Vapourware The problem with Speech (and your ideas for solutions)

[TheShadyLurker]

Here is an idea for a relatively simple solution.
You have an involved branched dialogue system where you need to select the correct answers to convince people, but you can only access those branches to begin with if your speech skill is high enough. The dialogue option that starts that discussion off to begin with has a speech skill gate. This means that you require the high skill to even get access to that, depending on how difficult it would be, but you then need to pick the right options, based on what you think would make sense based on the character and situation.

That's what we already have. I'd like to have a dialogue system where a particularly charismatic character can convince people of obviously wrong things.

Think of cult leaders who convince people to give working people all their money; of men who marry rich women and live lives of luxury; of politicians who lie, corrupt and break promises and still get re-elected; of women who charm their way out of jail, etc. That's what I'd like to see.

The reverse would be a loathsome character who wouldn't be believed even when telling the truth, who would get arrested by being in the wrong place at the wrong time, etc.

The problem with what we have is that there's often involved dialogue options and only a few of them have speech checks involved, often being the obvious right solution.
I'm switching the order around. The speech skill check is what gets you to even start trying to convince someone of something, and from that point forward, there's the involved dialogue options where you try to achieve just that. They are all options that already take into account your high speech skill.
If you wanted to divide it into different stuff like intimidation, sweet talking, bartering, etc, the process would still be the same. The gate is at the start, then you have to pick the right options to intimidate, barter or sweet talk someone.

 

[KeighnMcDeath]

Joke? Like… an RPG, a Visual Novel, a Puzzler and an Adventure Quest walk into a bar……

 

[Nifft Batuff]

Disco already solved this problem.

 

[gabe1010]

It depends quite a bit on what the appeal of the specific game is. BG3 is a cinematic, licensed DnD experience, so including dice rolls in conversation checks, and having a huge quantity of content in terms of dialogue, voice-overs and sequencer cinematics is basically required, regardless of whether it is the ideal speech implementation. More open-ended sandboxy type RPGs don't need, and pretty much could not produce the mass of content for, that kind of speech implementation (BG3 had like 400+ employees working on it full-time at peak, took many years to make, and included many years in early access, and even for that is pretty constrained/on-rails compared to a sandbox fully open world, and is also still quite buggy).

I lean more towards favoring the semi-sandboxy open world game you get from Morrowind or Daggerfall, where the story is what you make it, instead of the BG3 cinematic approach (and Morrowind was made by ~30 people 20 years ago). As I read over this thread, I kept thinking that the Morrowind way was the least bad of all those mentioned so far, especially if we're talking about a standard combat RPG, and not some kind of slice of life roleplay.

Morrowind already had individual disposition, a personality ability score, a speech skill, racial modifiers, faction modifiers, and reputation.

AFAIK Morrowind didn't utilize its well-realized disposition system and its various modifiers to its best potential. Hence, I wish that another game could've learned from MW and taken its already great base and expanded it.

It's a logical and intuitive system, Someone's willingness to give up information almost always depends on how much they like/dislike you and that's what the disposition system is meant to represent. if you have a high personality stat, People are naturally going to like you more due to the innate traits that make you more likable as a person - as much as I can excuse Fallout and similar games' use of the speech mechanic, I don't like how sometimes a character who might utterly despise you will still give up valuable info based purely on a single line of dialogue...

The above makes the most sense to me. Say individual NPCs have an out of 100 disposition towards you that defaults to 40 for the sake of the argument. You then have numeric alterations like:

-Base Charisma attribute bonus/detriment to all NPC dispositions (or whatever system the game uses, maybe charisma/speech is a trait that can be none/major/minor)
-Same Race bonus / Rival Race detriment
-Same Faction bonus / Rival Faction detriment
-Same Religion bonus / Rival Religion detriment
-bespoke bonuses for specific quests or favors done in game
-dialogue options to increase bonuses based off information you learn in-game (or decrease disposition if you are rude/aggressive/taunting)
-bribery to smooth things over (but it's a risk as it decreases the disposition of NPCs who can't be bribed, or who wanted more money than you offered)
-maybe charm spells (just not broken like in Morrowind)
-perks that balance bonuses/detriments to disposition based on class, profession, wealth, whatever of the NPC in question (like bonus to women but detriment to men or vice versa, or bonus to disposition from the poor but detriment to the wealthy, etc...).

And perhaps a few more things, and all this disposition boost really does is open up more dialogue options as it goes up. Maybe below 10 they won't talk to you. From 10-24 they give curt/rude answers. From 25-49 they provide basic personal and world information like directions and shop locations. From 50-74 they share rumors and secrets, and will participate in most quest specific dialogue. From 75-89 maybe they offer specific bits of help, or are required to be in that range for more advanced quest dialogue. And >= 90 is for the rare times you need someone to straight auto-win the quest for you, or agree to appoint you head of a guild, or follow you as a companion, or share info about hidden treasures, secret doors and such, or some other kind of very high disposition thing. It would also affect merchant prices. I think it would probably be a nice simplification if options always fell in these discrete ranges, and the player knows with decent certainty what kind of disposition they need from who for what (ie 0-10 is just "hates you" and 90-100 is "loves you" and everything in-between labelled, and you just know they need to "like" you for whatever).

If it was just a better balanced, more sensible, slightly evolved form of Morrowind, and used well by quest writers, it would be more than good enough for most combat/exploration/questing based RPGs. I think it fits well with the wiki-type dialogue as well, so you don't have to click through the same dialogue tree repeatedly as often. You come to an NPC to get info or advance a quest, and it's nice to just immediately select what you want, or see what's new, instead of going through tedious dialogue sequences over and over, especially many hours into the game.

As for going much further than that for speech options in an RPG, you could go all the way down the path of different types of speech, and many non-combat options with minigames and other complex mechanics for non-combat builds, but that then strikes me as bordering on a different genre of game. If you can be a dedicated diplomat, or run a store, or be a travelling merchant, or just a pillar of the community around town, then you're talking about the mechanics of The Sims, The Guild, Potion Craft, Recettear, a bunch of social/romance games, etc...basically a different genre to an RPG (although with "RPG elements" oftentimes). A given game needs to pick a focus, even an open world game, and including a full suite of both combat and non-combat play/lifestyles in an RPG is gonna be like making 2+ games at once. Who is the audience for that game? And who would make two games and sell it for the price of one?

 

[Lemming42]

As for going much further than that for speech options in an RPG, you could go all the way down the path of different types of speech, and many non-combat options with minigames and other complex mechanics for non-combat builds, but that then strikes me as bordering on a different genre of game. If you can be a dedicated diplomat, or run a store, or be a travelling merchant, or just a pillar of the community around town, then you're talking about the mechanics of The Sims, The Guild, Potion Craft, Recettear, a bunch of social/romance games, etc...basically a different genre to an RPG (although with "RPG elements" oftentimes). A given game needs to pick a focus, even an open world game, and including a full suite of both combat and non-combat play/lifestyles in an RPG is gonna be like making 2+ games at once. Who is the audience for that game? And who would make two games and sell it for the price of one?

The problem is that in recent years the cRPG genre seems to be simultaneously obsessed with ever-increasing amounts of dialogue, C&C, NPC interaction, and story, but devs are still unable to come up with a non half-assed way to implement those things. The general template devs use today is the one Fallout set out back in 1997, which is just absurd - over a quarter of a century of failure to innovate.

Obviously the issue is that TTRPG combat is very easy to replicate in a videogame because it's simple, predictable and rules-based, while everything else in a TTRPG is complex, imaginative and spontaneous, and thus extraordinarily difficult to replicate. A lot of older cRPGs were correct to eschew anything other than combat and dungeon exploration entirely (or offer it in a very abstracted and brief form, like the NPC encounters in Eye of the Beholder). This worked well; the games consisted almost entirely of actual gameplay and anything else was just a bonus that endearingly made the game feel a fraction more like a tabletop campaign. Modern cRPGs that have taken the same combat-first approach have been similarly successful, making sure that there's no half-implemented or shoddy mechanics dominating half the game (like, say, New Vegas' awful [Skill Check] quest solutions).

In other words: if a whole game consists of combat, and that's done well, then you've got a good game. If a game however, as many do nowadays, consists of lengthy dialogue, C&C, NPC interactions, non-combat quest solutions, and long periods without combat, and the game doesn't offer any real mechanics for anything other than combat, then you've got a visual novel/walking simulator where the player only spends a small percentage of their time actually playing a game. This is bad IMO; ideally all aspects of a game should be of equal depth and quality, as far as possible.

To take the Morrowind example, the problem is still that the player gains 100 Disposition with an NPC by just clicking "Admire" over and over. It'd be unacceptable for combat to consist entirely of the player opening a menu and clicking "Sword" repeatedly until a meter filled up, with no other mechanics involved. I think the same should be true of dialogue, especially in games where it's given equal or greater focus than combat, or can be used to the same effect as combat (eg solving quests). Otherwise, you've offered two paths through the game - one that involves actually playing the game (combat) and one that essentially involves clicking "win" (speech), and the two routes are treated as equal.

I don't think it's necessary to recreate The Sims though, only to find some kind of way to apply a set of satisfying mechanics to NPC interaction. Fallout took the first step and did so brilliantly, but nobody seems to have really taken the second step in the intervening 26 years, which is insane especially when you consider that developers seem increasingly interested in those very aspects. Your suggestions for evolving on the MW system are a good start; I think the solution lies in a degree of abstraction and letting speech be determined by mechanics rather than by hand-written dialogue options.

 

[Harthwain]

The general template devs use today is the one Fallout set out back in 1997, which is just absurd - over a quarter of a century of failure to innovate.

I would point to NEO Scavenger as a game that supports exploration (and crafting) through gameplay in very RPG-esque manner.

Streets of Rogue is also interesting, because it allows for some degree of creativity when it comes to finding solution to a quest (and interacting with a larger system on each floor, albeit that system is fairly simple, truth to be told. Still, that is a very good starting point for creating something unique, in my opinion).

Unforetold: Witchstone (formerly Project Witchsone) is a game I am eyeing from this perspective as well.

These are all examples of games that do not rely on heavily hand-crafted content when it comes to their core gameplay loop, while also being more than just "RPGs = combat".

 

[deama]

Actually, with the rise of AI, you could implement a straight up skill system.
Assuming the game has a AI npcs setup and you talk to them by typing it out.
What if the idea is that, e.g. if you're speaking to a lord in the castle and you type in "gimmy quest", the AI will write out an appropriate dialogue that your in-game character might have said, like if you're speech skill is crap it would say "gibby me quest lordy lord", and you'd basically be killed for insulting the lord. But if you're speech skill is high enough, the AI would turn it into a masterfully crafted sentence.

 

[gabe1010]

To take the Morrowind example, the problem is still that the player gains 100 Disposition with an NPC by just clicking "Admire" over and over. It'd be unacceptable for combat to consist entirely of the player opening a menu and clicking "Sword" repeatedly until a meter filled up, with no other mechanics involved. I think the same should be true of dialogue, especially in games where it's given equal or greater focus than combat, or can be used to the same effect as combat (eg solving quests). Otherwise, you've offered two paths through the game - one that involves actually playing the game (combat) and one that essentially involves clicking "win" (speech), and the two routes are treated as equal.

I don't think it's necessary to recreate The Sims though, only to find some kind of way to apply a set of satisfying mechanics to NPC interaction. Fallout took the first step and did so brilliantly, but nobody seems to have really taken the second step in the intervening 26 years, which is insane especially when you consider that developers seem increasingly interested in those very aspects. Your suggestions for evolving on the MW system are a good start; I think the solution lies in a degree of abstraction and letting speech be determined by mechanics rather than by hand-written dialogue options.

I agree that Morrowind's implementation often falls short, but it's more the base disposition system I like as a starting point. Admire should be an option, among those I listed, that would in a better implementation a) only be available (or at all effective at least) when used by certain builds, b) would have a pretty low cap of repeated effectiveness (spamming admire makes no sense in real life, and like bribing it should have a negative effect if taken too far), and c) only be effective against certain types of characters vulnerable to charm (and have the opposite effect on characters that dislike flattery, also like bribing). I think that would be adequate for most disposition type stuff in an otherwise quest/combat/exploration game like Morrowind, but with at least one eye towards some semblance of balance (unlike Morrowind). It would provide a few extra options for roleplay, and support some worldbuilding (like how in Morrowind Dunmer like bribes as a systematic cultural thing), and that's enough for that sort of RPG. The lockpicking minigames of TES 4 and 5 are tedious and boring, and the designers knew it so they included easily accessible skeleton keys, and they even harm roleplay as contrasted with Morrowind where the character does the lockpicking, not the player. I feel that speech minigames would just be that problem on an even larger scale.

If you want speech to be another way of playing the game altogether, and not just an augmentation of largely combat (or at least stealth) builds that takes place in dialogue windows, then yeah you'd need minigames et al, but now you're losing focus. It is perhaps insane that so little progress has been made on this front relative to combat in the past quarter century, but maybe it's just not the sort of thing that's as readily gamified, and those genres in which it is well gamified (The Sims, Romance games, etc...) tend to be either niche or have a big but non-traditional gaming audience like The Sims (ie the average The Sims player I suspect does not play Call of Duty or Morrowind). IE, it brings us back to the question of who would such a game be for? Or, to put it another way, there has been progress made on this front, but just in splitting into multiple other genres, ie you don't find it that much in games calling themselves RPGs.

The obsession with ever more dialogue, C&C, NPC interaction and story is simply a result of video games, gamer culture, and RPGs spilling out more into the mainstream, and what the mainstream wants is movies that you click through with a little bit of CYOA branching. Basically, Baldur's Gate 3 is a long animated movie shot through all the way through multiple branching plots and then stuck together with a CYOA option set, and between cutscenes you can play turn based battles (that I am finding every more tedious to get through, and I just got to Act 3). It won every award just about, but imo the "roleplay" aspect of it can't hold a candle to a game made 20 years ago.

Many people dream of AI being the solution to this (and many other problems), and there are some promising leads in that regard, but I'll remain wary until I see it implemented really well. Ultimately though, the AI would just be filling in flavor text and matching player inputs to existing possible options. It wouldn't really be an open ended conversation that could also map to a limited number of gameplay options.

 

[Daemongar]

I always thought of speech as an OP mechanic indicating too much with one skill.

If one had a 100 speech (on a 1-100 scale) and with your 100 could converse with kings, how could you possibly converse with some low tier beggar or parasite of society? Find anyone with a high speech. What, some talk show host can coax the safe combination out of some old bag at a bakery? I'd see Speech as a scale at what level you can communicate in society, not all encompassing. The higher you go, the less you are able to communicate or contact those who are too far off on the scale. Genius orators can't calm mentally handicapped people - compassionate old ladies who finished 8th grade can reach them better than any scientist or scholar.

So, I'd see speech as some manner of scale where one can be most effective with someone of the same speech score. That is, you have a speech of 50, you may be most effective with folks with speech between 25 and 75. A player may raise a score to 100 and communicate with judges and kings, but lose that "common touch" and be unable to sway regular slobs. Raising Speech too high would lock someone out of side quests, while having it too low would get the shittiest rewards for completing a quest to a king.

 

[Funposter]

re: Morrowind implementation, Tamriel Rebuilt's sway, barter and debate checks take good advantage of the framework that Morrowind provides with proper depth and an actual use for Speechcraft in a game where normally you just bribe everyone until they're at 100 disposition. It comes up in a lot of TR content and makes it WAY more useful to create a proper Speechcraft focused character. In an ideal world, you would still have Etiquette and Streetwise, and then these would also take into account social classes that are flagged for each character.

 

[Lemming42]

Coincidentally, Tim uploaded this today.

Summary: he agrees that clicking [Speech] to win is dumb, but he doesn't like minigames either. He thinks the solution is needing something tangible to back up a speech check (like the holotape evidence of mutant sterility in Fo1), and he thinks the game should telegraph when you'll need this so you know what you're looking for. He says getting evidence to back up a skill check could involve:
- finding some physical records in the game world (like the holotape), which itself should form a quest in which the player can bribe/sneak/etc to acquire the evidence
- talk to other people and piece together your upcoming speech check from things that they tell you. He thinks this can be supported with a Persuade/Intimidate/Lie system like in TOW, which creates a "pure speech solution"
- create the evidence yourself (by "running experiments with your science skill" and other such things), and present it to the bad guy.
He thinks that these solutions should take as much time and be as involved as the equivalent combat or stealth paths, and should be just as fun.

He's also put up a video about pacifist playthroughs which I'm about to watch, might have some other relevant stuff.

(btw Tim if you're reading this, I love you)

 

[HarveyBirdman]

I have pretty much a whole game in my head. It incorporates the dopest speech skills. It would be the best game ever.

But I can't code so it will never be made. Still not giving you my intellectual property for free.

 

[Zombra]

I just had an idea that eliminates the "auto-win" nature of some speech models, uses dice rolls, yet gives the player a certain amount of tactical agency in taking risks. It can work with either very coarse (1-5) or fine (1-100) skill spectrums.

In this model, dialogue trees are simple lists of options, just like we're used to. The complicated part is all "behind the screen".

In every exchange, every dialogue option is classified by the designer as leading towards either good, bad, or neutral/indeterminate outcomes. (Season to taste.)

For every option, roll against the appropriate social skill, modified by how hard it should be to figure out what the outcome is. ("Fuck you" leading to a bad social outcome will not be a difficult skill roll; however it may be hard to tell whether complimenting a king's shoes will please him or annoy him.)

No dice rolls are shown; this is all done "under the table".

For each option, on a successful skill roll, negative outcomes will be highlighted in red text; positive in green; neutral in blue.
For each unsuccessful skill roll, options will be in plain white text ("You don't know"), or on a failure, the wrong color.

So you might get a list like this:
1. Fuck you. (Critically failed roll; this will actually lead to a bad outcome.)
2. Nice shoes. (Successful roll; the person will cooperate with the PC.)
3. If you won't give me the stone, I'll just take it. (Successful roll; you might think this would be a negative outcome but it leads to further dialogue.)
4. <Attack.> (No color coding as this just goes straight to combat.)

In this way, just like with combat skills, a socially skilled character can be confident in their choices most of the time if they decide to follow the color coding, but they will never be 100% sure (even a guy with max sword skill misses sometimes; same thing here). So no options will be obviously "auto-win". Players will still be encouraged to actually read the text to see if they think it makes sense to say that, while their social skills will give them strong (but fallible) clues as to which is "right".

 

[Poseidon00]

Arcanum solved diplomacy, in terms of mechanics. You don't click the (Speech: 100) button. You have to navigate dialogue choices, plural being the key word, in a sensible manner in order to achieve the outcome. It never gives you any "meta" knowledge to guide you on your way like most games do. Sometimes it requires knowing the character of what kind of person you are talking to.

However, speech needs to stop being something that you use on the Quest NPC to fast track things along. If I have high enough speech, I should be able to make NPCs go turncoat. Let me slowly, over the course of the game, have chances to influence and interact with the Big Bad's number 2, and maybe they will eventually betray them. Why can't I form a gang of my own or engage in some crazy scams? Be imaginative with it.

 

[Maxie]

I just had an idea that eliminates the "auto-win" nature of some speech models, uses dice rolls, yet gives the player a certain amount of tactical agency in taking risks. It can work with either very coarse (1-5) or fine (1-100) skill spectrums.

In this model, dialogue trees are simple lists of options, just like we're used to. The complicated part is all "behind the screen".

In every exchange, every dialogue option is classified by the designer as leading towards either good, bad, or neutral/indeterminate outcomes. (Season to taste.)

For every option, roll against the appropriate social skill, modified by how hard it should be to figure out what the outcome is. ("Fuck you" leading to a bad social outcome will not be a difficult skill roll; however it may be hard to tell whether complimenting a king's shoes will please him or annoy him.)

No dice rolls are shown; this is all done "under the table".

For each option, on a successful skill roll, negative outcomes will be highlighted in red text; positive in green; neutral in blue.
For each unsuccessful skill roll, options will be in plain white text ("You don't know"), or on a failure, the wrong color.

So you might get a list like this:
1. Fuck you. (Critically failed roll; this will actually lead to a bad outcome.)
2. Nice shoes. (Successful roll; the person will cooperate with the PC.)
3. If you won't give me the stone, I'll just take it. (Successful roll; you might think this would be a negative outcome but it leads to further dialogue.)
4. <Attack.> (No color coding as this just goes straight to combat.)

In this way, just like with combat skills, a socially skilled character can be confident in their choices most of the time if they decide to follow the color coding, but they will never be 100% sure (even a guy with max sword skill misses sometimes; same thing here). So no options will be obviously "auto-win". Players will still be encouraged to actually read the text to see if they think it makes sense to say that, while their social skills will give them strong (but fallible) clues as to which is "right".

If you colour-code the options then it makes no sense game-wise to gimp yourself if the result you can achieve is success or failure. Players will always choose the best yield, and you've wasted time trying to come up with sarcastic dialogue options. It's as if you took Disco Elysium's dice roll governed dialogues, but removed the idea of dozens of skills you can roll for, leaving it in place for cargo cult reasons.

 

[Zombra]

Arcanum solved diplomacy, in terms of mechanics. You have to navigate dialogue choices, plural being the key word, in a sensible manner in order to achieve the outcome. It never gives you any "meta" knowledge to guide you on your way like most games do. Sometimes it requires knowing the character of what kind of person you are talking to.

But ... that "solves" the problem of poor speech mechanics by simply removing all mechanics. Your PC can't be good or bad at speech; it's just the player. It's the equivalent of removing combat stats and letting combat outcomes rest strictly on player twitch skills. It's no longer an RPG at that point.

 

[Poseidon00]

Arcanum solved diplomacy, in terms of mechanics. You have to navigate dialogue choices, plural being the key word, in a sensible manner in order to achieve the outcome. It never gives you any "meta" knowledge to guide you on your way like most games do. Sometimes it requires knowing the character of what kind of person you are talking to.

But ... that "solves" the problem of poor speech mechanics by simply removing all mechanics. Your PC can't be good or bad at speech; it's just the player. It's the equivalent of removing combat stats and letting combat outcomes rest strictly on player twitch skills. It's no longer an RPG at that point.

(Speech: 100) You're wrong

 

[Skinwalker]

Everyone agrees: Speech and Diplomacy skills suck. They haven't evolved in about 25 years, and all too often they're an auto-win button. Click the [Speech] option, win the quest. In addition to this, they often actively restrict players from accessing content - if you persuade an NPC, your "reward" is usually to skip ahead in a quest or just have it end right there, meaning you've just missed out on part of the fucking game.


solutions:

There isn't one and Speech/Persuasion/Diplomacy shouldn't be a skill that you level-up at all. Make interesting and varied dialogue trees that let you figure out how to best to talk to people by paying attention to the dialogue, character, context, etc. This doesn't translate into numbers and mechanics, but it does require some writing talent.

 

[deama]

I believe colon ship did a pretty good job. You have to navigate the dialogue and your speech skill determines how much +mood options give, some options naturally give more, some less, and they are enhanced by your skill. After a few dialgue talking, if the mood isn't over a certain check, you fail, or pass if it is.

 

[KeighnMcDeath]

I always thought of speech as an OP mechanic indicating too much with one skill.

If one had a 100 speech (on a 1-100 scale) and with your 100 could converse with kings, how could you possibly converse with some low tier beggar or parasite of society? Find anyone with a high speech. What, some talk show host can coax the safe combination out of some old bag at a bakery? I'd see Speech as a scale at what level you can communicate in society, not all encompassing. The higher you go, the less you are able to communicate or contact those who are too far off on the scale. Genius orators can't calm mentally handicapped people - compassionate old ladies who finished 8th grade can reach them better than any scientist or scholar.

So, I'd see speech as some manner of scale where one can be most effective with someone of the same speech score. That is, you have a speech of 50, you may be most effective with folks with speech between 25 and 75. A player may raise a score to 100 and communicate with judges and kings, but lose that "common touch" and be unable to sway regular slobs. Raising Speech too high would lock someone out of side quests, while having it too low would get the shittiest rewards for completing a quest to a king.

How does batman communicate with scum/criminals or ultra elite to near gods and understand them, and utilize them for his own gain?

How many ways was different dialogue used in games (not just rpgs but ALL).

What are these "mini-games" mentioned?

 

[Daemongar]

I always thought of speech as an OP mechanic indicating too much with one skill.

If one had a 100 speech (on a 1-100 scale) and with your 100 could converse with kings, how could you possibly converse with some low tier beggar or parasite of society? Find anyone with a high speech. What, some talk show host can coax the safe combination out of some old bag at a bakery? I'd see Speech as a scale at what level you can communicate in society, not all encompassing. The higher you go, the less you are able to communicate or contact those who are too far off on the scale. Genius orators can't calm mentally handicapped people - compassionate old ladies who finished 8th grade can reach them better than any scientist or scholar.

So, I'd see speech as some manner of scale where one can be most effective with someone of the same speech score. That is, you have a speech of 50, you may be most effective with folks with speech between 25 and 75. A player may raise a score to 100 and communicate with judges and kings, but lose that "common touch" and be unable to sway regular slobs. Raising Speech too high would lock someone out of side quests, while having it too low would get the shittiest rewards for completing a quest to a king.

How does batman communicate with scum/criminals or ultra elite to near gods and understand them, and utilize them for his own gain?

That's a good question. I suppose Batman would take a Feat to gain additional speech strata, if you will. So, as Clark Kent, he speaks with billionaires and the like. As Batman, he speaks with the dregs of society with easy, while beating them up. So, part of his alter ego framework is being at a 100 at home and at 30 while on the job.

What are these "mini-games" mentioned?

I suppose most people are Assuming he's talking about the roundly criticized Oblivion speech minigame.

 

[KeighnMcDeath]

Hmmm..... I ask because it does make me wonder.

Some games had talk + direction or a hot key like b to buy or interact. mostly you get a little descriptive chatter and nothing more. It doesn't do anything and some never change that bit of line (early ultimas). Looking tells you what you see. Buying/selling is hardly chatting but sometimes they add text lines but most are like the "Nurse Joy effect". ALL THE SAME EVERY TOWN.

Sometimes like in japanese rpgs the npcs will say something different as the game story marker is progressed. Most NPCs are still relegated to fetch quests, delivery, or give you something or activate a quest. The only personality is what is written.

You later had "keyword" phrases like ultima. These are more like adventurer "gate keeping." If you know what to say you can bypass a lot but generally they reveal a person, place, clue you follow to next npc. They might have different dialogue upon talking again. This reminds me of text adventures or just adventure games period.

Some games had that text tree option or A B C choose your own fate. Meh. Also just a dancing through choices.

Many of these older games or those that try to replicate have no real affect to the npc or you as far as relationship. They either have a use or are just scenery you can interact with.

Was Fallout the first to use its version of A B C based on stats & skill & circumstances?

There are a lot of games and probably a proper way to describe the methods of dialogue each uses.

I'd be interested to see a detail of this from oldest to present and where devs are stuck.

I notice jap games love to use affection levels between characters be they able to be romanced or just increase/decrease whether they like you enough to give some bonus, reveal a secret, etc. Watching that Harvest Moon videos and the like made me ponder that. Even Wizardry Daphne has trust levels with your talky party members.

How complicated should dialogue become?

 

[Zombra]

How complicated should dialogue become?

At least as complicated as hitting skeletons with a sword

 

[Alex]

I just had an idea that eliminates the "auto-win" nature of some speech models, uses dice rolls, yet gives the player a certain amount of tactical agency in taking risks. It can work with either very coarse (1-5) or fine (1-100) skill spectrums.

In this model, dialogue trees are simple lists of options, just like we're used to. The complicated part is all "behind the screen".

In every exchange, every dialogue option is classified by the designer as leading towards either good, bad, or neutral/indeterminate outcomes. (Season to taste.)

For every option, roll against the appropriate social skill, modified by how hard it should be to figure out what the outcome is. ("Fuck you" leading to a bad social outcome will not be a difficult skill roll; however it may be hard to tell whether complimenting a king's shoes will please him or annoy him.)

No dice rolls are shown; this is all done "under the table".

For each option, on a successful skill roll, negative outcomes will be highlighted in red text; positive in green; neutral in blue.
For each unsuccessful skill roll, options will be in plain white text ("You don't know"), or on a failure, the wrong color.

So you might get a list like this:
1. Fuck you. (Critically failed roll; this will actually lead to a bad outcome.)
2. Nice shoes. (Successful roll; the person will cooperate with the PC.)
3. If you won't give me the stone, I'll just take it. (Successful roll; you might think this would be a negative outcome but it leads to further dialogue.)
4. <Attack.> (No color coding as this just goes straight to combat.)

In this way, just like with combat skills, a socially skilled character can be confident in their choices most of the time if they decide to follow the color coding, but they will never be 100% sure (even a guy with max sword skill misses sometimes; same thing here). So no options will be obviously "auto-win". Players will still be encouraged to actually read the text to see if they think it makes sense to say that, while their social skills will give them strong (but fallible) clues as to which is "right".

Not saying this is a bad idea, but it doesn't solve the issue at all.

To be clear, the problem with an "auto-win" button isn't that it is too simple or direct. Having the auto-win button be hidden under a slightly more complicated clothing isn't going to fix the most important issue here: there is no gameplay. Or, to be clear, there is little gameplay during dialogue that wasn't there already. Your example, like any game with dialogue skills that matter, there is of course some gameplay related to dialogue, but it happens during character building, not during dialogue itself. During dialogue, the player will try to guess, from what he knows about the NPCs, the options themselves and, in your case, what colour they are from his roll. The nature of dialogue didn't change (though since this is how your skill works, it will probably be designed to be more involved than just selecting one option).

If you wanted dialogue to be as involved as combat, you would need to design all sorts of gameplay elements around it. When I am playing Wizardry, for instance; during a battle I can find out all sorts of things from the systems in place. For instance, consider the positioning system, the use of status effects, the mechanics of "hide", etc. All of these subsystems allow you to fail, learn and come up with strategies as to how you play the game. Dialogue, on the other hand, might be interesting, but it is much simpler by necessity. For one, you couldn't really come up with different results for all kinds of subsystem you might add. Oh, if you talk with the duke using this voice tone, this level of respectfulness, using this bad cop/good cop dynamics, then he will spill the beans. But if you are more respectful than this, he will lie because he thinks you are a rube. Etc. Rather, the nature of dialogue is that rather than dealing with each situation using specific strategies you develop during the game like in combat, you have to approach each dialogue as unique for that specific NPC, more akin to the gameplay of adventure games than that of turn-based combat.

You could try adding some kind of abstract mini-game to combat instead, but I hold this is antithetical to CRPGs. In a CRPG you want what is happening during any moment of action, in system, to be easily mapped to what it is supposed to mean "in game". Getting hit during combat can be easily visualised as being wounded. Being petrified by a spell usually means your character is now a statue, not that he is sitting around in the reserve bank waiting for someone to cast stone to flesh on him. Abstract systems on the other hand are difficult to translate to an imaginary state, if not impossible.

As such, I hold that the way to making dialogue more interesting from the gameplay perspective is to make it more like an adventure game. If you want to convince someone of something, ask around and try understanding that person. What can you offer to him that might make him like your or want to help you? What can you use to threaten him that will make him comply without trying to kill you afterwards? How can you get him in a more pliable state for the negotiations? I hold that the key to making dialogue more involved gameplay-wise is: first, bring back keywords. This makes it so the player won't just click every option when trying to learn about stuff, but has to pay attention and figure out who might know about what. You can still have dialogue trees for complex interactions, but have keywords be necessary to learn more about the world around you and maybe open some of these tree interactions. Second, make accomplishing something through dialogue require preparation in the manner I mentioned. If you want to use your diplomacy skill to make the enemy general cancel his attack, you have to approach him with a certain level of reverence appropriate to his rank and you have to offer him what he truly wants from this all.

 

[Habichtswalder]

The main problem with speech is indeed the "auto-win" aspect of it. Therefore it is bad to only show certain dialogue options if they are guaranteed to be successful.
For example, the player should always have the option to intimidate someone even if the respective skill is only 1. It would be a stupid choice, sure, but it adds a certain level of uncertainty. So speech must be designed in a way that includes failure.

But how do we decide whether a dialogue option is successful or not? Here I like the idea that every NPC in the game has an invisible resistance value for the respective dialog ability. Similar to how it is in Kingdom Come Deliverance, for example. So if the player character has a intimidation value of 5, but the NPC has a resistance of seven, then the dialog option fails.
However, the NPCs must be designed in such a way that you can work out which dialog option could be successful with which NPC by thinking about it. So if an NPC is very self-confident, you can deduce that its resistance to intimidation is probably high. And if an NPC speaks very well but appears cowardly, you can deduce that intimidation would be a better option than an intelligence check.

To make the system even more complex, I think it would be good if the player character also had a skill such as Knowledge of Man, which allows you to see the range of NPCs' resistance depending on your skill level. Not an exact indication of resistance, but a rough estimate with a certain degree of uncertainty.

In addition, the dialogue should react dynamically if certain language skills have already been used once. If the player character has already threatened the same NPC once, the NPC's resistance increases on the second attempt at a threat.

This is not a particularly innovative system and there are certainly more complex systems. Bad for most games it is sufficient and adds a little bit of spice and thought.