Variant Social Rules

Here are a couple of variant rules for social interactions. Their goal is to spread around who gets to be the party face, both by lowering the incentive for the character with the highest Cha to make every social roll, and by occasionally forcing other characters into the front seat.

Languages

One thing that struck me when watching The Bad Sequel is that:

• whenever there's an alien language, one of the protags can pretty much always speak it and translate for the others
• unless C-3PO is around, the translating protagonist is more-or-less randomly chosen

And I like this. It's simple and quick. It's no longer "everyone but the bard can safely dump Cha", since you never know when you'll be called upon to handle negotiations.

So here's a variant language rule:

All PCs, being widely-travelled oddball sorts, are multilingual, but language proficiencies aren't determined at character creation. Don't write down any proficiencies you might have from race, class, background or a high Int score.

When a character tries to understand or communicate with an NPC who isn't speaking Common, the GM decides the following:

• Is the NPC even capable of language? One GM might rule that a reanimated skeleton understands language, another might rule a skeleton magically follows its creator's commands but otherwise doesn't understand language. This is fine.
• Do outsiders share the NPC's language, or are they so culturally insular or isolated that this isn't possible? Obviously, secret languages like Druidic fall under this category too.

If the answer to both questions is yes, then the GM selects a single PC at random. Optionally, hirelings and the like can be included in this random selection process. This can be done with a bag of tokens, rolling a die and counting clockwise, whatever. I like the bag of tokens.

That PC can speak the language, perhaps not fluently, but enough to hold/follow a simple conversation.

Persuasion

There are no skills like Persuade, Intimidate, Diplomacy, etc.

You might decide to leave a Sense Motive/Insight skill in place. I would limit its uses to learning an NPC's emotional state, one of their ideals or bonds, or, in combat, judging what their immediate intent might be.

The player roleplays their character's pitch, then someone rolls 1d6 to determine the outcome. On a modified 5 or greater, they succeed.

The GM might raise or lower the difficulty. A Peaceful Villager might only need 3, a Haughty Archmage might need 7. Or the GM might apply the Powerful Enemies rule instead. This is meant to be simple though -- don't overcomplicate things trying to capture every nuance.

Modifiers to the roll:

• +1 if you have a relevant background
• +1 for a good social stat, e.g. 13+ in Cha. No further bonuses.
• +1 if you offer something of value*
• +1 if you appeal to one of the NPC's ideals, bonds, factional alignment etc.
• +1 if you act your heart out (can be hammy, doesn't have to be "good")
• -1 to 3 for demands that put the NPC at risk. -1 is "risk a parking ticket", -2 is "risk divorce", -3 is "be your torchbearer in the dungeon"

*Obviously the thing you offer needs to have value somewhere in the same ballpark as the request -- this is left to GM fiat.

Options

Use a Powerful Enemies rule if you like -- if the NPC has more hit dice than the PC, the difference is added to the difficulty of the roll. The reverse doesn't apply -- a 15th level fighter doesn't magically make every turnip farmer bend to their will. For 5e, use the difference in CR instead of difference in Hit Dice.

Certain NPCs may be immune to swaying the odds in particular ways, or may be more susceptible. Maybe a Magpiefolk gives you +2 to the roll if you offer a shiny geegaw of value, while a Angel gives you -1 to the roll instead, righteously furious that you would dare offer a bribe.

You can simulate regional language groups, loanwords, pidgins etc by putting an extra token in the bag when choosing who knows the language. This token represents whichever character, based on the GM's judgement, is most likely to know a language. e.g. the token represents the party dwarf when adventuring underground, the elf when adventuring in forests, and within the cosmopolitan City States of Arthea, the PC with "Background: Merchant". This isn't really a mechanical benefit, as any increase in one character's odds of knowing a language come at the expense of everyone else's odds. That is, don't make this a feat or anything.

I've presented this as a d6 roll-over, but feel free to reskin it to d6-roll-under, or modify this into a PbtA move or something. You could also multiply all the modifiers by 3 and use d20 roll-over, or use SotDL's boons & banes mechanic, whatever takes your fancy

Rationale

Yes, this means the polar Snow Elves and the Aquatic Elves of the tropical coral atolls probably don't speak the same "Elven" language. This is intentional.

I don't like how much weight 3e/4e/5e, or The Black Hack's d20 roll-under Cha, gives to possessing a high social stat. e.g. low-level 5e has a difference of about +5 (or +7 with Expertise) on d20, between a trained character with 16 Cha, and an untrained character with 10 Cha. All other factors (bribes, appealing to the NPC's beliefs, good roleplaying, etc) are typically handled with advantage/disadvantage, which doesn't stack.

This just gets worse as characters level up -- their proficiency bonuses increase, they may even find magic items that raise Cha beyond 20. For 3e D&D (faster scaling, stats don't cap at 20, magic skill bonuses are commonplace) the math is even worse, though Rich Burlew did write a nice, if crunchy, replacement Diplomacy rule a looong time ago).

So I want to reward clever manipulation and willingness to perform, but also allow a bit of room for characters being better at social stuff than their players, and not have the outcome be entirely DM fiat. Having a die result to point to, with really simple modifiers that are all worth the same 17% of the die spread, hopefully strikes the right balance for me.

Offering something of value is only ever worth +1. Why? Consider the following two examples:

• A stranger in the street offers me $10 to do some errand. Sure, I'll probably do it. If he offers me $10,000 for the same errand, he's probably a scammer.
• You're medieval nobility. A representative of the Pope says if you go on crusade, you'll be absolved of your sins (and oh, how you've sinned). This is theoretically a reward of infinite value, the difference between paradise and eternal suffering, yet not every nobleman went and did a Richard I. Humans are not rational appraisers.

Intangible Resources, a houserule for Knave

This is adapted from Throne of Salt's rules for Backgrounds. It's written for Knave, but is easily applicable for D&D.

A knave might have knowledge or social standing that they accrued before they join the campaign. These rules are intended for open-world, sandbox game with significant time spent in settlements. But in the event your dungeoncrawl one-shot spins off into a longer campaign, you can always go back and generate these resources for whichever knaves survive.

Generating a Knave's Resources

1. Add together your knave's Int, Wis, and Cha bonuses.

2. Roll that many d6. For every natural 1, put a point in Authority. For every natural 2, put a point in Renown, and so on.

3. Further increases or losses will come about as appropriate during gameplay.

Resource Roll Table

Die Result	Resource Type
1.	Authority - Influence within regions/social spheres where you have political power. May let you requisition resources, get better treatment from officials, etc.
2.	Reputation - As Authority, but where you don't have direct political power.
3.	Languages - Can you communicate with the goblins without resorting to pantomime? Can you decipher the ancient scroll? Can you understand the metaphor behind the sphinx's literary references?
4.	Secrets - Knowing the truth behind the curtain.
5.	Lore - Knowing the location of a forgotten temple, the properties of a legendary magic item.
6.	Divine Favour - Are you in good stead with the gods? Will they pull your arse out of the fire when your plan goes to shit?

e.g. Susan rolls up a new knave, getting Int +1, Wis +2, Cha +1. She then rolls four d6, and gets 1, 1, 6, 1. She notes an Authority score of 3, and a Divine Favour score of 1.

Susan decides her knave is the bastard daughter of the local lord. She cannot inherit title (hence turning to adventuring) but has almost as much social standing as a trueborn daughter would. She prays regularly but isn't particularly pious.

Using Resources

Some possibilities:

1. Roll-Under for Success. Roll a d6 when you want to draw on a resource. A roll equal or below your rating is a success.

2. Burn a Point for Success. You cannot roll, the only way to draw on these resources is to permanently cross off a point.

3. Burn a Point for Advantage. Permanently cross off a point to gain advantage on an ability score roll. You can retroactively use this on a roll you have failed.

4. Roll Under or Burn. Roll a d6, as with variant 1. If the roll fails, you may decide to succeed anyway by burning a point, as with variant 2.

The referee should decide on one of the above variants, or mix and match depending on situation and the risks/rewards.

e.g. you might decide that "the cheesemonger is having an affair with the town priest" is a pretty mundane secret, knowable with a d6-roll-under against Secrets. The True Name of a vengeful river spirit is a much more valuable secret that may require burning a point.

Consider:

• How long is this planned campaign? If it's only a few sessions, then perma-burning points may not be a meaningful resource.
• Are the players comfortable with the added complexity of remembering that ability score rolls are roll-high on d20 while want a high roll on a d20 while resource rolls are roll-low on d6?
• How often do you want players to solve problems with resources, compared with equipment or spells? Variant 4 is relatively powerful, and shifts the game to be more about managing social connections.
• How cool are you with a character who rolls a lot of one Resource effectively having a monopoly on that roll until they die or do something within the game fiction that reduces it? Requiring characters to burn Resource points is more likely to let everyone contribute.
• Depleting Resources eventually means characters will need to adventure specifically to regain social capital.

Note that adding Resources to the game should never take away the ability of any knave to bribe the town guards, find information in a library, or any other action resolvable with an ability score roll. Ability scores are, after all, king.

Knave Houserules

I intended to publish this shortly after Knave's release. Oh well, hopefully it'll be of interest to the growing community.

Houserules covered below the cut:

1. Odds of Success & Vocations
2. Death & Dying
3. Races
4. Backstab / Sneak Attack

The next post will cover Spellcasting and Cantrips. But first, some context.

Knave is a new rules framework (i.e. not a complete game) by Ben Milton. You can check out its product page or his  youtube video explaining some of its features and rationale.

@valzi has an excellent blog post listing various hacks, houserules, and resources he and others have made.

I've written a spreadsheet that randomly generates a 1st-level knave.

I think with some tweaks, Knave has real potential to appeal to people who otherwise wouldn't be into OSR games. Basically, I think a lot of people out there, perhaps introduced to D&D through streaming/podcasts or because friends are playing, are looking for a game that avoids the character build and tactical combat aspects of 5e, but who also want rarer lethality and more frequent spellcasting than the typical BX clone.

Knave could be the foundation for that game. It has enough mechanical similarity with 5e that it's easy to learn, but discards everything that bogs down the game (as judged by dead air on podcasts I listen to). Knave gets rid of, among other things:

• modifiers to damage
• bonus actions
• a skill list
• spell levels not lining up with character levels
• long-winded class abilities and spell descriptions
• any rules distinction between ability checks, attack rolls and saving throws
• proficiency bonus

So below the cut are some ideas to fulfil this design goal.

For now the following calls all d20 rolls "saving throws" in line with Knave regardless of whether they'd be an attack roll, saving throw or ability check in 5e parlance. I would change this for a published standalone product, but for a bunch of houserules I think consistency is important.

Adapting Troika!'s initiative system to 5e D&D

Troika! is a neat little game that is very loosely derived from the system behind the Fighting Fantasy gamebooks. The rules are easy to digest & teach, the setting is a weird, sometimes silly, sometimes poignant mix of Warhammer Fantasy and Spelljammer, characters' stats & background are randomly generated but they can learn new skills in play. You have no control over who you were, but you decide what you will become.

If I were to become even more of a cliche by starting an actual-play podcast, Troika! would be one of the few games I'd consider running.

It currently has a Kickstarter for a revised "Numinous" edition with better print quality, and there's a free no-art version of the current edition.

But seriously, check out the art :O

from the revised edition, art by Sam Mameli

Troika! Initiative Rules

Here are the rules verbatim from the no-art version:

5.1 Assemble the bag. During combat or at other times where it is important who goes first you will need to assemble the initiative bag. To do this get a container and a selection of coloured dice or other convenient markers. Each player will be assigned two dice of a single colour, all enemies will share one colour of the number specified by their entry or the GMs whim, and a final single token of a distinct colour will be added to mark the end of a round when drawn (5.3).

5.2 Using the bag. The GM will remove a token from the bag at random, the colour of which will determine who holds the initiative and takes a turn.

5.3 End of round. If the end of round token is drawn then all tokens, including the end of round token, are put back in the bag. Resolve any per round or end of round activities such as magic effects, fire, poison or bleeding out, then draw another token and carry on.

5.4 Henchmen. If you have any hired help that are willing to fight for you treat them as their own character that only gets 1 initiative die in the bag.

5.5 Rationale. The random turn length adds a degree of uncertainty where you never know how much time you have left. When actions are not taking place it represents hesitation, panic or other incidental delays that can happen in a tense encounter where every second counts. The goblins have few dice because they are cowardly, not because they are slow; the dragon has many because it knows exactly what it wants, not because it is fast.

Adapting for 5e D&D

• Disregard Initiative modifiers. Dex is already too much of a god-stat anyway.
• Characters with a class feature or feat that would give a bonus to Initiative get an extra token in the bag each round, but only act on the first two tokens drawn. Multiple "improved initiative" features don't stack. A non-exhaustive list of such features:
  • Alert -- feat
  • Feral Instinct -- Barbarian 7
  • Remarkable Athlete -- Fighter (Champion)
  • Dead Ambusher -- Ranger (Gloom Stalker) 3
  • Ambush Master -- Rogue (Scout) 13
  • Rakish Audacity -- Rogue (Swashbuckler) 3
  • Tactical Wit -- Wizard (War Magic) 2
  
  Exceptions can be made for very high-level or situational abilities. e.g. a 17th level Rogue (Thief) gets an extra two tokens in the bag on the first combat round, and acts on the first four drawn.
  Note that as no d20 roll is made, the following don't grant an additional initiative token:
  • Enhance Ability -- spell
  • Bardic Inspiration -- Bard
  • Portent -- Wizard (Diviner) 2
  
  
• Beast Master rangers don't have to spend their action to command their animal companion to attack. Either give their companion one token in the bag, as with NPC henchmen, or let the animal companion act on the ranger's first token drawn each round.
• Since everyone gets up to 2 turns each round, I'm not sure how to handle spell durations that specify "1 round, until the end of your next turn". For now, I'll treat it as "until the end of your next turn", regardless of when your next token is drawn.
• Similarly, I'm not sure whether spellcasters are restricted to one non-cantrip per turn, or per round. For now, it's per turn.
• Instead of the suggested DMG method of representing a mob of monsters by cross-referencing number remaining, attack bonus, and player AC to see how many auto-hit, do the following:
  • Multiply a monster's HP by about half of however many there are? You can multiply it by the full number if using a Morale rule.
  • Give them one or two extra Initiative tokens, and advantage on all attacks. They can't all get good positioning to attack at once, some might be cowardly or willing to let others be the vanguard, but they can all assist whoever's attacking.
  • Single-target attacks deal all their damage as though it were one big monster, i.e. if your greatsword hits for 15 damage against a mob of 5 hp hyenas, you cleave through three of them, not just one
  • Area attacks may deal double or even triple damage.
   

That's all that occurs to me for now. It'll probably be a while before I get to playtest these changes (my current 5e game deliberately has almost no houserules to keep it accessible to drop-in players), but I'm keen to switch up the rather... clockwork nature of modern D&D initiative for something a bit more exciting and scary.

If there's any glaring issues I've missed, let me know in the comments :-)

0th Level Characters in 5e

Note these 0th-level houserules are intended more for an adventure like the classic N4 Treasure Hunt, not a DCC-esque funnel game where every player controls multiple 0th-level PCs with the expectation most of them will die.

Waxing Lyrical About N4

(forgive this indulgence, skip down to the 0th level rules if you want)

My first attempt at DMing 5th edition was an adaptation of N4. I adore this module ever since I played in it under Pathfinder rules, though it certainly has its flaws. As-written it's a massive railroad -- my Pathfinder GM did a very good job of concealing that from us -- and it leans a little too much towards verisimilitude. There are two long-abandoned buildings whose room keys are largely iterations of "ruined furniture", "tatters of tapestries", and "chests of moldy garbage".

But the module has its cool moments, and playing at 0th level was my first real exposure to old-school paradigms of "avoid combat unless you can make it completely unfair in your favour", "emphasise player creativity over a long character sheet of class features", and "de-emphasise ~builds~, let characters grow organically based on their choices and mistakes".

I really recommend more new-ish players (I started playing in 2005 under 3.5 rules) give 0th-level play a try. After all, if you end up hating it, you'll soon be at 1st level, and you'll have some sense of the party dynamic so your GM will know what adventure material (official or homebrew) would be most fun to run for the group.

The Korinn Archipelago is also a great mini-setting that can be standalone, or segue into most of the official D&D hardcover campaigns. While the adventure doesn't mention the Realms, surviving PCs are canonically just a few days' longship travel (260 miles) from Waterdeep.

excerpt of "Map of Faerün", by Mike Schley

0th Level Rules

I've tried to explain my choices so that you can better decide what to keep/change.

For each ability score in order, roll 4d6 and drop the lowest. You may then swap one pair of scores, before applying racial adjustments. This is to discourage too much pre-planning about what your character will become. To completely minimise preplanning, I'd remove the ability to swap two scores, and run a game such as Swords & Wizardry or Whitehack where ability scores contribute very little to combat or spellcasting capability. Or a game where ability scores advance faster, e.g. The Black Hack and its various descendants. But this is for use with 5e, and I didn't want to completely remove customisation.

Choose a race and background, but no class. You get any corresponding features and proficiencies. However, if this would give you an at-will spell (e.g. high elf) it is only usable once per short rest. Once you reach 1st level, the spell becomes at-will. This is because N4 starts the characters in a shipwreck situation where they have to scavenge for weapons, and I didn't want to trivialise that with at-will magic attacks.

Starting HP = 6 + your Con modifier. You have no Hit Dice for short-rest healing.

Your proficiency bonus is +2. You have proficiency with simple weapons, but no armor or shields. Racial traits may give you extra proficiencies.

You start with the normal equipment for your background (i.e. no shopping from the equipment list) but a character with a variant/custom background might start with modified equipment, subject to GM approval.

Anyone can attempt to cast from a found spellbook, pray to deities for a miracle, or do other class-specific things, with a successful ability check. Attempting such class-specific things, or wearing armor & martial weapons in combat, is the key to leveling up into a class. Note that success/failure is not important, only whether you've tried in situations that have stakes.

The GM may decide to model characters slowly learning proficiency in something. Perhaps after two combats wielding a looted greatsword, you add half your proficiency bonus to attacks with it. Perhaps after enough study/experimentation with a spellbook, or immersion in the sacred poisonous waters of Lunulata Lake, you can cast a single cantrip.

You become 1st-level when the GM says so. They may restrict your class choices based on what you've attempted. GMs, be lenient with this. It's no fun saying "you can't be a wizard despite studying those spellbooks, because you also tried wearing gambeson and swung a longsword", though you could say "have you considered becoming a bard instead?". A character classing into druid might have to forsake metal armor from then on, but let a nature spirit appear and offer them that ultimatum. This might mean characters end up with an extra weapon proficiency pr something, but that can be their reward for reaching 1st level the hard way.

The GM might instead choose to start you at, say, negative 300 XP. But with combat so dicey at 0th level, this is only recommended if there's plenty of opportunities for non-combat XP.

When you reach 1st level, revise your HP to the "starting" HP for that class (and gain a Hit Die), note any new class features, choose some proficient skills. Your GM might let you choose skill proficiencies based on your actions, even if they're not on your class list. Your class's starting equipment doesn't pop into existence, so hopefully you found some neat stuff as a commoner :-)

Paring Down 5th Edition #2: Simplifying Spell Points, and a Dice-Based Variant

tl;dr:

idea 1: spells cost MP/mana/spell points equal to their spell level. This ends up being surprisingly close to the DMG costings.
idea 2: instead of having fixed MP/day, casters have spell dice/day. For each MP needed, casters roll a magic die (a d6). A natural 1 or 2 means the die is expended for the day. Different conditions, caster specialisation, or defiling/preserving may change these odds. Additional mechanics for particular results (doubles, triples, runs, etc) may be invented as desired.
See the Longwinded Explanation (after the cut) for reasoning and advice on which column to choose. Sorry for the lack of CSS.

Table 1: MP and Spell Dice by Level (with and without Arcane Recovery)

Level	MP	MP (+AR)	Spell Dice*	Spell Dice** (+ AR)
1st	2	3	1	1
2nd	3	4	2	1
3nd	8	10	4	3
4th	10	12	5	4
5th	16	19	8	6
6th	19	22	10	7
7th	23	27	12	9
8th	27	31	14	10
9th	36	41	18	14
10th	41	46	21	15
11th	47	53	24	18
12th	47	53	24	18
13th	54	61	27	20
14th	54	61	27	20
15th	62	70	31	23
16th	62	70	31	23
17th	71	80	36	27
18th	76	85	38	28
19th	82	92	41	31
20th	89	99	45	33

* Dice expire on a 3 or less
** Dice expire on a 2 or less

Paring Down 5th Edition

I play and enjoy 5e, but it really needs paring down of its unnecessary complexity and constraints. So this is the first in a series that explains some houserules I'm working on, and what I intend to accomplish with them

Many of these ideas are lifted from various OSR blogs and systems, but one of the goals here is to maintain the "heroic fantasy" power level of vanilla 5e, instead of the grittier feel many OSR games go for. Also I want to keep the player-facing rules pretty similar, so someone could potentially rock up to my table with a vanilla 5e pre-gen ranger and not have to translate their sheet before getting to play.

Ability Score Generation

Note that what I refer to here as "ability score" here corresponds to what vanilla 5e calls "ability bonus". It just reads a lot better that way.

Random Method

1. Roll two six-sided dice and ignore the highest result (i.e. "2d6 with disadvantage"). Subtract 2 from the remaining die and record it as your Strength score. Repeat this process for the other five ability scores, in order (Dex, Con, Int, Wis, Cha).
2. If your highest score is +1, you may re-roll scores of your choice until you attain a result that's +2 or higher.
3. You may swap any two scores.
4. Do not adjust scores up or down for your choice of race. Instead, choose one ability score that would be adjusted under vanilla 5e. Humans may choose any one ability score. Reroll that score, keeping the original score if it's higher.

e.g. Becky is rolling Lucretia, a dragonborn fighter. She rolls the following die pairs:
⚂⚃, ⚂⚄, ⚄⚀, ⚀⚅, ⚂⚁, ⚃⚀
for the following ability scores.

Str +1, Dex +1, Con -1, Int -1, Wis +1, Cha +2
She is not entitled to step 2. For step 3, she decides to swap her Con and Cha scores and hopes that step 4 will be kind to her. For step 4, she notes vanilla 5e gives dragonborn +2 Str and +1 Cha. She chooses to reroll her Str, getting ⚅⚁.

At the end of it all, Lucretia's scores are:

Str +2, Dex +1, Con +2, Int -1, Wis +1, Cha -1

Tweaking the Odds

Ignoring racial adjustment, getting an 18 on any one roll of "4d6 drop the lowest" is a 1/54 chance. Getting the equivalent +4 here is 1/36. You could tweak these odds in all sorts of ways:

• High-Powered. Roll 2d4, keep the lowest
• Low-Powered. Roll 2d8 keep the lowest, subtract 4
• Ummmmmm? Roll 3d6, keep the median die, subtract 2

 
Semi-Randomly Assigned Array

 Alternately, the GM may allow the following method, which semi-randomly allots the same set of ability scores to every PC:

1. You have six bonuses, +3, +2, +1, +1, +0, +0. You may assign the +3 to any ability score.
2. Randomly select a remaining ability score (e.g. by pulling out of a hat, or rolling d10, subtracting 5 if needed). Assign it +2.
3. Assign +1 to a remaining score of your choice.
4. Randomly select a remaining score (e.g. roll d6, subtracting 3 if needed). Assign it +1.
5. Assign +2 to the remaining two scores.
6. Do not adjust or reroll any scores for race.

Alternately the GM may just allow players to assign stats in whatever order they choose.

Design Notes

I only like traditional 3-18 ability scores for roll-under systems (Whitehack, Sharp Swords & Sinister Spells). The rest of the time it's this unnecessary step in character generation. It confuses new players when it's the bonus, not the score, that gets used most of the time. This is made worse by the few rules that do refer to the score, e.g. long-jump rules.

This approach is inspired by the playtest of Knave, which is an excellent classless system for classic D&D or OSR material in development by Ben Milton here. The main differences are one fewer die is rolled making it easier to roll high, some limited customisation after rolling is allowed, and this system adjusts the scores down to account for 5e's proficiency bonus.

The proficiency bonus may end up nixed from this set of houserules -- while it's a good unifying mechanic and certainly an improvement over 3.5's mess of different scales, there's a good chance it'll end up unnecessary here. If that happens, this post will be revised to remove the admittedly clunky "-2" bit. The only reason 2d4's not presented as the default is that they're a pain to roll, and one of my nonessential goals is to stick to just d20 and d6 for all player-made rolls.

I've also grown to hate racial adjustments to ability scores except in the most extreme cases (e.g. playing a giant). They unnecessarily constrain race/class combinations, unless the player is dead set on playing against type. I've seen anecdotes from various DMs who ran the D&D Next playtest (where for a while, racial adjustments were +1 at most) about how they loved the variety of character concepts they were getting at their table. Now that diversity's measurably died back.

The "reroll" idea is borrowed from this Coins and Scrolls post, spoiler alert, racial abilities will be handled in not-dissimilar way. I basically agree with the reasoning there -- elves as a population can be more nimble than humans, dwarves more hardy, but individual PCs can be whatever the player wants without getting too nerfed.

edit: here's a couple of quick charts showing the likelihood of rolling various scores. Note these include the +2 proficiency bonus at 1st level, as "to-hit" and "total bonus for trained skills" generally matter more for determining an appropriate power level for a campaign.

Effect of die-keeping method

Effect of die size