The Last Clockwinder post-mortem

The Last Clockwinder: Diachronic Script Development

My goal here is to present a snapshot of how the script for one scene evolved over time and for what reasons those changes happened. The only context worth having for reading is that Jules is the player character, and Levi is her old friend and colleague, talking to her over the radio. (Though you can check out this other post for more context.)

I’ve used images in-line with my doodled notes on them. If for some reason images aren’t good for you, I’ve replicated these and my scrawling on this Google Doc.

Alright. Here’s Version 1. This is the first typed-up version I found, so is the oldest barring whatever I scribbled in my notebook.


Iteration 1 (First Draft) 350 words, 2 pages

Three takeaways at the end, which mess with the rest of my numbering system, hah. 

  1. It’s shorter than I would have guessed for the first draft (though we’ll see why later)

  2. The skeleton of the final scene is here from the very beginning. That’s one of the reasons I chose this scene over some of the others that kept the same title, but whose content entirely changed – we’ll actually be able to chart a course through the revisions. 

  3. For all that it’s short, and in some ways more successful than a few of the later versions, I’d call the style quite rambly and unfocussed compared to where it ends up. 

On to version 2


Iteration 2 (First Revision): 500 words, 3 pages

The scene fills out quite a bit in this version. Key callouts (from this point, my numbering will be incremental and will help draw things together across versions): 

  1. Added some lines for flow – Jules needs Levi to coax out what’s bothering her. More dramatically/emotionally interesting BUT it very much feels like padding compared to the leaner final scene. 

  2. A revision of the previous version that doesn’t really change anything yet. This line will become a theme – it changes in almost every draft, as I was clearly dissatisfied with it in some way, until I land on something that actually works.

  3. Lots of lines added here – the wordcount soared by 150w! – which do a decent job of fleshing out the characters, building in some world details, and injecting some humour. But at the cost of bloating the scene significantly.

  4. I use placeholders like this a lot when drafting, though you won’t see many here, because I generally resolve them, even if not perfectly, for a turned-out versioned draft. 

  5. More added, as in point 3. It does flesh things out in ways that get lost in the final draft, but there just ultimately isn’t the space for this kind of thing in this form of narrative. 

  6. ‘and now our new friend here’ is an artefact of a point where the player character might have been a third party other than Jules or Levi.


Iteration 3 (Version 1.3): 575 words, 3 pages

2. Yep, still playing with that line.

4. An attempt at a ‘fictitious swear’. Didn’t fit well; rightly removed from here and the script overall after this.

7. A new closing monologue from Levi. This is playing with theme and what the story is grappling with and has to say. It comes off a bit bald, but even more than that it just makes the scene even longer (up another 75 words!).


Iteration 4 (V2): 300 words, 2 pages

8. Now we’re talking! This was part of the Big Cut Pass across the whole script when it became clear just how untenable scenes of the previous length were. This shaves off 175 words, and takes it down to barely 2 pages. It makes a massive, positive difference. You can see we’ve lost some detail, but considering how the scene flows now, it’s clear just how desperately needed that was. 

Slightly extended tangent on that point: the bloating and then shrinking of the scene could be taken as evidence of a certain amount of ‘faffing about’, given that it’s come back to a form much closer to where it started. BUT I think that’s a flawed reading. First off, the process of adding then subtracting is – if done judiciously – more like reducing a sauce than pushing a rock up a hill. Things may end up looking a bit like where they started, but it’s a lot stronger and more intense. 

Second off, the redrafting was part of the process of figuring out more about the game as a whole – exploring character and theme, and strengthening what things end up on the page, even if other things gets left out. I will say that this was an intensely good-feeling thing to do, and also a bit of a luxury in games writing. The ability to iterate like this – to test and extend the writing and make it better, was very, very valuable, but is not always possible. 

(Add to that that some of the iteration came specifically from playtesting and seeing what people were and weren’t getting about the story and story-world – what was boring them or going over their heads. Length played a big part of that – saying too many things just meant that nothing stuck, no matter how compelling the individual details were.)

9. This line is an interesting one to compare directly across versions. 

Original version: Over the years? A dozen? At most. There are – were, I guess – supply runs every half-year. I picked up a fair few of them. Heh. But I guess you know that better’n most. Why d’you ask?

First revision: Over the years... A dozen? At most. There are – were, I guess – supply runs every half-cycle. I picked up a fair few of them. Heh. But I guess you know that already. Why d’you ask?

Very minor wording changes. Notably, ‘half-cycle’ feels a lot more ‘on genre’ than ‘half-year’. 

New revision: Over the years... I brought a few dozen shipments, maybe? Why d’you ask?

Those words – 13, count ’em – say as much as the 36(!) of the original line and give up precisely nothing of value. 

What’s even more important, actually, is how this line looks in the final script:

Final script: 

That’s right, zero words. Because, ultimately, with space at a premium, that line was achieving absolutely nothing. 

2. THERE it is! Finally, a version of this line that actually feels like it snaps. The space dedicated to it feels fairly sizeable in this revised scene (a line, a reaction, then a line), but I think this was worth keeping because it was amusing and good for showing the relationship between Levi and Jules. The humour angle (I’m not going so far as to call it A Joke, but it’s certainly a minor gag) gets goodwill from the player BUT crucially also gives some nice dynamic lines for the voice actors to play off, and really works tonally with the shift into the emotional core of the scene at the end. 

10. Another set of lines to compare to the previous version. First off, they’re the only bits that remain of a much longer exchange. Second, the lines themselves:

Version 1.3: I never really knew her. We’d speak, a little, when I brought shipments. Small talk. She wasn’t very good at it. I guess when you’re stuck out on the edge of the world with nothing but a bunch of plants for company, you don’t get too good at small talk. This life wouldn’t be for me, is what I’m saying.

Version 2: I never really knew her. We’d talk, a little. But when you’re stuck out on the edge of the world with nothing but a bunch of plants for company, you don’t get too good at small talk. 

About half as many words, and I think nothing lost that’s not a) not that important or b) communicated just fine elsewhere. 

Version 1.3: She was brilliant, though, that much was obvious. I don’t really understand how this place works, why it’s here, any of that, but she clearly knew every living breathing bit of it. And she loved being here. This was her place, y’know?. There’s few enough find that in their lifetimes. Got to envy her that.

Version 2: She was brilliant, though, that much was obvious. She clearly knew every living breathing bit of this weird place. And she loved being here. But I think she was lonely. 

Similar reduction. We lose some of the more naturalistic rambliness and wordiness, but we knew by now that was absolutely not the way to go with this script. 

7. Also note how much has come out of the ending. No more monologue; slightly different emphasis, but striving for the same emotional/thematic notes.  


Iteration 5 (Final script): 175 words, 1 page

There we go. Less than a page. 

11. The first line is now Jules. This is the result of a match pass – these scenes were always triggered by the radio ringing, and the player answering it. Having Jules speak first didn’t make sense! So, this is purely functional to cover that. 

9. Right to the point. Some of the same lines, but no working around to the point of the scene. There just isn’t the space for that. 

12. We switched this from a chuckle to a groan on the basis of trying to tamp down any read that Levi and Jules were an item. If that’s what you got from the game... well, you do you (but also: no). 

13. Slightly simplified version of this gag. 

7. Same ending content, but much simpler. No monologues, just a few lines with the characters’ own points of view in brief.  


These were by no means the only interations, just the ones notable enough to reproduce here. If people find this valuable, I might dig into the passages I quoted at length and why I think each of the cuts to it worked (the process of figuring out what to take away at a word-by-word level). And, potentially, look for another scene to dissect in this way. Say ‘hi’ in the comments or on Twitter if that’s interesting to you.

The Last Clockwinder: Refining the Writing

The Last Clockwinder's playtesting process provided an excellent means to probe and iterate on the narrative. With scratch audio we put together in the game, the managed playtests Pontoco ran were a chance to see how players were responding to the narrative.

A big focus for this process was comprehension. The story-world of the Clocktower, the Clockwinder, and the history of Jules and Edea in particular had a lot of different elements which we needed to get across without stating them too baldly or losing the charm and characterful dialogue. We were able to make a lot of well-reasoned changes in response to seeing playtesters actually play the game, at a stage when the script wasn't polished, the audio was scratch (e.g. me doing my best 'Alex from The Expanse' voice for Levi), and the game wasn't playable end-to-end.

This also gave us insight into the ways the players were interacting with the environment and the narrative elements, which helped us reposition or resequence the audiologs between and within rooms to manage that flow a little better.

Here are some key ways in which we responded to this feedback.

Simplifying the story

Things started off too complicated. This was mostly about the makeup of the story-world and 'rules' of the Clocktower. I think this was partly just a desire to make the story seem maximally interesting on paper during development, and partly as we tried to link together key threads from the story with what was being depicted in gameplay.

We did a lot of work, sometimes quite painful and difficult, to rationalise the story-world and the story itself. Collapsing or omitting plot beats; coming up with new explanations for things; working out what things could afford to be ambiguous to the player and what couldn't[^1]. The key here was not being attached to any one vision of it over time -- accepting the necessity that things would change (while still caring deeply about the story). Though it could be particularly difficult to really 'see' a new version of the story, since it was a kind of palimpsest of all the versions that had preceded it.

Being more explicit

I said before that we didn't want to state things too baldly or lose the character/charm and feel like we were hitting the player in the face with LORE and MOTIVATIONS. But there had to be some give in that. As always, what's obvious (or even banal and uninteresting) to the people immersed in a thing is not so to a first-time player. Recognising that was important, as was realising that some things that had become routine to us could be interesting revelations to the player taking their first steps into this world.

In practice, this equated to a) working out what these essential factors were ('the tree is called the Clocktower and is meant to safeguard endangered plants'), b) saying these more explicitly, and c) saying them enough times.

This could turn into quite a mechanistic process -- 'find 3 places to say that thing in the first five scenes' -- or inform revisions to ensure we weren't missing opportunities to underscore key details. Both get smoothed out in polish, but literally doing script passes where the goal is 'use Jules's name five times to ensure the player knows who they actually are' means you have the right bricks in place.

Cutting, cutting, cutting

I mentioned in the last post that I cut the script down to roughly 1/4 of 'average pages per scene'. That was vital. Too great a volume of material, no matter how good or characterful, can end up obscuring what you need them to know to understand the 101 version of the story. Slimming down the number of lines and how long each one is helps the most important stuff stand out to the player.

Obviously, there are limits to that, and you don't want to suck all the joy and whimsy out of the script, but the limit is usually much closer to the bone that one would naturally assume. To paraphrase something Stephen King says in On Writing: 'But it's good' isn't an argument against cutting something. It's supposed to be good -- that's your job as the writer. 'Good' should be the baseline.

Cutting is valuable for plenty of other reasons. It puts more trust in the voice actors: they can make a line that seems simple, even boring on paper sound great -- usually more easily than a complicated line. And short is usually just better -- I think knowing how much you can say in how little space is a foundational skill of the craft of writing.

Squint Factor and Overindexing

Two additional factors in the iteration process from playtesting. Playtesters and their feedback were invaluable, but we had to account for squint factor, particularly with regards to the voice acting. Our scratch audio was of servicable quality, but it was not recorded by professionals. The presentation and quality of it was also different from how it would be in the finished games. We had to consider what feedback might just be 'taken care of' by the switch to professional VO. (But also not let that stand as an excuse not to address something that could be improved in the draft.)

Likewise, it wasn't necessarily the right move to make a change in response to every single piece of feedback.

  • Every player and playtester has their own unique context
  • It's resource-intensive to be constantly making these changes
  • Some feedback might contradict other feedback
  • Our own vision of the game and its story was an anchor point worth protecting to some extent

[^1]: And not 'ambiguous' in the sense of 'vague', but drawing the distinction between what we had to have them understand and where it would be okay if they got the wrong end of the stick/weren't clear on our own version of things.

The Last Clockwinder: Constructing the Narrative

One thing I skipped in the previous post on this was a description of The Last Clockwinder's story. You play Jules, a young engineer returning to her childhood home -- the Clocktower -- a giant tree in the middle of a waterlogged planet, and was previously tended by Edea, the Clockwinder. Jules returns with her friend and colleague Levi to save the Clocktower from sinking into the ocean. As she works on this problem, she encounters old recordings of her previous stay on the Clocktower -- when she was a wayward child crashing into Edea's solitude.

Our main narrative units for The Last Clockwinder wound up being audiologs and radio conversations -- two gaming mainstays. The risk with both of these -- audiologs in particular -- is that they can end up making the story feel like an adjunct to the gameplay rather than a cohesive part of it. Story, then gameplay, then story. I found myself wondering at some late stages whether we'd missed the opportunity to try for something more 'interesting' in terms of narrative delivery.

But I don't think that was the case. Not that we couldn't have done anything differently (something I may cover in a future shard) -- but they're mainstays for a reason. They can achieve a lot within many of the resource constraints I mentioned in my last post.

What's more, it's been very validating to see various reviews (user and critical) which have mentioned the 'just enough' nature of the story and narrative delivery. One of the reasons we took this approach was that fourth constraint -- balancing the player's attention. A player could skip every audiolog, multitask while problem-solving, or give them more of their attention. We made sure that nothing in the audiologs was critical to playing the game or having an understanding of the story, even if it would be an underweight one.

The radio conversations were, for the most part, the 'narrative golden path' -- the 'sine qua non' of the story. They represented more of an interruption (though most of the time multitasking was still possible), but were more important to motivating the player, clarifying next action/objective, and moving the story forward. (That's a very 'plumbing focused' take on them; I hope they're also characterful and entertaining in their own right. But, as far as possible, every bit of material had to pull its weight.)

One affordance of both audiologs and radio conversations is that they allowed us to think in terms of short, discrete scenes. That's not always a good thing in games, but the internal structure and placement of these scenes were very clear tools to play with for pacing out the narrative.

Here are a few further snippets that feel relevant here:

  • Before we knew more about Jules and how she'd map (or not) onto the player character, we knew things would lack a certain dynamism/interplay if there was only one speaking character in the present on the Clocktower. Levi (the 'old hand') was an early addition with this in mind, to provide a foil to Jules and/or the player character. (An alternative would have been to only tell the story through the audiologs -- but this would have wound up remote and unsatisfying, and limited the scope for shaping an emotional arc across the game.)

  • The original concept for the story was more 'hands off' and about striking mood and tone than about interfacing so closely with what the player was doing and motivating/directing them more explicit. I'm glad we didn't pursue this route in the end, for a host of different reasons -- iteration through playtesting was a key step in validating this. Moving away from that made it easier to give the player more of an emotional anchor on and understanding of the characters and world and better support the gameplay (helping motivate and direct them as the game unfolded). But there are a bunch of details of the world and characters that got left on the cutting room floor for that reason. 100% the right decision, but still interesting to reflect on now.

  • My original script for the game was significantly overlong. I'll definitely write this up as its own post and maybe even do some side-by-side comparisons on my cuts. The first complete script draft I remember averaged about 4 pages per scene. I got the scythe out and cut that in half, and things still felt too long, so I had to cut it in half again. Really interesting (and vital!) process to undertake.

  • We underwent several iterations of the script before the game was playable end to end -- all fine and normal. We had to get things down on paper and sometimes with scratch audio before knowing final decisions on various elements (sometimes so that those drafts could inform the decisions themselves). But it was also vital that we actually allowed time for several 'match passes' amid the polish when things were playable. This didn't just mean fixing things that were straightforwardly inconsistent with final game flow -- it meant looking for specific opportunities to exploit that connected the script (which always risked feeling 'remote' and removed from the player's reality due to our constraints) to what the player was seeing and doing in a given space. Again, something I can perhaps showcase in a script comparison, but often it meant replacing a general detail in a line of dialogue -- something that declared or implied something about the wider world, for instance, or a joke -- with something that filled the same role in the dialogue but was rooted in what the player would be seeing or doing at the time. This was another way in which we could mitigate some of the common downsides with our narrative units.

The Last Clockwinder: Core Constraints

The Last Clockwinder came out on VR platforms last week. I worked on the game, on-and-off, from January 2020 through to right near the end. Specifically, my role (alongside Olivia Wood) was developing the narrative background for the game, iterating on that, and writing and polishing the game's script.

My plan now is to work up a series of these shards on aspects of the development from my perspective. In keeping with the nature of this blog, they're not meant to be finished, polished artefacts. I'm thinking of them more as 'thought sketches' -- a way for me to reflect roughly on the project as much as to surface anything externally. They may serve as a precursor to more developed posts elsewhere or pitches for talks.

One of our main early discussions for The Last Clockwinder was over how to tell the story within the project's constraints. In the main and in no particular order, these were:

Affordances of VR

Representing any text-heavy content in the game, for example, didn't seem like a viable approach. This also informed things like expected play session duration in terms of parcelling out content.

Development resources

These priorities, for instance, made it unrealistic for narrative to demand character models or custom animations. Written story could inform aspects of the art and mechanical design, particularly in terms of influencing things that had yet to be designed, but the visual and mechanical design would need to have a heavy influence on the written story.

(I'm drawing a false distinction here between 'written story content' and the narrative of the game as a whole, because the art direction, environment design, etc. are not separate from the narrative and storytelling; they are it. But: see my original disclaimer around this being a 'thought sketch'...)

We also knew from the start that we wouldn’t be implementing any kind of complex underlying narrative systems. We weren’t going to incorporate branching narrative or have space for the player to make any significant choice in the story. That’s not a good or a bad thing; it was a reality of the game as it could be made, and so it was our job to make the story as satisfying as possible within that frame. Which is to say: making the player still feel like an agent in that story where the nature of their agency had these constraints.

Financial resources

Into which I lump: funding levels, general budget constraints in indie games, and prioritisation of investment narrative vs parts of development that were more critical to realising the core vision of the game. This is relevant in as much as it determined the available 'narrative resourcing' as a function of Olivia's time and mine plus the available time from other members of the team to dedicate attention to narrative. Which, of course, has an effect on what can realistically be delivered.

Balancing the player's attention

This always is (or should be) an important factor in narrative and game design. But specifically here, the heaviest emphasis was on the strength of the core mechanic and puzzle design. Those elements would be giving the player a lot to think about (and marvel at), and the core audience the game was targeting wasn't necessarily one who were automatically invested in narrative. 'Storytelling vs mechanics' is a false dichotomy, but we also wanted to be realistic about the game's audience and bring that mindset to decisions around the storytelling.

In the next post, I'll look at our approach to some of these constraints.