I recently wrote about Bob Bates’ commercial parser IF game Thaumistry for PC Gamer. Bob was kind enough to speak with me about the project for context.
A couple of other observations came up in that conversation with Bob that couldn’t go into the PC Gamer article because they involved spoilers or too much detail about parser IF implementation, but I thought I’d discuss them briefly here.
I’ll do the spoilery bits last, with additional warning, for those who might not have played the game but intend to do so in the future.
- I also talked a bit about the game for Rock Paper Shotgun back when it was still in development
- Andrew Plotkin has shared his views on the game’s design
- Steam page with player reviews
- IFDB page, also with player reviews
TADS 3. Thaumistry is really solid from an implementation perspective — I didn’t run into any bugs in the course of play, and it feels smooth and clean. It is also not implemented in the style I usually associate TADS 3 games. Many authors who use TADS 3 do so because they like the detail and complexity of the provided libraries: the meticulous handling of senses and space, chains of implicit actions that are neatly rolled up and reported together, the provided conversation system that offers topic hints. (Mike Roberts’ own Return to Ditch Day demonstrates the possible richness here.)
Thaumistry isn’t really invested in having a very complicated world model like this. There are a handful of convenience verbs, like the ability to GO TO a location via the game’s main transportation method rather than having to manipulate the knobs and dials in detail.
(In fact, I would have liked it if the game handled GO TO more generally, for already visited locations, because a number of parts of the story involve revisiting a previous location to try out newly-acquired solutions.)
We now reach the bit with SPOILERS.
Endgame action sequence. There’s a passage near the end where the player needs to reach the ceiling of the museum rotunda, through a series of risky / foolish actions.
Puzzle action sequences can easily go really wrong: you want to maintain a sense of tension and forward momentum, so the puzzles have to be easy enough that the player doesn’t get stuck, but there also needs to be a sense of jeopardy… and if the player dies during an action sequence, they’ve often lost some of their sense of excitement on the replay. Often, what you want is an explorable sequence where the player always feels there’s a strong possibility of failure, but does not actually fail — and also never sees the game rearranging outcomes to prevent that failure.
On top of all of that, text makes these sequences even more challenging to construct, because an acrobatic action sequence requires the player to understand relative position and layout in a way that’s easier (usually) to communicate in graphics.
I can’t speak definitively to anyone else’s experience here, but at least for me, this section worked well. I knew what I had to do, and I was able to do it without messing up but also without feeling like the game had put me on Easy Mode.
(For other good examples of this difficult art, see The Shadow in the Cathedral.)
Characterization through play. Often, seemingly sympathetic characters would send the protagonist on a fetch quest before helping in really important ways: both Sarah and Theo the chief engineer demand additional help before doing anything to help save the bodgers, for instance. Theo asks that you get a collectable stock certificate, which requires you to commit a small robbery; Sarah, meanwhile, is concerned with protecting her husband’s legacy, even at the possible cost of lives for a lot of people in the present.
Bob described the decision about Theo as a pacing choice:
I think in the earliest designs of the game, the player would go to the Chief Engineer, who would just tell them what they needed to know, and the player would launch into the endgame. The pacing of that felt wrong to me. I felt that the player should spend a little more time at the beginning of Act 3 before being catapulted into the end sequences. So I needed some smallish activity that wouldn’t have them chasing all over the world, but would feel to them that they were closing in on the end, but not quite there yet. I also felt that putting a puzzle there would give the player some time to get to know Theo and bring out some more of the backstory of his relationship with Sarah and Henry.
All of that makes complete sense to me as a set of pacing decisions, but I did wish the narrative justification were a bit different, because what we’re finding out about Theo in fictional terms is at odds with his actual behavior in the presence. (Not quite ludonarrative dissonance. Maybe just narrative-narrative dissonance.)
Meanwhile, Sarah’s motivation is even more important. Bob invested most in Sarah of all the characters, and she’s meant to emerge as her own character, outside her husband’s shadow, in the course of play.
That means that this particular incident, where she is willing to harm others in order to preserve the memory of her husband, is potentially a key beat of character development. It doesn’t feel as arbitrary as Theo’s stock certificate request; it’s the kind of thing a real person might do. This character in general seems to care about people, so why in this particular case is she willing to harm the living in service of the memory of the dead? If we had the opportunity to explore Sarah’s motives further, that contradiction would yield nuance and depth.
But we have no opportunity to interrogate this motive or push back on it: Sarah sets the fetch quest, and that’s that. It’s a design move that keeps her firmly in adventure game territory, rather than feeling like a fleshed-out person.
The dog. There’s a bit in the game where you need to interact with an invisible dog, and it’s a guess the verb puzzle intentionally: can you think of three dog-related verbs to perform in order to make the dog like you?
For my tastes, this was one of the best bits of the game: a little unusual, playful, and in character. (Or, in my schema of things I value in puzzles, it hit the targets of Fairness, Originality, and Narrative Integration.)
This worked because the coverage of dog-related verbs was pretty solid, as far as I could tell — nothing that I tried was misunderstood by the game, which is critical for something like this. Usually guessing the verb is a bad thing, but there are games that make it work when they constrain the guessing space and provide lots of support for alternate inputs. For instance, Ad Verbum gives the player constraints like “verbs must start with a given letter to be understood in this room”.