The sea frightens me. I know I’m not alone in this.

It frightens me for numerous reasons. One is just that the sea is bigger than anything else on the planet, and it contains any number of hungry, poisonous or simply revolting creatures. Another is that I saw Jaws at an impressionable age. A third is that, because of the first two, I’m not a particularly strong swimmer.

A fourth reason is that on the 25th of June, 1973, a little before midnight, my father flew his Phantom FGR.2 fighter-jet into the North Sea. He evidently didn’t do this on purpose, but because neither he nor his navigator survived, the details remain unclear. Much of the aircraft was recovered, but their bodies never were.

I was a year and a half old at the time, and I have no memories of him or of the event, but I think it’s safe to say that it had an influence on my development. When I decided to make a game about a black sea which eats captains, where death is telegraphed as the likely end of every voyage – even when I decided that one of the core story-lines was about the recovery and disposition of the protagonist’s lost father’s bones – none of this was at the front of my mind. I caught on pretty quickly, though. I think I slapped my forehead and went ‘GAH’.

Sunless Sea is not a tribute to my father, nor is it an autobiographical game. But the writing is influenced by my experience, and some of the themes are important to me. It’s one of the reasons I’m scared of the sea, and that’s one of the reasons that the Unterzee is such a menacing place. It’s the reason that the core story-line is about a father’s bones, not a generic mentor’s or parent’s bones. It’s why The Tempest and Peter Grimes stay with me, although I’ve never lived by the sea, and why you’ll find fragments of them here and there throughout the game.

A player tweeted me yesterday and asked if a Fallen London dream-card called ‘Night Flight’, about a dirigible crash, was a tribute to flight QZ8501, the AirAsia flight on which 162 people died last week. It wasn’t, because it was written several years ago. I don’t really have any insight into the pain of the people who lost someone on that flight, or on MH370, earlier this year; as I said, I have no memories of my father’s death. But if by an unlikely chance you’re one of those people and you ever read this, you have my sincere and heartfelt sympathy. Pain fades with the years; I hope you find yours survivable.

I’m casting about for a conclusion to an unusually personal post. I think I want to say that death is the only thing that everyone experiences – the only universal. Parenthood isn’t, the love of parents isn’t, even birth isn’t. I think one of the most important things that stories do is prepare us for our death, and for the death of others. I think that goes double for games. I think that death and the sea are hungry, but that doesn’t mean we have to wait for them to come to us. In the meantime, we’re alive. Happy New Year.

C.M. comments, on a post where I said ‘content spreads thin‘:

“On content quantity: it depends on the game design, I think. The reason Sunless Sea still needs more content is because it speaks to the main design problem. Most of the content seems to be backloaded instead of frontloaded. The game desperately needs varying storylines at the beginning of the game because the most punishing thing about death is repeating the same early stories every time. All you need is 5 to 7 near-London islands that each choose from one of 3 early-game stories, and interact with London, and the game’s chief design problem is solved.”

I think this is worth a post rather than just a comment reply, cos ‘it just needs more content’ is a common reaction from players, and I want to talk about why on the writer-designer side of the fence this can be a mistake. And, for novice writer-designers (man, I’ve been there), quite a dangerous mistake.

I once worked with a guy who told me a parable about a Buddhist master who went to a monastery overrun by mice. The master (the guy said) refused to let the monks put down traps, and instead watched the comings and goings of the mice, and found the hole in the wall where they came in. He instructed the monks to put down food just outside the hole, pointed out that the mice didn’t come into the monastery any more, and went on his way, presumably beaming smugly with the sun shining on his saintly shaven head.

Then presumably in the following weeks the monks found that the mice

(a) multiplied enormously
(b) kept coming back
(c) ate all the food the monks put down, and when that ran out, came back into the monastery
(d) meanwhile swarmed all over the monastery walls and found loads of new entrances and got into the grain stores and oh my god can you believe that guy, I don’t think he was even really a monk, I bet he was like a mouse spirit or something
(e) shit, now we have rats, too

I want to be careful about this analogy, because players aren’t mice. The point of monasteries is monks, and the point of games is players. But what the monks, of course, should have done, was: try to fix the problem by changing their design, not adding more resources. To put it another way, they should have got a fucking cat.

Back to Sunless Sea.

“On content quantity: it depends on the game design, I think. The reason Sunless Sea still needs more content is because it speaks to the main design problem. Most of the content seems to be backloaded instead of frontloaded. The game desperately needs varying storylines at the beginning of the game because the most punishing thing about death is repeating the same early stories every time. All you need is 5 to 7 near-London islands that each choose from one of 3 early-game stories, and interact with London, and the game’s chief design problem is solved.”

(C.M., I’m not beating up on you – I understand why it would look like that, and, this is a much more polite way of saying it than I’ve seen in three dozen forum posts).

Well, first, it’s not (IN MY OPINION) Sunless Sea’s chief design problem. That problem is that I vacillated between roguelike and CRPG design. If you’re interested in seeing me flagellate myself about this, it’s in the post-mortem here. 

And, second, if it were actually all we needed, we’d have done it. 🙂 Here’s why it’s not what we needed.

The content in Sunless Sea is heavily frontloaded. The farther you get from London, the sparser the content becomes. Partly that’s intentional. Partly it’s because that’s the order I mostly wrote it in (and, later, commissioned other people to write it in) – I got more hurried as we approached deadlines, and I generally went back to earlier ports to add more content.

But it doesn’t seem like it’s frontloaded, because although there are ten officers and a bunch of quests on the left side of the map, you see them over and over again. Content spreads thin.

So couldn’t we add more branching content? And you’d see different things each time? Well, there is some of that – the Father’s Bones story, for instance – and it does help. But it addresses the symptom, not the underlying problem. It doesn’t have a multiplying effect on quality of player experience – not even a 3x effect, and certainly not a 3x(5-7) times effect. No-one but a robot plays through the game lawn-mowing all the options with perfect equanimity. If the choice between the options is completely arbitrary, it’s not a very interesting choice; and players often get annoyed because they’re locked out of content until the next playthrough. If the choice is not arbitrary but depends on personality preferences or other story elements, they’ll often choose the same one on repeated playthroughs; if there’s an optimal choice, they’ll often go back to that.

But if you’re playing a roguelike or an ARPG, what about going back and playing with different builds? Absolutely. But now we’re well out of content or even narrative design, and firmly in system design. I’ve played the first level of Spelunky or Crypt of the Necrodancer I don’t know how many hundred times. They both of them have less story content in the whole game than you’ll find in one island in Sunless Sea. But Derek Yu and Ryan Clark are better game designers than me, and those games win on their mechanics.

tl;dr: If we had added more content to Sunless Sea, it wouldn’t have fixed the design problems; we’d just have someone saying, right now ‘you need to have six, not three, variant stories on each of those five-seven islands’.

Am I really saying that adding more content to Sunless Sea or Stellaris would not improve them? God, no! If you like it, you’ll want more of it (and you should totally buy Zubmariner when it comes out, and I’m not in the loop on current production but you might also find, shhh no one but us here right? that it addresses some of the design issues).

My gut feel is that Stellaris is not yet past the point of diminishing returns on adding content. (I think – just from published dev diary stuff, no inside commentary here – the Paradox guys are also very well aware that there are design changes they could make that would make more of the existing content, and they’re on that.) My gut feel is also that Sunless Sea is past the point of diminishing returns in adding content to the vanilla game. We could have kept adding content for another year, and never launched. Here’s the thing.

The right option for the monks in my original analogy was to get a cat, because cats work pretty well, are cheap, and don’t require specialised handling. It may seem like it would be a better option for them to get a lion, because lions are awesome, but lions are expensive and eat people as well as mice.

In other words, design problems are always about what’s possible as well as what’s best. This isn’t some sort of whining, bean-counting, creativity-killing quibble, it’s the difference between life and death for any studio. So this is the part of the post that’s directed at people who want to make a game, or are making a game for the first time, and it may save your life, seriously. Run the numbers. Don’t trust instinct. Estimate. Run the numbers.

‘5-7 islands with three stories each’ – it sounds so innocuous, doesn’t it? But let’s say each of those extra stories is about an island’s worth of content. I was writing SS content pretty fast by the end, so call it 2.5 days each, so that’s 2.5 x 2 x 6 = 30, or a month and a half’s work. Except that’s if I worked interrupted, which never happened, because the studio needs running and my kid gets sick and I need to work on the combat redesign and yadda yadda, so more like two months, except it also needs testing and debugging and reworking after feedback, which is hard to estimate but easily that’s 2.5 months now.

(There are also subtle but serious effects when you’re adding more content to an area. I’ve described it elsewhere as being like a snowball: you can’t just keep rolling it bigger indefinitely before things start falling off. It may seem weird that there are so many things going on in one area; there are more likely to be continuity issues; it’s harder to come up with ideas that feel distinct and interesting; and so on.)

Could we have eaten another 2.5 months? Maybe. It’d have been a pretty serious risk. I mean, for the size the studio had grown to, that would have been, like, 70K GBP, and we didn’t know then whether Sunless Sea would be a success or not, and if it hadn’t that 70K GBP would have been essential for keeping the lights on a little longer. But remember, this wouldn’t have actually fixed our problem, it would have just helped a bit more with some of the symptoms.

If you’re an indie full of vim and enthusiasm, it’s very tempting to think ‘I’ll write more, and that will fix it’. Sometimes it’s true. It’s true if you actually haven’t written enough. But if you’re a writer, you’re probably going to write more than you need. Your first instinct will always be to fix problems by writing more content. Don’t feed the mice. Don’t buy a lion. Find the right cat.

If you like it, take a look at my previous columns. They’ve included pieces on gravity in games, an alternative history without D&D, the People’s Crusade as it relates to Kickstarter, and six reviews of nonexistent games.

Fallen London’s an oddity. More than a million and a half words of sort-of-multiplayer online interactive fiction, free-to-play but polite about it, kinda grindy but absolutely crammed with story: a videogame with no moving pictures at all.* I originally built it, and I founded Failbetter Games, who still run FL. Yesterday I left Failbetter, so I can finally use Fallen London to illustrate a point without feeling like I’m plugging it. I don’t want to talk about Fallen London, exactly: I want to talk about endings.

For years (FL has been running for seven) people asked: how the hell is it going to end? It’s a story-based game, and one of the defining qualities of stories is that they have an end as well as a beginning and a middle. There are exceptions, but one of the defining qualities of a horse is that there’s a leg at each corner, even though some horses have three legs. Stories, basically, end.

How do you do that with a free-to-play game where players want to keep going forever? If a player can still play, their story isn’t over; if they can’t still play, they’re upset (and, candidly, they won’t make a free-to-play game any more money). This is a problem that (eg) MMOs face, too, but the story in MMOs is generally an afterthought. Fallen London, notoriously, has almost no gameplay. It’s all story.

We did settle on a solution – in fact, a couple. I’ll talk about that in a moment, but I want to say a few things about endings generally first.

One of my favourite, and Frenchest, quotes is something Balzac said: that coming up with ideas for stories is the fun bit like ‘smoking enchanted cigarettes’. Beginnings are fun; they’re barely more than an idea, and a promise of an ending. It’s much easier to make a promise than it is to fulfil it. To finish a story you need an ending, and endings are hard. (Middles are even harder, sometimes, but that’s another column.) There are a bunch of reasons for this – fulfilling that initial promise, being both surprising and inevitable, all that good stuff – but also, just finishing something, deciding when to call it done, is something that only seems easy to everyone who’s never tried to do it.

Deciding when to call a story done is even harder when player action can alter what happens. What if they do something that takes the plot elsewhere? The first-pass response to this, the one people usually expect, is to add multiple branching endings. The first problem with that is: when the hell do you know when to stop? You’ve just gone from a flat to a three-dimensional problem, and every writer of non-linear narrative knows the devouring temptation of adding just one more ending.

But in any case, multiple endings generally aren’t multiple endings, exactly, if the player keeps going after the first ending to look at the others. The first (second, third…) endings become late middle, epilogues to the real ending. Or, sometimes, the first ending is the Real Ending, the one that you chose in your personal headcanon. So even multiple-ending games still have one ending, and then one kind of epilogue or another.

Some games lock this down hard. Big CRPGs branch their endings based on stuff that happens earlier, so you have to replay the whole game to see something different – and that feels more like a reboot or a retelling, an extension of the same story. No one sane will reload a game from five hours back to view an alternate ending (although there are plenty of non-sane YouTubers to whom I am very grateful for their ending videos).

So if you have a big, long, slow feedback loop, you can make multiple endings feel like different endings. But that still doesn’t help with a game that can’t afford an ending at all. So Fallen London took two distinct approaches, with two different strategies: one, make sense of the desire to choose have an ending, but allow them continuity. Two, make players want an ending anyway.

The first approach was Destinies: what some people have called an ‘equippable ending’. At rare times of the year, characters can experience a dream of the future. They get to choose a really cool but also horrible thing that will happen to them as the first scene of the last act of their life. They bring back the memory of that dream, and equip it as a trophy item that gives stat buffs. They can change the dream later – exploring multiple endings – but in the meantime, they get to continue in the knowledge of the ending, seeing it have a real effect. “If you can’t solve the problem of players wanting closure or continuity, give them both.”

And the second approach was FL’s most demanding, aggressive, ridiculous storyline: the Search for Mr Eaten’s Name. Players are told that this quest will not end well, and that they’ll destroy themselves and everything they love (especially and mostly, their cool tools and phat lewts, but also their friends, spouses, past, future and more). If you know anything about human nature, you’ll have guessed that this is a tremendous hit with a minority of really core players – because of, not despite, the fact they’re working hard over a long period of time to set fire to their character and watch it burn. It’s not a con or a trick. There’s a huge amount of story in the Search. But the Search isn’t joking around when it gives the player a quality called A Bad End. “If you can’t solve the problem of players wanting closure or continuity, make them want closure.”

Non-linear stories work like and unlike linear stories. We’re still learning the unlike things that make them work. Even endings can be unlike. Sometimes, with a non-linear experience, you need to embed the ending in the middle, or, as it were, crush it up and stir it into the rest of the game. It’s even harder to decide when you’re done, but if you can’t decide when you’re done, you don’t have a story, just a mess of offcuts. And not all stories are tidy, even linear ones. Sometimes you just have to stop, right in the middle of everything.

* I was at the ceremony when Fallen London won Best Browser Game in 2009. The games in other categories were things like Warcraft and Arkham Asylum, so the looping ceremony video cut from high-end fight scenes to a camera panning optimistically across a beige web page. I couldn’t look anyone in the eye.