Non-linearity in a cascading tag cloud – 10-20-09

Putting this up raw, without a second thought, more in the nature of online archiving than for presenting as a polished bit of thinking.  I was mostly concerned with trying to untangle the threads that were winding about.

It feels like there’s gold somewhere in here…much slag to clear away first.

So, first write the whole story, many threads, typical KSR, but broken out into their separate threads.  Duplication where overlap, rather than shorthand.  Then, tag the hell out of it, both hidden and visible.  Define several meta-tags that will be prime orgainzers, and several hidden tags that will relate importance to a coherent narrative.  Player, at the start, chooses one of the meta-tags, and the game then performs a random sorting, taking into account the hidden tags to maintain overall narrative coherence, and giving the player a unique balance of the meta-tags that will create a specific narrative.  This isn’t branching narrative, nor parallel branching narratives:  it’s narrative defined by a central set, and many other partially overlapping sets, with each moment of interaction defined by the tags that connect them, however closely or loosely.

The difficult equations come where sets/tags intersect, depending on what other values/sets/tags are in this particular instance of the overall narrative.

Possible that each decision moment reshuffles the weighting of various tags, which would also require a redefinition of those intersection moments based not just on current states but past shuffles.

And branches—because those are unavoidable—are determined not so much by specific user choice as by the collocated value of the intersection where the choice is made:  that’s what determines the weighted values of the reshuffling.

Here’s the example:

3 very basic meta-tags for our prime story, election of a new mayor:  Government, Press, Business.

Each meta-tag has anywhere from a few to a few dozen possible story threads within them.  Government allows you to pursue a role as a candidate (for which party, or a fringe 3rd party, or independent), or an advisor, or lobbyist, or independent PAC leader, or other governmental employee (Police, Fire Department, etc.), etc.  Journalism allows you to pursue a role as a reporter (beat or investigative…and at which paper/tv station/website), columnist, editor, business/advertising/marketing scum, etc.  Business allows you to pursue a role as a real estate developer, organized labor leader, governmental sub-contractor, etc.

Each of these story threads is conceived and written as part of the larger, overall story of the election…and realistically, there are only a few possible outcomes to the election, and for each of the roles depending upon (or independent of) the election’s outcome.  The non-linearity being pursued here is not the exact final destination, it’s the path taken to get there, and the unexpectedness of that outcome depending on what’s come before.

So, you’ve got these meta-tags, like so:

non-linear meta 1

And then each meta-tag has a whole bunch of sub-threads, like so:

non-linear meta 2

And of course there’s a wide variety of possible ways each sub-thread could go:  an advisor to a candidate could be angling to help win the election, so they get an appointment in the government, or to get a job with a business that supports the election, or to tank the election in hopes of a position with the winning side afterwards…a labor leader could decide to run as a 3rd candidate, a blogger could angle for a position in one of the campaigns…these possibilities would have to be somewhat limited, just for feasibility, but that’s more a function of the initial concept for the overall narrative (i.e. choose something somewhat simple) than the inherent nature of the choices.  You want the player to not feel restrained by the choices available to them—if they can think it, they can do it—you just want to make sure that the amount of possible choices and thoughts are reasonable based on what initial states are given to them.  For example, the deputy chief of police could angle to support one candidate to get a promotion, or another because he’s sincerely in favor of his policies, but can’t decide to quit his job and join a bluegrass band…because the game’s about an election, not a damn Altman film.

So here’s where the non-linearity happens:  within each sub-thread, each and every moment where a choice is made that is more affecting their course than “turn left” or “turn right” (though in the proper circumstances, those could be profound choices), is tagged with a variety of values:  not numeric, at least not at this point, but contextual.  So, you can figure that 99% of the possible moments within the game will be tagged with “Money”, or “Power”, but only some of them will also be tagged with “Zero-sum outcome”, and even less with “Positive-sum outcome”…while all possible moments will either be tagged with “quid-pro-quo” or “selfless” or “advantageous”, but only one of those three.

Each moment will have multiple tags, as many as makes sense.  And it is the resulting tag cloud that will impact the progression of the player through the narrative.

The player will choose initially one of the three meta’s, and a random shuffling of all tags will occur, with weighting being given to those most prominent within the meta they chose.  This shuffling determines what possible moments in each sub-thread are available to them (the hidden “coherence” tags determining how likely one is to be kept or not”), with the outcome being weighted but inherently random.

Each moment they make a choice, the tags that apply to the chosen option will be given higher weighting, and those that apply to the non-chosen option will be given lower weighting…a lot of this will cancel out, as both possible choices might have the tag “Money”, so while it will both be higher and lower weighting, regardless of what is chosen, those will cancel each other out…but other tags will not…for example, “Money” might cancel out, but “Profit” might only be tagged to one of the two choices…and “Long-term Profit” vs. “Short-term Profit” would be a black-white type choice…or both options could be tagged with “Short-term Profit”, but only one with “Long-term Profit”…the key is to make sure that all possible tags are meaningful, and properly applied.

Then, at key moments where larger decisions are being made—the first primary election, or the first debate, or what have you—the accumulated value of all weighted tags is used as a modifier to reshuffle the basic tag cloud.  So it’s possible that you could play through the game twice, making exactly the same decisions each time, and get different meta results each time, or you could play through twice making entirely different decisions each time, and end up with the same basic meta results each time.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s