up only 5% of the table).
• Of those in the PHB, I want humans to be most common (65%).
• Of the remaining races, the latter five are specifically called out in the PHB as being uncommon, so elves, dwarves, and halflings must be the "secondary" races to humans (20%), where the latter five are tertiary (10%).

A DM can easily take these same percentages and change the races associated with each to suit their setting. If the game is OD&D, I would omit everything except humans, elves, dwarves, and halflings. In AD&D, I would also include gnomes and half-elves, and the final 5% might be the various race options found in the edition's many splat books (if I wanted to ruin my game by allowing players to make firbolgs, pixies, ogre magi, and the like, which I've done in the past). 

My guiding star is that whatever's in the PHB should be most common, and all supplemental options should be relatively rare, but this can vary by setting. For example, I have a separate setting where these "common" races are all but extinct, and "monstrous" races are the predominant inhabitants, and I use the more granular d100 to create a table with a greater array of race options inspired by the Monster Manual. The point of providing a table like this is that the players can instantly get an idea of what races are most common as well as what others are present in the setting. The exact composition of races and their rarities is important only to the individual campaign.

The actual percentage breakdowns are modified from certain tables in AD&D, namely random encounter tables (where common, uncommon, rare, and very rare monsters are typically broken up almost exactly like this, 65/20/10/5). I use this same breakdown for my own random encounters. 

At the DM's option, the percentages provided to players can also be used when creating NPCs, ensuring that the racial distribution of NPCs is similar to that of player characters. 

I grant a +1 to an ability score for rolling on the table because I want to incentivize players to randomize character creation. Most of my character creation rules attempt to incentivize this. This is an attempt to make party composition more interesting and varied than the characters the players might tend to create. (Interesting and varied, in this case, is relative to the usual choices made by the players - obviously this method creates a lot of human PCs, which is certainly less varied, though not inherently less interesting, than a party consisting of a tiefling, dragonborn, automaton, half-hag, and owl person. But if the players in that group always make those exact characters, this method is meant to encourage them to try something else.)

I've been in too many games where each player essentially makes the same character every time, and I wanted to try to break up those habits. If the players want to continue playing the same types of characters anyway, they can still do that, of course.

Determine Ability Scores

Players can determine their characters' ability scores using the standard array or point buy methods presented in the Player's Handbook (page 12) or by rolling dice. If the player chooses to roll, they can roll 4d6 and drop the lowest die six times or roll 3d6 six times. If the player chooses to roll 3d6, they can choose a feat of their choice at 1st-level (in editions which don't include feats, I would just choose whether players roll 3d6 or 4d6, or I would offer both and grant some other appropriate benefit for choosing to roll 3d6).

Rolling 3d6 may seem odd in 5e, but I think 5e characters are robust enough to cope with having more average ability scores. I have had a few players take me up on this in the past - the extra feat is too tempting to pass up. More new-school players who are concerned about having "bad" ability scores are free to choose 4d6 or one of the more dependable ability score generation options from the PHB.

Whether the player rolls 4d6 or 3d6, they can either assign the scores in any order they like, or assign the scores in the order they are rolled (Strength, Dexterity, Constitution, Intelligence, Wisdom, and Charisma in 5e, but the order might differ by edition, for example in OD&D I believe it would be Strength, Intelligence, Wisdom, Constitution, Dexterity, and Charisma). 

If the player chooses to roll their ability scores in order, they can add +1 to an ability score of their choice after determining ability scores (again, I want players to roll their stats in order, so I'm trying to incentivize it by offering a mechanical benefit in exchange).

The player then adds any ability score adjustments from their chosen race.

Players can also choose or roll to determine their character's age category, as follows:

d20

1-13     Young                    

14-17   Middle Aged    -1 Str, -1 Con, +1 Int, +1 Wis

18-19   Old                   -3 Str, -2 Dex, -2 Con, +1 Int, +2 Wis

20        Venerable         -4 Str, -3 Dex, -3 Con, +2 Int, +3 Wis

If the player chooses to roll to determine their character's age category and gets a result of Old or Venerable, they can add +1 to an ability score of their choice after determining ability scores (the benefits to Intelligence and Wisdom don't balance out the penalties to Strength, Dexterity, and Constitution in these age categories, so I'm offering another mechanical incentive to mitigate the hurt a bit).

The age categories are pulled from AD&D, and may seem overly punitive for a 5e game, however all this really amounts to is that older characters are less likely to be physically-oriented and more likely to be clerics, druids, and wizards, which I think is fine.

I have thought at times that players should choose their class before determining their age, thus allowing for the possibility of an "old fighter", for example, however I'm of the opinion that character creation represents only a snapshot of the character at a moment in their life. After all, according to Gary Gygax, Conan's class levels change with age.

At this time, the player adds any ability score adjustments accumulated from rolling to determine race, age category, or rolling ability scores in order. No ability score can exceed 20 (if the DM wants a more "balanced" 5e experience, they might cap it at 16 instead, but I personally don't care if a 1st-level character starts with a 20 - the faster the players maxes out their character's important ability scores, the sooner they can consider actually interesting ways to use their ability score improvements, such as by choosing feats or preparing to multiclass by increasing secondary or tertiary ability scores).

Choose a Background

I have no special rules for this, but one could easily use the percentage spread I've used so far to create common, uncommon, rare, and very rare backgrounds determined by the setting, if they wish.

Choose Languages

When choosing which languages a player character knows, exotic languages count for two language proficiencies. Thus, if a background allows you to learn "any two languages", a player can choose for their character to know two standard languages or one exotic language.

Standard and exotic languages are laid out in the PHB (page 123), but a DM can further customize these lists by adding or subtracting languages or switching standard and exotic languages according to the setting. For example, in my setting where "common" races are extinct, those races' languages are considered exotic, and other, traditionally exotic languages are considered standard.

Choose a Class

At this point, character creation can proceed according to the chosen system without much or any changes. In 5e, I use the following prerequisites, meaning a player character has to have at least the following score(s) in a given ability in order to choose the class (these are based on the ability score requirements for multiclassing into a given class, which are included in the 5e PHB):

Barbarain    Strength 13

Bard            Charisma 13

Cleric          Wisdom 13

Druid          Wisdom 13

Fighter        Strength or Dexterity 13

Monk          Dexterity and Wisdom 13

Paladin        Strength and Charisma 13

Ranger        Dexterity and Wisdom 13

Rogue         Dexterity 13

Sorcerer      Charisma 13

Warlock      Charisma 13

Wizard        Intelligence 13

The DM could take this a bit further, using something like the ability score requirements from AD&D if for example they wanted bards, druids, monks, paladins, rangers, and the like to be especially rare. I haven't done this in my 5e games, but I like the idea. I'd have to come up with something for warlock, since as far as I know they're the only class to never appear in an edition where ability score requirements were the norm (I didn't DM 3e or 4e, so I could be wrong).

That's pretty much it. The point of this is to customize character creation to facilitate the kind of game I want to run. I like truly randomly generated characters over carefully crafted ones, and I don't mind more or less powerful characters that may result from rolling ability scores, aging penalties, free feats, and extra ability score bonuses. 

Rolling ability scores in order produces more varied characters with interesting arrays (the buff wizard or the smart fighter). It allows for more interesting multiclass combinations that you wouldn't normally see, and starting with higher stats allows players to spend fewer of their ability score improvements on maxing out their primary ability score, which allows for more robust character customization much earlier in the level progression.

It's not for everyone, and I've had more than one person claim that this somehow completely "breaks" 5e, and that's fine. They can find another game to play or run their own. The customizability of procedures like this is one of TTRPGs' greatest strengths!