This is the second part of our extensive interview with Arcen Games' Chris Park about their forthcoming procedurally-generated action-adventure survival game, A Valley Without Wind. Read on...
RPS: You've described AVWW as a 'procedurally generated adventure game', but from some of the other descriptions it seems to have more in common with RPGs than your typical adventure game.
Park: In terms of player/character progression, it is indeed a bit more RPG-ish. In terms of what you're actually doing while you're out adventuring and encountering monsters, that's all action-RPG, though. You have to cast spells or hit monsters with weapons or lay traps for them to step in. This is all done in realtime, like any adventure game. It just happens that how much damage you deal when you strike a monster is based on your player stats and levels, and the stats of the weapon you're using. And the monster's health values are based on the level of the monster.
RPS: What's going to be the meat of the player experience? Is it going to be puzzle based or combat based?
Park: This is where the uniqueness of the game really comes in, I think. It's an adventure game, but the focus is neither on puzzles nor combat. "What's left," right? In our case, the core of the experience is going to revolve around impacting the world. You do these various deeds, and the world changes for the better or the worse in the affected regions. Kill a big overlord bad guy, and everybody who had been oppressed by him is happy to now be free. If your character dies, then a grave marker goes up and people that knew him/her reminisce about them (or are glad they are gone, I suppose).
When you help other survivors out, you can convince them to join your settlements, and over time those settlements grow and change and provide new equipment and abilities for you. Also, when your current character dies, you get to choose from among the characters that you've added to your settlements (and while your overall stats persist, each character brings their own flavor and crafting specialties). As already mentioned, you can make certain areas a lot safer by setting up wind shelters, too.
RPS: Despite being a post apocalypse, this is sounding like there's a fair bit of player interaction with non-hostile NPCs, who are presumably not all horrific monster AIs. How deep is the player interaction with other people?
Park: Most NPCs will be pretty decent people just trying to survive, like you are. Most of the enemies are non-human monsters out to get all of you. Of course, you'll run into some enemy humans, or entire settlements of enemy humans, but they're more the exception than the norm. Most NPCs won't care much about you at all until you do something to help them out and win their favor, and then they might join your proposed settlements or might not.
In terms of interaction with other NPCs, in early alpha it's going to be extremely, extremely limited. Exactly what that would entail at that early stage depends on what we have time to fit in, honestly. For core gameplay reasons we need the ability to speak to NPCs to enlist their crafting services, but just about everything else lifts out if we have to for time reasons in early alpha.
In terms of where we'd like to take this post-alpha... the idea is that each NPC will have skills, desires, and problems, and you'll be able to help them out with their problems or to realize their desires, and that has benefits from both of you. Or not -- you can also just murder every NPC on sight if you want to. Probably not the best way to win, and the character that does all the murdering will get a pretty bad reputation (that dies with that character, conveniently for you), but the flexibility is there. Since the story is non-centralized and somewhat emergent, there are no "key characters" that the game can't progress without.
If the game really takes off, then this is one of the aspects that we'd like to push further in new directions. But to start out with, our focus is on the adventuring, and the NPCs exist in service to that.
RPS: And how easy was it to move from AI behaving like an AI, and AI behaving like a human?
Park: Well, that's one aspect we haven't particularly gotten to yet in our implementation. That said, in the Alden Ridge game I was doing before AI War, I had exactly this sort of system for NPCs. And I've worked with that in numerous other past projects over the years. Honestly, most of this sort of thing is only what I would tenuously call "AI." It's all a matter of branching decision trees that are fairly limited and scripted. If you play long enough with any sort of RPG game, you start getting repetitive conversation at some point.
In AVWW, this is another area where player content submissions will be welcome, so that should help really minimize the amount of repetitiousness and make the characters seem more real and varied than they otherwise would. We'll see how that works out. But honestly I don't have hugely grand plans beyond the standards of the RPG/adventure genres when it comes to the NPCs -- our goal is for a SNES-like poetic brevity of dialogue, and for a lot of unique-feeling characters, without going too crazy into a Heavy Rain direction or whatever (of course, that was heavily scripted, too).
RPS: But combat is still going to be a necessary part of the game?
Park: Combat is certainly still a key part of the game, but it's a means to an end and avoiding some fights is just as valid as engaging in them. You don't have to fight every monster just because you see them. In terms of puzzles, I think some degree of puzzles are also inevitable, but that's not something we've thought a lot about yet.
Right now we're still in pre-alpha, so we're mostly focused on getting the art the way we want, and the general world-building aspects in there. It's still very early days, but we've gone through a lot of intensive design and are really excited about what's coming. The first public alpha versions will mostly be exploration with little motivating purpose (no settlement building, limited interaction with NPCs, etc). Those higher-level features will be built-up with player involvement through beta and the 1.0 release and hopefully beyond.
RPS: With the popularity of games like Minecraft, the experience of a world unique to each player, yet persistent, has struck a chord with gamers. What do you think the appeal of that is?
Park: This is actually something that happens with the galaxies in AI War, too, though that's less of a creative situation and more of a historical "I remember anecdotes about this place that are unique to just me and the other players in this galaxy." Other players have talked about this effect that AI War has on them, and I've certainly seen it, too. Remembering the scene of some epic battle if I load up an old save is pretty cool, because generally that battle played out in a way that was unique to me and that map -- it feels more like a real life memory, rather than just another chapter in a book that someone else wrote.
So I think the appeal is multifold. In Minecraft, of which I'm a huge fan by the way, you get these massive worlds that you can reshape to your will. Starting with a world that is unique to begin with, and then adding on your own personality on top, is just a really special feeling. My wife and I have a private Minecraft server that we've played on since first encountering that game. We've never created more than one world, we've just kept expanding in that one single world. That's particularly cool, because we can go back to places that we constructed six months ago, and which are just a distant memory now.
AVWW will be a mix of those two concepts when it comes to world persistence and growth. You can't reshape the terrain or your surroundings like you can in Minecraft, but you can still affect the overall world in a a lot of persistent ways: settlements and wind shelters and stuff you do for NPCs being the three most obvious ways, but also including things like it remembering that you knocked down a bunch of trees in one area, if that's what you wanted to do. And as you explore around, you can leave spraypaint markings to show where you've been. And no matter what happens to your specific characters, their deeds and their legacies live on. You can go visit their graves or even memorials to the bigger heroes, and the things that they accomplished have a lasting effect on the game world.
Our goal with AVWW is that individual players will never create more than a single game world. Why would they? Every bit of content, even future content or expansion content, can become available in that world simply by exploring into new regions. There's a huge disincentive to ever start over, because all those older marks you made on the world would then be lost. Instead, even if I wind up exploring way out into the nether reaches of my world, I can always go back to the original starting area if I so desire, months or even years later. To me personally, that's really exciting.
RPS: One of my problems with Minecraft is that I create all of these huge monoliths, and then there's no one but me to witness their majesty. This makes me sad. Who will bear witness to my accomplishments in AVWW? Is there any thoughts towards multiplayer?
Park: Multiplayer is something we're hoping to have by alpha. It will probably be buggy and rough right at first in alpha, just fair warning -- but the player testing is what is needed to iron out those rough edges early on. Last week I'd said that I couldn't confirm if we were going to have multiplayer or not, but as we've been progressing in the technical implementation of the game that's been something that we just couldn't ignore any longer. I think that the game really needs it, and I'm always a huge proponent of co-op. If we delayed on implementing it too long, we'd run into the lengthy sort of refactoring period that Minecraft has seen, and I wanted to avoid that sort of pain.
The reason why multiplayer was ever a question mark for this game was that, as an action game, its networking model will be entirely different than that of AI War or Tidalis. RTS games in general use a really unique lock-step-synced sort of model, and for Tidalis we were able to make that work there, too. With an FPS, racing, adventure, or other action-oriented game, the networking model is based on an entirely different premise. Supposedly its easier to do than the RTS method, but it's new to me and so makes me wary for the time being. We've been preparing for the new networking model this week, though, and my hope is to have some early versions of that next week internally, and then fully ready by the time we hit alpha.
Of course, with any new networking model, we're going to run into unexpected problems and places where we have to optimize when a variety of players get into the game on a variety of kinds of networks. That will all shake out in alpha. My tentative goal is to support at least 16 players per world, but in reality it will be limited by whatever your network supports, I think -- so it could be more or less than that, depending on where you put the server. The game itself will be able to be run in sort of a headless server mode, and then the other instances of the game can connect to it. You'll be able to run both a server instance and a client instance on one computer, if you want.
All of that networking stuff is subject to change, of course -- it's new territory for us, some I'm cautious about promising too much for sure. But it's looking promising, and we should have some early working versions pretty soon to raise our confidence levels that this is definitely viable. I think it will turn out well, I'm just not sure how much bloodshed will be involved in getting it there!
RPS: If you look at what you guys have done with AI War, with near constant updates, and really huge expansions, there's a really deep level of interaction between you and the community of the game that isn't typical in game development. Has AVWW been developed with that kind of involvement in mind?
Absolutely! And actually, this game is designed to make that a tighter connection than even we've had in the past. Players are always wanting to help us out with direct content creation, but in AI War that's tricky because of balance concerns being that it's a strategy game. Even so, we've had donations of player art and sound effects for that game, and literally hundreds of design ideas, ship or feature concepts, etc, that we've implemented. With AVWW, we'll be doing all that, but also trying to take it to the next level for the core community.
There is a pretty cool component of custom procedural-content-script creation that advanced players can partake in (and then get their creations added to the official builds of the game, potentially, or just release them as unofficial mods). Part of the problem with having huge spaces in the game is that you don't want to be exploring Office Buildings A, B, and C hundreds of times. That's boring and feels stale.
The goal is that with AVWW we hope to hit an unprecedented number of unique places. We've already developed out techniques for a hybridized model of procedural and hand-crafted elements, all tied together with procedural seeding. Instead of literally designing "Office Building A" by hand, we instead make a "chunk script" that defines the overall design of that office building, and then which components of it can be randomized and in what way. That alone makes for dozens or hundreds of possible office building variants from just a single script. But by having lots and lots of scripts, that gets multiplied into the thousands or tens of thousands even for just a concept as discrete as "office building."
When you start adding in other concepts like scripts for "suburban house" and "rich person house" and "old shack" and so forth, and even "types of clumps of trees," the number of possible emergent designs quickly gets ridiculous. My goal is for Arcen to create a few hundred of these sorts of scripts by the time 1.0 hits (and numerous dozens by the time we even reach public alpha), and then my hope is that we'll hit many more hundreds of scripts that are player-created. The end result is hopefully worlds that are orders of magnitude more unique than in any other game ever seen.
RPS: Something like Spore? One of the problems that seems to plague procedurally-generated content is that while each building or level is unique in general, there are recognisable parts that pop up again and again. The same table, the same rock, the same colour palette. How are you setting about adding unique flair to the world, where the threat is just creating the same building in a million different ways?
Park: That's fair enough -- and to some extent, is unavoidable, for sure. Yes, you'll be seeing the same chairs and tables and rocks, the same pieces of wall, repeatedly. That's absolutely unavoidable and I wouldn't want to pretend otherwise. That said, AVWW is going to have the same advantage over Spore that AI War has over other RTS games: it's 2D. That means that the RAM cost, and the production-time cost per item, is vastly lower than in 3D. So if we want to have 10 different variations of trash can, we can and it's not that hard. If we want to have thirty different types of brick wall: also possible.
I think that Minecraft is a really good example of how many different things can be constructed out of even a simple toolset, though. How many blocks are in Minecraft that people commonly use for construction. A dozen? Yet the end product is often really more than the sum of its parts. With AVWW, we have less "atomic" parts compared to Minecraft, so the variations won't be so endless for each individual part. But given our 2D nature, and general art pipeline flow, one of my personal goals for the game is just having an enormous amount of possible parts to use.
Does that mean that, at some point, you're still exploring the same house with windows in a different place, stairs shuffled around, different tables and chairs in slightly different spots, and a different texture for the walls, door, and roof? Absolutely. But that's true of real life, too, I think. Our homes are not all that unique except in the details. You and I might have the same TV, or the same armchair, or both.
The cool thing about a randomized procedural scripting system, though, is that it can take broad hand-designed outlines -- such as a split level home, a duplex, or even a duplex with these general features or eaves or whatever -- and then it can randomize selected components such that you wind up with a surprising number of combination. Even how you design a duplex and how I design one would be different, if we each make a script for it. If you have all that variance in designs just for houses alone, and a suitably large number of props and textures to fill in the blanks with... that complexity is multiplicative. There's obviously a lot more structures for the game than just houses, anyway, right?
The result won't match real life any more than any other game does, but if we wind up with enough content and enough scripts, we will be able to push this further than most games. I think that some games that have come before, like Graal, Minecraft, and even Spore do show what's possible going in that sort of direction. The proof is in the pudding, and our pudding isn't done yet, so I'll be as interested as anyone to see how far we can really push this. But seeing what we've been able to accomplish with AI War, in as barren and blank a place as outer space, I'm really encouraged. I really think that we can do for procedural adventure games what AI War did for strategy games.
RPS: Right, so the players will be almost as important as the developers, ideally.
Park: Yeah, and how far we're really able to push that procedural generation is largely based on player support and involvement, but past experience is that this is the sort of thing players will be all over. Worst case is that the game will be larger than most, but given our past experience with AI War in making one of the most (if not the most) content-rich strategy games around, I think we're well suited to this job.
In terms of why we're doing an early alpha, it's all about the players: we're wanting to let the core fanbase (and anyone else who wants to join in) get at the game early either by preordering or demoing it. The larger gameplay elements will still be very much in the process of being added at that time, but there will be enough to be exciting and for players to help with content scripting if that's their thing. That way we get lots of testing, and hopefully lots of player-created content to make the world feel even more huge and varied than we can do on our own.
Our past experience has been that some players really love getting into our games as early as humanly possible. Most likely, their ideas and suggestions will take the higher-level game mechanics in directions that I can't currently foresee, as happened with AI War. What we have presently designed on paper is really solid and already really huge, but new ideas will be occurring to us as we go, and players always come up with really innovative stuff we'd never have thought of.
And given that we're self-funded for all this, we need the support of those preorders in order to fully realize this game. We're able to coast on our recent releases like the latest AI War expansion up to alpha and a bit beyond, but to make it all the way to 1.0 in the way that I've laid out here, it's going to take player support. Our job is to make even the early alpha versions so compelling that folks want to get involved right from the start, and just can't resist playing it even before some of our cooler world-persistence ideas are in there. It's worked for Minecraft, Dwarf Fortress, and in a lot of senses AI War as well, so we'll see what happens.
Right now we're focused on making the art as amazing as possible and something that will get folks excited to play the game, and then it's on to packing in as much content and functionality as humanly possible before we get to alpha. Given the early response on our forums and other sites, the future looks bright for AVWW. But time will tell!
RPS: Thanks for your time, and your many, many words.
Expect an alpha release in March!