Interview: Moved By Mod -- Dear Esther's Dan Pinchbeck

[University of Portsmouth researcher Dan Pinchbeck is the creator of Half-Life 2 interactive story-based mod Dear Esther, somewhat of a sleeper hit in the modding community, and in this interview he discusses its curious inspirations.]

I've had Dear Esther on my radar for a while. It sounded incredibly interesting; you are left on an island in the Hebradian range, with a nameless narrator and what is supposedly a ghost story. 

I never really took the plunge, though, waiting until the reports of invisible holes in the maps and ways to break the mod were all ironed out. It wasn't until Lewis Denby's piece on Rock, Paper, Shotgun arrived that I finally downloaded it and played through. 

It's hard to describe Dear Esther without ruining it, and even then it's difficult to put it into words. I'm of the firm belief that it's a large step forward in game narration, finally moving beyond feeding us cutscenes and expository dialogue. 

Before you read the interview, be aware that there are what could be considered small SPOILERS below, and so, I urge you very, very strongly to go download Dear Esther, play it (it required Half Life 2), then come back. It takes about 30 minutes to play through, and it is absolutely, entirely worth it. (You can get a brief view of what the mod looks and sounds like by checking out a YouTube gameplay video, but try not to spoil it for yourself.)

Can you explain a little about yourself and what you do?

Dan Pinchbeck: I’m a researcher and lecturer based at the University of Portsmouth, UK. I teach games and interactive media, but spend most of my time now on the research side of things. This splits into two major strands. Firstly, I study first-person gaming, particularly the relationship between game content and gameplay. I’ve just completed a doctorate on that, looking specifically at how story functions as a gameplay device. 

There’s a lot of historical antagonism between the two and I don’t agree with that. When story and gameplay bang up against each other, to me that’s just bad design, not a fundamental incompatibility. I spent the last four years doing a big analysis of major FPS titles of the last ten years, looking at their worlds, agents, avatars and plots to back that up. 

The research mods spin out of this: basically, there are questions about games you just can’t answer by looking at commercial titles, but they are really important questions nonetheless. So what do you do? Forget them, or theorise about them without any real evidence to back up your ideas? Or you can get on with it and build them yourself. That’s the real impetus behind thechineseroom. 

In my other major strand, I’m involved in game preservation, mainly through a big European project called KEEP, which is developing this emulation architecture that you’ll be able to plug all kinds of existing emulators into to run obsolete titles. I won’t go into that here, but there’s a lot more info at www.keep-project.eu.

Where did the idea for Dear Esther come from?

The basic idea came from this question about what happens when you ditch traditional gameplay out of an FPS space and what that leaves you. So you have nothing but story to keep a player engaged – is that possible? What kind of experience does that leave? What does the space you free up by losing all those gameplay mechanisms and activities allow you to do? 

Secondly, it was the idea of just how abstract and problematic you could make a story and retain it being this engaging experience. Can you have a really quite non-linear, non-literal story, that’s almost more of a mood piece, rather than a traditional narrative, and, basically, get away with that? Bottom line is that FPS spaces are exceptional environments for anchoring very intense experiences, so why not gear the story around that immersiveness rather than trying to tell a very tidy, closed down plot. 

And that led to the randomization and open-endedness of the plot. I guess I was really interested in what people made of it, how they joined up these dots that may not actually have any grand scheme behind them. How much they will create a story from all these pieces, where I only have limited control over how they fit together. And as writer, that’s really interesting to me. If I wanted a tidy, linear plot with clear characters, I’d write a novel. Games give us the opportunity to do far more interesting things than that, why limit yourself to formats that work better in other mediums anyway? Dear Esther’s story wouldn’t work half as well in anything other than a game engine, and that’s a really interesting thing for me.

I'm loathe to call Dear Esther a game, not least because it doesn't seem at all trivial, and its interactivity is in question, but also it isn't fun, instead moving onto something far more poignant. How do you think games can move beyond merely being entertainment, and towards something more emotive?

I’d agree with the first bit of that. I have no idea if it is a game, or art, or anything else. Even ‘interactive story’ is a difficult one because, as you say, the interactivity is on one level, very trivial. I’m not sure if I’d ever want to come down on a final decision about any of that, and it’s kind of not my call anyway, that’s something for users to work out for themselves, if they want to. 

The idea of fun is a really interesting one, and it’s great to see it being really chewed around by commercial developers right now. The historical gung-ho, guns&ammo schlock is being played around with by recent titles. Not that I’m anti a good bit of schlock, per se, but sometimes you want Chuck Norris and sometimes you want something with a different mood or flavour. And that’s not a quality judgement either, it’s about games offering a wider emotional range. 

You’d hardly call the end of Far Cry 2 a celebration of gun culture or victorious forces of good over evil in the mould of, say, Timeshift. Half Life 2 Episode 2 isn’t exactly Happy Days. And that’s terrific really. I love the fact that S.T.A.L.K.E.R. is mainly characterised by unease, punctuated by moments of panic. Bioshock is really quite sad and poignant, when you’re not hammering away with the plasmids. I loved that moment near the beginning where you can hear and see the splicer with the pram. That’s a great little moment. I think we’re in this transitional phase at the moment. 

You can see this expanded emotional range in commercial, major titles, and you’ve got more indie stuff like The Path and, I guess, Dear Esther. There’s clearly a niche market for that, but what’s interesting is how these multiple layers of emotional range can get into bigger titles: so you can have a quite deep, affective experience or you can just charge about with a railgun. The player can define the extent to which they engage with that. 

But when you look at it, a lot of contemporary FPS games have been chipping away at the idea of pure, escapist, fun, or changing how it fits with the genre. We don’t demand that any other form of media has to make us feel great and happy and powerful all the time, it’s quite the opposite often. So it makes sense that games are getting into this, that there’s no contradiction between wanting to continue an experience and the experience itself being quite harrowing or unsettling. That’s drama, and people have been doing that since before they were writing, it’s as old as human culture. Of course games are going to tap into that and I think it will make them better games as a result.

Ambiguity runs rife in the mod, from the moment you start, leaving the player unsure of anything they see or hear. Was it your intention to set the player at unease?

Yes, pretty much. Ambiguity was always central to it. It was always going to need to create unease in order to get away with the abstractness and fractured nature of the narrative. We had to have something where the player was drawn to go deeper into the story, but it wasn’t like an episode of Columbo or something, where it fell too heavily on being a mystery to solve. It was important to get the atmosphere in early on that this place was almost beyond a direct answer, that you’d never really get there, but be left with these pieces that almost make sense, but you’re never fully sure of it. 

So then, if you go back to dig out more or re-visit, you then get this different version of events. I really wanted to achieve that, this idea that two people could talk about it afterwards and have quite different ideas of what it all meant (if anything) as a result of hearing two different stories essentially. But like I said, I also see it as being a mood piece, as much as (if not more than) a plot in the traditional sense. I think you can go through Dear Esther and not really understand it, not really have a clear sense of what happened, but still have this engaging experience. 

That’s something game spaces do really well – you don’t have to know everything, it’s about how you feel during and after it. In a very weird way, it’s a bit like Tetris in that regard, it’s trying to make people sit back and lose themselves in this world that doesn’t make literal ‘sense’. I always find it a bit disappointing when game writers and developers feel they have to tie up all the threads at the end of a game, it’s often clumsy and unnecessary. I remember playing the original Silent Hill for the first time and just loving the end. 

So I’ve played for six hours or whatever and I’ve been on this quest for my daughter and she just turned into an angel and I shot her and now I’m dead behind the wheel of a car and I have no idea what just happened, but I do know that it’s hit me really deeply and I can’t quite quantify that. It’s a stunning piece of narrative design and should be used more often in games. Personally, I have a hunch that gamers would prefer a narrative that is left open and unanswerable than a pat, Hollywood, clumsy resolution. 

Gamers are more sophisticated than that. We need less bloody awful pseudo-heroic resolutions with sub-Shyamalan final cutscene twists to open the door for a sequel, and more unfinished business, unanswered questions, unresolved emotions.

Dear Esther explores the negative end of emotions, evoking sadness, horror and confusion. Was this your primary intention, or just the easiest way to explore presenting the player with the ambiguities you sought?

Maybe I’m just miserable. I think I’m more drawn to that as a writer. The types of environment the engine leads you to also play a part in that. That’s one of the things we tried to do with Dear Esther, was to develop the whole thing quite organically, so I was writing all these notes and fragments at the same time that Josh (Short, now at Frontier) was building from my sketches and Jessica was composing the music. 

So a lot of the story actually came out as a response to the island as it developed, and the music that was responding to that as well, then both of those would adapt to the writing. You’d have to be mad to try and build a commercial game like that, but it was a fun experimental way of doing things. The idea of being abandoned on this island and this lost history that keeps leaking back into the present was there at the beginning, but even that came from discussions about what kind of environment we could have that was really enclosed but open, isolated. 

In a way, the whole of Dear Esther derived from the availability of assets in Source, given our time and budget constraints in putting it together. The emotional feel of the piece was a natural, quite organic development of the process. I have to say though, I hadn’t really clocked how affected people would be by that side of things. I knew it was pretty odd and sober, but I hadn’t really considered it to be particularly frightening or massively traumatic when we released it. So that’s quite a nice outcome for me as a writer, that it has really affected people; not just in this trawl through despair, like car-crash TV, but that people have taken this quite deep experience away from it.

Another of the major themes in the mod is religion, and in specific Paul's journey to Damascus. Was this used just to place another layer of interpretation, or was it there for a definite purpose?

It was quite a late addition actually. Paul was named as a character before, and there was no conscious idea to tie that in to the biblical character. Then, afterwards, we already had the visual image of the white lines on the cliffs, which is actually historically accurate, it’s from a text I found when I was looking for Hebridean islands to base the modelling on. This had led to the electrical diagrams and nerve cells and I thought about writing on the rocks – sure I’ve stolen that from somewhere, but I can’t place where – and it kind of clicked. 

Then there was this great synchronicity between the Damascus story and what was developing in the narrator’s journey. I knew I wanted it to end on this idea of transformation – how do you even arbitrarily close off this story – and it just seemed to stack up like dominoes. Then this led to Lot’s wife, and this idea of never being able to look back. That just seemed so appropriate given the crash theme. So I’d be lying if I said it was a carefully planned plot stream that was hardwired into the thing from the outset. It was a product of way we developed it, this more free-form, almost stream of consciousness approach to building it. 

The whole last section of the script, pretty much all of the north shore and ascent, was written in one burst and I tried to just let it go where it went, rather than having this whole planned thing, aside from knowing the set-up (the infection) and the basic resolution (throwing himself off the aerial). That was one of the really liberating things in using the randomised cues, I could go off at tangents without feeling I had to get all these ideas and images into a comparatively tiny space of time. So the whole religious, kind of transcendental images in the text, they can come out really strongly, or be there in the background, unresolved, depending on which set of cues you trigger.

The music in Dear Esther is one of, if not the most, important aspects of how well it succeeds in telling a story. Who composed the pieces? Did you see them as an important narrative tool?

The music is probably the thing more than any other aspect of the mod which absolutely lifts it. It’s stunning, and I can say that quite happily because I didn’t compose it! It sets the emotional tone, it glues all the elements together, it adds this landscape to the experience that just wouldn’t be there otherwise. I’d worked with Jessica Curry, the composer, a few times, collaborating on these art works, and knew her work really well. As soon as the island started coming together in my head, I knew I wanted to use her, knew her work would fit the emerging tone. There’s quite a bit of her earlier work on her website, and you can hear all the forms and moods in those pieces and really see how she was just the natural choice for the mod. 

I think the music fundamental as it sets up a kind of emotional landscape for the voice-overs. Music bypasses so much of our interpretative preconceptions, it just hits directly to the heart, and that’s why it’s so fundamentally important to games, and enormously undervalued and underdeveloped. There’s far too much banal, boring, background ambience and not enough foregrounded, completely self-conscious use of music still. 

You look right back and take something like Tomb Raider; I was reading this article by Tom Armitage in InfoVore about Far Cry 2 the other day, and he talks about Africa being almost a character in the game, and the music in Tomb Raider was really like that, it had its own identity and it wasn’t just this bland, subtle wash. Halo promised that too. Why not have massive, epic, emotional themes? After all, games are all about manipulating players, especially emotionally, so why not? 

So yes, the music in Dear Esther is absolutely key. My only issue with it is that we never got deep enough into Source (didn’t really have the time or skills) to move it on from straight-out start/stop triggers to something more dynamic, more procedural. That was a bit frustrating, especially the caves section, where you never knew if players would get the whole of “Always”, because you could get through faster than the track. I think it’s such a beautiful piece of music that, perhaps more than any of the other cues, encapsulates the whole of Dear Esther in one track. That’s why we extended the caves in version 1.1, not because I had images I thought were missing, or wanted more voice-overs, I just wanted to increase the chances of players getting the whole of that piece.

The writing in the mod is beyond impressive, garnering emotion without really forming any sort of cohesive story. With Dear Esther being a research project, did you struggle to come up with something so compelling? Did the setting and tone come easily?

Thanks! I’m really pleased with it. I’ve got a couple of unpublished novels rotting under the bed and used to write MUD environments way back when, so I must admit, the project was really fun for me as it let me flex those muscles as well as handling the research side of things. It’s very much my style of writing, I’m not a great plot person. I love Philip K. Dick and what I’ve always really admired about his work is how it just keeps firing off on these completely mental tangents so you never really know where it’s going next, or why, or how it will resolve. You read something like Lies, Inc., and you’re just left reeling, but it works so well on this sub-interpretable level, if that makes sense. 

And musically, I’m a big fan of Sigur Ros, Godspeed You Black Emperor and early Throwing Muses, where you have these things that feel like narratives, but they don’t fit traditional models. With Throwing Muses, you always had this incredible sense of ‘feel’ of a song, even when the lyrics didn’t really add up or make any sense when you considered them coldly. And that’s what is amazing about games, you can’t come at them cold, everything is geared towards manipulating you so the system is partially in control of your reflexes, your behaviour, your interpretations. That naturally fits with my style of writing. I’m pretty awful at putting together coherent plots, I’m more natural at mood and throwing out odd ideas. So it came quite easily, but I was absolutely sweating when it was released as I knew I’d take any criticism hard and personally.

Comparisons between such works as Wuthering Heights and The Turn of the Screw are obvious, and not at all unjustified. How do you think Dear Esther can only work as a game, rather than a short story or novella? Wow.

Thanks again. I didn’t really imagine it would ever get much attention at all, to be honest, so the way it’s been picked up and talked about is still a massive surprise to me. 

Could it be another media form? I think it probably could, but I don’t think it would work as well. I mean, you could monkey around with the script and batter it into a proper story, or do the whole thing as film, but what I think is fundamental to it, the reason it works, is because you are in it controlling the movement and perception. The fact that it operates so much on not knowing really who these people are, and what their relationships are, and where you are, and if its real – so much of this is anchored within the relationship between the player and the avatar. And you find this technique, although maybe not pushed to the same extreme, in a lot of commercial games. 

Amnesia is really quite a common theme in FPS games, and this idea of being an outsider who is finding their place in a world that is unknown. What I think is a bit different is that you usually go through this process of normalisation, you get to know the world better and better and become more powerful, more in control of it. In fact, that’s probably one of the hardest things to balance in terms of gameplay progression, controlling that process, so the player gets the emotional reward of increased knowledge and power while the game is holding back, not letting them get the full story. In Dear Esther that never really happens. 

You are almost as in the dark at the end as at the beginning, or at least, you can’t trust any of the understanding you’ve developed over the course of it. So that, to me, is really central, and that’s a device, or a mechanism which is pure game, it’s not something that works in the same way in any other medium. And you have this real investment in it, because you are implicated, it relies on you acting the world in order for things to progress. You have a real sense in games that what you do, what you see, where you point and move your avatar has an absolute impact on the world and the action. Dear Esther plays around with that, but at the core, that’s what makes it more like a game than any other form of media, even if that’s a problematic relationship or definition.

You're next project, Korsakovia, is looking at how to disconnect the player from the expected cues and place them in an unfamiliar situation. Can you explain a bit more about what you're trying to achieve with it?

It started from a very simple idea. Normally, agents in games are anthropomorphised to an extent. They all look pretty much humanoid or animal. You can see which direction they are facing, you have all these cues to predict behaviour. Plus, you import from real life a whole set of expectations about what they should be able to do, or are likely to do. I wrote this paper recently called Trigens Don’t Swim, and part of that was saying how it’s OK and reasonable to stop Trigens from going into water (or drowning if they do) because they are basically hairless gorillas and gorillas don’t swim. Or, you can have stupid zombies in Doom 3 and it makes sense because we all ‘know’ zombies are stupid. 

So you wrap a level of representation around the finite state machine and if you synchronise the two effectively, you justify the actual AI behaviour. Korsakovia is about what happens when you get rid of that, and you have agents that are basically just a standard FSM but you don’t have any of those cues, so your ability to predict is reduced. That’s it, basically. 

From there, I also wanted to go back to explore that idea of a reality coming to pieces, which again is a really common device in FPS games, from Doom 3 to F.E.A.R., and so on. So, on top of these agents who are really hard to figure out, you have a world that doesn’t make sense and is starting to distort in unexpected ways. That naturally lends itself to a type of story that has resonances with Dear Esther in some ways, but is more direct in being disturbing. Korsakovia is much more of a survival horror game – I wanted to take some of the ideas we’d explored in Dear Esther and see how they work in something which is a ‘proper’ game, if you like. 

I’d first come across Korsakoff’s psychosis years ago in the William Gibson novel Mona Lisa Overdrive, and it’s just such a terrifying illness and fits so well with this idea of a world being broken into these fragments that you just can’t put together again... it was a natural development from there really. 

So in terms of what I want to achieve, well, to create a game that really undermines the player’s sense of security. To create this world that is genuinely frightening and aggressively beyond comprehension, and to drop in this story that only comes in as fragments. Most the script takes the form of scattered pieces of dialogue between a psychiatrist and this deeply disturbed patient who has created this entire fantasy apocalypse of the world ending, but can’t remember this in any more than fleeting bursts, so you have these two realities colliding. 

Rather than Dear Esther though, where there’s this question of “what is actually real?” in Korsakovia it’s ultimately about how that’s just a redundant question, as everything and nothing is, at the same time, if that makes any sense. It’s a difficult time to describe it really, as I’m still writing the script and will be bringing in voice actors during July, so quite a lot could still shift. But that’s the basic premise anyway.

With these mods as research projects, how much do you see yourself as a developer now? Are there things you'd like to do that are divorced from research?

That’s a good question! I see myself as a researcher who makes experimental mods. I suppose that could extend to mod developer, but I’d be cautious of going any further. Game development is such a difficult and complicated business and really what we do is a very stripped back version of that. I have an ongoing grand scheme to roll out some of these experiments and package them in a much bigger experience, moving from the one hour mark to maybe 4-6 hours and testing whether you can sustain these experimentations in a near full length game, but there are cost implications to that I’m still struggling with. 

At the end of the day, I don’t know that anything we’ve done has any direct commercial application, even if ideas we’ve played with get picked up by professional developers, which is obviously something that would justify the whole process in a single go... And I’m not so interested in working commercially. As an academic, I’ve got a freedom to play, to experiment, which is fantastic, but I also feel that given that we’ve almost got a duty to games as a medium to use that freedom to push at ideas that are uncommercial but still really interesting. 

We’ve also absolutely got a duty to put our money where our mouths are and experiment, by developing, to look both at the bigger picture of what you can achieve with this medium, and to try and get answers to questions you’d otherwise just be making educated guesses about. When we made Conscientious Objector, the Doom 3 mod, I wanted to know if you have non-lethal force at the centre of an FPS, so you can’t actually kill anything (not necessarily because I have a political problem with that, but because if you do that you mess up the basic premise of the genre, i.e. removing things from an environment), can you retain the full-on, adrenaline rush, visceral gaming that characterises a run & gun shooter? You can’t answer that unless you go ahead and build it. 

As far as non-research projects go, we’ll see what happens really. Everything we’ve developed has been part of my day job, basically, which means it’s funded from somewhere. A non-research project would also have to be funded, to be frank, or I wouldn’t have the time to commit to it. If that’s another way of asking whether I’d work in industry, then the short answer is that it would very much depend on the project. The opportunity to play within a commercial development is obviously really attractive, and again, would justify what I’ve been doing for the last two years, but against that are the constraints of that kind of working environment. 

In a way, I get to operate at a quite glamorous edge of the reality of writing for games. Finding fifty-eight different ways of writing ‘you shot me!’ doesn’t really appeal. Plus, generating the sheer quantity of dialogue and script required for a commercial game, and keeping the quality level of that high is a real skill. I’ve got a huge amount of respect for game writers, as it’s an enormously difficult job that still has some way to go in terms of the recognition it deserves, that is can really make the difference between a mediocre game and a great one. 

Also, I’ve got this really privileged position of being able to just work within game formats that I find interesting, rather than having to take on jobs to pay the bills. But you never say never. I’m open to suggestions, inspiration, bribery, coercion, etc, as much as the next man.

While you're obviously making these mods with very specific purposes in mind, are there any other mods or games out there that you find particularly interesting from a research point of view?

I end up talking about S.T.A.L.K.E.R. every time I do an interview. One of the many people holding their breath and keeping their fingers crossed that GSC do a u-turn and head back to all the things that made the first title so damn good in Call of Pripyat. Oblivion Lost is a truly great mod and anyone with S.T.A.L.K.E.R. should be playing it. 

More than anything, I want to see ambition and experimentation, and I’ll sacrifice it being a great, great game for that. I got put onto Pathalogic recently and that’s a must for anyone interested in FPS games. Portal, obviously. Mirror’s Edge is one of the most intriguing fusions of genres that’s come out recently: I keep boring people with shouting about how it’s more like a driving game with combos than an FPS, which is probably a really banal and obvious statement, but it’s really interesting to me... 

It’s the end of the [school] year here, so I’m buried in marking and finishing off Korsakovia so my playing time has suffered. I’m in two minds about The Graveyard and have yet to play The Path, but they’re on the list. I’ve failed to play Braid yet, which could probably get me barred from future games conferences. Basically, there’s a stack of games about four feet high on my desk at work just waiting for a chance to get ‘researched’. 

If I had to put together a list of FPS games which need looking at by anyone interested in how the genre is changing? It’d probably be pretty much the same anyone with half a clue about the industry would come up with: S.T.A.L.K.E.R., Far Cry 2, Bioshock, Portal, Mirror’s Edge, maybe Fallout 3. Near misses would be Penumbra, Pathologic, Call of Cthulhu: Dark Corners of the Earth. XIII, Boiling Point, Far Cry, System Shock 2, if you’re going further back into ancient history. And the Half Life franchise for a masterclass in how to ‘do’ story, although I really hope Valve haven’t backed themselves into an impossible corner with that one. 

Weirdly, the game I’ve most enjoyed recently as a player is Syphon Filter – downloaded it on the PSP the other week and although my thumbs are now knackered, it’s really quite good...

Is there anything else you'd like to add?

Bludgeoning a zombie with a crowbar is fun. Existentially bludgeoning an invisible zombie with an identity crisis – that’s got to be worth a pop.

This article was first published on Gamasutra in 2009 and was lightly updated in 2023.