Random Stats: Story vs Creativity vs Math

As I'm gearing up to try to run a game or two, I've been thinking of how to generate ability scores. For my tentative 2nd Edition Al-Qadim game, I've got the following penciled in:

Roll 4d6 and drop the lowest die in order for each ability score. If you are human, you may rearrange them in any order you choose. You must fit race and class minimums and maximums for any given score.

This gives us randomness while, at the same time allowing people to play the class they desire at the cost of an interesting race. I waffle over whether to somehow encourage a few other races by allowing them one ability swap to a "favored class" sort of thing: something like an elf may swap Intelligence with one other score, a dwarf may swap strength with one other score, a halfling may swap dexterity with one other score, etc. "Rare" races could be not allowed a swap so you either pick the race you want or the best class but not both. Less relevant for a game with a limited slate of races than an open-ended game like 4e which had a Mos Eisley cantina of races.

I like the idea of random ability scores, but the post third-edition math makes rolling potentially really bad. That is to say, a few high scores can be slightly unbalancing, whereas older editions didn't assume you'd have particularly high scores at all.

Now, I wouldn't encourage rolling abilities for anything third edition or later. The bonuses are just too important. One high stat can make a great OSR character, but you're not hindered by mediocre stats.

What's the benefit of randomness? I think, beyond some fetish for tradition and dice rolling, I think it forces a little creativity. We often let players make characters ahead of time, or else they have ideas of what they want to make before arriving at the table. But once you're stating at a 16 strength and 12 intelligence you've got a choice: Do I make the wizard I want who struggles with wizardry, or pick the fighter because I've got the stats for it? Because fewer things are based on stats in older versions of D&D, you can make a 12 intelligence wizard who is relatively functional. You cannot do that in modern editions of the game.

However. Some of the randomness may come at a price. Namely, it may be best to start with a coherent party rather than character concepts. And once the players decide they want to be crusaders fighting against demonic old gods and founding a new city to dedicated to their ideals... Those paladin-preventing scores could be rough. If I'm asking players to decide before they create characters what sort of alignment or ideals the party might have, that may limit some character options that randomness would encourage or possibly discourage.

I'm hoping my rule above has struck a reasonable balance. By allowing humans to swap scores around, it allows people to be the rarer classes of their choosing. It also will make demihumans less common: non-humans are more likely to have unexpected high and low ability scores for their class which or the player might opt for a class that fits the race better.

One possible problem I see, however, is that 2nd edition (like pretty much all editions) needs a healer. It may be beyond randomness in character creation though. If no one rolls well for wisdom or wants to be a priest, there's not many healing options. I'm hoping to alleviate this decision by including the Spells & Magic crusader and monk (revised as Dervish) alongside the more native Al-Qadim classes. I.e. more priest choices so one might be more appealing to players.

None of this, however, deals with the problems of a possible 5e game though. If you roll 4d6 drop the lowest the average character will have at least a 15 as their highest score and its reasonably likely to having between a 14 and 17 as their highest on either end of the probabillity curve. These numbers are high at low levels, but the cap at 20 probably helps. Most ability scores will end up 9-15 (within one standard deviation of the mean of 12.2), which is at least in a reasonable range for modern D&D. Maybe its not worth worrying about it and using the same rule will be fine. Or maybe the rule should only be by-the-book humans rather than variant humans. But here's the real problem: low scores aren't a huge deal in second edition, but in modern editions (including 13th Age) they are.

Low scores can be damning in modern editions. The fact that the 13th Age designers hadn't encountered characters with a negative constitution modifier before making the necromancer in 13 True Ways highlights this. Stats under 8 make for some really shitty characters. This is obviously mitigated by point-buy options, but rolling a 6 isn't going to be super uncommon.

The possible answer: Modify the rule above to allow humans to choose the standard array. This would ensure no PCs have crazy low stats, but doesn't prevent luck from giving someone three 17s, which are bumped up by the chosen race. I might just have to live with a little luck, I suppose, unless I want all the characters to include no randomness whatsoever. And I like how my rule encourages humans at any rate.

I suppose I could reduce the role of luck by changing the rolling method if there's a point buy backup for humans. 3d6 in order, or 3d6 twice in order might do. There's still a chance that luck will grant an exceptional character to one player while the others rely on a point buy.

One possible problem remains with randomness: The players decide on a mercenary party attempting to raid the tombs of old gods to find the wealth needed to found a new city and rule over it, yet one player rolls perfect paladin or monk stats. Maybe that's just a roleplaying or creativity challenge if the player really wants to play the character they rolled who doesn't quite fit the party. Or maybe the party needs to be designed at the same time as ability scores are rolled.

Ultimately, I still like my rule. Modulo it might need to be adapted for a specific version of the game and specific racial desiderata. Minimums or rearranging ought to prevent absolutely useless characters, and randomness can nicely constrain creativity. However, it means one does need to tread carefully with the party story perhaps.

Gussying up monsters (the Ogre!) and crits

The Antagonizer lamented that running 5e monsters was a bit lackluster compared to 13th Age or Dungeon World. Like: he wanted an ogre to grab two PCs and smash them together, giving me a vague impression of that scene from the Avengers where Hulk whips Loki around like a rag doll. So obviously I challenged him to a contest of gussying up some of the basic monsters in D&D. I sorta said I'd work on the Ogre. This has definitely sprawled beyond the Ogre, but I think in a good way. Ultimately my thoughts have moved a lot more to critical hits and fumbles over fixing the Ogre, it seems.

But I said I'd do the Ogre, so here's some thoughts. Also we've now agreed that we should both write up our thoughts separately on this topic, so I'll come back to this after I read what the Antagonizer has written. tl;dr: There's stuff for ctitical hits and additions to the Ogre at the bottom.

First up, 5e Ogres are boring, but 13th Age ones aren't all the inspiring to me. Though I do like a few of the story implications of 13th Age ogres, the stats are still a bit lackluster even with the rock-solid—if not completely incredible—13th Age Bestiary.

Second up, I want to respect the rules lawyers out there. Being a bit of one myself, I totally understand wanting to know how I escape the ogre's grapple. So the restriction I've got here is one that keeps the basic combat rules intact, but adds a few options. So this can't be full-on Dungeon World style narration where anything goes.

Third up, gussying up combat might have some interactions with critical hits and weapon properties that I've toyed with before.

Fourth up, there is no fourth. All hail discordia. Or some nonsense like that.

Mechanical and Narrative Options
So gussying up the Ogre. What you really want is something that makes combat more interesting by adding complications in:

• Dungeon World does this by giving mechanical and/or narrative penalties for failed rolls (you attack the Ogre but roll low, so it damages you or knocks you prone) and bonuses for high rolls (sweeping attack, you damage the ogre and knock it perilously close to the cliff's edge).
• 13th Age (often, but somewhat irregularly) does this by using the technology of the d20. Low rolls against certain creatures open you up for attacks or effects, natural odd/even results can trigger secondary attacks or less common effects (essentially replacing generic: it hits you, save vs whatever to avoid some condition), and sometimes specific numbers (a generic high roll of 18+, natural 2, natural 5, 10, and 15) can also trigger effects. This makes 13th Age a bit more exciting for the GM because you don't know when a creature will unleash its bigger power, and makes combat a bit more random. Its a little unevenly applied in some monsters get this treatment and some don't, and some get a lot more of it than others.
• 4th Edition did it by adding rider effects to just about all your powers. Damage wasn't the only thing you did, but you also moved creatures about and inflicted conditions. Now 13th Age does that a little bit as well, but 4e really made tactical combat shine. And, I suspect for some, changed the game a lot beyond just "I hit it with my axe." Before 4th edition (and the 3rd edition Book of Nine Swords, I'll wager) this sort of awesome combat was largely found in indie games. Now it's becoming necessary in the mainstream stuff.

So what we want is a way to add in some of these mechanical and flavorful benefits into 5e or even OSR combats. I think we see the lack of this in the 5e Critical Hit rules.

Critical Hits
In 5e, these simply do double damage dice on a nat 20. How unexciting is that? I think the 4e method was slightly better: max damage plus maybe an extra die. Thing is, people like rolling dice but double dice is statistically equivalent to max damage in terms of averages, but has lackluster low possibilities (2 damage on a crit?!?) and unbalancing highs (one attack is effectively two really powerful blows). I've mused before (though apparently not written about?) altering critical hits to include some of these tactical options. Like: a crit can either do max damage or you get to shove the enemy and do normal damage. I included this as an option in some weapon properties (disarming, sundering, tripping) to make combat more interesting.

One could make a table of these sorts of effects:

1. Knock the opponent back
2. Knock the opponent prone
3. Disarm the opponent
4. Sunder the opponents weapon
5. Sunder the opponents shield
6. Sunder the opponents helmet or +1d6 damage
7. Grapple the opponent
8. Inflict a level of exhaustion

Ultimately it might be satisfying if the ogre could grab or knock back an opponent (or some other weapon-appropriate option), but crits are pretty rare. And an extra table roll for a crit isn't a huge game stopper, but if we want this to be flavorful and more common than a natural 20 you don't want to always roll on the table.

Interestingly, you can make this happen twice as often simply by allowing the effects to happen when the opponent rolls a natural 1 as well. So now we're getting into a 10% chance that some of these effects are going off, even if the players are benefiting from them quite a bit as well as the enemies. No longer is that natural 1 just some random "You horribly miss" but "You leave an opening, the ogre uses its reaction to grab you."

Templates for creatures
One idea to apply a 13th age style mechanic to creatures is by adding templates. These were a third edition idea where you simply increase the power of a monster and likewise increase their challenge rating and XP value. So we could give our Grabby Ogre the ability to get a free grapple in on a successful even attack, and when an opponent is grappled the Grabby Ogre gets the ability to fling the opponent 10' feet (preferably into a wall or an ally). Given that this is random though, and that the ogre would need to survive for about 3 rounds to have a good chance of having this go off (and the PC probably gets one chance to escape the grab), I'm not sure it really necessitates modifying the XP values in this case, but if you add in something else it might. Like if this were instead a scourge-wielding priest of Ishishtu who is knocking players prone on a field of caltrops on natural even hits... You can see why this might be something to consider in encounter building if you're the type to actually count XP or even just want to inform PCs that "It is obvious that... you are outmatched in this fight."

But assuming we want to go the template route, I think it follows on those weapon properties to some extent. We can build a few templates that can be applied to give 13th Age style mechanics to these creatures. Like:

Grabby X.
Natural Even Hit - Target is grappled. (Limit the number of grapples by the creature's number of appendages probably).

If you have a grappled target, you can make the Dirty Hands attack.
Natural Even Hit - Squeeze the target for extra damage and inflict a level of exhaustion.
Natural Odd Hit - Fling the target at a nearby ally. If you hit, both take X damage.
Miss - Fling the target 10' away, can make a dex save to avoid X damage.

Scourge-weilding X
Natural even hit - Free grapple
Advantage when knocking grappled targets prone. Target takes normal damage from attempt.

Avoiding the Rules Lawyers (a.k.a. Nastier Specials)

I would apply these templates post hoc and liberally. First off, many humanoids will be equipped with weapons and should benefit from the same types of weapon properties the PCs get. So you can easily apply a Pokey X template to the three orcs with spears (meaning maybe they get to be 3 abreast in the 10' corridor because they're using spears rather than big axes). You might not describe the weapons of each combatant at first, so just deciding to give someone a Sundering Axe midway through combat isn't the worst. You can also just apply a template when a creature is enraged (or bloodied?), so the ogre could be grabby once it gets disarmed or gets hit by that annoying ranger. I'd just be careful to remember that I decided that this combat I've got a Grabby Ogre that uses odd numbers because that's when I decided to apply the template.

Its also important, I think, that these templates only change the rules for monsters. That is, they do get a chance to escape the grapple or maybe a dex save to avoid being tossed off the edge of a tower. Because you want to know what your character can be expected to do and you want to be able to possibly minimize some effects that you hate.

Some effects may be better off as the PC makes a save after each attack, such as poison. Dwarves get a bonus against poison, and if I played a dwarf I'd like to make use of that. So that's not as good of an option as knocking someone back, disarming them, or sundering gear. Extra rolls can slow things down, but in the case of poison and dwarves, its worth it to show off how awesome dwarves are. This might also apply in other cases, such as frosty enemies slowing you down or whatnot.

Types of effects to apply

So what types of effects can be readily used and reused?

Forced movement & prone. A staple of 4e, but I'd restrict it to shorter (5'-10'?) increments in theatre of the mind in general: some of those 4e effects could be ridiculous (beguiling strands pushing opponents 25-feet!?!). Falling rules can be applied in terms of giving additional damage as needed.

Grapple. There are already monsters in the book who auto-grapple on a hit. Gives PCs a chance to waste some actions to attempt an escape as well, and those with athletics or acrobatics type proficiency get a chance to show off. Obviously restrained is harsher, so I'd reserve that for actual criticals maybe?

Sundering and disarming. I like these, just because it teaches players to carry a back-up and also changes the nature of the combat a bit. Obviously sundering is a bit toucher compared to disarming, and would be possibly quite harsh with magic items. It might be worth saving sundering for real bad-asses and critical hits, or both of these for just natural 20 options. Then again that mending cantrip would easily fix a sundered weapon or shield, and who ever thinks mending would be a useful cantrip to have in combat? Given that this is a special action for PCs though, it seems a little less appropriate for villains unless they're wielding crazy special weapons (flindbars!!!) or are weapon masters. 

Exhaustion or wounds. I've considered using something like this with criticals in the past. Its a good way to remind PCs that combat is dangerous. These should be reserved for major bad asses though. Or possibly exhaustion for level-draining types of undead. Cause level drains are fucking nasty, but exhaustion at least is kinda similar. Aside: Obviously I think the exhaustion rules might be a good way to simulate some lingering injuries, but it may require a bit more thought.

Blinded, charmed, deafened, frightened, stunned. Sure, but it probably needs a end of the round or save ends thing. Also not at all sure what creatures would apply these effects, they're probably already built into the creatures.

Paralyzed, petrified. I feel like these effects are probably already integrated into the creatures in question by and large, more so than the others.

Poisoned. Again I really feel that, since dwarves get advantage on saves against poison it should require a save.

Difficult terrain. Don't overlook changing the battlefield. But I wouldn't let the monsters do this in ways that PCs can't (i.e. no free overturning bookcases generally) but maybe some that leave huge footprints or trails of slime...

Reduced movement. That ray of frost spell does it, so I don't see why some monsters couldn't reduce your movement. Though again... harder to imagine when to use this.

Looking at all these effects we can kinda categorize them into a two groups: normal and unusual. Forced movement, Prone, Grapple, Sunder/Disarm, and maybe exhaustion are pretty normal. You can imagine these happening a lot. They're good candidates for templates. But Heck, even a slime trail template could involve leaving oozy puddles for difficult terrain. The others though: particularly petrified and paralyzed, are poor choices for these templates. They require really specialized monsters (medusa, basilisk, etc.) and are already built into the design for those creatures.

So a lot of this kinda comes down to a few normal-ish monsters doing many of the same normal-ish things.

The end game
I'm looking at a two-way solution here which involves critical hits and fumbles, but also possibly 13th Age style templates. I'm eager to see if the Antagonizer came up with some similar thoughts as I did, but I expect to revise some ideas after we bounce things around more.

First: On a critical hit or fumble, mirroring the weapon properties I advocated for earlier (whether they're whole-heartedly adopted or just ad-hoc adapted) I like the option of doing max damage or normal damage with the appropriate condition. These would be a free grapple, shove, disarm, shove aside, mark, tumble, or whatever attacks. Your weapon should take that into account, and canny players should ask things like "Can I do normal damage and also topple the bookcase over onto them?" I think you'd also make a ruling about whether things auto succeed or not based on how likely they are or if they do lots of effects. Like: knocking someone down with a whip or large staff seems like it'd happen no problem. Knocking someone down with a longsword or axe might require the roll. Knocking the bookcase over to block an escape or create rough terrain seems fine, knocking it onto the guy to possibly trap him or do extra damage might allow a save. The reverse should be allowed on a fumble: the knight fumbles his attack against the scourge-wielding priest and the priest gets to trip the knight. This may warrant a little revision of my weapon properties, like the sundering property activates on a 19 or 20, rather than just a 20. I can live with that probably.

In terms of fumbles, I like these less. So I'd generally advocate for something where the fumble involves a die roll (i.e. shooting into melee might force you to attack an ally) rather than simple bad things happening (your bow breaks). Also you may give the player an element of choice to avoid the fumble (you rolled a 1. Your shot is awful, you can either lose your successful attack against the ogre or you have to roll against the fighter). Still thinking about this a bit.

As an aside on probabilities, the third edition critical confirmation is actually genius, though it does slow things down. It takes actual critical hits from a flat 5% to 5% of hits, meaning they're much more common for fighters than wizards. In the flatter math world of 5e I'm not so sure it matters. Combat & Tactics gave crits on any natural 18+ that also beat the AC by 5 or more points which at least doesn't require another roll and eliminates the thing where you can only inflict critical hits against the dragon because you can't hit any other way.

Critical Hits. If you roll a natural 20 in combat, your attack deals either: a) maximum damage or b) normal damage and allows you to do something interesting. The DM may suggest some standard interesting things based on your weapons or spells (i.e. staves and flails tripping an opponent) or you can suggest something interesting yourself (i.e. disarming an opponent or knocking over a bookcase). The DM will tell you if your interesting thing will automatically succeed or if it may require a roll before hand so you may still opt for maximum damage instead.

Critical Fumbles. If you roll a natural 1 in combat, you've made a mistake. This may open you up for reactionary attacks from nearby enemies, force you to attack a different target, or make a save against some effect.

Second: Templates for more common effects. If you want a grabby ogre, or a slime that trails difficult terrain, or a frosty sword that slows enemies down... just let that happen on a natural even or odd. Well, ok, you might need to work out a slightly more complex template if you want to let the dice tell you what happens after he grabs you (rip off your helmet and try to eat your face!). But half a dozen templates or maybe even a dozen will probably go a long way and see lots of reusability. Even Grabby could be applied to the Orc when he's fighting the gnome and halfling!

So without further ado, additions to the Ogre's normal stat block based on the Grabby Hands template. I'm not 100% sure I like it all yet, but its a start and could be re-used for other creatures.

Actions:
Slam. Melee attack: +6 to hit, reach 5 ft., one target. Natural Even Hit: 13 (2d8+4) bludgeoning damage and the target is grabbed. Natural Odd Hit: 13 (2d8+4) bludgeoning damage.

Dirty Hands. Melee attack: +6 to hit, reach 5 ft., one grappled target. Natural even hit: 13 (2d8+4 bludgeoning damage and inflict a level of exhaustion. Natural odd hit: Throw the target 10 ft, 15 (2d10+4) damage if the target hits a solid obstacle. If you fling the target into another creature the target takes half damage and that creature must make a dexterity save (DC 14) or take half damage. Miss: Fling the target up to 10 ft. Target must make a dexterity save (DC 14) or take 13 (2d8+4) damage.

Reactions:
Quick Grab. The ogre can make one grapple attack against one target within 5 ft. that rolls a natural 1.

Must buy: SlaughterGrid

Christ on his cross. Sometimes I really debate telling other people about a glorious item that is 100% amazing because I have dreams of running it. (I had this problem with Far Away Land.) But sometimes a thing is so good you have to spread the word. This thing I SlaughterGrid. Read the review here.

Let me be frank. This is a little too weird/gonzo for me. I'd have to really detail the crazy hex-crawl nonsense a bit and even then it wouldn't quite fit into any campaign I'd run long term. And this is coming from a dude who recently ran a game that included an onion witch whose crazy onion eyes let her see things from afar but only during the lead-up to orgasm (and probably a bit post orgasm)... And that was, in part, because the players opted to avoid the river rafting which was kinda the railroad of the module. For gonzo/weird shit, I have a hard time keeping all the crazy in mind and making it meaningful crazy. I know when it looks good though, even if its not quite for me.

Nonetheless. I agree that this adventure seems very well done. For me, the adventure is unuseable as it. But still 100% amazing. You get enough details to run each little encounter. Specific attention is payed to smells, writing, and the like. I <3 the bullet points, I've long debated if I could get away with a bullet point format (not just for myself but when gifting things to others) and this answers with a resounding yes! This thing is basically wonderful. Buy it. Read it. Dream of running it and realize it was still worth the cost for all the ideas you'll steal from it.

In a way, it seems overdone. Each encounter is kinda meaningful. Almost Torchbearer/Dungeon World meaningful. Like its a big deal, more than 2d8 gnolls. I think the real trick is figuring out how to use this as inspiration to make 2d8 gnolls meaningful.

The cleric is broken: or, how the eff do I give clerics (and others!) new spells...

I'll admit it: I legitimately believe that the 2nd edition specialty priest was one of the best forms of the Cleric class. But we live in a post third-edition world (and first-edition and before are part of that world too), and I have a legitimate dilemma. How the eff do I give clerics new spells..? (I've noted this before.)

Clerics, since time immemorial, have always had access to their entire spell list. The problem is that every book that expands the cleric spell list directly expands the cleric's versatility and therefore power. The conflict is: certain gods should grant certain spells, but the cleric class doesn't have a great mechanism for gods granting spells.

The genius of second edition (at least following on the heels of the 1st edition Dragonlance Adventures) was dividing cleric spells into spheres. You could easily assemble a list of spells known based on spheres, and new spells were simply added to the list of spheres. It wasn't perfect, but at least each book with new spells didn't automatically add to each priest's selection. And those that did add were reasonably appropriate.

Now, I wish 3rd and 5th edition had expanded domains a bit, such that each domain granted more than a couple spells at each level. Ultimately that would be like an expanded second edition sphere of all but more distinction between deities/domains. It still suffers because there's no way to grant appropriate spells to a cleric of Isis and not to a cleric of Horus or Set. The problem remains: expand the cleric list, you do it for all or none (or specifically just to a few).

So I'm considering running a 5th edition assorted Saturdays game here in faraway and need to decide how to integrate spells from the Elemental Evil stuff or the Necromancer Games' Book of Lost Spells. I'm picking on the cleric here, but most D&D classes are now like this: bards, clerics, druids, fighters (eldritch knights), rogues (arcane tricksters), sorcerers and warlocks [Edit: forgot paladins & rangers!] all have the same basic problem. However, there's a few options.

1) Just fucking add them for everyone. This is the least satisfying option.

2) Let some appropriate spells overwrite generic spells. This is say I offer detect disease or putrefy food and drink in exchange for some spell on the regular list. This is reasonable at each spell level, and requires the DM (i.e. me) to decide which spells belong to a particular deity, patron, or whatever-the-fuck bards and sorcerers use for gaining spells. Its reasonable, but also unsatisfying.

3) Discovery. Through mystic tomes, journals from dead mages, runestones, or whatnot I could add spells to particular characters' lists. This is still in the DM's hands, and could be combined with option #2 I suppose. The thing is... there's no real reason why bards couldn't have found a particular spell anyway. For bards and sorcerers this is a bit more satisfying as they basically just add to a set list of spells known. For clerics and druids, this is less exciting because, by the rules, it seems like their god should have either granted the spell from the get-go or not.

4) Replacement. This is like #3 but without discovery. Here I (i.e. the DM) just arbitrarily replace some spells on a PC's list with others. Doesn't motivate anyone to discover new shit, but it is reasonable.

5) Bonuses. Like #4, but based on spellcasting stats. Essentially, the DM gives bonus spells to particular characters based on their spellcasting list. So a Cleric of Ptah with a +3 wisdom bonus would add 3 additional spells of the DM's choice to their list. I like this option, but it might be hard to come up with that many spells per level. Simply replacing the bonus/level with the bonus spread across levels this would be pretty manageable, or maybe double the bonus across levels (so that +3 bonus translates to six added spells of whatever levels are appropriate). The problem with this option is finding the right number of additional spells for any given priest. That's 5 or 10 depending on wisdom scores, I suppose. Thankfully the Book of Lost Spells does provide quite a few, but there might not quite be enough great options for any given priesthood (or bardhood, druidic cult, sorcererous bloodline, or warlock patron).

I'm obviously still brainstorming a bit, but option 5 seems like its the best. Its DM-sensitive, but gives some customization to different types of cleric (or bard or druid or sorcerer or warlock) without stupidly expanding each class's spell list. I'm still looking for something better. I'm thinking combining #3 Discovery with #5 Bonuses, but ultimately anything that could be discovered also feels like it could be a bonus. If deities are active in the world, I suppose one-off spell options could be given as well: Isis knows her agents may need a Detect Curse spell and offers it on one particular occasion. 

Hidden in the depths of 2nd Edition...

After going through a bunch of second edition stuff (and even scouring forums and blogs and the like) I've discovered a few things I've never really realized 2nd edition had. Many of these come from the campaign settings, which were, in many ways, the laboratory of the game. Designers could craft crazy nonsense and at least if they were in one campaign setting they weren't "polluting" the entire game line. There's a tension between game designers wanting a stable core game with players wanting lots of crazy options and expansions, and the settings of 2nd Edition are where good ideas were tested out. I suspect we'll see this philosophy a bit in 5e now, such as Adventurer's League only allowing people options in the core books and the most current setting/player's guide. Heck, we did it once during our first Empire of Man game, where we let each player bring one book to the table. Anyway, three things of interest, though #2 and #3 are kinda related.

1) Racial proficiency groups. This might only be in the Domains of Dread book for Ravenloft, but it is such a superb idea. Just like your class gives you a list of proficiencies you get a discount on, your race could do the same thing. Maybe the groups aren't perfect and maybe races don't need more than 0-3 options to make this a thing, but its kinda nice that elven warriors might take astrology or elven wizards may take bowyer/fletcher.

2) Extra classes. Second edition was obviously full of a lot of nonsense, but they did craft a good number of additional classes, mostly restricted to the campaign settings. Monk and Assassin classes? Added in the Scarlett Brotherhood book (And the internet thinks they're decent too!). Berserker and Runecaster are added in Vikings (and again, the internet thinks they're on-par with the other classes, though runecaster is a bit weak and might could be augmented with the Giantcraft runecaster for Forgotten Realms). The later Forgotten Realms also gave use Harpers, Crusaders, Spelldancers, and Shadow Walkers hidden away in splat books. These last four were news to me just this past month, so I had to update the Wikipedia list of alternate D&D classes.

3) Unlimited slow/weaker spellcasting. Linked to extra classes, both the Al-Qadim Sha'ir and the Forgotten Realms Spelldancer (Wizards & Rogues of the Realms) are spellcasters with unlimited casting potential at the cost of speed. They're both odd options where you have a decent chance of getting just about any spell you want as long as you're willing to wait. Not too useful in combat, but you're a pretty grand utility caster. I'm not 100% sure I think this is fair and balanced or would work well in a low(er) magic setting that I'd prefer, but given that this has appeared a few times (and the interned hasn't shat all over the Sha'ir) it might be reasonable. Presumably in-play the characters also end up getting a number of wands or something for a few things to do in combat (though I might opt to multiclass rogue or something). It does sort of green-light some crazy ideas though like making all spells have a longer casting time (in rounds/turns) and just letting wizards/priests cast at will. Likewise, two versions of the runecaster show up (Vikings, Giantcraft) which has lower-powered but essentially unlimited spells as well. Someone felt this nonsense was balanced at least.