A Chinese Room Postmortem

Firstly, sorry this is so late! It’s been a hectic time (whoop whoop Seedcomp!) but I think I’m finally ready to write what I think will end up being a relatively short postmortem for a relatively long game EDIT: Morgan Freeman style narrator voice “It was not a short postmortem”.

There is very little I can talk about without spoiling some or all of the game. Ultimately I’m not sure how much that matters, as I think the main ‘reveal’ is the bit of the game that I’ve bungled the most, judging from comments and reviews. So in what follows I’m just going to go into all of it. I’ve sectioned the bits off below, however there will be story spoilers throughout. But before people immediately turn to leave, I just wanted to say how much all the feedback for ACR (A Chinese Room) has meant to me. Writing this game felt important to me, something I was called to do, something that I felt was right to do on a moral level. With everything else I have written I was, on some level, adding to a portfolio. ACR wasn’t like that. I wanted to make people think, maybe differently, maybe just think, about something that I believe it’s important that we all continue to think about. Did I succeed in that? Was it as needed as I thought it was going to be when I started writing it? I’ll talk about that below. But either way, this game meant a lot to me and it was incredibly gratifying to see people really giving it care and attention. Thank you all.

(2 other quick notes:

I believe that readers are a huge part of the creative process. Everything I’m about to say was merely intent. If that’s not what you read in the work, then it’s not true of the work for you. I’m not writing this to say what you should think, what is ‘right’. I don’t believe there is a ‘right’. What I tried to put in and what you actually got out need not be the same thing and I think that the latter, for each reader, is the only thing that really matters.

Secondly, I’m going to get really tired of writing (see below), so if you’re curious about a thread I leave hanging, hopefully I’ll pick it up later. If not, nag me about it.)

The ‘reveal’.

The ‘reveal’.

Several comments and reviews have said that they think that ACR would have worked better as two separate games. Maybe that’s true, but that was never the intention. The central idea around which I built the game was that of a government using Searle’s chinese room set up to attempt to pass the buck of the moral responsibility for illegal actions they carried out, on to someone else. This was one of those ‘3 in the morning just trying to fall asleep’ thoughts. The next thought that followed, as I opened my eyes to stare at the ceiling, was that this might make a compelling core around which to build my next 2-player game. The third idea, fully awake now, was that I had to set it in Russia. In retrospect, there is a correlation between how awake I was and the quality of the idea. I’ll get to the other two later, but I’ll start with the first one.

ACR was always supposed to be 2 stories in one, or rather, 2 stories which, when placed together, revealed the larger story which they both made up. Like two halves of an ancient pot glued back together. They were not originally intended to be played separately from each other and it was intentional that key plot questions could only be answered if both parts had been played. But since this is a postmortem, I’ll start by doing the opposite of what I did in the game, I’ll describe the pot rather than the pieces (or rather, what I first designed the pot to look like):

Unnamed country A has just conquered unnamed country B. In order to facilitate its occupation of country B, country A has produced a list of potentially disruptive elements within country B’s citizenry. In order to both test whether the people on the list are actually worth executing, and also to obfuscate responsibility for crimes against humanity, country A sets up a black-site within country B’s borders. In this black-site, military operatives question the people on the list and then pass their sentencing recommendations to a computer system, which then encodes them as colours and sends those colours to a citizen back in the homeland. This citizen is not told what she is doing. She is not told, explicitly, what to do. She is merely placed into a room which contains the suggestion of guidelines for what buttons on a number pad she should press when presented with certain colours. Unbeknownst to her, each number is associated with a certain action (release, torture, execute) which will be carried out against the person on the list. Since the designers of computer systems cannot, in certain circumstances (self-driving vehicles) be held responsible for certain outcomes of those systems, and since what country A has created is in many ways identical to a computer system, as outlined by Searle in his philosophical Chinese Room argument, country A is not responsible for the final actions that result from the citizen’s choice. At best, no-one is responsible, but at worst, responsibility lies with the citizen, who, of her own free will, pressed the buttons.

This was the original premise of the game and was supposed to only be fully visible to players who played both halves. This failed for two reasons:

  1. Anyone who played Leon would have basically all of this.
  2. It’s dumb.

1 isn’t a major problem. If someone is playing single player, I recommend that they play Caroline first. That way the player starts on the side with less information and then moves to the side with more. For people who start with Leon, the experience of reading Caroline is then changed from a slice of life/love story with a mysterious element, to a tragic story of someone doomed from the start and through no fault of her own. I honestly don’t know if I succeeded in this. I am only aware of one reviewer who read Leon first and I don’t believe he has read Caroline, or at least he hadn’t when he published his review. I’ll leave it as a question for the reader.

2 is a much larger problem and one I only fully realised over half-way through writing. The plan is, fundamentally, stupid. The kind of thing a young philosophy graduate would think of while lying in bed half asleep. No-one would take it seriously for a second for so many reasons that I can’t even be bothered spelling them out. The concept is wholly absurd. I probably shouldn’t even admit that I started writing this project playing it straight, but if you’re reading this you’re interested in the process so I owe it to you to be honest about it. It took me too long to realise quite how fully ridiculous this ‘plan’ was.

So I thought of a fix. I think I stand by this fix. It filled in the game’s most major plot hole (the plan just being really stupid), it reinforces one of the central themes and it, hopefully, adds to the horror of the story. The problem is, I executed this fix really, really poorly. I hid it: the final twist which explains everything that is going on is only really vaguely implied, and even then, only in some of the endings. I think this is why even the reviewers who gave ACR more time than I could ever have hoped for missed it (I don’t believe anyone commented on it?). So here in plain english is the ‘pot’:

The government of unnamed country A has just conquered unnamed country B. It has already committed multiple war crimes and in facilitating its occupation it intends to commit many more. It knows that the media, both at home and abroad, will find out. So it designs a plan, a farce, something so complicated, absurd, and unjust that that the global press will eat it up. They will, in a maddeningly complicated way, frame a more or less random woman for a small handful of extrajudicial killings. Obviously this is a bad look for them, but it is nothing compared to the other things they have done and are planning to continue to do. So leaking this ‘made for TV’ scheme allows them to control the narrative, making them look simply corrupt and mad, as opposed to totally evil.

In other words, the entire thing is a set up for a show trial which the government will use to distract international attention from all the much worse things that are actually happening (as a slightly self-defensive side note, I found it interesting to see that the first thing to replace the Russia-Ukraine crisis in my newsfeed was the Depp-Heard trial).

I genuinely don’t know if this is a good final twist. I feel that it is. But I do not know, because no-one commented on it, because, I believe, no-one realised it was there. Perhaps this is an artefact of interpretation, but I feel in this instance it is more likely simply the result of bad writing. This twist was a relatively late day addition. It only really appears in one of Caroline’s endings (be accused of war crimes and flee with Matteo) and the process of it, without the explanation, appears in only one of Leon’s ending sets (those where he is discharged), where it receives less than 200 words (in a 101,000 word game). In writing this I realise that another of Leon’s endings (the one where he flees with Hayes) would have been the perfect opportunity to show the actual abuses ‘country A’ is committing, but I missed that opportunity entirely. I have excuses for this, but none of them narrative. I am not sure if I will bother going back to ACR and updating it, but if I do, making this final twist explicit will be the main change I will make. I don’t even think it would be particularly hard.

What this all means is that, despite some very interesting speculation on the nature of the legal argument made in defence of the use of the Throne, I don’t want to talk about it any more than I have. @DeusIrae, I loved reading your actually informed thoughts on this. I know that it is probably very disappointing to hear that the author’s own thoughts on the plot point are: it’s dumb. But there you go.

2-player vs 1-player.

ACR was always going to be 2-player. Anyone who played Alexisgrad (my IFComp entry last year) will know that this is something that interests me. I won’t go into the details about that, if you want them you can go read my postmortem for that game (which, as it’s more than 1 year old, I am too embarrassed to reread myself): Alexisgrad postmortem. I do see, however, that the only review I linked to in that was Sam Kabo Ashwell’s, which fits because it was that review which informed the main design choices of ACR.

Ashwell was the first of several reviewers to sort of imply that there are two separate ways that 2-player IF could be played, which I’ll call high interaction and low interaction. High interaction games are ones with rapid fire interaction, best experienced by players who are communicating quickly, like on a video call. Ma Tiger’s Terrible Trip and the conversation endings of Alexisgrad are examples of this. Low interaction games, then, are ones with much slower or fewer interactions, best played over message or email, since there would be long, painful stretches of silence on a call. ACR is one of these games (if you’re interested, I believe most of Alexisgrad was mid-interaction, which ended up just inheriting most of the worst of both types). Why did I pick low-interaction? Firstly, it better suits my ‘walls and walls of text’ writing style. Secondly because it’s much easier to do: fewer interaction points means less branching. But the main reason relates to another point Ashwell made: the openness of information. When I released Alexisgrad, it said “do not share information with your partner” during the instructions. Ashwell thought this detracted from the experience, as hidden information was neither important to the plot or the mechanics and it simply meant that the two friends who got together to play a game were pointlessly told to have less fun chatting together. It was a good point and perhaps the obvious response would have been “oh okay, I’ll make sure my next game encourages/allows open communication”, but I guess I’m just a problem child and I saw it as a challenge.

So ACR is a low interaction, hidden information (although I believe I never explicitly say that in the instructions…) 2-player game. But what does this add? I’m not sure. I know what it was supposed to add, but I don’t have enough feedback to know whether it succeeded. On the Leon side, it was supposed to create a feeling of responsibility. When you are required to choose a card which will be used to administer ‘feedback’ on your civilian, I had hoped that knowing that there was another, real human on the other end of that choice would make it more impactful. On the Caroline side, I had hoped that knowing that another, real human was feeding you these colours would increase the feeling of unease and paranoia. Did I succeed? Will enough people ever play multiplayer for me to have a large enough sample size to know (since I’ve had reviews that seem to imply one way and others that imply the other)? Who knows!

What I do think I know is that the other reason for making the game multiplayer did work. It’s fun to play with friends. That’s it. And it’s fun to discuss games, especially games with lots going on (like I hope ACR has). And especially when one player saw one thing and the other player saw something totally different. If you’ve read the section above, then you’ll know what I mean when I say that I like to think ‘putting the pot together, together’ is an enjoyable experience.

But what about the actual levels of interaction? Well they’re there. I’m just going to copy paste directly from something I’ve already written about this here

“On the Leon side, what happens to each of the ‘subjects’ is entirely dependent on the number that Caroline chooses. Of course, which number the player playing Caroline chooses is likely to be strongly dependent on what the Leon player chooses. I don’t know if either you or your friend played Caroline ‘disobediently’, but that can lead to the Leon player making choices and those choices not being carried out. Which in turn affects the Leon ending, as which ending Leon gets is (mostly but not exclusively) based on what happened to each of the subjects.
Caroline’s ending, meanwhile, is much more dependent on Leon. There are 4 endings for Caroline (given in-text differences it’s not so clear cut how many there are for Leon), with each run giving the player the choice of up to 2 of them, but which 2 depends entirely on what ending Leon got. It sounds like you got the same (more likely) choice both times. (sidebar: it seems that the players who are playing single player are getting the other choice (where Caroline is never accused of war crimes) disproportionately more often. I’m not sure if this is a good or a bad thing but it was an unintentional outcome of my use of RNG in the single player version). Other major plot points in the Caroline story (such as Daniel being arrested) are also dependent on choices the Leon player makes.
I won’t say any more than that this (im)balance of power was intentional.”

A few months on, I’m still not sure where I stand on that quirk of the RNG. In hindsight, I think that the ‘Caroline is never accused of warcrimes’ is strictly the worse of the two endings (least interesting, least thematic) and I should maybe have made it much less likely. But on the other hand, it already is very unlikely (there’s a just over 50% chance that a player who follows the guide for every instance in the room gets it, but that chance falls dramatically for a Caroline who is ‘disobedient’ even just once) and I like that they’re there as an extra set of ‘rare’ endings for players to get, which means that they have to be at least possible to get. And I do think that they have a certain horror of their own, but a horror that only works if you then either immediately play the Leon half and find out what you were unwittingly doing, or talk to your 2-player partner who just played the Leon half (obviously doesn’t work for single players, who are the ones most likely to get the ending).

I also think that there might be a problem on the Caroline side that certain plot points are not clearly related when playing single player. Most specifically, the outcome of Christopher’s party and what happens to Daniel (is arrested, is appointed to his dream job, or just carries on) is entirely reliant, in 2-player, on decisions the Leon player makes. That means that people playing 2-player are more likely to pick up on something going on during those sections, as they will have had to stop play to exchange codes. People playing single player do not have that speed bump and therefore probably have no way of knowing that those outcomes have anything to do with what they are doing in the room. This is a problem I am unsure how to fix.

Practical problems.

One of the guidelines for the Interactive Fiction Competition is “Resist the temptation to submit an entry you know is unready”. You may be asking yourself: “Milo, if you’re able to talk about all these problems now, then why didn’t you delay release and fix them?” There are 3 answers to that: I didn’t realise they were there, I couldn’t delay this specific game, and I’m just not built like that.

I didn’t realise they were there: above the piece of advice just quoted on the IFComp website it says, in big, bold, underlined letters: “Playtest your game.” I wholeheartedly agree with this and left a whole month spare for this process. What I didn’t do was find appropriate testers. I am fully aware that many authors ask for playtesters on these forums. I did not. I love this space, but after having barely been here since the last IFComp, the idea of asking the real heavyweights to playtest my game was intimidating. And more than that. I’d written something political, something about a still live, still absolutely devastating political issue. My sleepless nights after having released ACR and before feedback started coming in wasn’t that people would call it badly written or a mess or any of that, but that people would think it was offensive, demeaning, and trivialising to the very real and very great current suffering of the Ukrainian people, potentially even that it glorified Russia by framing the narrative from their perspective. I thought people closer to the issues and places would point out inaccuracies, insensitivities, and stereotypes. Russia, its culture and history, is something that has gripped me for years and that knowledge was a strong part of why I felt like I had a responsibility to write about what is happening. But I am neither Russian nor Ukrainian and I am still not sure that it is appropriate for me to have written what I have (but then again, what was my alternative? To have sat by and said nothing? I believe I did right, but I was ready to listen to people telling me I hadn’t). I wanted to run it by people I knew first, people who I knew and could trust to be gentle with me when they told me that the last 6 months of my life were spent working on something I shouldn’t have made.

On a quick tangent here – this is also why I never explicitly name the countries. While I like to think that I am quite knowledgable on these subjects, it is as an impassioned amateur, a mere “Russophiliasis”. I am an outsider and not a formally trained one at that. So I felt the need to distance myself from the actual facts, which is also why I deliberately got some details wrong. This is also the main reason behind the names (although I also had the idea here of drawing parallels between Russia and the West, but this idea very quickly fell by the wayside as the setting became more and more reflective of Russia (especially in Leon)).

But returning to the subject of playtesting, the problem was most of the people I sent it to simply didn’t read it. My ‘first testing wave’ was 7 people, of whom 1 (the person who had already helped me edit it) read all of it and 2 only read the Caroline half. I could complain here, but I won’t. I am grateful that those who read it did read what they did. And I am grateful that, after I panicked just a few days before the comp and finally posted something on the forums, @AmandaB swooped in and tested the Caroline half. She offered to test the Leon half as well, but I didn’t want to take advantage, although purely selfishly I think maybe I should have. Her feedback was excellent.

I couldn’t delay this specific game: why should be obvious. ACR is a statement, one made out of the current political moment, about the current political moment, and for the current political moment. I wanted it to have the biggest stage I could give it, so I had to enter it into IFComp. To do otherwise would have been simply to let it instantly sink on Itch. Waiting for SpringThing would have been to wait too long. So IFComp it was.

I’m just not built like that: I like to finish things. Simple as that. I am acutely aware that no piece of art will ever be perfect. I could tweak absolutely everything about every piece of every piece forever. But I won’t, so I cut it off when it’s ‘good enough’ so that I can move on to the next thing. I’ve got more ideas than I have time and I want to give as many of them as possible a chance to exist.

The difficult to write positive part.

There is nothing I hate more than being positive about my own work (at least, outside my own head). So I’ll keep this short, but I think it’s important to note that despite the last almost 4 thousand words of self criticism, I am actually immensely proud of ACR. I find it very difficult to not simply dwell on the problems, but this game obviously touched people. Intellectually or emotionally, I don’t really know, but I’m not really sure it matters. Obviously I’m very proud that it did well in IFComp and I think that the Caroline story really does stand on its own quite well (although I do agree with reviewers, if it was a stand alone story, it would be best without the Chinese Room elements. But it’s not). And I’m very happy with my writing: I think I did a good job of characterising and humanising and I’ve been told that my descriptive writing was, at least in parts, good.

But the point of this work was to make people think. Maybe think about all the social pressures on the average Russian citizen. Maybe think about the nature of control and state manipulation of the population. Maybe think about the political, social, and religious reasons that this war happened in the first place. But most of all, just think about the war. I started writing ACR just a few days after Russia invaded Ukraine. At that time I thought that by the time I had finished writing the war would already be over and international discourse would have already have moved on. I didn’t think that that was morally right. I wanted to write a story where the horror is the way that war affects everyone, even the victors, even those on the sidelines, even those who feel they have nothing to do with it. That human suffering is not and should not be something we can simply look away from. That to understand an issue, one has to understand both sides, one has to listen, think, and empathise (which does not, crucially, involve agreeing). In a world that I assumed would already be forgetting, I wanted people to remember.

But, of course, the world hasn’t forgotten. The Ukrainians haven’t lost. My voice is one of far more than I worried it would be. ACR is a hopeless story about inevitable defeat, suffering, and how even when something is lost, it can still get worse. It’s pessimistic. We are far, far from the end and there has been far, far more pain, cruelty, and death than ACR even hints at. But that I’m writing this in December and saying that there is still hope; that is something I wouldn’t have dared believe when I first sat down to write.

What’s next for Milo?

Okay, pure self-indulgence now. What am I up to next? Well, on an artistic level, I’m going to find it difficult to go back to non-political fiction after ACR. This was a story with a message, something I felt I had to say. I’m foreseeing motivation issues with writing things ‘just for fun’. But then again, something like this, something I felt so obsessively passionate about, doesn’t, can’t, come along too often. And besides, I’ve had other logs in the fire and I do really feel like I need a break after 6 months of depressing news articles and weird research crawls through history and geography.

The one I’ve already started working on is a very weird TTRPG I’m currently calling Council. I built the first prototype in 2019 but shelved it due to lockdown but I’ve recently picked it up again. The players play as leaders of a city state, with each play session split between a more traditional (but still far from traditional) roleplaying section and the second a fully in character budget meeting for setting the political agenda for the city. The game involves swapping between multiple characters, blackmail mechanics, the world’s flattest tech tree, and rules for multiple full playgroups to play competitively or cooperatively with each other. All wrapped up in an optional setting containing hyper capitalist space centipedes and techno ferrets who are actively debating enslaving humanity.

I’m also still working on Last Night of Lotosk, a character-focussed Ren’py visual novel about a town trying to survive after the sun didn’t rise one morning. It’s set in the same world as Last Night of Alexisgrad, but has a very different feel (it’s neither military, 2-player, or Twine, although it is heavily choice based). If you’re interested in that, we’ve got a little standalone demo of it here.

I’m also working on a psychological horror-ish novella, drawing from my life experiences, about mental health and living with someone with CFS.

And I’ve got a piece of interactive non-fiction in the works about cosmology, ontology, and how they can be used to find a place within the universe, again born from my own experiences but in a very different, and much more hopeful and optimistic, way. I wanted to release it alongside ACR at IFComp this year, but ran out of time, so aiming for Spring Thing.

OH, and of course, how could anyone forget, Seedcomp! is happening! It’s very exciting and everyone and anyone should enter, still plenty of time to get your seeds in the first round!

(Also I keep toying with the idea of doing audience participation writing streams. I might play with the idea during round 2 of Seedcomp! although I’m also tempted to do it more fanfiction style. I want to put those thousands of hours in Final Fantasy XIV to use…)


Thanks for this! I’m glad to read you have more stuff in the pipeline. I’m also glad you sent out ACR even though you realized/felt it was unfinished, according to your standards.

Heck, I only got through the Leon part, and I enjoyed getting lost in the weeds there. But I think a lot of people got a lot out of what was there.

I was quite disappointed in myself I didn’t look into ACR more, and this speaks more to be trying to be completionist than anything else.

It seems like I say this about a hatful of entries in IFComp, but ACR is the one I most want to get back to and explore more, because the Caroline side should be quite revealing and rewarding to me. (There are other more comedic ones I need for other reasons.) So I’m glad you wrote this postmortem as a reminder to follow through. In the case of many IFComp entries, I worry I forgot to cover a few rough edges that stuck out or ask a few miscellaneous questions, but in the case of ACR, there’s a whole world still to explore.


Thanks for posting this! I enjoyed reading your process, and perhaps oddly, I think your decision to keep pushing ahead after the “this is dumb” realization was the right one: the “twist” doesn’t really make sense, but even acknowledging that it helps establish the Kafkaesque nature of the regime, and the personal-level stakes are established so well that the fact that the big picture kinda falls apart doesn’t matter that much.

(I will say, I did get the ending that gestures towards the idea that this was a giant troll/distraction, but I didn’t fully understand that that was meant to be a full replacement explanation for the convoluted I-can’t-believe-it’s-not-a-war-crime! rationale that’s directly stated – I think this is maybe because I seem to recall it being the case that commentators in Western countries seemed to be taking the argument seriously? Though maybe that was just meant to indicate that the distraction was working, now that I’ve seen behind the curtain a little bit more).

Anyway I think there’s a real dearth of serious, political IF – let alone serious, political IF that engages with the issues it raises in anything other than a highly-didactic way – so ACR was a real treat for me, and I’m very excited to see how all the projects you mention turn out! Definitely feel free to hit me up for testing next time (albeit I do usually need a little more warning than a couple days to turn stuff around :slight_smile: )

EDIT: one additional thought: you didn’t mention this, but as someone who played Caroline’s part in multiplayer, one of the advantages of it being a two-player game was the impact on pacing. Taking action but then having to wait to see what consequences were going to come elevated the sense of dread, and played over a couple of days, the game wound up taking over more of my head space than it would have had I tried to just bang through in one fell swoop!


I regret that I never tried the two-player implementation. I thought it was an extremely clever way to interlock stories, and to maximize the use of the dystopian setting.

I certainly rebel at your ‘dumb’ characterization. I would have embraced the solution you provided, had I found it. Even absent that, and while it didn’t work straight-faced for me, it did not feel inconsistent with governing contempt for the governed. I humbly submit there was another ‘spin’ available that was equally thematic: an autocratic government so sure of its power it belligerently created ‘alternate logic justifications,’ confident that its hold on the populace did not require even lip-service coherence. And being right.

Also, I am sorry you did not add your character work to your list of positives. When I reflect on ACR, the well-rounded and vivid character study of Caroline is a critical component of what worked so well!

Lastly, let me echo the call for MORE political IF! You created a worthy inheritor to A Mind Forever Voyaging. Hyperbole? Fight me, I will die on that hill.