This interview is part of our Road to the IGF series.
Heaven’s Vault follows an archaeologist trying to translate a lost language in an ancient Nebula, finding its hidden meanings one symbol at a time. It’s up top the player to help her do that. But will they ever know if they translated things correctly?
Gamasutra had a talk with Jon Ingold and Joseph Humfrey of inkle, chatting about the challenges of creating an entire language for players to unravel. The game is nominated for the Independent Game Festival’s Excellence in Narrative award, and earns an honorable mention in IGF’s Seumas McNally Grand Prize category.
Archaeologists of the stars
Ingold: I’m Jon Ingold, and I’m the Narrative Director at inkle. I wrote the script, built the history of the world, and designed the systems for dynamically building the narrative. I also do a chunk of the game design, including the translation mechanic.
I grew up with a terrible computer at home, so while I played a few of the 90s platformers and Lemmings, I mostly played Infocom’s parser-based text adventures. But that led to me writing my own text games in Inform in the early 2000s, and since founding inkle with Joe, I’ve been writing them professionally for the last 8 years, so it all worked out neatly.
Humfrey: I’m Joseph Humfrey. I’m the other co-founder at inkle. Beyond co-designing the game with Jon, I’m the lead programmer and I’m responsible for art direction and UI design. I’m a bit of an odd-job guy – I do many of the bits and pieces that a triple-A company would hire specialists for!
I spent my childhood playing with game modding tools and creating levels for existing games. My family always had Macs, so I enjoyed making little adventures with HyperCard (the tool that Myst was made with), giving me a real nostalgia for Obra Dinn’s art style! I preferred to stay on the art side of things as a teenager, getting to know Photoshop by making texture maps for Bungie’s 3d innovative tactics game Myth 2 and making models in Strata 3d – another of the tools also used by the Myst guys. But after studying Computer Science at university, I went straight into coding at Rare, and then Sony’s Cambridge studio, where I met Jon.
Finding new archaeological mysteries in space
Humfrey: Heaven’s Vault has been a long time coming. What we’ve always been most proud of in our games is the way the narrative flows – making the world feel organic and alive rather than a fiddly mechanical box which constantly prevents you from progressing until you solve a combination lock. Part of this is the fundamental way that we design the structure of the game, and part of it is about the surface layer – how we visually present the story, and how the player interacts with it.
After creating a series of predominantly text-based games with the narrative architecture we believe in, we wanted to try something more visual. Could we create a huge world with the depth and breadth of our previous games, while also making it possible to explore them in 3d rather than purely in text? And of course, could we do this with our tiny indie team?
Ingold: The core concept was the idea of archaeology as an activity for the player. Something like a detective game, but without the fail state of accusing the wrong person and having to start over – something like an exploration game, but where everything you discovered had a narrative meaning.
We quickly realized that an archaeology game set in, say, Egypt, could never quite capture the feeling of discovery that we wanted – we already know so much about Ancient Egypt – so we moved the setting to somewhere completely, bafflingly new, and so the Nebula and it’s 5,000 year history was born.
Tools of discovery
Humfrey: We use Unity as our core engine. And, of course, we use our own narrative scripting language ink, as well as the editor for it – Inky. Both of these are open source and free to use! We couldn’t have built a project with the narrative complexity of Heaven’s Vault without them.
On designing the hieroglyphics system of Heaven’s Vault
Ingold: It was really, really hard. The basic idea of the player learning a few ideas that allow them to decode longer and more complex strings is straightforward, but in practice, the balance between making it trivial to crack or completely incomprehensible was very difficult to strike. We tried semi-random languages, but they felt too arbitrary; we tried non-English grammars, including some “structural” ones where glyphs were arranged above or underneath each other, but the results were unreadable and unsatisfying.
We finally settled on the design of Ancient – a language that is completely logical and built up from simple principles to form complex ones – but is also full of poetic, expressive details: the word for God is “Person of knowledge”, the word for value is the same was as the word for water, and so on. The logical framework meant every discovery the player made was valid, even if it was something we hadn’t thought of ourselves: the authored dictionary meant we could bury vast amounts of world-building and lore into the fabric of the puzzles themselves.
Humfrey: We were also keen to make the language a visual one. We were strongly inspired in particular by Egyptian hieroglyphs as well as Chinese and Japanese writing systems, which often use symbols that resemble real-world concepts. Even the Roman alphabet was originally evolved this way, even if the original visual meaning has been lost.
We love that there are lots of different ways for players to deduce the meaning of an inscription in Heaven’s Vault, and visual recognition is a great way in that requires no special knowledge or puzzle solving skills.
Capturing the feeling of learning a language
Ingold: When we were designing the language, we didn’t think the idea of not confirming translations was that radical, to be honest! If we’d confirmed the player’s translations, then players would have brute-forced them, and never looked at the construction of each word. Because the whole language is the puzzle: each individual inscription is just a part of that puzzle. You’re never really finished with Ancient, just as you can’t finish learning German or Chinese: there’s always something new to figure out and discover.
We built that into game’s replay loop: we have players currently on their seventh or eighth replay of Heaven’s Vault, decoding longer and longer inscriptions, and still discovering new words – and of the course, the story adapts differently to the new discoveries you make on longer translations.
Letting players get it “wrong”
Ingold: Following up on mis-translations fell out of the language mechanic’s design. But it also felt right: we wanted to make a game about archaeology, and archaeology is not about getting things “right”. You can’t ever know if you’ve interpreted a find, or a culture, correctly. You have to make the best guess you can; but you have to be open to changing your interpretation if new evidence comes to life.
Games are often very cautious of allowing that kind of ambiguity: there’s this urge to let players “complete” a game – to “master” it. We wanted to push back against that and say: here’s a world for you to discover, explore, contemplate, and never really understand. My favorite science-fiction has that sense of mystery: whether it’s in the hints of an ancient history in Star Wars or the bizarre, hard-to-parse worlds of Gene Wolfe and Jack Vance.
We’ve had a really vibrant Discord community since launch, piecing together the world of Heaven’s Vault, proposing theories and counter-theories. It’s been a joy to watch it unfold.
A real world instead of story branches
Humfrey: What I love about the structure of the story in Heaven’s Vault is the way we’ve almost entirely dropped the structure of branching entirely. It is literally impossible to draw a big story graph of the game, because it doesn’t exist – it’s not the right structure. It’s more organic than that: half way towards being a story simulation, albeit a tightly-authored one. The ultimate aim has never been to create thousands of branches or to make players feel like they should investigate every single corner of the game’s possibilities.
Ingold: As Joe says, the game doesn’t branch: you couldn’t generate the narrative flexibility we needed with branching trees. Instead, every single piece of dialogue, every interaction, every moment, is available all the time, but is gated by the conditions it needs to occur (and the conditions that make it irrelevant). The game then makes a guess at what the best things to offer next would be. So, no two playthroughs of Heaven’s Vault are ever quite the same, and we still uncover new wrinkles and details when we play.
It means the player can go and explore whichever site in the Nebula they want to. They can ignore parts of the plot entirely. They can dig deep into other parts, and the characters will always talk, specifically, about what you’ve seen so far.
Agency in a real place
Humfrey: Our aim is simply to make the player feel like they have agency within in a living world, and that their actions result in a story that truly makes sense, without any of the silly robotic responses that are all too common in games.
Ingold: It’s about raising the bar on what game narratives can do. Let’s have real characters. Let’s have a protagonist with a name and a past. Let’s have worlds that move forward when you’re not there. Let’s have a game where you can’t get stuck, and you don’t need to backtrack and revisit areas because you missed something over there. A game where when you don’t want to do something, the game respects that and integrates that decision.
Let’s ensure whatever happens, things are always moving forwards, without being on rails.