Choose Your Own Adventure

“Choose Your Own Adventure” was a series of paperback “interactive” stories published by TSR in the 80s. The basic idea is you take a story which branches (based on reader choices) and then scatter the pieces into a book, but put the starting point first. When the reader is faced with a choice you tell them which page (or paragraph in some cases) to go to next.

This idea did not originate with TSR (of course), and in fact TSR’s books weren’t even close to the most interesting expressions of the general concept. I’m hardly an authority on the subject, but there are two standout examples: GrailQuest and Alone Against the Dark.

GrailQuest

The idea in GrailQuest is that you’re a knight in search of the Holy Grail. Arthurian England is navigable by following paragraph links (and it manages to be something of a “sandbox” game despite comprising a meagre 132 paragraphs. This is because GrailQuest allows you to learn the landscape and move around in a consistent manner, and provides meta-rules for using the content, including things like random encounters, and assumes you’re willing to administer the combat rules yourself (which works just fine). In fact the rules content of GrailQuest is larger than the content itself — there’s a lesson to be learned here.

Amazingly, GrailQuest not only worked as a multiplayer “headless” scenario (i.e. you could play with friends and no GM) but I actually ran it as a Chivalry & Sorcery 2 campaign. I simply changed the encounters to match the new ruleset and embellished the descriptive content. It’s truly a remarkable piece of work. (Who knows? Maybe the author will see this and feel warm inside for a minute or three. Every time I get an email about Prince of Destruction it makes me happy.)

Alone Against the Dark

Alone Against the Dark is simply a tour-de-force. I can only assume that it (along with the nearly-as-good Scorpion Hall) made very little money (or was incredibly hard to write) because there were never any follow-up products. AAtD manages to recreate the spirit and flavor of a Lovecraft story, perform the neat trick of embedding a detective story in a solo adventure, and break all kinds of new ground in terms of solo game design, and offers many interesting lessons for game designers. I can’t remember the arc story — you’re tracking down a missing person or something and then it’s like every Lovecraft story — relics, cults, ancient gods, death, insanity.

To begin with, AAtD manages to have sandbox qualities to it, although unlike GrailQuest it isn’t one large sandbox. Instead, AAtD has an overall arc structure that’s basically one-way, and then individual areas are sandbox-like (such as when you’re essentially free to “explore” cities). True to Lovecraft, there’s ample opportunity for your character to die or go insane and then be followed by another investigator (who is assumed to track you down via diaries, etc.).

The key mechanic that AAtD introduced beyond GrailQuest‘s sandbox + arc structure was to provide essentially open-ended exploration of large areas. You might find yourself in a city (Arkham, say) and be told that there are a bunch of points of interest and you can either visit them or phone them, and be given paragraph entry points for doing so (and a time cost for doing so).  E.g. you might decide to phone a place to save time, but then end up having to visit it anyway because the phone call is so unproductive (or suspicious).

Indeed, a big feature of AAtD is the introduction of time-tracking and giving the player a huge sense of urgency. (The bad guys are plotting to destroy the world, of course, and you need to stop them. The slower you are the deeper the shit you end up in.)

Please bear in mind that I am writing all this from memory, and I haven’t looked at my copies of either of these games for at least fifteen years. GrailQuest I played in high school (the C&S2 campaign was in college) — so that’s over 25 years ago.

Writing a Solo Adventure

One of the things I really admired about Victory Games’s James Bond 007 rules was that they included a pretty good solo adventure intended to help introduce the game’s core mechanics, and it was not only fun to play but instructive. So, one of the first things I did after finishing my ForeSight rules was produce a similar solo adventure for ForeSight — Solo on Simiolus. (I see some references to Solo on Simiolus online, but no-one seems to have scanned it for me. I have soft and hard copies coming out my ears so maybe I’ll post it here I’ve posted a bare bones version with no illustrations here.)

Solo on Simiolus was smaller than but structured similarly to GrailQuest/AAtD — there’s an overall sandbox structure and lots of entry points. The idea is that you’re a consular official working for the Federation (part of my ForeScene: The Flawed Utopia setting — think of it as being like a less powerful version of Iain Banks’s Culture) when the planet you’re working on collapses into civil unrest, and your job is to track down some missing personnel and get them to the “enclave” (think “Green Zone”) ASAP. You can simply do the minimum (rescue the two missing people) or you can poke around and figure out who’s behind the unrest.

The way I wrote Simiolus was simple enough. I just wrote a “node”, decided what the reasonable options were, and inserted placeholders for each one. I kept going until I was done. And then I randomized the nodes and renumbered them. I did all of this in a long evening (and then spent a couple of weeks fixing mistakes). I don’t think I wrote a masterpiece, but it’s not bad and — I think — better than the James Bond 007 adventure at least.

Lessons for Modern Game Developers

If you were to represent a “quest” from a modern RPG as a solo adventure it would probably have at most 10 nodes and perhaps three branches. (Most nodes would simply be something like, “You ride your horse to Olduvai Gorge, kill twenty albino cheetahs, and skin them. Go to page 7.”) Quests in WoW or EQ would have no branches at all — you’d need to go to a truly outstanding and original game such as Fallout 3 to actually find a quest that can have more than one outcome.

I’ve quoted (or paraphrased) this line many times, but some guy at EA defined a good game as being “an enjoyable activity and a reason to keep doing it”. It’s one of those definitions I found instantly irksome and then liked more and more as I thought about it. Sure, some games have more than one enjoyable activity and/or more than one reason to keep doing it and blah blah blah but yup, that’s it. In an MMORPG the “enjoyable activity” is generally “killing stuff” or “dressing your doll” or sometimes “exploring”. The reason to keep doing it is sometimes story, but more often than not it’s “getting prettier things to dress your doll with”.

My ideal is for “story” to become a major component of CRPGs (massive or not).

Thus far, story tends to fail in single-player games because the story is static and more-or-less linear. In MMOs it fails completely because there’s so much time to fill and since players are focused on level and gear that they become frustrated by any attempts at inserting story and route around it (e.g. by reading spoilers or just skipping it). The latest incarnation of WoW so greatly facilitates story skipping (e.g. by mapping out the places you need to go to finish quests for you) that it’s often a rude shock to actually have to read quest text to figure something out.

GrailQuest is a stunning example of how a very simple piece of content can function as a huge “reason to keep doing it”. But in order for story to become important it has to become a player goal — if the player doesn’t play for the story, why have a story? And players will only play for a story if the story is interesting. And, by and large, a story will only be interesting if the player can actually impact it. And for the player to impact a story, it has to be able to change based on player actions.

The Gold-Plated Tunnel of Fun

One of the major reasons for linear plots in games is the cost of creating assets. It goes something like this: you want the game to look good because graphics sell games. But it costs money (and time) to create nice graphics and you only have so much. It follows that you want to create a game that looks as good as possible for as little as possible, and you want players to see as much of the good stuff you’ve created as possible — otherwise the time and money are wasted. It follows that any branch in story will lead to unseen content and is thus a bad thing.

Perhaps the most surprising case-in-point is WoW. When WoW “classic” first shipped it had two continents and two factions. It was perfectly possible, and in fact highly likely to go from level 1–30 with one character seeing one set of content, and then play another character of the same faction and see a completely different set of content. (It’s quite common to encounter veteran players who’ve never huge swathes of classic content.) But as more expansions came out the options narrowed, until both factions would get channelled through nearly identical content.

Today, a typical level 60-85 “zone” comprises a bunch of common quests along with a Horde base and an Alliance base (and maybe some Horde, Alliance, and neutral quest givers). Anyone leveling in that zone will basically do the same quests — Horde or Alliance. Indeed some of the newer zones are so linear that everyone does pretty much exactly the same set of quests in exactly the same order. And, not coincidently, the most offending such zone has the most cut scenes in it.

None of this is a new phenomenon. I’ve always found games where memorizing levels is the most important skill in getting good at the game to be stupid, and there are many such games. Others clearly disagree! (Interestingly enough, my favorite old school video games were always dynamic in content — e.g. Time Pilot, Q*Bert, and Elevator Action.)

Aside: a Twitter comment suggests that Elevator Action is static in content. I strongly disagree. The building layout is fixed, true, but the location of the red doors (which you must visit to collect secrets and complete a level) is dynamic and so is the behavior of enemies (unlike, say, Xevious or 1942, where everything basically flies patterns). In fact, the random door positions have such a strong impact on gameplay that it was quite a while before I realized that the building layouts were static, and when I mentioned this to a friend (we used to sit and play Elevator Action solidly for hours on end) I had trouble convincing him of the fact.

(If you’ve ever played Elevator Action you’ll almost certainly recall that if the door that’s alone in a small room and accessible only by escalator is red this becomes as hard to deal with as rest of the level by itself. Similarly nastily placed doors in the final “all elevators” section of the building.)

I don’t think anyone would have any trouble recognizing the static nature of WoW quests and dungeon content. Even those minor random elements (e.g. some WoW dungeons had random “rare spawns” in them) have virtually zero impact on gameplay.

Randomly Generated vs. Static

I’m not arguing for “randomly generated content” games. (I just discussed some of these thoughts I’ve been havingwith a WoW-playing colleague and he raised Minecraft as an example.) Ultimately, those are every bit as annoying as completely static games. The key is compromising somewhere in between, rather than going the easy route. To put it simply: GrailQuest manages to produce a highly replayable and satisfying role-playing experience with a little over a hundred paragraphs by adding a few pretty simple “meta-rules”. A little bit of randomness added to a whole bunch of static content adds a huge amount of replayability and interest — it leverages the effort of all those expensive game designers and artists and keeps players paying subscriptions.

EverQuest’s “Lost Dungeons of Norrath” expansion was the best part of EQ despite having many obvious (and easily fixed) flaws. Basic idea: you go to a quest giver who says “I need you to go to some place X and kill some stuff Y | get some stuff Y | rescue someone Y” where X is a randomly selected static dungeon map (with static population), and Y is a randomly selected target, thing, or named NPC. I think I had more fun playing in LDONs than any other EQ content despite their “afterthought” quality. If even a small amount of attention had been paid to making it more interesting (e.g. better/more/randomized maps, better/more/randomized content, more variations on objectives, and more care on doability — low numbered collects were usually impossible) it could have been absolutely awesome.

The first two “Elder Scrolls” games were Arena and Daggerfall. Arena had a huge world with static “high level” detail (e.g. city names and locations, overall geography, a few NPCs) and all the detail randomly generated. There was one arc quest line and then all the other quests were randomly generated. Daggerfall was essentially the same but with horribly confusing dungeons. Both games were, modulo graphic quality etc., more fun to play and far more replayable than anything Bethesda has done since (at least relative to the time of release). It’s easy to imagine Arena scaled up to modern computer capabilities being an absolutely compelling game even without improving on its pretty silly random quest generator (“Go to the [Thieves’ Guild] and kill [10 rats]”).

The amazing thing to me is that no-one seems to go the extra yard and try to produce random/dynamic content that’s just a little bit interesting and just a little bit tied to the underlying game world. Here’s a really straightforward example: in the latest WoW expansion there’s a quest line you have to enter in order to visit one of the new zones (Uldum). At some point a little ways into this quest line you need to kill a bandit. He’s basically sitting on a camel in the middle of the desert (helpfully marked on your map — I shit you not).

This is a classic WoW “assassinate” mission. I very much doubt that one in ten WoW players will read the quest text, know who he is or why he’s being killed. You literally run up to a quest hub, click everyone with an exclamation point, then check you map and start running to the different nodes and killing/looting whatever the game interface tells you to. So this guy appears on your map, you run over, kill him, yellow completion text appears and a new question mark appears on the map, you run over and get some money and xp.

How might such a mission be improved by adding a small amount of randomness? I have to ask: how could it not?

Instead of there being one static target there are N options. Each player gets given one of the N as the target.

Instead of there being one static target with a fixed name, each player sees the target as having a unique name.

Instead of there being one static target with a fixed name, each player is given a unique name and the game engine has a chance of deciding that certain enemies will have been the player’s quarry after the player kills them (and the player is given clues as to whom to go after). This mission might be possible to fail (if someone else gets the target first) or the game engine might create new targets until the player is able to succeed.

Final Thoughts

Clearly there are plenty of people happy with mindless static content. There’s no reason for game designers to ignore them. You can put dynamic and static content side-by-side in a game and create both in-engine and organizational feedback loops to decide how much of what to put in the game.

(It occurs to me on reviewing this post that there are a tiny number of somewhat random quests in the game, e.g. the craft dailies, and perhaps these are sufficiently unpopular that Blizzard feels dissuaded from trying anything more challenging. Unlikely on both counts, though.)

E.g. imagine if there’s a “Thief Taker” in Stormwind who gives out daily quests for a fixed static wanted guy and a dynamic wanted guy, the second requiring more time and attention but providing a concomitantly greater reward. You could measure to see how many players choose to do one vs. both. You can dynamically change the awards based on how much time players seem to take to do the two quests. And finally, at an organizational level, you can start allocating development resources to dynamic quests if they generate more player engagement per hour of development and testing.

It’s not like WoW is a static game engine. The random number generators are hard at work determining AI behaviors and resolving combat outcomes. I don’t think that carefully adding dynamic aspects to content is going to make the game logic stupendously more complex or hard to test.

Different people play games for different reasons. I’m sure that one reason WoW has a pretty high level cap is that many players are trying to get to max level as their goal (either with one “toon” or as many as possible). For some players “reaching the level cap” is the game. For others being the best arena player or getting achievements or having the best gear or whatever is the goal. I’m an old school role-player which means that for me the journey is the reward. It follows that I want the journey to be interesting and — thus far — the journey in CRPGs, and MMORPGs in particular, is pretty awful. As far as I’m concerned, having a lower level cap would actually make the game more interesting because it would mean that more content would be available “a la carte”.

Imagine if WoW’s level cap were 20: then pretty much everything except starting zones could be accessible to players in “the end game” and there’d be no “oh it’s just a level 60 dragon, ignore it” nonsense. This begs the question as to whether having such a low level cap would lose players who were fixated on reaching “max level”.

WoW has something like 13 million subscriptions. It dwarfs its predecessor, EverQuest, at its peak by a factor or 20 or so, and as far as I know it dwarfs any single player RPG (at least in the Western world). Unless you count The Sims. (And what’s The Sims’s level cap?) So, I can understand if WoW’s developers think that they’ve found the “sweet spot” of the market where they can please the most potential players with the least investment in content. Let me suggest, however, that a lot more people watch TV or read books than play WoW, and most people don’t watch TV or even read books because they want to get to the end.

Figure out how to make the journey interesting and you might just beat WoW by a factor of 20. In any event, you don’t need to get a huge proportion of WoW’s 13 million players to justify investing in a truly better game. Currently a lot of people are hoping Bioware’s The Old Republic might be at least marginally less stupid than WoW.

P.S. similar thinking can be found here.

Foresight 2010

Foresight received a thorough revision in 2003-5 including blind testing, but the results were never published — although you can download a near final version here. I’ve considered self-publishing it through lulu.com, and just discovered that it’s very easy to publish to the Kindle (which I do not plan to buy until I find out what Apple has planned).

Back when the Newton first came out I planned to produce a version of Foresight for the Newton based on the deliriously optimistic idea that Newtons would become ubiquitous. I also started work on a pure HTML version of ForeSight some time ago.

The obvious thing to do at this point would be to convert the latest version of Foresight entirely to XML, and then build an XSL presentation layer on top of it — allowing instant delivery as HTML (which is also the best way of getting stuff onto the Kindle), and relatively easy porting to any other platform. The beauty of XML delivery would be the ease with which tools could be delivered or even embedded in the rules (e.g. the character sheet could simply be driven by XML skill and background data).

Foresight always pushed the envelope in terms of what was easy to lay out via DTP (e.g. its tables exceeded the maximum complexity supported by most word-processors of a given generation). Electronic delivery seems like the way to go — assuming anyone wants it.

Let me know if you have any interest.

Resolution Systems: Sweet Spots

Ideal resolution systems for RPGs ought to have a large “sweet spot”. What’s a sweet spot?

Well, if you use 3d6 (the way GURPS and Hero System do) then your “sweet spot” is the range of “difficulties” represented by having to roll 8 or less to 13 or less… The probability curve is very narrow, which means that in GURPS the range of human capability ends up being pretty narrowly defined (a character with an ability score of 9 is pathetic, while 15 is awesome). In Hero System almost anything you want to do is trivial or impossible.

The ideal system would be exponential in generating probable outcomes, something along the following lines: if score A opposes score B and they are equal, A has a 50% chance of prevailing. For every point by which A exceeds B, the chance of failure is reduced by 5%. For every point by which B exceeds A, the chance of success is reduced by 5%.

Well that’s RuneQuest opposed resolution, right? NO! RuneQuest success chance goes 50%, 55%, 60%, etc. I’m talking about 100 – 0.5 * 0.95^n where n is A-B. In other words, if A exceeds B by 100, going to 101 reduces the chance of failure by a further 5%.

Of course this is a pain in the neck for dice-based games, but it works jim dandy for computers 🙂