Other Wise

Evaluating evolution

Note: The below was originally written as an email to a friend just starting out with an 18xx design project. I’ve copied (and lightly edited) it here for preservation.

More directly, I think you are losing sight of the game, or at least where the game lies within the larger system.

The 18xx are fairly simply structured from two and a half almost-complete games bolted together. There is a spatial route building game and a stock market game, and they are glued together with a creative destruction technology race to make one complete game. Each of those components is damn near a standalone game in its own right, and certainly other games have been made from just those subsystems (Greentown, Lokomotive Werks, etc), but at the end they are just mostly-isolated subsystems with the actually interesting bits in their connections to each other (the players sit in the connections).

The internal structures created by that glueing process are also boldly simple:

  • There’s an evaluation function which each company uses to generate a number using as inputs the network on the board and the trains in the company, and we call that evaluation function “running trains”. You can get all sort of complicated here about the details of track and tokens etc, but at the end of the day there’s just a function that does something and returns a number that is input into the stock market game. All the rest is internal implementation details. Sometimes the number is a bit special in that it expresses a liability for the president (emergency train buy), but it is still just a number communicated from the evaluation function to the stock market game.

  • The stock market game takes the numbers that the network game generates and thence changes its own numbers in well known ways.

  • And, as a back-flow, the stock market game moves some of its numbers back into companies (withholdings, etc) via the creative destruction technology race glue to change the inputs to the network evaluation function (train purchases and rustings).

And thence of course those new numbers go into the stock market game and so forth.

And so we have a triangle: The network game informs the stock market game which informs the technology game which informs the network game which…etc. And the players sit in the middle fiddling with the dials in all three sub-games: moving the numbers around in the stock market game (shares), moving the numbers around in the technology game (trains), and moving the numbers around in the network game (building/changing routes), all while the loop keeps cycling as a feedback loop, round and round and round.

Which I assume you already know — there should be no surprises there — but you might not have articulated in such a stark format. More usefully and more to my point, what this really basic deconstruction does is to highlight where the game is. The game is in the players fiddling at those three key junctures in the triangular feedback loop while the game spins underneath them:

  • Players can fiddle with shares

  • Players can fiddle with technology

  • Players can fiddle with the evaluation function inputs

And that’s pretty much it. Everything else is in the feedback loop orchestrated by the rotating sequence of SRs and ORs stepping around and around the feedback loop and relentlessly driving the game forward.

What this means however as a game designer, is that it outlines where your interests and activities lie. In order to do something interesting in the 18xx world you have to either alter one of the three interaction points (shares, technology, evaluation function) in a way that substantially changes player concerns, or you have to alter the properties of the feedback loop itself (1880 did this latter with its new intertwingled OR/SR model; 1846 did this by fundamentally changing the feedback loop of money with how its incrementally capitalised companies work; 1860 did this by allowing entities to enter, leave, roboticise and re-enter the player interaction-space, etc etc etc).

My general sense is that at the litmus-test level, in order for any change to be interesting, it must substantially affect at least two corners of the triangle in ways that provide both substantial opportunities and problems for the players to address. Just touching one corner isn’t enough, as that’s almost instantly an internal implementation detail rather than anything materially interesting. And so you need at least two player touch-points to change in a way that’s substantial and different and interesting.

But more usefully (I hope), that deconstruction provides a set of analysis tools and litmus tests for your candidate changes. You can look at any candidate change and ask how it affects those three contact points, how that change to those contact points significantly alters the three stages of the feedback loop and thence how it changes the player’s competitive lives. And if you come up with a good answer, your idea potentially has some good legs under it, and if you don’t come up with a good answer, then your idea is more likely just shuffling the deck-chairs around.

And, shrug, I find that useful, as it sure weeds out a lot of options that fiddle little numbers inside one of the touch points without actually doing anything structural. Oh look, now this little internal number that isn’t actually a primary contact point is a little larger or smaller or different or has little brass bells and is painted red…but everything else is exactly the same…and…this…is…interesting…WHY? I do that a lot, and then I slap myself on the back of the head, say “Doh!” and move on.


PS BtB this deconstruction has an amusing side effect of also dropping out the four basic types of 18xx by extrapolation — which is kinda cute and unexpected (by me).

Notes on a vantage

Three quick quips which I’ve had cause to reference multiple times and thus seem useful to keep stashed. First a comment on how and why I live:

Meanwhile the universe is interesting, delightfully interesting even, and I apply and dedicate my awareness and interest to that end. I assume, generously and with scant evidence, that it is a Good Thing to Have Understood, or at least to have Striven to Understand. I don’t know if that is in fact true, it seems quite likely that it isn’t, but it also seems a fine and possibly even wonderful conceit to live under.

And thence on the critical structures and requirements of something being a (viable) game:

That the answer is complex.

Primarily I see it as a function of cognitive ability. Oh, not in any elitist or superiority sense, but simply in terms of capacity. Tic-tac-toe is a fine game for younger children as they simply don’t yet have the cognitive abilities necessary for recognising what’s necessary to remove the game. It is rather less interesting if you or I played it. Similarly, the game of Go is very interesting for people precisely because we don’t have the cognitive capacity to remove the game. In short: in order to be a game the system as presented must fundamentally exceed our ability to comprehend the system.

How it exceeds us, by complexity for instance, or by stressing analytical forms we humans are inherently weak at like conditional probability, or by requiring modes of thought that we’ve not yet fully developed (and thus a lot of games for pre-adolescents stress symbolic thought as they’ve only just developed that capacity (it comes in around age 11)) — how it exceeds us really doesn’t matter, just that it exceeds us and is hauntingly close to the apparent edge of our capacities so as to provide the taunting illusion of almost-graspability. And it will remain a game only to the point that no matter how hard we study and analyse and work it, that we will still not fully understand the system represented or implied by the game.

(Somewhere in here there’s a fine rant that I’ll skip for now on the necessity for ambiguity in games, and how far too many so-called “games” are not in fact games because they don’t contain ambiguity: they just have game-states that are hard or laborious to parse)

Now a kicker in this is that most games rely on the fact that as humans we cannot completely model another human (of comparable capacity to ourselves). Godel’s incompleteness theorem guarantees that, and provides the primary reason I rarely ever play 2-player games. The problem here is that the modelling problem is also a cheap out for lazy game designers (ahem — there’s no subjectivity in this declaration, no sirree!). In this I find that the more interesting games exceed their players in both their systemic demands and in their demands of modelling the other players (and thus in the intersection of multiplayer interactions and the system, an implicit third factor).

After that basic, well, it gets murky and subjective and ever so much more complex. A lot of the subjective preferences there are arm-wavingly discussed in my profile text here on BGG. And, not to short shrift you too badly, I need to get ready for a gamesday with one of the top 18xx players in the world (Todd vander Pluym) who is in the area for a few days…and fascinating as this question is (and it is truly interesting), that’s a time-bounded opportunity and this question isn’t. Sorry.

And finally on the activity of playing games:

I find that face-to-face games provide context and variety which is largely not available in solo study. That alone more than makes face-to-face play worthwhile. Just participating in a game with other actors, and observing those actors, actors not in my mind, suffices a lot of my requirements for playing a game as you say. Much of anything I may think during that time or later will be catalysed and informed by my observation of those actors during the game — which is the great thing about playing games rather than merely thinking about games.