Sunday, January 06, 2008

Information Handling

Raph had a post that got me thinking.

All of my games feature some kind of information system.

That's a bit misleading, though. All games are entirely about information processing. You see what information is available, then you try to figure out how to combine it so you can react to it.

If you have to decide which weapon to take, you call on your experience with earlier levels and similar games, then you extrapolate by what you expect later challenges to contain. If you're playing a shmup, you analyze the trajectories of everything on the screen, taking into account your ship's size and speed as well as what the various things are likely to do (change direction, explode, fire shots...)

Obviously, you're working with partial information on a complex system. There's a lot of room for leeway: some people are very good at collecting and analyzing specific kinds of information. For example, when I'm playing a shmup, I'm really good at analyzing a few very complex trajectories, but I'm bad at analyzing lots of simple trajectories. Some people are just the opposite.

Every kind of play I've ever seen is about collecting fragments of information and piecing them together as effectively as possible. In situations with time limits, more, simpler information tends to be the watchword rather than fewer, complex information.

Now, when we think about quests, information is usually simple window dressing. It's an excuse for the designer to make the player run through various challenges. Often, the information is partial. For example, telling you who to go to, but not what they will give you; or telling you what to get, but not where you can find it. This can be done either for narrative effect (getting you deep into a plot before twisting it into something nasty) or for play effect (making you spend time searching for the missing information).

In a MMORPG, using "static information" is a problem if you use partial information, because any player that's figured out the missing piece can share it instantly with everyone. Obviously, this short-circuits the whole purpose of using partial information.

To me, it is without question that static information is not a good idea in such a game. However, saying "use varying info" is not quite as simple as Raph has suggested. There are a lot of methods. Here are a few I love.

Game Chunks

You can get around the flaw with static info while still using static info if you make your game (or subsets of your game) have a definite end point. This is not always suitable in a MMORPG, but it is not unknown: Tale in the Desert is an example.

In this sort of situation the information is usually extremely fragmented, and the player(s) who assembles it first will have a big advantage. This is actually a fairly delicate system.

The problem with game chunks is fairly obvious: someone has to build them, and that can mean a huge amount of GM overhead. Fortunately, there are alternatives...

Player Affected

A lot of competitive games use player-affected data. For example, every time you play a level of Team Fortress, it's the same. Except that the other players radically affect how you play the game. Effectively, playing Team Fortress is less about the information from the designer, and more about the information from the other players.

This is really easy to design, but if you're aiming for persistence, extremely hard to balance. The core problem is that players participate at different levels, and the players who participate more (and are more skilled) will gain an advantage... and that advantage will continue to grow the more they play. If it doesn't, they'll get irritated and stop playing.

It's a very difficult challenge, but there are ways around it. One of the big ways around it is to partition levels apart from each other, and another is to add an outreach/mentor system where the powerful players are rewarded for helping weak players. For more information on this kind of thing, study WoW's level partitioning, City of Heroes' mentoring system, and Eve Online's corporation system. I can write more about those if I need to.

Fundamentally, player-generated content is simply a particularly feisty subset of this.

Player Adapted

One type of information generation system is slowly growing more popular, and that's when you generate information specifically for each player. Because of the amount of shared space most MMORPGs have, adapted information is generally very shallow so as not to disrupt the game for other players. But it's not shallow in and of itself: it's just a design restriction.

Examples of this abound and overlap with player-affected systems. For example, in many games NPCs won't give you quests until you are a certain level or unless you are a certain class. This is a mechanism for weakly enabling player-affected information by partitioning certain quests for certain kinds of players.

Of course, it's certainly possible to have much deeper algorithms. These almost require you to have a very dispersed player base, though: if the player density is as high as in most games, then players changing the NPCs or terrain of the game will radically change other players' play experience, which is generally a poor idea.

Still, assigning each player a chunk of game for himself is not going to be as rare as you might think. As computation and memory get cheaper, it will be more and more feasible to give each player his own little piece of reality in which deeper adaptive algorithms can run. Think of it as instancing gone mad.

Marching Information

Another variation is to have marching information. This is a situation where the information in the game is continuously in flux as it is replaced - every day or week - by new information which is somewhat different.

As an easy example, this could be that a merchant changes locations to another part of the city, or the Swamp of Boggyness stops producing Jewel Toads and starts producing Psychic Crocs.

This can be done with any level of automation. It can be done entirely by hand, entirely automatically, or some combination of the two, depending on the ratio of writers to players you're willing to pay for.

Many designers shy away from this idea because it denotes what feels like a lack of control. They may claim balance problems, but the fundamental problem is that it is more erratic than most designers really like. For some reasons, designers like the idea that what they design will still be in the game a year from now.

I think that's kind of silly, but my games are always studies in controlled chaos. Ask anyone who has played one.

Fundamentally, static writing is not the wave of the future. Designers may want the Palace of Peace to stay exactly as they have built it forever, but that's simply a poor idea in a MMORPG.

As to balance issues, it's certainly the case that the new bits will be out of balance. However, as they change every week or so, the unbalance is only so much. You can spend your time polishing the balancing algorithm instead of releasing ten thousand content patches.

...

What do you think?

16 comments:

Craig Perko said...

This essay was definitely fully caffeinated. A bit rambly. :P

I should note that what players are intended to do with information is also very important. For example, in many of my games, information is an extremely valuable resource that gives an edge (usually by helping you assemble other information). Sharing the information is a sign of great trust, since it gives away your advantage...

Anonymous said...

Hi Craig,

You said:

"In a MMORPG, using "static information" is a problem if you use partial information, because any player that's figured out the missing piece can share it instantly with everyone. Obviously, this short-circuits the whole purpose of using partial information."

I do know that Raph seems to share this idea. But I don't completely. Static information has not only bad qualities, it has good very qualities.

I.e. there isn't necessarily a short-circuit, but rather the revealing and sharing of partial information is the design and intent of the game.

In the real world we operate with partial information all the time and a lot of it is in fact not fully dynamic but rather static. Our own world is persistent and we can rely on a lot of information be reusable. In the real world when you figured out how to get from Santa Fe to El Paso that information doesn't change. And on top of it it is a natural real world problem solving skill to ask for directions and if someone has figured it out it is sharable.

It does seem to me that Raph now things that static information is kind of categorically broken but really, what matters much more is what you are trying to do.

If you want deep and intricate riddles and the whole game enjoyment concept is centered around revealing information then you will have to protect that information and a static design is too prone to be descriptive and sharable.

But if your goal is to to design a game that is about exploring and sharing that has a natural feel then there has to be partial static information, just like in the real world that people can find, share and talk about.

In fact I would argue that the persistence of information and the dialogue about this information is a big part of MMOs. It's a "Meta-game" that one can play even outside the game. Some MMOs are very good at tapping into this meta game of having information discovered and shared and discussed.

Rather than a "short-circuit" they have achieved a different type of value out of revealing information and engaging the player. Of course a game runs pale if all information is revealed and nothing is added, but MMOs today add information and hence are not globally static.

If you want to enable the social aspect of information sharing you may well absolutely have to have static information, which has to be partial, so that it is something that can be shared and be of interest.

Craig Perko said...

Hey Moroagh. Good comment.

"If you want to enable the social aspect of information sharing you may well absolutely have to have static information, which has to be partial, so that it is something that can be shared and be of interest."

Static information comes in a lot of flavors. What's your turnaround time? How long does it take for the fun player-to-player sharing of information to turn into a "RTFFAQ" comment?

The idea isn't to make the game into an ever-shifting morass of information quicksand, but to keep the discussion lively and prevent it from stagnating into a FAQ.

Obviously, various kinds of information would vary as to how static they were. How to get to El Paso? Completely static. What jobs are available once you get there? Not very static.

Sort of like the player marketplace: where it is? Constant. What the best deals are? Always changing.

Does this make things a little clearer?

Anonymous said...

Craig, I'll be honest I do think it's an old design issue that we have been doing/seeing for a long time, maybe since games come about.

The fact that players make a world less predictable and dynamic goes back to Pong. And ever since when one wanted to guarantee dynamic content, one gets players involved. In MMOs that's often PVP.

In games that want to have longevity there has always been addition of information (content patches/expansions). So no longevity game already is not globally static. The world grows and changes in a way and hence there was new partial information available.

Even the question "what jobs are available" exists already in form of looking-for-group routines with job bulleting boards (LFG-UIs) and self-advertisement (town crying).

Some quests are already only available under certain conditions hence also have the quality of not boing completely static.

Looking at MMO forums, the discussion about the games is extensive and I don't see them starved for information content to discuss.

That randomization helps avoid too easy repeatability isn't that new either. Already the appearance patterns of enemies in old arcade shoot-em-up games was randomized to increase replayability and unpredictability and challenge.

So to be a little too harsh, in a way we are already doing it all.

But I do understand that there is a specific context to this, which is to actually protect information so one can have replayable information based achievement. I.e. so that one can solve a riddle and the next time it's not the same. If you go to a sage and he asks you a question to answer, he may not just be able to come up with one question, he may have many and this isn't static.

So yeah in this setting I can see a problem with all too static design, but it's a specific setting.

But to be more constructive some ideas how to make riddles less revealable:

* Include information from the player's personal history. This is information that can be remembered but is difficult to share.

* Include randomization that doesn't change the riddle idea. For example one can randomize mazes quite well. There won't be a guide that gets you through the maze, just hits how to do so.

* Include many branches of static riddles. Make it unlikely that all variations of what can happen is quickly discovered by using a large selection of information.

All these are nice because they are static design ideas and are not overly complicating. I can imagine that many of them have actually been used before.

But to go back to being a little contrarian, I don't feel the topic is new and as far reaching. And I don't think that static partial information becoming public is a really big problem for game designs. People still enjoy these designs immensely and stick with them for years evidently.

On top there isn't even much evidence that many players enjoy not having the option to get outside help.

I think it would be very interesting to see an MMO which is full of hard riddles that are designed to not be walk-throughable and see what kind of player reaction one gets to it. Is it the Myst market or does this kind of design where riddle help isn't available something that can have broader appeal?

I can actually imagine that I can dig it, but I do dig the current designs as well and don't have any problems that spoilers exist or people are able to share solutions, rather I also see good qualities in this.

Craig Perko said...

First off, most of your comments seem to be under the impression that I'm advocating that all games need to use my shiny new theory.

Well, I don't think that, and it's not shiny or new. I simply wish there were some less conservative game designs.

It's not hard to think about information - how it will spread, what it will do, what it means to the players. But few designers do. They are happy to stick with a static system that stretches a bit.

Fine, they can. I just want to see some games that don't. I'm sure I'm not alone in wishing that my actions in a game had some kind of actual effect.

"In games that want to have longevity there has always been addition of information (content patches/expansions). So no longevity game already is not globally static. The world grows and changes in a way and hence there was new partial information available."

I tried to gloss over this, because the reply is fairly long. But here it is:

Mudflation is not simply a statistical phenomenon. Any massively multiplayer game that's been running for very long suffers from severe content mudflation of every kind.

Looking through older games - even WoW - you can clearly see large patches of barren content: content nobody uses, nobody plays through. This is because other content is more efficient, more fun.

It's not simply that expansions replace old content: often some of the content in the expansion doesn't measure up to the original and gets forgotten, too.

This is the nature of continually increasing the size of the system: huge chunks of system become no-mans land, and not the interesting kind. The kind nobody uses because there's just no point: the dungeon over the next hill does it so much better, the merchant in Morrowhall sells them for 15% cheaper...

Aside from a very few casual players that don't care at all about efficiency, these wastelands are uninhabited. They are, at best, useless. At worst, they are actually detrimental since they represent tens of thousands of dollars of wasted development. More, if your team is struggling to rebalance them.

Let alone the fact that for each expansion it gets harder to compete with the winning information from previous iterations!

On the other hand, adaptive content will not suffer this kind of problem nearly as much, and will also be cheaper in the long run than building massive expansion packs.

"Looking at MMO forums, the discussion about the games is extensive and I don't see them starved for information content to discuss."

At what cost to the designer? At how many millions of dollars per content pack?

And how openly? So much content globally released: what if I prefer to talk to my friends about content only we know about?

How impersonally? What does it matter that Golgoth the Destroyer is best faced with fire spells? I don't care: what's Golgoth ever done to me and mine?

...

You gave some comments on how to make riddles less revealable, but I think that you're misunderstanding my line. I don't give a flying rat's ass about riddles.

I care about actual content.

However, your riddle suggestions are pretty much concrete examples of the kinds of changing information I suggest: changing the nature of a challenge based on the player, letting the player change the nature of the challenge, and changing the nature of a challenge at pseudo-random.

"On top there isn't even much evidence that many players enjoy not having the option to get outside help."

Now you're misunderstanding on purpose. NONE of these options requires any kind of isolation or forces any kind of solo play.

Even in the event that you give the player their own land, there is no reason they cannot have other players on their land!

Personally, I can't stand MMOGs because "persistent world" really translates to "you can't do shit".

Allowing players to interact in a more personalized way would go a long, long way towards making these games less pointless.

Patrick said...

I've never been a fan of persistence, so my favorite approach is player affected. Not coincidentally, those kind of games are orders of magnitude cheaper to produce.

Anonymous said...

Craig, my apologies I do come with a lot of baggage from Raph's blog where this really wasn't about not so static worlds but about a bunch of other things.

That you don't share any of the setup that we had in the discussion at Raph's blog is quite understandable.

I don't think anyone minds less static game setups, I think even game companies, if they could get scripting less buggy and the content creation time down and flexible and dynamic content well-tuned and controllable, they'd be very happy to have it all.

And yes I'd love worlds where there is loads of meaningful things to do.

And as said I'd love to see an MMO that tries the hard puzzle route.

But none of this was really how Raph arrived at his claim that static partial information doesn't work anymore.

I think we actually pretty much just agree here.

Craig Perko said...

Sorry: I read many of Raph's posts, but never the comments. So, no, I'm not really basing this deeply on Raph's stuff.

Anyhow, I'm glad you agree. Sorry if there was misunderstanding.

Patheros said...

Hey Craig,
You said "For some reasons, designers like the idea that what they design will still be in the game a year from now."

I agree that this is seems like an outdated mentality. They should be interested in designing the systems that generate the dynamic content, rather than the content itself.

You give two "easy examples" of such systems: Merchants moving and location production switching. These are like an example of an RTS AI that just mines. That's perfectly fine when your trying to give a taste for what these systems can do, but there is wide range of harder examples: NPC cohorts that are smart enough you don't feel like killing them yourself, realistic economies, player involved politics, continually evolving tech tree. The list goes on, but these harder problems are where designers should be spending there efforts rather than generating the next static level.

Craig Perko said...

Pantheros: I completely a profoundly agree. I've written on those topics before, and the only reason I didn't include them in this essay was length.

Christopher Weeks said...

Hey Craig, would you be willing to unpack the following comment for me?

Personally, I can't stand MMOGs because "persistent world" really translates to "you can't do shit".

I'm a game consumer, not producer and I'm trying to catch up on terminology and stuff. I'd have thought the opposite about the meaning.

Craig Perko said...

"Persistent world" technically means that the things you do carry over from session to session. Therefor, to keep control over the game world, designers have to make sure that you can't do anything meaningful.

So you can gain a level, or move from one tier to the next... but you can't found a city, or build a wall, or actually defeat a boss. Everything has to be the same for all the other players, so they can't let you do anything that has any actual effect on the game world.

Christopher Weeks said...

Oh, I see. So you're beeef is primarily with the common implementation. I mean, some games do allow you to affect the world. Wurm Online seems to accomplish that particularly well (with other faults, certainly). Is it tremendously rare?

Craig Perko said...

It is very rare. The only "big" MMOGs that let you affect the world are SecondLife and Eve Online. They both have severe flaws.

There are a small number of tiny games which allow you to affect the world, such as Tale in the Desert and, evidently, Wurms. These are generally either extremely chaotic or excessively grind-heavy. Or both!

Also, all of these games have very strict limits on how you can affect the world. For example, you would find it almost impossible to build the ruins of an ancient civilization. In most of them, you cannot build an army unless it is made entirely out of other players, etc, etc.

Patheros said...

Craig, you say "The only "big" MMOGs that let you affect the world are SecondLife and Eve Online. They both have severe flaws."

I'm curious about their flaws. You imply they are related to the fact they allow there players to alter the world. I was wondering if you could elaborate (or just point me at something that explains)

Craig Perko said...

I made a post: http://projectperko.blogspot.com/2008/01/user-content-and-tradeoffs.html