Dragon Age - Preview @ CVG

[Arpyjee]

The DA team is apparently pursuing the low fantasy route, where magic, recovery, buffers and immunities are greatly reduced. I wouldn't call it realistic (well, it shouldn't be, anyways), but I would call it minimalist.

This should result in a more lethal and demanding experience in combat, where penalties and consequences for bad or careless decisions are not so easily overcome, or "wished" away.

 

[Steel_Wind]

What I'd prefer is a game where no NPC is essential. No one was essential in BG or any of the sequels or spin-offs that I'm aware of, or Fallout or a lot of other games with great stories. Obviously you couldn't solo it in BG, but if somebody died the game didn't compel you to reload, did it?

Look, you just can't do this and have the world feel real and the NPC's in the tale respond to conversation and be part of the way the story is advanced.

You either have a Solo exploration MMO style game like Oblivion - or you have KotOR style NPC's that can't be killed.

BG2 style game design is not going to cut it. I know that's a shocking thing to say and bordering on near heresy in these parts - but it happens to be true.

It's a matter of design. The NPC has a role to play - something to do - something to say at various critical parts of the game. If he or she is dead - the NPC can't do that and the story may break.

The options are:

1 - Don't do it. Don't give that NPC a critical role at all. Leave out any of this interaction. Revert to an older style game where that never happened. Ignore the fact that when those older style games were released - people bitched long and hard to put in real NPC's to interact and talk with who were part of the story.

This requires a significant change in the tastes and expectations of the player base. It's just not going to happen. It's not 1995.

2- Go Half Way: Make your NPC's key dialog generic. Any NPC available in that slot can do it or say it. If one is dead - another will do or say it just as easily. This restores some element of the suspension of disbelief, insofar as NPC's can now die. But at a cost of writing NPC's in a generic fashion, without personality behind them - or at least much of their key dialog. Moreover, in this age of voice acting, it VASTLY increases your voice acting costs, increases QA by an order of magnitude and doesn't change the story in any measurable way - other than the fact that some avatar is "dead" and some other avatar is saying the words written for somebody else.

To what end?

If you think this is a superior design and rationally justified cost increase - I urge you to go found your own game company and make a game like that. Because you won't be seeing it from mainstream CRPG developers. QA costs are way too high for this.

3- Make plot critical NPC's unkillable: This is the current default design. Key NPC's cannot die in the party. Differing games all face the same problems in WHY key NPCs cannot be permitted to die, and many will take a different road in how they try to maintain a suspension of disbelief. They all reach the same destination though.

Modern story based game design favours this third option for very good reasons. How this is finessed in the game's design can make the unkillable mechanic more or less palatable, but the consequences are the same. It becomes an exercise in subtlety and spin.

Players demand and expect NPCs they can interact with in game, who play a role in the story, who have personalities and who have voices of their own - and well acted voices at that - throughout the tale.

That's the expectation - and there is only one reasonable means of meeting those expectations.

There it is.

 

[Alrik Fassbauer]

[satire]

"Woohoo, I gave in my bloodlust and killed an plot-critical NPC ? Noooooo !!!!!"

[/satire]

 

[Squeek]

That's the expectation - and there is only one reasonable means of meeting those expectations.

How is explaining how something works now the same as explaining how it can't possibly work in the future? In my opinion it doesn't, and that's just short-sightedness. I don't mean that as a personal attack, either. It's an easy mistake to make, an easy way to think, especially when you're trying to imagine future technologies.

Here are some examples:

"640K (memory) ought to be enough for anybody." -- Bill Gates, 1981

"Heavier than air flying machines are impossible." -- Lord Kelvin, President, Royal Society, 1895

"There is no reason that anyone would want a computer in their home." -- Ken Olsen, President, Chairman and founder of Digital Equipment Corp., 1977

"I think there is a world market for maybe five computers." -- Thomas Watson, Chairman of IBM, 1943

"The telephone has too many shortcomings to be seriously considered as a means of communication. The device is inherently of no value to us." -- Western Union internal memo, 1876

 

[Prime Junta]

How is explaining how something works now the same as explaining how it can't possibly work in the future? In my opinion it doesn't, and that's just short-sightedness. I don't mean that as a personal attack, either. It's an easy mistake to make, an easy way to think, especially when you're trying to imagine future technologies.

I thought the discussion was about Dragon Age in particular and games design using currently available technology in general. If there are breakthroughs in AI research that will let you create agents that dynamically adapt themselves to the player's actions to keep the plot alive, that'll be a real revolution -- but it ain't gonna happen within the next five years, and pretty certainly not during the next ten.

 

[Squeek]

That's a valid response, since my comment made it out like that. But I don't think it would take new AI as much as it would simply take more work. The real limitation here is the $60 price point.

You're a programmer, so I'll ask you. How hard would it be to moniter progress and decide if it's necessary to insert NPCs on an as-needed basis in order to advance a plot or sub-plot?

 

[chamr]

You're a programmer, so I'll ask you. How hard would it be to moniter progress and decide if it's necessary to insert NPCs on an as-needed basis in order to advance a plot or sub-plot?

Not technically difficult at all. Just lots of coding to cover lots of possible outcomes. IOW, more time spent coding the AI. This is one of my long standing beefs with/hopes for the genre. I think there is much ground left uncovered in the AI area. The focus, for a long, long time, has been on graphics and size. Compared to how far they've come, AI is barely out of the gate. It would not be hard to advance this aspect of cRPG's dramatically. It's just that no one's really doing it.

 

[Steel_Wind]

You're a programmer, so I'll ask you. How hard would it be to moniter progress and decide if it's necessary to insert NPCs on an as-needed basis in order to advance a plot or sub-plot?

Wrong question. Try these:

How hard would it be to convince the money behind development to fund a game premised upon that logic?

VERY hard.

How hard would it be to QA such a game?

VERY hard.

How unlikely is such a game?

VERY unlikely.

A Triple A title is $15-25 million in development dollars. An adaptive story AI is just not in the cards at this stage. Games are a business - and this would be far too high a risk for little payoff.

As part of a MMO dev cycle where failure means you simply never hear about it and the game carries on? Maybe. But developing this for a Triple A Single player RPG where the game's critical path is premised upon it?

Not a chance.

 

[Prime Junta]

You're a programmer, so I'll ask you. How hard would it be to moniter progress and decide if it's necessary to insert NPCs on an as-needed basis in order to advance a plot or sub-plot?

The programming wouldn't be hard -- we're just talking about evaluating a few new parameters and adding a few new branches to the decision trees. The difficulty would be in the writing: if an NPC got killed off, and the plot has a role for them, you'd have to dynamically replace the NPC with someone else.

Consider some of the roles NPC's typically have. Take KOTOR -- a game that had interesting and varied roles for NPC's. Now, suppose you got Zaalbar killed before you got to Kashyyk. The entire Kashyyk plot is written around Zaalbar's relationship with Hanharr. There's no way you could get it to work without that particular relationship, and the nature of the relationship dictates that the individual is a wookiee. So, somehow, you'd have to arrange a sequence where *another* wookiee joins your party, and then steps into Zaalbar's shoes. And if this wookiee gets himself killed halfway through, you'd have to produce *yet another wookiee* -- and this in such a way that the plot wouldn't be disrupted.

Put bluntly, it can't be done -- not within that particular narrative structure. You'd have to make both the plot and the NPC's generic enough that one NPC can do another's job, and that IMO would be a big loss. I'm not saying you couldn't make a game like that -- but I am saying that the stories in it couldn't be anywhere near as detailed, deep, and interesting.

Edit: in fact, such games exist. They're called squad-based shooters. One squad member gets shot, you get another squad member to step in. They may even have individual personalities, appearances, and lines, but for the purposes of the game, a rifleman is a rifleman is a rifleman. We're no longer in RPG territory.

 

[Squeek]

Not a chance.

Maybe. But I've worked with about five dozen startups, most of which were pursuing new approaches to unsolved problems. How about you?

EDIT: We posted at about the same time, so I missed yours, PM. Good answer. That's why I keep thinking cRPG makers should try to get out of the business of making individual games and get into the business of making and then continuously modifying game worlds.

That would take some thinking, and there's no shortage of folks who would love to tell you how it couldn't be done business-wise. I'm a fan of the genre, and that's why I keep throwing these things out. All we can do on our end is voice our opinions, after all.

 

[Prime Junta]

One idea I've been toying with is a single-player RPG that tracks what the players are doing, and has a team of writers creating content that fits wherever the player populations are. The content would then be delivered over the Net on a subscription basis; the modules downloaded would match wherever your player character happens to be in the "opportunity cloud."

The platform would have to be insanely robust and insanely easy to write for, so you could get the new content on-line with a delay of days rather than months or years, and the content modules would have to be very small, and it would have to be possible to mount them dynamically as the player is playing.

The upshot would be a world or plot lines, characters, locations, and what not that gets richer and richer over time, as all content would be available to all players, assuming their characters hit the mission bits, of course.

Wouldn't be easy, I'm sure, but it just might be possible -- and it'd be a novel business model.

(There's no reason it wouldn't work in a MMORPG either, of course.)

 

[chamr]

The programming wouldn't be hard -- we're just talking about evaluating a few new parameters and adding a few new branches to the decision trees. The difficulty would be in the writing: if an NPC got killed off, and the plot has a role for them, you'd have to dynamically replace the NPC with someone else.

Why? You're limiting your thinking. Why must it be an NPC that moves the plot forward? Perhaps you take some identifying "thingy" from your fallen commrad, and when you show it to the right person, the plot moves forward. In a different way, perhaps, but it doesn't come to a halt without a replacement NPC. Or, if it's a side quest, perhaps it does. Or maybe it takes it down a whole other branch.

Again, it's not impossible, or even a stretch. It would just take time and making it a priority.

 

[chamr]

A Triple A title is $15-25 million in development dollars. An adaptive story AI is just not in the cards at this stage. Games are a business - and this would be far too high a risk for little payoff.

As part of a MMO dev cycle where failure means you simply never hear about it and the game carries on? Maybe. But developing this for a Triple A Single player RPG where the game's critical path is premised upon it?

Not a chance.

Again, limited thinking. You're assuming that all things in the development stay the same and we add more complicated and involved scripting and AI. What I'm saying is spend less time on graphics and, especially, less time on making a world that is too large with too many repetitive quests. I think it would be a good risk to make a game that was perhaps half the size of Morrowind with graphics that do not push the envelope (read: license an existing engine) but was packed to the gills with a truly adaptive world & storyline plus startling NPC and monster AI. I don't know about you, but I'd snap a game like that up in a heartbeat.

 

[Steel_Wind]

Again, limited thinking. You're assuming that all things in the development stay the same and we add more complicated and involved scripting and AI. What I'm saying is spend less time on graphics and, especially, less time on making a world that is too large with too many repetitive quests.

No. I'm assuming that you need to finance the game you are developing, because I develop CRPGs in the real word - and not on a discussion forum.

Securing that financing is premised upon, inter alia, graphical strength, perceived market demand and risk assessment of the feature set you are pitching. All assessed with a view to profit return and measured against risk.

You want to cut eye-candy in search of a risk heavy adaptive AI, that will take you an additional 8 months to a year to QA with no guarantee of success?

Nice pitch. Good luck with all that.

 

[chamr]

No. I'm assuming that you need to finance the game you are developing, because I develop CRPGs in the real word - and not on a discussion forum.

I've been project managing software development for a living for over 10 years. Some of my projects have been at the multi-million dollar level. I know a little bit about managing risk in this context. In fact, as a PMI certified PMP, I'm willing to bet I've got more skills at it than 95% of the folks running game development.

What I don't have is industry vertical experience. I'm more than willing to concede that the sell would not be easy. Impossible, however? I doubt it.

You want to cut eye-candy in search of a risk heavy adaptive AI, that will take you an additional 8 months to a year to QA with no guarantee of success?

- How much essential "eye candy" would really need to be cut? Really.
- Where do you get an additional 8 months of QA? I've managed some pretty complex development projects. If you need 8 months of QA for a game, you ain't doin' something right.

I've always suspected that project management in game companies are primarily run by two types:

1) Suit with an MBA. No formal software development project management training, and little to no experience. Good at fiddling with budgets, making employees feel like they matter, and making the occasional, high level decision. But God help us if they even try to run the project on a day-to-day basis.

2) Developer that's been promoted to project management based on what's called the "halo effect". IOW, "Boy, they are a great developer! I'm sure they'll make a great project manager!! Let's hand control over to them!!!". As I'm sure you can figure out, this often results in some level of disaster. I can personally attest to the fact, having started out as a software engineer, that having intimate knowledge with the work at hand is invaluable to a project manager. However, being a great coder does not make you a great project manager of software development. Not even a good one. They are very distinct skill sets, and this is very often not recognized by companies.

And, finally, guarantee? Who the hell can offer a guarantee, even with the safest AAA borefest? That's right: no one.

Nice pitch. Good luck with all that.

Being an armchair quarterback on this, of course I'll never have the pleasure of trying. However, like every other industry vertical I've worked in, I'm sure game development is no different: those willing to take the risks are the same that have the best chance at a smashing success. Guaranteed? Of course not. Safe? Nope. The foundation of all innovation? You bet your sweet ass.

 

[Prime Junta]

Why? You're limiting your thinking. Why must it be an NPC that moves the plot forward? Perhaps you take some identifying "thingy" from your fallen commrad, and when you show it to the right person, the plot moves forward. In a different way, perhaps, but it doesn't come to a halt without a replacement NPC. Or, if it's a side quest, perhaps it does. Or maybe it takes it down a whole other branch.

Again, it's not impossible, or even a stretch. It would just take time and making it a priority.

I never said it's impossible. I didn't even say it would be technically difficult; technology isn't the problem, writing is the problem.

I did say that it would require making both the story and the characters generic enough that one character could step into the next one's shoes. That would be a pretty big trade-off, and personally I would rather see a rich, deep, involving story about fully fleshed-out characters, even if it means those characters can't be allowed to die at the wrong time.

IOW, you could have written another plotline for Kashyyk that would have revolved around some artifact carried by Zaalbar -- but that would have meant cutting out the interactions between Zaalbar, Hanharr, Mission, and the rest. And I just don't believe that stories about objects can be as involving as stories about people.

 

[Prime Junta]

This is an interesting topic, and I'd be happy if we could dial down the antagonism a bit -- we could arrive some place new.

Thinking about your proposal, @chamr, it seems to me that what we'd need is something that dynamically rewrites the storyline in response to unpredictable events arising from player actions.

That is, not just something that spawns a new NPC with a fixed baggage of dialog trees should one die: that would clearly break the suspension of disbelief pretty quickly. (Oh my, another wookiee drops from the tree and introduces himself as Zaalbar's second cousin twice removed. Yay.)

I'm not aware of any AI research that has gotten close to the point of being able to produce a believable and emotionally compelling story from dynamic AI, set parameters, and "building blocks" -- and certainly not one that does it every time. I don't know much about the state of AI research, though; perhaps one of you guys have better information about this?

However, if it were possible to make such a beast, that would open up a whole new world in gaming, and it would revolutionize the way games are written -- the writers would set the parameters for the "story engine" and then fill it with "story modules" that the engine would connect together as it goes.

I heard Mozart did something like this with music: he produced a few pages of phrases that you could connect in any order you like, and actually get something that sounded like a coherent composition. At least some of the time.

 

[Prime Junta]

Finally, a note about writing: the problem here is that we're dealing with limitations of human creativity, not about inventing clever new algorithms. It's hard enough to write a linear story that's exciting, believable, and emotionally compelling. It gets harder if the writer has to think of choices, branch points, and different possible outcomes. And it gets exponentially harder the more branches and variations there are.

In other words, to create the "storyteller machine," we'd also need to find some way of addressing this complexity. I have a feeling that we're already approaching the limits of the human mind in storytelling complexity with the current crop of games.

For example, take the discontinuities in The Witcher's storytelling in Act 2 if you did things in different orders are a reflection of this as much as they are of QA. They're not problems with code quality; they're problems with story continuity. If your investigation already revealed the murderer before the autopsy, and you proceed to do the autopsy, the dialog proceeds as if you discovered the murderer during the autopsy. IOW, to avoid that discontinuity, the writers would have had to write the scene two different ways, depending on what you had done before. More to the point, the same consideration applies to every point in the story where you can potentially discover the identity of the murderer. Since there are several such points, you get a whole another layer of complexity -- but you don't affect the outcome of the plot at all. The decision trees in the script logic look exactly the same; only the phrasing in the dialogs changes.

This, I believe, is the main difficulty in creating the adaptive story engine. Not to mention some secondary but equally difficult problems: for example, if the engine dynamically changes dialog lines to fit the character's state, you'd either have to cut out voiceovers -- or have speech synthesis good enough that you can maintain suspension of disbelief, emotional inflections and all. I don't know what the state of the art is in speech synthesis, but this sounds like a non-trivial problem as well.

 

[Squeek]

Reading this, I'm not sure everyone's on the same page. Conversations like this one, where some people want to discuss how a goal might be reached while others want to consider obstacles first can be frustrating enough to have in person. They're awfully hard to have in writing.

So I think I'll bow out.

 

[abbaon]

I did say that it would require making both the story and the characters generic enough that one character could step into the next one's shoes. That would be a pretty big trade-off, and personally I would rather see a rich, deep, involving story about fully fleshed-out characters, even if it means those characters can't be allowed to die at the wrong time.

For Dragon Age and games like Dragon Age, I agree. In general? I wonder. Dispensing with a stable cast of characters would cost you storytelling depth, but perhaps you could make it up in other areas. Like scale.

If you only wrote for races or personality types, not fleshed-out characters, the player could travel with a group the size of your ship's crew in Pirates. Or a fleet in the X series. Straying back toward RPG territory, your mercenary company in Mount & Blade.

It seems to me that you could tell stories about a party like that which wouldn't work with a fighter, mage, cleric and thief. You wouldn't have to build your worlds around the idea that a handful of "high level" people can defeat armies. You could offer the player experiences that he just never sees at the moment. No RPG has ever asked me to sustain heavy losses to defeat an enemy. I've never faced down a mutiny.

With the genre in its current sickly state, a game like this would probably have to replace one of our proper party-based RPGs, and we can't have that, but I hope someone finds the courage to explore this avenue of development in the future. If Taleworlds ever gets around to sticking a story onto M&B, I'm there.