Quote:

Original post by Wysardry Okay, I'll try answering the updated version:- Quote: Original post by Alpha_ProgDes And unfortunately, now I have to backtrack. If the game doesn't end then we run into the problem of procedural content and events. The game is creating the game. That's massive. If the PC was free to do his/her own thing, yet could still influence events at any time (s)he chose, then the underlying system would already be capable of dynamically changing and adding events etc. to some degree. My previous reply still applies to the remainder of your post. Quote: Original post by Trapper Zoid I was toying with making my system episodic. Since I'm interested in modelling the structure of stories, I was planning on having events mimicing the form of adventure stories, sagas or possibly comics; i.e. problem set-up, quests for testing the protagonist, problem resolution. Then there's no reason to not start another story with the same heroes and villains. That sounds a lot like the way individual PnP adventures are played, and they can be experienced in isolation or as part of a larger campaign. A computer should be able to allow a player to experience several of these simultaneously.

Episodic gaming on open-ended gameplay is something that cannot happen.

I'll explain.

If the end of one episode can be ANY situation, how do you choose the initial situation of the second episode? Let's say you DO have a limited three-choice ending per episode. On the second episode, you must have nine stories going parallelly. On the third episode, that's twenty-seven, and by the fourth, you've reached the 81 figure. ON the fifth episode of your semi-open-ended you've had to produce a hundred times more content than if you had completely foregone the semi-open-endedness of the episodic gaming.

Unless you deliberately decide to let the player choose his own experience, episode per episode, and then impose on him one of the endings as the one you chose for the following episode.

I think that the branching storyline of pre-scripted events is definitely the way to go. But it requires, as a background setting, some sort of urgency, for things have to happen on time. Any delay you can bring about will be a node and a branching along the storyline.

But, instead of enforcing the choices of the branching upon the player, let the branching happen as a background occurence, so that the player doesn't feel like he is being led somewhere, but instead that he is making his own choices.

The example of the warring factions and the bridge is an excellent one.

But the choices must be made more subtly. Let's have a meeting in a public place, like the market place, or an inn, ingame. The playing character comes through, and hears, or reads the comments and opinions of the different NPCs. Some might say the war is a good thing and one side should be helped as much as possible, some might say that war is never a good thing, a Chess player might say that the choice of the bridge as a battlefield is a stupid one and someone should force the armies to go fight somewhere else, maybe the other valley.

From there, the PC can CHOOSE (and that is important) to put his nose where noopne asked him to, and make a choice about his course of action (be it blowing a hole in a general's tent, in the brisge, or in a side's army, or maybe leave the armies battling in the valley, over the bridge while something more urgent and personal than the safety of a village requires him to be elsewhere.

In this respect, you can either

_ Explicitly display all mutually excluding possibilities of quest accomplishment. (But this may be too explicit and make the player feel led)
_ Hide them and leave the player in the fog (but I'm afraid of players feeling lost and disgruntled)


Always the same old topic of the choose-your-own-adventure Vs. dynamic storytelling generation...

Quote:

Original post by Fournicolas Episodic gaming on open-ended gameplay is something that cannot happen. I'll explain. If the end of one episode can be ANY situation, how do you choose the initial situation of the second episode? Let's say you DO have a limited three-choice ending per episode. On the second episode, you must have nine stories going parallelly. On the third episode, that's twenty-seven, and by the fourth, you've reached the 81 figure. ON the fifth episode of your semi-open-ended you've had to produce a hundred times more content than if you had completely foregone the semi-open-endedness of the episodic gaming. Unless you deliberately decide to let the player choose his own experience, episode per episode, and then impose on him one of the endings as the one you chose for the following episode.

That is true if the episodes are all hand-crafted pre-scripted storylines similar to that in most RPGs, but that approach can only be applied superficially to open-ended gameplay anyway; you'll always get that exponential explosion in content if you craft the storyline by hand regardless of whether the story is episodic or not.

The approach I am aiming to implement (eventually!) is going to have to involve some form of algorithmic story generation, using some form of lower-level story "building blocks" to construct the plot. With that system, it should be possible in theory to generate an infinitely long chain of stories. Then the only drawback will be the inevitable repetetiveness that will arise from repeating the story elements [smile].

Quote:

Original post by Fournicolas If the end of one episode can be ANY situation, how do you choose the initial situation of the second episode? Let's say you DO have a limited three-choice ending per episode. On the second episode, you must have nine stories going parallelly. On the third episode, that's twenty-seven, and by the fourth, you've reached the 81 figure. ON the fifth episode of your semi-open-ended you've had to produce a hundred times more content than if you had completely foregone the semi-open-endedness of the episodic gaming.

This is not a new problem. Script writers for episodic television shows have been facing (and dealing with) it for years. The trick is to ensure that there is some sort of closure and no complex changes at the end of each episode.

Modern computers can keep track of huge amounts of data, and as long as each change is to something the program already has to keep track of (such as whether an NPC is still alive or where an object is), there shouldn't much more effort required than there would be for the average CRPG with multiple quests.

Take the first Diablo game as an example. In the single player game each quest could be treated as a separate episode, as they could be completed in almost any order and the program chose a selection of them from a larger "quest pool".

Expansion packs for games such as Morrowind or Oblivion could also be viewed as additional episodes.

Alpha_ProgDes: I sincerely hope that this is the future of RPGs or even MMORPGs. And to be honest, with all of talk about this type of game in the last 6 months or so, I'd say theres at least a market for it.

We've all grown up on RPGs that led the player through a story. These are sometimes good stories, sometimes not, but the problem is, in order to play out the story, the gameplay is required to be linear. Sure there may be 'side quests' (Baldur's Gate) or an 'open-ended world' (Morrowind), but in the background, theres a story that has an end goal that you must perform X to get Y to complete the story.

Do RPGs need to be story-driven? Absolutely not! RPGs should revolve around the character, not the game. Thats why the latest advancements in RPGs have allowed the players to pick their own classes through repeated actions that bring skill advancement. The game doesnt define what the class the player should be, the player decides by how they play their character.

I think RPGs can be taken a step further. If the player were provided with a dynamic world, one which could exist without them, then the player gets to make the story. This is in line with the current evolution of RPG games. Developers have begun to realize that players do not want to have their hands held while developing their character. They want to control 100% of what their character does. The only way to do that is to get rid of the storyline and let players create their own.

Allowing players to simply enter a dynamic world and affect it any way they want would be the ultimate in immersion. A player is a blacksmith in a town where the magistrate is very greedy and makes the business owners pay unfairly high taxes. By making the world dynamic, the player would be able to rally a mob of people to revolt against the magistrate, or pay an assassin to kill him, or make the magistrate beautiful weapons to cause him to lower just the players taxes. In a dynamic world, all paths are possible.

The only problem with this so far, is that to make the world truely dynamic, the developers would have to construct this world on very large, complex algorithms that can take several different types of player responses into account. Not saying that it cant be done, hell, Im already attempting it myself. But it certainly won't be easy. And I think until someone can find an intuitive way to accomplish this, this type of game will be nothing more than a pipe dream.

But I certainly hope not. ;)

As far as feasability goes, I'd like to point out that one of Oblivion's big selling points was the Radiant AI, which they had to tone down, because it kept doing things they couldn't anticipate.

There are two big concepts here that need to be separated for discussion. First is the NPC AI that is necessary to react in a completely unscripted environment. Secondly, the dynamic quest system, which hasn't been attempted with any large degree of success that I am privy to.

I think that Luminous AI^tm (which is similar to but legally distinct from Radiant AI) would be an incredibly fun project, but only on the academic level. If Bethesda had to scale back their NPCs, I think that it would take a hobbyist years of just working on AI to top them. Probably only a handful of people on this board have the wherewithal to complete something like this, (and I hope you're one of them!)

A dynamic quest system, as mentioned by TZ, would be much more achievable in the near-term. It requires more content creation, but probably a lot less fine-tuning for the AI.

In short, I recommend you start with dynamic quests and see where that takes you. I'm actually much more excited for the AI, but I doubt anything I can do in my spare time will top a company that can make a game like Oblivion.

Quick quip. I'm no AI genius and honestly I didn't think, initially, the AI would play such a major role. Yeah, that's why I'm not a Game Designer.

Quote:

Original post by templewulf As far as feasability goes, I'd like to point out that one of Oblivion's big selling points was the Radiant AI, which they had to tone down, because it kept doing things they couldn't anticipate.

As I alluded to a bit earlier, I think the primary problem that Bethesda ran into with Radiant AI was that they hadn't really thought through how they were going to present "what just happened" to the player. Sure, it sounds cool when you talk to another developer about having three NPCs in a room with different emotional levels and goals, and one of them gets jealous and backstabs another because he's having an affair with his wife, while the other guy's greed takes hold of him and he steals the jewels from the safe, and then the first guy finds out and hires an assassin to strike him down, etc, etc, etc. But if all this happens while the player is out putzing around in some cave, then all the player comes back to is a room full of corpses with no explanation. In Oblivion's case, the only way to know what happened is if you read a debug trace log of the AI. The game is entirely voiceover, and of course they couldn't record voice dialogue for every possible thing the AI could have done, so instead you basically have no idea what anybody is doing unless you're sitting there watching them. Even if you do catch them doing something "interesting", you can rarely figure out that one guy is jealous and one guy is greedy, or that they have any sort of past relationship or feelings to eachother, because again, there is no recorded voice dialogue, no explanations, just a bunch of AIs doing apparently random things to eachother.

In the end, I found it very questionable whether or not Radiant AI actually added anything to the game. Honestly, beyond a technical programmer interest in the AI, I just was utterly uninterested in whether NPC Bob was jealous of NPC Jim, or whether NPC Steve stole an apple from NPC Susan. I was only interested in things that affected my character directly. Perhaps I'd be interested in a non-interactable (is that a word?) story between NPCs if it was really interesting, say, more interesting than your average pre-scripted cutscene, but if it's not, I'd rather be putzing around in the cave killing something. And I don't think that AI will be producing stories more interesting than pre-scripted ones any time soon.

Quote:

But if all this happens while the player is out putzing around in some cave, then all the player comes back to is a room full of corpses with no explanation.

This would all be a part of the way a dynamic world would work, however. If these NPCs were all a part of a city, then most of them had jobs right? Now theres vacancies for these jobs. That could affect the player. Plus, I would think that other NPCs who had relationships with the dead NPCs would find out about the deaths, and want to take some sort of action. If this action is something the player could do, it could turn into a new quest, or line of quests really, for the player.

As for the whole explanation thing, the next NPC to find the corpses would call a guard, who the programmers could give a Sherlock Holmes level of deduction and suddenly the guard knows what happened at the scene and the word spreads through town.

Quote:

Original post by makeshiftwings Sure, it sounds cool when you talk to another developer about having three NPCs in a room with different emotional levels and goals, and one of them gets jealous and backstabs another because he's having an affair with his wife, while the other guy's greed takes hold of him and he steals the jewels from the safe, and then the first guy finds out and hires an assassin to strike him down, etc, etc, etc. But if all this happens while the player is out putzing around in some cave, then all the player comes back to is a room full of corpses with no explanation.

This is also part of why I don't feel the simulation approach is the best way to achieve interactive stories: firstly because events will happen where the player cannot experience them, and secondly because they have a tendency to spiral out of control; if the backstabbed guy happens to be integral to a subplot, then the whole thing just got wrecked by no fault of the player.

My preferred approach is to build the system from the story perspective, rather than as a character simulator. Although I find this hard to explain briefly, under this approach the events that happen in the world revolve around the protagonist (i.e. the player character). Every major event would have to happen for a reason, the ends justifying the means. This is somewhat the reverse of most approaches I've seen to the problem.

I'll try to explain using your example, working backwards. If the player comes back to find a room full of corpses, there has to be a reason for it. If the corpses are all strangers, maybe there is a mystery to solve, or the corpses are there to signify some other threat. If the one of the corpses is a member of the protagonist's family, maybe that is a trigger for a quest of revenge. I probably would not even consider making one of the corpses the chief love interest or the antagonist in an RPG, because that would be too hard to trigger off an appropriate quest.

Under this approach, since the player is putzing around in a cave, given they cannot see what is happening in the village the system could generate anything that they come back to. Thus the system can consider not just the events leading up to the situation the player meets on their return but what the player could do next. Given a storytelling system would wish for the player to do interesting things, it can tweak circumstances to make anything within reason occur in order to keep the plot going.