The Versu Galatea

In addition to Blood & Laurels, in the late days of Versu we built — and came very close to releasing — a Versu remake of Galatea. The idea was that it was a piece that some people were familiar with, but which could be more accessible in this form; and releasing it this way would tell an interesting story.

Doing the conversion was a strange project. For one thing, I myself have a kind of weird love-hate relationship with Galatea at this point — a lot of people love the piece, but it’s pretty much the first thing I wrote that ever got any widespread scrutiny. I would write it differently now, in many ways and for many reasons. Parts of it strike me as flippant, parts clueless, parts overblown. I’ve gotten some great fan mail, art, and even music about that game, and also more creepy and bizarre email than about anything else I’ve written. And I’m also grateful, as that single piece is probably responsible for my career, a lot of my friendships, even my marriage. I remember it fondly but I almost never replay these days. So revisiting it long enough to reimplement all the text in a new context was strange. I disciplined myself not to change too much of the original dialogue, even when it wasn’t what I would now write.

From an implementation perspective, it wasn’t difficult to move the text of the original game into a Versu format. Versu’s conversation implementation is strictly more powerful than the one in Galatea while at the same time being much easier to author; often all I had to do was strip the strings out of the original code and do some minor formatting in order to feed the results into Prompter, Versu’s dialogue-authoring tool.

There were a lot of other considerations, though. This wasn’t, it couldn’t be, the kind of port that preserves the gameplay experience of the original. The original is a parser-based game with a lot of noun-hunting, and people get stuck on it sometimes, not sure what to do next or how to drive things forward. Versu is designed to surface those affordances, not to hide them; to produce forward movement, not a halting sense of difficult discovery. Versu characters talk, a lot, even if the player doesn’t say much.

There were some tweaks I made in order to emulate a little of the original Galatea’s reticence, to provide some pauses. But this Galatea was forthcoming in a way that felt very different. In the early prototyping, Richard Evans remarked that the game felt somewhat less magical because of its increased fluidity, and I did know what he meant.

At the same time — this Galatea felt more active than the old one ever did. And there was also the point-of-view shift. The way Versu works, a protagonist and all the NPCs have to be instantiated as characters in more or less the same way. There was therefore a parity in Versu between the player and Galatea that never existed in the original code base. The original code models topics and things to say about topics; the character Galatea’s emotions and reactions are hung off of those, triggered by the player’s questions and gestures, and only ever very rarely by additional daemons that add one- or two-turn-later follow-ups. In Versu, Galatea and the protagonist were both modeled as agents with a range of social possibilities open to them, and it was a matter of run-time choice which of those agents were driven by the player. What that meant (among other things) was that I could make Galatea a playable character and go through the same scenes and the same dialogue as her, for the first time.

It was an astonishing experience, playing as Galatea. The protagonist came off as this tone-deaf jerk, since so much of his dialogue consisted of endless nosy personal questions.

This disorienting view of the conversation seemed relevant enough that I decided to keep it as a feature of the game rather than just a weird experience I had had in testing, though I figured I’d set it up so that you could only unlock playing as Galatea after you’d played the regular way once or twice. I created some extra memories for her — “memories”, from Prompter’s point of view, refer to short expository paragraphs that are triggered to fold into the story when a given topic of conversation comes up for the first time. So this offered me a way to give Galatea some additional interiority never exposed by actual dialogue. And I wrote in some additional endings, some possibilities that were reachable only if the player took action as Galatea, because I wanted her to have that kind of agency as well.

Then there was the issue of character art. I wanted to preserve the feature from the original that Galatea turns more towards the player as an expression of increased intimacy, but I wanted to reconcile that with the Versu engine feature of having images that showed a character’s expression. So we needed a whole bunch of pictures of Galatea, with different expressions and poses (though the Sad back of her head looks the same as a Neutral or Happy back of head — so getting her to turn towards you not only expresses greater intimacy but makes her emotions more legible):

The protagonist’s gender is never specified in the original, though people tend to assume he is male. (That he is/can be attracted to Galatea is of course no indication of his own gender. The way he talks to her might be thought to represent some gendered tropes, though.)

The art and the fact that Versu is in third person rather than second meant we needed to make a decision, so I went with the sense that most people already had: he became explicitly male, though rendered mysteriously as a picture of a hat and shoulders. His face is never explicitly shown and he doesn’t have visible emotional states.

But I didn’t like that there was only a male protagonist, so I also wrote an alternate, female viewer with a different backstory: she is a collector, considering buying pieces from the gallery. The Buyer had a lot of fresh dialogue of her own, though she also could use some of the same conversational gambits as the original protagonist. But she will also talk to Galatea at some length about the presentation of art, and self-presentation; she was sort of an excuse to add some thoughts about performative femininity that I hadn’t really thought about when I wrote the first version of the game. Attaching those to a new selectable protagonist seemed like a way to open up the game to those ideas while still leaving the original experience semi-intact for those who wanted it.

Then there were the endings. In the original version of the game, a lot of the endings consisted of a long piece of non-interactive text triggered by whatever mood or situation or conversational spot you’d reached with Galatea. In Versu, that felt unnatural. The player has so much control in other parts of the game that taking it away there was uncomfortable. And getting a big chunk of text, in the non-parser implementation, no longer felt like a reward for unlocking something difficult. So instead I expanded a number of those scenes to allow the player a bit more choice about how they played out, in some places making an additional major decision or having a few lines of epilogue followup.

The result (at least somewhat) addressed a pacing issue from the original piece. People often mentioned feeling a little surprised by how something happened or by a sudden outcome, and I think it plays a bit more naturally if you’re allowed to experience some kind of crisis point (such as becoming friends or making her angry) and then play a few more turns before the story completely resolves.

In the original, it would have been difficult to make the end scenes interactive, because it would have required so much state-tracking to keep track of what dialogue was supposed to happen when. The original code base didn’t include a concept of “scenes” or narrative chunks of any kind. Versu’s conversation design made it comparatively trivial to establish dialogue associated only with one ending scene. In fact, some of what I’ve written here was also present in the game, as part of an unlockable Director’s Cut conversation. (Meta? Yes. But the game was always like that.)

So. When we went to release this version under our own name (rather than under the Linden brand, where it had already been accepted once), Apple rejected the app for some reason to do with client memory handling; and as it wasn’t possible for us to fix that, it was never in fact released. (None of our interactions resulted in Linden owning any of the IP for the original, however, so there’s no issue on that front, lest anyone be concerned.)

In any case, since it doesn’t seem likely that anyone is going to see this item again in the future, I wanted at least to record that it existed, and what was unusual about it.