Wednesday, August 03, 2005

Interactive Storytelling and the Gamer, Part II: Nuts and Bolts

In part I, we left off with a Social Simulation-Enhanced Role Playing Game (SSERPG) that had spawned the following situation:

A player had gone to Merlock the Magnificient looking for a 'Magma Mallet' spell; Merlock offered the spell in exchange for the killing of his enemy, Splitwick the Sage. Merlock didn't know 'Magma Mallet', though. He lied because he wanted the player to think more favorably of him, because he knew he had the trust of the weaponssmith, who's daughter Merlock hopes to marry. Splitwick knows Merlock is a liar, and has been using this to blackmail him to the sum of 100 gold monthly.

Now, let's take a look at how a social simulation would create this situation. I'm loosely basing this on my limited understanding of Chris Crawford's Erasmatron engine as I've read about it on his site. It's complicated, but not as nearly as complicated as it looks because we humans immediately draw all sorts of conclusions when we hear a string of gossipy facts like the previous paragraph. The reality is pretty shallow. It's certainly no Turing test.

Merlock, like all the actors in this social simulation, is nothing but a few lists and tables: values he's trying to maximize, and values he holds for other actors. These values are suggestively labeled 'trust', 'obligation', 'esteem', etc., but have specific mathematical relationships to those of other actors. The simulation provides various methods by which actors can end up expressing these relationships, and these include textual representations that make sense to the player but are irrelevant to the simulation.

By luck of the draw, Merlock has a want called 'love', and in semi-random fashion he drew a 'marry suitable woman' method from the simulation's dictionary for methods on gaining love. His indicators of socio-economic status matched a number of other actors classified as “women”, and in semi-random fashion he was assigned the weaponsmith's daughter as his ’suitable marry' target.

'Marry' is simply a word that, to the game, denotes a comitment that will not be accepted without some combination of high values of esteem towards the one asking, or obligation, or payment, etc. With different weights on the values, it could be 'share family recipe' or 'give footrub' or anything else. The name is actually unimportant, except where it helps the player understand the mechanics of it.

Because the weaponssmith's daughter did not have enough esteem/trust/obligation towards Merlock, she refused his ‘marry’ request. But by default, most actors give a 'what could I do to increase trust/obligation/esteem with you enough for X' query when they are refused something on such grounds. The dice rolled again, and the simulation selected the reply ‘gain the trust of my father', in part because her trust value towards her father is currently quite high, and in part because trust-increase replies to a man’s ‘marry’ request are inherently skewed towards a trusted father. So Merlock went to her father, the weaponssmith, and gave the 'what can I do to increase trust/obligation/esteem with you' query. The weaponsmith said ‘money‘ (you and I might call this ‘dowry’) or 'gain esteem of my friends', and gave a list of friends for whom his own trust value was high. This list included the player.

Since Merlock was low on cash, he could not immediately satisfy either of these requirements. His want for money increased, as did his want for the esteem of everyone on the weaponsmith’s list.

Enter, finally, the player. When the player came knocking, Merlock recieved an “I want Magma Mallet’ request. According to the rules of the simulation, actors will sometimes make false claims, or promises they can’t deliver, when they wish to increase their esteem with someone else right away. In other words, they lie, especially when their 'integrity' and 'intelligence' attributes are low relative to their degree of want. And when an actor in this situation has an 'i want this' request from another actor, he will not only claim he can provide it, but offer the same kind of trade he would if he were not lying. Otherwise, the lie would be pretty transparent -- and immediate -- making it useless for a temporary esteem boost.

This is not the first time Merlock has lied. It is a character defect for him, because his 'integrity' attribute is low. He had previously told a lie to another actor, because he wanted 'money', something the actor had and was willing to exchange for ‘Fireball Level 2', something Merlock did not have.

You may not be surprised to learn that this other actor was none other than Splitwick the Sage, who's wants included 'magical power' – that’s how he became a wizard in the first place. ‘Fireball Level 2’ was one of the items in the list of things that would satisfy Splitwick’s 'magical power' want. So Splitwick was asking other characters if they had ‘Fireball Level 2’. Merlock said he did. Splitwick asked for the spell. Merlock refused, because he did not have enough trust/esteem/obligation towards Splitwick. A “what can I do to increase trust/esteem/obligation with you enough to get [‘Fireball Level 2']” query was made, and Merlock responded with '100 gold', because, as mentioned, Merlock wanted money.

Splitwick gave Merlock the money, but Merlock did not give Splitwick the spell because he did not have it. Splitwick's trust and esteem of plunged, weighted by the worth of the unmet agreement, which was '100 gold'. But the relevant part for our story is that the simulation also computed that Splitwick had caught Merlock in a lie worth 100 gold, because Splitwick was privy to all the information neccessary to know this -- Merlock had said he had the spell, took 100 gold based directly on that claim, and then denied he had the spell, all in Splitwick's presence. Now, the simulation allows actors to respond to lies in a number of ways beyond the automatic drop in trust. One of the responses that can be chosen, especially for actors with low 'integrity' values, is 'blackmail'. It turns out that Splitwick isn't the embodiment of integrity himself, and this is what happened.

'Blackmail' is really just a name for a kind of threat transaction, which is when an actor tells another that unless he is given some want, he will perform an action guaranteed to lower the attainment score of what he believes is one of the others' wants. When the ‘blackmail' threat is drawn as a 'caught someone in a lie' response, the threat is to reveal the lie to a list of other actors -- an event which is sure to lower their trust towards that actor. Splitwick wants money. How much? The lie was worth 100 gold, and the simulation says that's worth 100 gold a month for 10 months. Ouch!

The simulation computes whether the lost trust will be worth more than the blackmail amount to the actor. In this case, it was determined that Merlock valued the collective trust of the people on the blackmail list more than 1000 gold, so he agreed to the blackmail. But as is the consequence of accepting a 'blackmail' threat, Merlock's esteem for Splitwick dropped to an abysmal low. Given Merlock’s lack of integrity, this score is low enough that he might, given the opportunity, contract an assassin.

Such an opportunity arose, of course, when our player arrived asking Merlock if he had the ‘Magma Mallet’ spell. Once again, Merlock lied. He lied because, as was shown, he had low 'integrity' and wanted to increase his esteem with 'player' in order to increase his trust with the weaponssmith in order to gain the trust of the weaponssmith's daughter, in order to obtain from her the 'marry' commitment, in order to satisfy his 'love' want. And the simulation required that he 'hide' the lie by behaving just as he would have if he were not lying, by naming a want that the player could satisfy that would be equal or greater to Merlock's theoretical value of the ‘Magma Mallet' spell. The simulation determined that 'kill Splitwick' was a suitable match, for reasons which we now know all too well.

So when the player made his 'what can I do to increase trust/obligation enough for 'Magma Mallet'' query, he was asked to kill Splitwick the Sage.


See? There's nothing magical about it. In fact, Merlock had no clever plan for dealing with you once you had taken care of Splitwick. You would've caught him in a lie worth whatever you decided it was worth, because you're the player. He may have responded in one of the ways people caught in a lie can sometimes respond when they know they've been caught, like trying to kill the person who caught them. It may have looked to you like a clever plan, but it was really a simple causal reaction.

Or, if he had no such reaction, you might have tried a blackmail threat on Merlock yourself. And if you picked a low enough amount, and gave a good enough actor-tell list, he might have agreed. You came in looking for ‘Magma Mallet’, expecting a run-of-the-mill 'fetch me this' request. Instead, you assasinated another wizard and are now getting blackmail money every month – it’s unexpected, but logical, interesting, and helping you kick butt. In a word, fun!

Perhaps this all sounded needlessly convoluted to you.

In fact, it wasn’t complicated enough, as I gave only a peek at how a social simulation can give rise to such encounters. I omitted many details – most of which I probably haven’t even thought of.

But the point I hope I have made is that the social simulations behind interactive storytelling can solve the general case of creating interesting situations in a fictional universe.

Gamers may be inherently repelled to textual content because of how it is traditionally used, but in a SSERPG, the social simulation adds valuable depth and longevity to a tried-and-true game stye. I believe this is a recipe both for comercial success and for reaching out to non-gamers; The same mechanics that push you into the assassination business today could turn you into a collector of decorative swords, a marriage counselor, or who-knows-what-else tomorrow. And it could all still be in the name of helping you kick butt – if that’s what you want.

Tuesday, August 02, 2005

Interactive Storytelling and the Gamer, Part I: Hearts and Minds

Even if you're only a casual gamer, you may have played Myst, back when CD-ROMs were new and gullible PC shoppers thought the word "multimedia" meant something.

If you're "old school", you might have played Zork, back when floppies were starting to overcome cassette tapes as the removable media of choice.

And even if you're a non-gamer, you might have read "Choose-your-own-Adventure" books.

None of these have anything to do with interactive storytelling. A modern interactive story is not really a story at all, but a social simulation in which stories arise naturally.

So get it out of your system and repeat the "simulation" mantra until it sinks in. Repeat it until you see interactive storytelling as something that is, in fact, unheard of in every game you've ever played, because none of the games you've played have ever simulated much of anything in a way that could create or change the plot.

This state of affairs will change. Soon, hopefully. But the fact that I felt compelled to address these likely misconceptions about interactive storytelling (IS) should give a sense of the difficulty IS products will have gaining initial market share among gamers.

It is true that non-gamers represent a huge untapped market potentially receptive to IS products. But IS isn't going to sell itself. Even if a huge swath of non-gaming adult women were born to play a new IS title, they're not going to go stumble into a game store and pick it up. They're not going to pay attention to advertisements for video games. Gaming is antithetical to their current sense of self. How will they ever get to the point of sale?

I see two major avenues. The first route is by way of entertainment they're already consuming. "Can't get enough of [novel, tvshow]? Enter the intrigue at [web address]." Web-based game play is something this audience could handle long before they could bring themselves to visit a game store, and probably more appropriate to the serialized media crossover model anyway. Web-based interactive stories, while benefitting from authorial and graphical styles, need not – and probably should not – be as flashy as their pc/console counterparts.

The second major avenue into the hearts of non-gamers is through their gaming friends and family. Consider a female college student, home for the summer. "I bet YOU would even like this game," says her obnoxious 15-year old brother. "I don't think so," says the young woman, just like she always does. But she steals glances at it. She becomes intrigued, then hooked, and eventually finds a way to get more.

The challenge for the developers in this case is creating products that will appeal to gamers and non-gamers alike. While graphics and explosive settings aren't the selling points of interactive storytelling (IS), an IS product that is not web-based had better capitalize on the technical advantages and create a very rich client that gamers will accept and non-gamers will be able to grasp just by watching someone. I would strongly argue that The Sims, a crude-but-addicting social simulation, owes its amazing success with women to pulling off this procedure. (And coming out of the established Sim franchise certainly didn't hurt with the get-in-with-the-gamers-first part, either. There's probably a lesson there, too.)

But rich interactive storytelling is going to be even harder to sell to gamers without the extensive camouflage of a graphically-rich client. It’s a problem of past conditioning. The text/dialogue portions of most games are typically the parts we gamers try to skip. Chatty cut scene? Pass. Domestic side-quest? Nein. Expositionary lump? C'mon, let me kill something!

This is not because there is anything inherently wrong with dialogue or text in general, but because it has a cosmetic or janitorial role in most games -- generally to give the impression of a greater scope than actually exists, or to impart information that the interface was unable to communicate on its own. It's a facade. It's duct tape. It's filler. We gamers see right through it to the underlying mechanics, where words are scarcely to be found.

So how do you bring textual content back into a game in a meaningful way without triggering the gamer’s reflex? You camouflage it.

One way to make essential game text more palatable is to use an ‘iconized’ mini-language, something pioneered almost fifteen years ago in Chris Crawford’s inventive flop, Trust and Betrayal. The player builds statements by assembling icons that represent “words” understood by the game. The Sims could be said to use a drastically-simplified version of the same scheme: much too limited for a meaty interactive story, but easily understood by neophytes.

But iconized mini-languages of any complexity are difficult for the non-gamer to understand just by watching; indeed, Trust and Betrayal is said by some to have failed on account of its “confusing” interface. Today’s best compromise between plain-text and iconization is probably Deikto, the “tinker toy” language Crawford has devised for Erasmatron 4, his latest interactive storytelling engine. It’s not as visual as icons, but it’s still sleek, precludes the need for an ambiguous text-parsing prompt, and is a system a non-gaming observer can latch on to because it still uses actual words.

Wrapping up the language with a sleek GUI will not be enough, however, to turn today’s gamers into IS consumers. I think the best way here will be to retain elements that gamers know and enjoy. This idea is probably perverse to some IS purists, but I think that IS can, and should, be used from the start to add a much-desired level of depth to classical role-playing games – like the kind where monsters are slain, spells learned, and treasure discovered. Though I see nothing wrong with such hybrids in their own right, gamers who appreciate the IS dimension will have a good chance of being won over to “purer” IS titles. This will be especially true if the game includes non-violent means of achieving objectives that are at least as fun as the violent ones. Plenty of games have claimed to be revolutionary for including non-violent alternatives, but the non-violent paths have sucked to play.

So let's suppose we've got an IS-enhanced game that the gamer can now at least recognize as kin. How will the IS portions be used? Can diplomacy be as interesting as spellcasting? Can conversation compete with armed combat? When the words -– iconized, Deiktoized, or something else -- start scrolling up, what will keep the gamer from taking this opportunity to use the bathroom and check his email?

First, the words will be integral to the that which the gamer cares most about -- increasing his ability to kick butt.

In most games, Merlock the Magnificent doesn't feel like teaching you the ‘Magma Mallet’ spell. He sends you out into the swamp to find some rare reagent before he’ll help you.

But today, in the IS-enhanced version, he's going to try and talk you into killing his enemy, Splitwick the Sage, instead. The problem is, Splitwick knows other spells you also want to learn. And in this particular game, there is no "charisma" stat. If you want to reach a favorable agreement with Merlock, you're going to have to do it yourself.

Being clever, you may discover that Merlock does still, in fact, want that Screaming Rotweed. And you might convince him to teach you in exchange for the weed, which you will then seek in the Swamp of Slimy Horrors, with adventurous results.

Or maybe not. Maybe you'll agree to kill the Sage, but just prior to the killing make a false promise to spare his life if he gives you his spells. Or maybe you'll offer to dispatch a different rival who means less to you. It all depends on the scope and flexibility of the game engine, but these kinds of options are exactly what an IS engine is designed to provide.

As we know, a modern interactive story is not really a story at all, but a social simulation in which stories arise naturally as players choose their own paths.

[In fact, I believe the term “interactive storytelling” should be reserved for games in which the social simulation is enhanced by a “Virtual DM” (Dungeon Master, Drama Manager – take your pick) who follows and adjusts a player’s journey in order to promote the dramatic structure and suspense-arcs we instinctively associate with storytelling.]

With a robust enough social simulation, you can have an economy where the participants can exchange not just gold for spells, but commitments for trust, trust for quests, quests for gratitude, and gratitude for gold. You'll still get your spell, if that's what you want, but the path will be more interesting, because you found it and chose it yourself.

Now, some of you will say that you have had these kinds of options before. Yes, I did play Morrowind, and there were indeed occasions where you could take a diplomatic tack, or a stealth tack, or a brute-force tack to increase the trust NPCs had in you. And this trust did indeed translate to new spells, items, or abilities. And they did indeed allow me to kick butt.

And in my growling gamer voice, I shout, "Yes! And it was good!"

But I also say that it was good because the world builders put in so many thousands of hours scripting up these specific situations. It was obvious that there was no underlying social simulation that could've spawned these options dynamically. The hand-sculpted approach has the advantage of allowing every situation to be rich in intelligent, creative details, if someone is able and willing to add those details. But the social simulation approach offers other advantages. The first is obvious: vastly greater numbers of situations. More interesting are the possibilities of tragic, humorous, or convoluted situations that make perfect sense within the simulation but would never have been anticipated by the designers or the players.

In our SSERPG (Social Simulation-Enhanced Role-Playing Game), It turns out that Merlock didn't even know the ‘Magma Mallet’ spell. He lied because he wanted you to think more favorably of him, because he knew you had the trust of the weaponssmith, who's daughter's hand in marriage he is seeking. Merlock the Magnificent is a con. He asked you to kill Splitwick, because the Sage knows he is a con, and has been using this to blackmail Merlock to the sum of 100 gold monthly.

It all makes sense now, doesn't it?

But maybe it sounds too good to be true. Perhaps you are thinking, like many game developers past and present, that no social simulation could never create this situation without the management of a human-level AI.

If so, stay tuned: In Part II, I show you how its done. Mostly.