No matter how good or bad the argument/lie a player attempts, the most they can achieve through role play (using Pathfinder's core rules) is a modifier of some kind to their skill roll. Certainly, it can be entertaining when a character with high bluff skills steals something right in front of a guard, then says "You didn't see me steal anything," and rolls a 35 on their bluff check. Truthfully, I've had a lot of fun with those kinds of players in the past. But it's a somewhat less entertaining when the players approach a local regent, present maps and evidence of an incoming attack on the regent's town, but are ignored because they rolled a diplomacy of 8. Even if good role playing & evidence gathering earned those players an unprecedented +10 bonus to their roll, they would still fall short of the DC of 20 required to affect an attitude shift from unfriendly to indifferent.
And none of that even mentions how the current structure of the diplomacy check slows games down. If a player wants to try influencing the attitude of an NPC, the GM needs to pull out the core rulebook, and find the spot in the chapter on skills with the Starting Attitude / Improvement DC chart. I suppose the chart could be memorized, but it seems silly to ask a GM to memorize a chart of arbitrary numbers. Diplomacy also harms gameplay because it turns NPC interaction (which should be part of game play) into a binary proposition based on die rolls. Either you succeed, or you fail. Either way, that NPC can't be influenced again for another 24 hours.
And on top of all of that, the skill check is so utterly and completely broken that playing it as written in the core rulebook allows a player to rule the world by level 13.
Of course, any of these issues could be softened with house rules, or simply avoided by throwing the entire diplomacy/bluff check system away. Let the players role play with the NPCs, and let the GM decide how NPCs react based on how well the characters influenced them. This is fine, it's why we have rule zero. It's what I've been doing myself for a number of years now. But that doesn't make Pathfinder's broken rules on this subject okay. Just because the individual gamers can fix the problem doesn't mean the problem doesn't exist. That's called the rule zero fallacy, and lets get it out of the way right now. We shouldn't be okay with our favorite games having bad mechanics, we should want those games to improve on those mechanics. Not just for our sake, but for the sake of new players who may not understand that they have the option of throwing rules away.
NPC interactions deserve a better mechanic. Modern video games have managed to turn conversations into a mechanically relevant and interesting feature, so it should be no problem for tabletop RPGs. Such a system must be simpler to use than Pathfinder's current chart-reference method. Yet it must also provide a conversational experience with more depth than simply rolling to determine a binary result. It should also be impossible for the system to completely break the game.
Below, I've pieced together a simple mock-up of where I've started thinking. Bear in mind that this is not intended to be a complete and functional mechanic. It is, as the title of this post suggests, only the very beginnings of the idea. My hope is that we can build a dialogue from this starting point, and eventually produce a usable system. I've chosen to present the initial version of this ideas as one which builds off of Pathfinder's basic structure. I believe that, eventually, a good system for NPC interactions will need to be divorced from any existing system. However, at this early stage, I think there is value in connecting the system to a frame of reference.
Simply stated, my rule is thus:
When attempting to convince, deceive, or intimidate through conversation, the speaker uses their CHA score (not modifier) to oppose their target's WIS score. These scores are modified somewhat by abilities, and heavily by circumstance and role playing. If these scores are relatively close at the end of the speaker's attempt, participants roll 3d6 and add the results to their scores to determine the outcome of the attempt. Otherwise, the higher score wins. At the GM's discretion, characters who make a particularly convincing attempt should automatically succeed, without the need for any number comparison. Succeeding will either achieve the speaker's goal, or allow the speaker to continue attempting to achieve said goal without penalty. Each failure forces the speaker to suffer penalties during future interactions. Particularly egregious failures may result in insurmountable penalties.
I understand that some of this is confusing, and some of it is downright vague. The above is meant to be the distilled essence of the rule. It is the only part of the rule which you will actually need to remember once you understand how the system works. Those points I left vague were left so because I really have no idea what the specific numbers should be. I'm hoping that someone with a better grasp of the mathematics of game systems can offer some insight on that. For now, just to facilitate the discussion, lets say that every +5 to your diplomacy/bluff/intimidate check (not counting Cha bonus) grants a +1 when you attempt any of those tactics in a conversation. For the purposes of the dice rolling, "Relatively close" will be defined as "within 5 points."
Henar the gnomish rogue would like to convince a guard for the city of Yedge to allow her passage into the city, even though the gates have been closed for the night. Henar's Charisma is 10, and the Guard has a Wisdom of 16. Diplomacy is a class skill for Henar, and she also has 4 ranks in the skill, as well as the Skill Focus: Diplomacy feat. Normally this would give her a total of +10 to Diplomacy rolls. Using this system, that number is divided by five. So Henar now has an effective Charisma of 12 to compare to the Guard's Wisdom of 16. Henar's player speaks to the guard:
"Please, good sir! It is cold and wet out here. I am but a gnome, what harm could I do to your city and its powerful watch!?"
Henar's player doesn't exactly make a good argument, but the GM decides that this guard will react favorably to flattery. The GM grants Henar a +2 for the role playing, bringing her effective Charisma to 14. Since this is only 2 points away from the guard's Wisdom of 16, the GM calls for a roll. Both parties roll 3d6. Henar rolls an 12, bringing her total up to 26. The Guard, however, rolls an 11, bringing his total up to 27. The guard "wins" the encounter, and responds:
"Rules is rules, gnome! I won't let you in for nothing."
Henar has failed, but the failure is not extreme. The GM decides that she will suffer a -2 penalty during the rest of the conversation. Henar decides to try again, picking up on the "won't let you in for nothing." cue. Remember that at this point her effective Charisma is 10 + 2 for her bonuses - 2 for her previous failure. She is again at a disadvantage of 10 to 16.
"Surely an upstanding gentleman such as yourself is due more prestigious duties than guarding a wall! You ought to be spending your evenings in leisure with your fellows. Let me ensure you have the means to truly enjoy yourself next time you are at liberty..."
With this, Henar offers a bag of 25 gold coins - a very generous bribe! The GM decides that the guard is not the type to become offended when offered a bribe, and grants Henar a +4 for this offer. Henar's flattery has also been laid on more thickly this time, and the GM decides to grant a +4 for that as well. All told, this gives Henar an effective charisma of 18 compared to the guard's charisma of 16. This is close enough to call for a roll, but this time, Henar wins. The guard allows her through the gate, and takes her gold.
I want to reiterate that this isn't a finished product. It's merely the beginnings of an idea. I can already see possible avenues of abuse, and I worry that it may be overly clunky, but it's a place to start. Maybe the end result of this process won't look anything like this. Regardless, I'm eager to hear any feedback. I would encourage people to use the comments on this post (rather than IMing, emailing, or twittering me) because any criticism made in private to me isn't one which is part of the larger discussion.