Post-mortem for Ma Tiger's Terrible Trip

Bottom line up front: if you’d like to make a multiplayer Twine game and want some help, talk to me - I’ll help you, and the code is (hopefully) (probably) usable without much more technical knowledge than it takes to make a normal Sugarcube game.

As for the postmortem - well, this is an unstructured jumble of words, not a postmortem, but carry on - my primary takeaway is this: if you want feedback on your game, post it in IFComp. There will be more, and it will be more critical.

I’m completely serious. I made Ma Tiger’s Terrible Trip as, more or less, a proof-of-concept, and as an experiment in seeing what kinds of techniques would work in a multiplayer setting. It was written in a little over a month, and so shoddily tossed together that the vast majority of the work was not proofread; I never did my customary out-loud readthrough. There was a fatal bug in the release version. I discovered this live as the members of the SF Bay Area IF meetup were playing through the game, which was kind of funny but also terribly vexing.

The goal of putting the game out in Spring Thing was to find out what seemed to work and what didn’t, and while the playtesters who helped kick it around gave some very helpful feedback, there wasn’t as much in the realm of reviews and the reviews tended to be too polite for my purposes, as befits the more encouraging nature of the festival. Now, some time after release, I’ve little more insight into what works in the format than I had at time of release (which is more than I had before I started writing, so lessons have still been learned). My conclusion is that, in Spring Thing, people are nicer than in IFComp. I will adjust my future expectations accordingly.

Anyways, postmortem. There will be spoilers.

Let’s see. Ma Tiger’s Terrible Trip is a real-time multiplayer Twine game in which the players play a brother and sister who haven’t seen each other in a while. Also, their mom’s doing shady stuff, and it’s very taxing on them. Everything in the game was written essentially off the top of my head; I went in with the following ideas:

  • The players needed to be nominally cooperative but also have friction between one another
  • There had to be a timed section, because I wanted to see how timed sections felt
  • There should be points at which each player feels like the other player made a conscious choice

Before, that, though, let’s get to the non-multiplayer stuff.

The Setting

A couple of reviews have commented on, like, the setting being cool, or seeming fleshed out. There’s no setting; it was just me going “haha that’s an amusing sci-fi thing I’ll throw it in.” King is a genetically engineered dog because I wrote the intro, I got halfway through the bar scene, and then I had to figure out why Ma Tiger’s companion in the intro might not understand her, so I made King a dog. Later, I realized that dogs don’t live very long so King would probably be dead by the time the game takes place so I was just made him a genetically engineered dog. It’s all that; cyberpunk pastiche, slammed together at the exact moment of writing with only cursory reconciliation. Ma Tiger’s car is simultaneously a relic from the pre-self-driving dark ages but also has a genetic lock integrated into its dashboard because I needed a last-minute excuse for the car not to work.

The Story

I made, in order, the character select screen; the intro screen; the title; the bar scenes, sans the Ebi/Jeku conversation; the chase scene; the endings; the Ebi/Jeku conversations. At no point did I know where the story was going, because I was focused mainly on…

Multiplayer Theory

How do you do multiplayer IF, and what is its purpose? What can multiplayer IF offer over a singleplayer game? What can it offer over a board game, or tabletop game?

My first attempt was a purely cooperative detective game with no intention of hidden information. Players interviewed different people, then were supposed to combine their information to come to a conclusion. Three issues: first, the difficulty was off. Second, there was no intention of hidden information, so it breaks down to sort of an escape room or gamebook vibe - which is good! people are talking to each other, it’s fun, right? but it felt like the kind of game where one person could ‘take over.’ Third, it wasn’t very narrative.

It also expected people to be on a voice call, which, eeeeeeeeh.

It was good, though! I think that the form could have legs, but to be honest, you could do it just as well as a gamebook.

So, here I thought: how about making this one narrative, and making it slightly oppositional? Not a lot, because you want the players to be having fun together, but enough that you can visibly choose to be nice or mean, and, importantly, that the other player can see what you do. Because the question is: what’s the difference between playing a game with another player, and playing a game with an extremely well-realized NPC? Well, in a lot of ways, nothing! If you’re not communicating outside of the game, the only thing that tells you your partner isn’t is the occasional “waiting for partner” prompt.

At the same time, you know that there’s another person on the other end, and that knowledge does some of the lifting.

The entire “so how you been” dinner scene was added after-the-fact because, before I added it, some playtesters were sometimes going “it’s cool but it’s not that different from playing a game with a cool NPC” - the idea was that by pushing a very obvious “you have 3 choices and they probably have 3 choices and sometimes you say yes/no” it feels like the other player consciously picked a choice - even when they didn’t. I actually think this went really well because when the SF Bay Area meetup group was playing it the player controlling Ebi at one point said over the call, “Oh, it railroaded me into picking that.” So, I think that worked well.

The timed portion, less so, because what I ended up doing was completely splitting the players up. This is because of technical and design limitations, not because I wanted to - originally I wanted each player to be able to tend to King and give chase to the attackers, but differently (Jeku could drive the car, Ebi could run, for example) and the idea was that the timed section would push them to coordinate. The issue with that is that reconciling two distinct player tracks which could intersect at too many different points was just too hard; what happens if Jeku hops in the car while Ebi is halfway through the chase? If Ebi catches up and is fistfighting the kidnappers, do I have to write a whole other set of text for Jeku when he arrives? Sufficient massaging could solve this problem but eh, don’t got time; I cut it so Jeku can’t chase and Ebi can’t do first aid. Therefore, the vast majority of players will hit Ebi’s chase action before the scene times out and almost everybody got the “Ebi catches the kidnappers” ending in playtesting.

As a consequence, Ebi basically got most of the plot advancement. Eh.

My main takeaway here is that giving two players a buffet of options, all of which impact shared state, means you gotta do a lot of extra writing, and anybody making multiplayer games where that comes up should either avoid it or go wild on tricks.

A Note On Popularity

Looking at the stats, not many people played this game. That’s of course deeply unsurprising - a multiplayer game needs a partner, and unless you have one already at hand, you’re unlikely to go through the trouble of finding one. Compounding the issue is that there’s little to no replayability for most IF, and certainly not for such a narrative-heavy work as this. I think if a similar game gets made doing something like an in-game signup sheet might be helpful, but it’s definitely going to have a smaller audience.

My Takeaways

  • Release stuff in IFComp if you want people to look at it with a critical eye
  • The technical aspects of multiplayer weren’t as hard as the writing parts (but only because I’m a software developer; this wouldn’t be true for the vast majority of authors)
  • Branching is a huge problem in normal choice games but it’s way worse when you try to offer choices to multiple people
  • Don’t have two players trying to mess with the same complex moving state; even if it weren’t real-time it’d be a giant writing snarl
  • Having the players have an in-character incentive for minor hidden information is funner than not

If anybody who played the game reads this and is looking at it with their head tiled and going “Well I don’t know about that part of the post-mortem” tell me!