Foreshadowing and Layered Storytelling

Spoilers for Life is Strange and Night in the Woods. Night in the Woods spoilers are hidden behind a button (in the web version) and you can skip them.

Recently I've seen a lot of games stumble over the concepts of foreshadowing and layered storytelling. Even really good games. So let's talk a little bit about it.

This might be a bit rough, it's my first take.

Games have twists. At the end, in the middle, twists everywhere. All these twists have to feel like they make sense.... but they usually don't. They often feel like they "came out of nowhere".

There's usually foreshadowing: nobody simply forgets to set up the ending. They just... foreshadow badly.

The two most common kinds of bad foreshadowing are the accidental reveal and NG+shadowing.

The accidental reveal is pretty obvious: the foreshadowing is so blatant that it simply gives away the twist.

NG+shadowing is the opposite. The foreshadowing is so minuscule or unrelated that you can only see it as foreshadowing when you're playing the game a second time.

You might think we're on the hunt for goldilocks moments: foreshadowing that isn't too blatant, but isn't too subtle.

Not me. I say both those kinds of foreshadowing are misunderstanding the role of foreshadowing.

Let's get... hoity-toity.

The twists that stay with me are the ones that are thematically relevant.

Let's say we're writing a game about a main character struggling through a rough time in their life. The backdrop to this is a series of mysterious murders: our MC lives their life, notices murders happening in the background. At the end of the game, the two plot lines come together and there's... some kind of cool twist.

If you're a writer, you need to think about how these two plot lines support each other. They're not just two stories happening in the same place: both need to be stronger because the other exists.

The first instinct is to make them related physically. Our struggling MC witnesses a murder. Or is suspected of being a murderer. Or one of the friends is a suspect. Simply moosh the two plot lines together and then stitch things up. It works. Ish.

A good writer will think about more than that. We need the two to support each other thematically. The serial killing is a twisted reflection of our MC's internal struggle. The twist at the end is how the two separate or merge, and it reflects not the meanderings of the plot, but the slow syncing of the two themes.

...

Let's discuss and compare Life is Strange and Night in the Woods. They both have the same main character and a similarly laid-back approach to gameplay, so they're a good contrast. Since NitW is new, I put the NitW spoilers behind a spoiler block. You can skip them.

...

In Life is Strange, our main character has psychic powers. These help her to mask her social anxieties by allowing her to simply undo social interactions. Initially, her personal arc is about coping with or overcoming her social anxiety - perhaps with a highlight on the awkward fact that she's attracted to her childhood friend.

This arc is probably strong enough to sustain us through the whole game. Our B-plots each episode could easily integrate in. We can talk about Kate, who faces social execution due to a video of something that was not her fault. We could talk about Chloe, who has no 'social' because people keep leaving her and she's burning on a combination of deeply loyal raging and deeply bitter self-isolation.

These segments integrate well with our personal arc. Our worries about our own inadequacies and mistakes are shown as funhouse mirrors, blown out of proportion. Some segments are minor, like Victoria's cronies briefly admitting their own social worries. Some segments are major, lasting almost the whole game and providing a constant thematic companion - Chloe, in this case.

However, our personal arc doesn't require any superpowers. That story could be told with no superpowers at all.

The role of the superpowers is not to advance Max's arc. Max is responsible for advancing Max's arc. Instead, the superpowers are a mirror of Max's arc, showing in more concrete terms how Max is advancing, and giving us a simple tool to help tie other arcs to Max's own arcs. Think about it: if Max's powers reflect Max's place in her own arc, then naturally they'll be good at interacting with arcs that also reflect Max's place in her arc. Max's powers can be a needle and thread to stitch arcs together. They give us, as authors, a tool to make Max's arc feel palpable and clear.

Let's make it plain:

We start Max with the power to briefly rewind time, which we initially use mostly to let the player re-do awkward social moments. This gives us plenty of opportunities to cement both who Max is and who she wants to be. This basically represents Max's ability to consider other people, rather than just wallowing in herself: via retrying, she can consider how other people will react and plan her conversation accordingly.

That said, the limits are more important than the capabilities.

We can meddle with other people's lives, such as stepping in for Kate and Chloe. But we quickly learn that no matter what we say, there are things that can't be fixed with a few careful lines of dialog.

The arc of the powers reflects her own personal journey: she's just starting off, so she is pretty limited.

We slowly ramp up her powers to interacting with physical things. First with miraculous timing, then by being able to hold items while rewinding. In the real game, these are all mixed together... but we're talking about our own theoretical version, so we can separate things out.

This growing power reflects Max's interest in being better at peopling. The things she can't do with a snippet of dialog, she can do with physical objects.

It's very possible to use these powers to tie arcs to our current situation.

For example, the "Victoria on the stairs" puzzle, where she blocks our way until we pour paint on her. We can't convince Victoria with words, so we convince her with actions. In the game, this is one of the worst puzzles, since it makes so little sense, takes so long, and is immediately rendered pointless with another nonsense puzzle. But thematically, it's a good idea. A similar puzzle could have worked well.

For example, the Warren Gets Beat Up scene could have been replaced with a Making Warren An Accidental Badass scene. That would have felt awesome, and would have allowed us to take Warren down a self-destructive path for later use in a different arc.

If we really wanted to have it be a Victoria scene, why not have it be something that makes sense? Victoria's goons found Max's satchel, and are saying that nobody can be SURE whose satchel it is, and that they should probably go turn it in to the lost in found... in Pensacola.

Both of these puzzles are about situations dialog can't resolve. Both of them are about people who get things done beyond the reach of simple dialog, and about people who are also suffering from concerns dialog can't reach. In fact, we could get away with outright stating that as a throwaway line. Max sneaks up on Victoria and friends, hears them talk: "It's all just talk! If talking could solve everything, my mom wouldn't be in the hospital dying." Etc, etc.

This section of Max's arc could culminate in the second episode, when Kate tries to commit suicide. In canon, this situation is beyond Max's powers of time control, and represents a chance for Max to step up and personally resolve the situation without relying on "cheats" - proving that she no longer needs them. It represents Max closing the circle: dialog spoken from the heart is powerful enough to do what her carefully calculated superpower dialog could not do. Max has outgrown her powers.

In canon, Max's arc changes from social anxiety to heroism. More specifically, the theme becomes "how far will you go to stay with someone you love", which is also a very potent arc. I have no problem with this arc change, because it builds off of Max's earlier arc in an organic way and is completely believable.

Unfortunately, the arc doesn't follow through: the author clearly didn't think it out well enough, and the story devolves into a macabre serial killer dream. While the individual elements are written well, they don't thematically tie in with either of Max's arcs, and the whole thing falls apart. The finale does drag it back around, but only after hours of serial killer nonsense actively detracted from it.

That weird serial killer sequence? It was a twist that came out of nowhere.

It was foreshadowed, but it still felt awkward and forced, because the themes didn't line up.

Creating gameplay based on your theming is a powerful tool for a lot of reasons. First, it gives you a great lead: no matter what you're trying to do in the game world, you know how it should feel and what it should be about, so it's much easier to develop. Second, it draws the player in and keeps them in.

We can craft moments of triumph and pain that work within our arc - either our social anxiety or our heroism. Finding that Rachel died and watching Chloe crumple will gut us either way, but subtle highlights on Max's reaction will tell us whether Max is struggling to support Chloe or rescue Chloe. Those are very different feelings and players will pick up on it, especially as Max's subsequent actions struggle to do one of those things specifically. IE, are we trying to go back in time to rescue Rachel? How about to prevent Chloe from meeting Rachel? How about just supporting Chloe with hugs, pot, and ice cream? How about taking advantage of Chloe to bind her to you emotionally? All of these are valid paths which support different arcs in different ways, and develop Max in different, interesting directions.

Which direction you choose is guided by what path you want to take. And, at the end of the game, the ending sequence will be about resolving that mess.

The canon ending is: sacrifice the town for Chloe, or visa-versa?

However, there are other, equally powerful possible endings. For example: wipe yourself out of existence. Chloe's dad wasn't there cooking lunch for you, never died. Choose which of two or three Chloes is "real", and face down your nightmares of the other Chloes - it's even worse because they forgive you. Rewrite history so many times that there is no "real" Chloe, only your Frankenstein's Chloe made up of a hundred different pasts stamped together with superglue.

Those are all "heroism" endings, but there are just as many "social anxiety" endings. Solve Chloe's problems, or let her work through them? Leave Chloe straight, or just tweak reality slightly so she's your love interest? Etc, etc.

Knowing our goal allows us to build up a path towards it, choosing gameplay, puzzles, B-arcs, and supporting cast that highlight our journey. Usually this is about building up Max's vector until it is out of control, then simply pointing out that it's out of control.

I say "vector" deliberately. This isn't about letting Max use her power to stop a serial killer or whatever. That's not part of either of these personal arcs.

Technically, sure her powers are being used. But not in a way that reflects her personal journey. It's just out of desperation - and live-or-die desperation isn't part of her personal arc, not unless it's on someone else's behalf.

The serial killer arc could have worked if Chloe was the captured victim - Chloe ran off to check the grave without you, and she was gone when you arrived. It would have been interesting to let Max go off the rails trying to rescue Chloe from ever-earlier points in time, watching reality disintegrate.

But I still hold that a serial killer wasn't even needed. Max's powers are the serial killer. Jefferson's role could easily have been to narrate Max's fall from innocence in a much less murdery way, by simply being a smooth-talking pervert. This would have been useful in the part of Max's social anxiety arc where she learns that she can't really know someone by simply collecting their responses.

In the end, Max choosing whether to sacrifice Chloe or the town only makes sense if Max's arc is "Chloe and I will be together at all costs". And that can be established best by creating this universal, stitched-together flow.

That's the best kind of foreshadowing and layered storytelling to me. The two episodes spent on serial killer stuff weren't stitched in properly, and that was the game's clearest failing... despite the fact that it was technically both foreshadowed and layered.

What I'm saying is all of the B-arcs, puzzles, and twists should come out of their thematic connections to the core arc.

...

Now, how about Night in the Woods?

Night in the Woods is really good and the same length as Life is Strange, go play it instead of reading spoilers!

Well I loooove spoilers! Click here to spoil NitW!

Night in the Woods has the same main character, but instead of mild social anxiety, Mae clearly has something very wrong with her. It becomes clearer with every passing hour that there's something broken in her head and she should really be seeing a specialist.

Mae's personal arc is harder to set up than Max's. With Max, you start off knowing almost everything and can just walk alongside as she moves forward. With Mae, a lot of the power of the arc comes from you not knowing precisely what is going on, either with her or the story. This approach relies heavily on mystery to keep pulling the player forward, supported by brilliant dialog and incredible visuals.

Night in the Woods paces the arc out by using a number of B-plots which advance alongside you. You're technically allowed to choose which B-plots you want to engage with, but the constraints of the game mean you end up choosing almost all of them and only advancing a set distance down each. Aside from some minor constraints, this is the same as using a linear story, so let's consider it as such.

In practice, these B-plots are mostly getting to know people around town. From the central cast (like Bea, your oldest friend) to super-minor characters (sports fans drunkenly shouting catchphrases). In the first segment of time, Mae's homecoming feels like a homecoming, reconnecting with the locals and her childhood. In the second segment, things start to fall apart, in much the same way Mae fell apart. In the third segment, Mae is rescued by her friends and comes to terms with her personal issues. In theory.

In practice, the first two arcs are really solid and the third is very shaky.

The solid arcs are solid because all the supporting elements are solid. She comes home, and the only changes that are highlighted are the pointless ones - one snack shop became a different snack shop. Highschoolers are now slightly older highschoolers. Her best friends still hang out and have a crappy band, not made any less crappy by Mae's return.

The arc moves forward, and things start to fall apart for Mae. She's more socially awkward than ever, in fairly inexplicable ways. She's getting sick, tired, and having hallucinatory dreams.

This is supported by all the B arcs. She sees a ghost kidnap a kid. She finds a severed arm. The town council is refusing to let the pastor help the homeless. Bea is angry at her, and has a rough life. Gregg is happy to see her but is in the middle of leaving. Her mom pressures her until they get in a bitter fight. She finds there's a freaky cave under her house and her grandfather's family treasure was a human tooth ripped out of a still-living human's head. She raises and releases a swarm of rats. Everyone recommends she see a doctor. Oh, and her cop aunt has unsettling things to say.

All of these are "things are going awry" arcs, and they work well to give the story a strong forward thrust. Without this combined tide of supporting elements, the game would probably be unplayable, since in the second arc Mae is pretty unlikeable and her illness is largely inexplicable. These arcs provide the thrust needed to move into the redemption arc.

It's a mistake to just say "these arcs" and leave it at that.

It needs to be clear: these arcs are the meat of the game. They do give the game the thematic thrust to move forward, but that's backed up by brilliant dialog and exceptional art. The thematic match stitches all those elements together into a compelling whole rather than leaving them as beautiful but disjointed vignettes.

Unfortunately, it falls short in the third arc. It's not a bad experience, but the thematic matching disintegrates and we end up with something that... comes out of nowhere.

Well, literally and figuratively.

Mae's third arc is when she properly disintegrates under the pressure. None of her friends believe her, even though they see the same things she does. It becomes more clear that things are - physically and mentally - Not All Right with Mae.

Unfortunately, there aren't very many helpful B-arcs. Most of the B-arcs are closed, or stop matching up thematically. Only two really make sense to me: your second hangout with Horror Film Girl is a fairly powerful sequence where you talk about wanting to be monsters. Also, when you accompany Bea to a party, not only is it clear that Mae is Really Not OK, but it's clear that Bea is Really Not OK, Either, and you're Really Not OK Together. Nice reflections in both cases.

There are other scenes. For example, you investigate various ghost phenomena. However, this is where things really start to disintegrate.

These ghost locations make no sense. A small building with more elevators than rooms, and something chases you from elevator to elevator. If it was Mae on her own, it's easy to explain: she's delusional. But Gregg is with her, and sees the same things, so now the question becomes: what supernatural thing is happening? Why is the building like that? What is chasing them around, since nobody knew they were coming and nobody should know where they are in the building even if they did show up?

All of the ghost hunting events are like that. There is definitely something very real about Mae's concerns. This is a huge change in the direction of the story, since up until this point there was no sign of anything supernatural, and Mae's personal difficulties seemed to be the center stage.

It's possible to entangle Mae in supernatural affairs and have it work. You just need the supernatural affairs to properly synch up with Mae's personal affairs. In the same way that all the plots synched up in arcs 1 and 2.

But it doesn't synch up.

Out of nowhere, it turns out that Cthulhu lives in the basement and there's a cult of dads hurling bad children into a pit. Moreover, all of Mae's mental problems are because Cthulhu is talking to her.

There are two big problems with this.

One is that it's really not great. People can simply suffer from mental issues. It's allowed. You don't have to have a paranoid delusional prone to psychotic breaks be right. They can just be psychotic and delusional. It feels oddly disrespectful to learn that none of Mae's problems are actually Mae's problems and she can just... shrug out of them by yelling at Cthulhu.

This could have been viable if it were an attempt to represent something - like faith can help you through tough times, or the universe is too big and uncaring to worry about things too much. But if that was the intent, it wasn't done very clearly.

The other problem is more structural: it doesn't line up. There's no thematic continuity.

Let's consider this sentence:

"There's no hint of a cult or that supernatural things even exist in this universe until - bang, dad cult out of nowhere."

This is a common complaint I've been seeing, and it's valid. But it seems to me to be missing the point.

There is some foreshadowing of the evil dad cult and Cthulhu. The town council gives some hints, and the dreams give some hints. Especially the sphinx dream - it is clear foreshadowing.

But it's not very good foreshadowing. It's NG+shadowing.

The dadcult isn't a good match for Mae's personal journey. It brings it to a close, in that the dadcult and related dadgod nonsense caused her problems and killing them all solves them, but it's a particularly unconnected plot arc that doesn't line up with Mae's personal journey.

It could have worked. Mae is a screwed-up person trying to make a place for herself and those around her. The dadcult is screwed-up folks trying to make a place for themselves and those around them. Even the dadgod could easily have been portrayed as a screwed-up eldritch abomination trying to make a place for itself and those around it.

Think about it. The sphinx dream was all about how annoyed and grumpy the sphinx was. Why not have the sphinx complain about things like how ever since the rip in the sky opened, it's been hard to talk with the other gods, since they're still eternal and unchanging but now I've been bound to this stupid flowing stream of useless thoughts and dreams.

The dadcult needs stitches to bind it to Mae. Over the course of the game, the only stitching attempts were to show that a ghostly figure was following Mae around and being creepily near-supernatural. When we reach the dadcult proper, it's just "here's a huge data dump blaaarrrrpphhhglglele".

Rather than foreshadowing, it needed flat-out time. The dadcult needed to be revealed more slowly, so it could start to sync up.

Obviously, in our magical machine of make-believe, we can throw as much time and money at the project as we like, so let's pretend we arranged it a bit differently. Instead of discovering the dadcult during the third arc and having a fun time falling off cliffs and limping around, let's presume we find the dadcult at the end of the second arc.

Rather than just being told that the police didn't care and our parents didn't respond, we could have been allowed to talk to aunt cop in person and have her dismiss our concerns. This would have given us a lot of opportunity to bind the situations together.

How?

Mae's personal problems are weird and largely ineffable. She can't explain them, and they're destroying her life by making her unable to understand things everyone else takes for granted. We could show the cop having the same problem, or Mae's problem extending to the situation, or both. And then we could do some supernatural/psychotic break hinting.

Mae tells the cop there's a cult. The cop says it's hunters, and Mae should be more careful, not go wandering around in places she doesn't belong. When asked why there was a massive group of hunters wearing deer heads and shooting stuff before hunting season, the cop says it's a small town and sometimes the customs don't make a lot of sense. When pressured, the cop says she can't explain something she can't explain. There's a short pause, then the cop's visuals glitch out. She becomes just shapes. For a moment.

This would obviously be a very different progression. In this version, Mae's not so badly injured, if she's injured at all. So the pacing has to be redone to allow Mae to get injured in another cult event.

At this point, dadgod hasn't been introduced. If we want to introduce dadgod, we would need to show some flat-out supernatural shit. If we want dadgod to be responsible for Mae's problems, we'd need to set that up a lot earlier with a lot more pseudosupernatural powers.

There's a lot of space for that. "Nightmare eyes" Mae? Yeah, just give her literal nightmare eyes.

Personally, I'd prefer it without a dadgod. The cult is just a cult, Mae just has problems. I think the supernatural element is not necessary, although I do like the idea of dreaming of ancient gods. I like the idea of the gods in Mae's head reflecting her personal struggles. Talking to cryptic monsters is a good way to work through your own cryptic thoughts.

Anyway, I hope this didn't sound like whining. Night in the Woods is superlative, and will probably be remembered as a masterpiece. Although I felt the dadcult was rushed, the game remains excellent.

I just felt compelled to write about this concept, because it's the one thing Night in the Woods falters at in the end. They "foreshadowed" the dadcult, but that kind of foreshadowing just isn't what they needed.

They needed to stitch the dadcult themes to Mae's themes.

Just foreshadowing doesn't make something feel like it belongs.

Galactic Line Missions

The mission system in The Galactic Line is the foundation that supports the personal-level play. Missions give a reason for the ships to exist and travel, gives goals to the crew, and throws complications in that give texture to their daily lives.

Missions are built on the resource system. So let's talk resources.

The most fundamental resources are the tech level economy points. Currently called "T0", "T1", etc, these replace both generic materials and scientific research. You don't unlock a new kind of superconductor, you get a T2 economy point. And these points get spent on building colonies and space ships. Simple enough.

There are also a variety of resources used to balance the construction of ships and colonies. Things like power, administration points, etc. These create a terrain for the players to build a variety of different kinds of ships and bases in a variety of different circumstances. Mods or custom settings can easily alter the playing field here.

But that's local stuff. Why do ships run around the cosmos? Here are some small possibilities:

1) Local rare resources. However, mining or research colonies are a better long-term investment.
2) Temporary situations, such as negative space wedgies or plagues.
3) Freight. Shipping resources around.

But those are not really enough to sustain a proper mission system. They're very basic and either too boring or too unpredictable.

Instead, the most critical trick is that resources can be local.

Local resources are simply resources that are about or from a specific place. For example, if you scan planet X-9, you get cartography points for planet X-9 specifically. Not generic cartography points.

There are many facilities which convert resources. Normally this is generic resources, like a solar array converting administration points into power. But a residential block converts cartography points into colonists, and a trade center converts cartography points into investment. These conversions maintain the localness: feed X-9 cartography in, you get X-9 colonists and X-9 investments. It takes time: conversions take some time, gathering resources takes time.

If you want to build a colony on X-9, you use X-9 colonists and investments.

This means your chain is to spend time at X-9, then to spend time at a colony world (or start those conversions up and come back for them later), then back to X-9 to found a colony. This has a fair number of gaps in it that can be used by other kinds of missions, allowing for multitasking.

For example, a simpler use of local resources is scanning a sun or solar system in general. Solar readings of the star X take some time to gather. You might be able to do it simultaneously as mapping X-9, depending on your setup. Either way, those localized readings can be used in a variety of ways. For example, they can be baked down on-board, converted into generic astrophysics points over a month or two. This doesn't require you to be anywhere specific, and can therefore run in the background wherever else you go.

Astrophysics points and local solar readings can both be used on any colony with a lab to create T2 economy points. It requires either a lot of astrophysics points, or readings from a lot of different stars. Either approach will pay off in the long run, and can be used at any colony you think needs more T2 economy. Of course, it takes time. You could drop those things off at the colony you're gathering colonists and investments from, do all three missions at the same time. Or you could start the colony missions, leave a crewmember at each, and fly off to take more star readings.

You also don't have to use those mapping points to build a colony. You could convert them into T1 economy points in much the same way. So your overall methods and objectives are up to you.

This is called an "open resource chain": a variety of resources that can be knit together in a variety of ways. This allows for mods and alternate builds to integrate in at any level. The key, from the dev perspective, is that local resources create a reason for star ships to keep moving. Also, during the missions complications arise, which gives players an opportunity to have a personal scenario where they interact with a bunch of people.

The actual interface for this is via "missions". This is a dense and complex environment, especially when the player doesn't know precisely what capabilities are available in which places. Missions offer a method to organize that.

A selection of missions are offered whenever you go someplace new, by simply batching up possible resource transactions. You go to X-9, you see some missions for mapping X-9, some missions for scanning the sun or the solar system. You also see missions for refurbishing your engines, studying for promotion, and a bunch of other ship-centric missions.

You can create more missions if you wish to customize things. It involves interacting with people that represent mission options, I'll cover that some other day.

After mapping, you go back to a colony. The residential district can convert local mapping into colonists, so you get at least one mission about that... and you can choose to look for more if you want other options.

These methods allow the player to have pretty good options offered no matter where they go, and they'll quickly learn "gut instinct" mission chains when they see certain resources can be converted in certain ways.

...

Galactic Line's core gameplay is scenes where you get to interact with people in a rather open way. This gameplay about choosing destinations and missions is a setup to help that happen in a convincing way, because mission complications and alternatives offer a method to unify and connect characters, as well as give them something to care about and a background event the player cares about.

This allows us to sidestep the problem of so many continuous-interaction games like The Sims. The player becomes inured to the soft, indifferent day-to-day pressures and begins to aim at the edges of the simulation. That's why so many people try to murder their sims, or get a ghost baby, or whatever. It's also why Rimworld players tend to vivisect their prisoners.

But if every "day" the player actually plays through is centered around a specific mission complication, the player will have a goal, the characters will have goals, and everyone will have a reason to share the stage and connect with each other.

As a demonstration of this, think about The Sims. Think about those specific events where you were throwing a party. You were probably focused on your goals for the party. For some people, this might have been "drowning everyone in the pool", but for most parties it was about raising friendship ratings, providing enough food, achieving sims goals, and maybe hooking someone up. There's a drive to get the party "running smoothly".

Imagine if playing the sims was done with much lower time pressures, but you didn't play each day. Instead, you had a weekly planner and only played the days when something important happens. It's a bit like that.

In theory.

Galactic Line's Battle System

I've been vague about the battle system in The Galactic Line for a few reasons. I knew what I wanted it to include, but not how I wanted the pieces to come together. I don't want the game to devolve into yet another space ship battler. But I have worked it out now, and here's a quick description from the point of view of a commander.

Your ship and an enemy are stuck out in an asteroid field somewhere. You don't know where they are. When you zoom out, you can see a hex grid around your ship: a hex for your ship and six more neighbors. Three are above the horizon, three are below. As you rotate, the hexes do not.

As your sensors scan the skies, they pick up traces of your enemy. A bit of leftover exhaust, a bit of radiation, a glint of light. Dunno where they are at the exact moment, but we know what general part of the sky they're in, and some of those hexes darken. "Not that way". Which of the remaining hexes is the right direction? Who can tell?

Eventually it's down to two, then just one. You accelerate in that direction, trying to close. You don't have a solid lock, just a general knowledge that they're in that part of the sky.

Then lasers come crashing in, immediately blowing apart your armor and tearing through your systems. Yeah, it came from that part of the sky, but what the hell?

The sim ends and your competitor asks why you didn't launch any sensors or drones or anything.

So you start over.

This time you realize that those hexes are destinations as well as directions. You launch sensor drones into a hex, and in a few moments they arrive and begin scanning. You launch some missiles and have them wait quietly in another hex. You fire a spray of water in a direction you think the enemy might be: it turns into a cloud of ice, preventing scans from that direction from picking up your heat signature, breaking up your profile. Later that hex goes dark - the enemy isn't that way. You replace it with a pack of missiles, patiently waiting for a chance to attack.

You realize you can zoom out again. All seven hexes shrink into a central hex, and you're faced with six new, bigger hexes. Three high, three low. You send a sensor drone out into one of them, and notice it'll take quite a bit longer to reach.

You zoom out again. Those seven hexes shrink into one central hex, and a new six appear. Now you're looking at a fair chunk of the asteroid field - there's actually an asteroid in one of those hexes. Deploying sensors to it looks like it'll take half a day, and you think about setting course, but get the warning that turning on your engines will more than triple your profile.

And then you notice there's a glowing red perimeter around those hexes. Mouseover: "Enemy scan warning". The enemy has figured out you're in this volume of space, although it's obviously a pretty weak lock given the scales involved.

Now you're starting to get a handle on this. You've deployed some offensive and defensive systems, you're winnowing down the enemy's direction. This time, you get a lock first. You have to decide whether to open fire or refine your firing solutions: if you open fire, the enemy's gonna detect you pretty much immediately.

Well, you fire those missiles you set aside. The enemy shoots most of them down, but you get a much better firing solution from them and fire your own laser barrage. This time you win.

The next fight begins to unfold the same way, but when you start to get a good lock-on, you realize the two of you are within the same hex space. Your systems overlap, things become a mess, and then there are bombers on top of you.

As you progress, you learn a lot more tactics. ECM drones specifically calculate enemy scan patterns and help you avoid detection. Anti-missile interceptors. "Buzzers" that draw off the enemy sensors and incoming missiles. Sensor drones arrayed in "shotgun" lines along different scales of hexes. Weapon drones. Moving drones to new locations, or retrieving them. Radiation storms or ice clouds to mask your profile and the profile of your drones.

All of this is on top of managing your own profile. The higher your profile, the quicker the enemy will narrow down their search. You have these fancy high-tech deflector shields, but if you use them, your profile spikes so high the enemy will immediately detect you. You find out through experimentation that you're better off actually leaving them off even while you're under fire, if the range is long. A low profile helps more than a barrier.

Then you get to fleet logistics, and begin to struggle with the idea of deploying a ship to a distant hex and then having them have local hexes within that hex... and you begin to work with noncombatants and fixed fortifications, requisitioning or hacking for sensor logs, resources, triangulation help.

And that's how The Galactic Line's battles work.

Even a one-on-one, ship-on-ship battle takes place at great ranges, on a strategic map. There are moments when someone warps in right on top of someone else and a slugfest starts, but no matter how epic that feels, ships are expensive. Hugely expensive. And the people who die? They're people. Your people.

You begin to appreciate the art of getting a hard lock and then pinging the target with targeting lasers, forcing them to surrender without firing a shot. You begin to appreciate the art of convincing a local politician to get the enemy fleet to stand down.

Hopefully.

Anyway, the same system is used outside of combat, especially when scanning planets or assisting colonists.

... the same system is used for building colonies, actually. That's what gave me the idea.

I haven't settled on this 100%, but it plays well on paper. I can't really justify arcade-style action.

Humanizing an Empire

As you may know, I really love games where you build things. And what is the biggest thing you can build?

An empire.

When I was younger, I was obsessed with 4X games like Civilization, Masters of Orion, and so on. As time wore on, their shine wore off. These days, I still play those kinds of games on occasion, but I prefer games where I build smaller things.

The reason is simple: those smaller things are more people-centric.

Recently, games about building empires have tried to become more people-centric. Civ animates their iconic rulers with loving care. Stellaris gives you a bunch of people with personal traits and names and asks you to assign them to colonies, star ships, factions. Everywhere, people are becoming more common.

The problem is that these games aren't about people. They never were, and they can't be: the structure of the game is oriented around nation-building, not personal dramas. This means interactions are all about bowing to the will of the empire.

Games like Crusader Kings II push this as far as can be reasonably expected, tracking hundreds of distinct individuals throughout lifetimes, letting you push them into service, crush them, or exploit them. This appeals to a certain kind of fantasy, I'm sure, but it's not really very good at letting me build something that means something.

This actively interferes with my enjoyment of these games. Civilization VI isn't about building a nation, it's about beating Alexander the Great as he weirdly and obsessively declares war against you every decade for a thousand years. It's humanized to the point where I don't think of that territory as another nation - I think of it as Queen Victoria's house. Vickie's front yard.

We've been humanizing these games for eons. Everyone remembers Gandhi's words backed by nuclear weapons. Everyone remembers the personality-filled cockpit views in Star Control. But these were very shallow, passive efforts - the difference between anthropomorphic tags and being a person.

Basically, when something is tagged, you can easily slip a personality onto it and ascribe a mood or emotion behind its actions. The Ur-Quan dreadnaught fighting the Spathi? It becomes the grumpy, remorseless Ur-Quan hunting down a hilarious, cowardly Spathi.

When something is portrayed as a person, things are different. The easiest example is Stellaris: a few minutes into the game, you meet a pirate faction made up of your own species in order to inject some kind of military action into the early game. You choose an admiral for your fleet. Do you pick Sara? She has a bonus to attack and a neat haircut. Doug? He learns fast and is wearing a space baseball jersey. Veronica? She can move the fleet faster and looks appropriately grumpy.

Whichever you choose, it's a person, not just a vague indicator. Instantly, you start thinking of your fleet ops as part of their career, part of their personal story. So you go to fight the enemy.

... and they have a fleet commander, too. A person. With a unique name, unique appearance, unique powers.

You kill them. You never learn anything about them. Ironically, you never even have enough info to give them a personality. They are the same species as you, so there's nothing distinctive enough to make you assign them the personalities you give to the other aliens you might meet.

This is distracting. Players naturally gravitate towards densely simulated things. You've taught the player that individual people matter, each one is simulated and tracked independently. Then you show them a person, as dense and complex and promising as the one they chose... and just kill them. Wordlessly, effortlessly, pointlessly.

Worse, the alien species cultures are largely randomized. When you do meet an alien, the anthropomorphic tags they bear don't correlate to their actual characteristics. This turns the emotional responses into mud. You've trained the player to respond to individuals, but then refuse to let them interact with individuals. You train the player to not care about species, because their appearance is unrelated to their traits, but then force them to interact with species.

Civ VI doesn't do as many things wrong. It has very, very strongly personified nations, but they have the correct personalities and feel reasonable. Unfortunately, you can't really interact with those people. You can only interact with their nations. Even that is in a vague, impersonal way.

On the other hand...

Let's say I make a game about building cars. There's a bunch of NPCs, they want cars that suit their personalities and needs. I build them a car. They go "whoaaaa! NICE!" They go drive it. They come back. "It was great, did X really well, thanks a ton!"

The scale is small enough that it feels like a personal interaction even when I don't do social things. The response is personal, about something concrete and part of their life. The choices I make are much more centered around facets of their personality, because each NPC can have wildly different taste in cars without becoming unbelievable.

It is already much more deeply personal than Civ VI, even if the NPCs are just a doodled pixel portrait.

Honestly, even if there were no NPCs, building the car is still more personal than building an empire in Civ VI, because I can easily imagine how someone would use and enjoy the car. That's a big part of why Space Engineers is fun! There's no NPCs, no personalities, but imagining how people would live in space, use my ships - that's fun!

In my mind, 4X games can't survive in this awkward place. They will naturally have to become either more humanized... or less.

"More" seems to be the trend, so how would that go?

The two basic paths are the Civ VI path and the Stellaris path, which is why I used them as examples.

The Civ VI path focuses on a few, highly detailed, carefully designed NPCs. Further down this path lies a game where you can interact with those NPCs in more detailed, complex ways. Expanding your empire probably matters less than establishing a good relationship with your neighbors. This will generally take the focus off the militaristic side of things and play up more complex interactions.

The easiest way to do it would be to make the lands of the empire play the role of the home and body of the other ruler. Rather than negotiate with the ruler as a lump sum, you interact with specific cities or lands in a wider variety of ways.

There's not really any way around this. We can't continue having a bleak, war-focused set of interactions if we keep humanizing the other side. I already have no interest in going to war, even with the relatively weak personalities in these games today.

The other direction is to track an impressive number of people. Rather than the ruler of each nation, you track everyone. Politics constantly evolves and changes who is in what positions. The rulers of each nation come and go.

I have a suspicion this is definitely going to spawn a genre, but I don't think I'll like it much. That many people means that each person will be disposable, temporary. That's not going to scratch my itch.

At the very least, I hope all these people have personality tags, so I can easily assign them a mood and emotion as they do things. At the moment it's all undifferentiated mush, and it's very hard to establish any deep connection to the various NPCs. It'd be interesting to try to improve that connection while still having a large number of randomly-generated NPCs. Maybe use some techniques to inject personalities into the flow of the game.

But right this instant, I have a car-building prototype to create.

Stats, Classes, and Fantasy Races

In Ye Olden Days of RPGs, you would roll stats straight. Only then would you start thinking about who the character was. You would choose a class to leverage whatever decent stats you had, and you might choose a race to push your understrength stats into a viable range. So the races and classes were less about building a character you wanted to play, and more about coping with brutal random chance.

These days, we build our character stats along whatever paths we want. There's no need to cope with randomization, but we still have the same fundamental structure. We've gotten used to it, but it's pretty bad. The added complexity doesn't serve any real purpose aside from making the game's high-level play be almost entirely about knowing which stats and classes and races interact in which ways.

If we want to make our game more approachable, that is a useless kind of skill. We want our players to excel at playing the game, not excel at setting it up.

This is a common concern, and many modern RPGs reflect this tendency. We've begun ripping out stats or classes or both.

But there is power in the gap between stats, classes, and races. The way they augment each other is quite dense and interesting. Just ripping out those cornerstones means you lose all that depth!

Without the resonance between the layers, you end up with a very shallow approach and extremely straightforward builds.

Easy example: Skyrim. No stats or classes, just a few skills. Worse, improving any skill is incredibly expensive, since the enemies scale with the number of skill points you earn, not your maximum combat skills. This means your approaches are going to be both extremely basic and extremely focused. Moreover, many approaches seem viable at first but at midgame peter out entirely - for example, pure archers don't end up working out, and having to fall back on melee means you're better off just building a melee character.

A few recent games have tried to keep the complexity of stats, class, and race, but they tend to end up feeling mishmashy and sodden. Pillars of Eternity is a good example: their attempt to make all builds viable results in all builds feeling largely the same.

Statistically speaking, that's what they have to do. Otherwise you're back where we started: learning to minmax would be the primary player skill, rather than learning to play.

Is there any other approach?

Sure! Here's a simple idea:

Stats and classes, basically normal setup. However, the stats don't give bonuses. Instead, they are capacity.

For example, if you have 8 strength you can carry 8 points of gear. Say, leather armor (6 points) and a broadsword (2 points). 10 intellect? You can focus on 10 points of passive skills. Say, detect hidden (3 points), monster lore for crits vs monsters (4 points) and deflect criticals (3 points). 6 points of spirit? You can equip 6 points of spiritual gear/enchantments/spells. Magic deflection (4 points) and a ring of sparks (2 points).

This is just one idea, but the core here is that stats don't have to offer statistical effects. By splitting the class and stats into orthogonal concerns, we can give both a much wider range. A magician with a high strength might seem like a waste, but in practice it allows her to carry 3 staves and 4 wands. She can just pull the right one out in any given situation!

It also allows for midgame optimizations. You want to shave a point off the strength price of that spear? It might be more important than giving it a magical +2 damage bonus! You want to merge a bunch of magic rings into a single artifact set with a lower total spirit cost but ruin them as separate objects? Sure, sounds good!

Races? Sure. Lower the price for specific categories of gear. Elves get bows one point cheaper. Dwarves get armor one point cheaper. Maybe you have a family heirloom? It's not great, but it's super cheap for you alone. No need to modify the stats!

In addition, I find this just flat-out makes for more memorable characters: their equipment variation will be cleaner and sharper.

If you're only controlling a single character, this kind of approach is a bit iffy. If you make it light enough for a beginner, it will usually be uninteresting after a session or two. Make it heavy enough, and people will forget their setup after a week away. But if you're controlling a party, it's very easy to make it work out.

The lightweight approach combined with gear serving specific roles means that players can tell who a character is and what a character does at a glance. Even after a week away, the gear will still be easy to see and you can slip right back into the flow of things. If you're worried about stagnation, a steady trickle of new gear variants or abilities that change the prices will keep things swirling nicely.

Anyway, I thought I should mention it.

Stats vs classes vs races: it was originally built for a completely different purpose. It's ok to not do things that way. It's ok to reinvent the wheel.

Just make sure your new wheel turns, or you'll end up with an oversimplified pile of mush.

The Stars of Stellaris

Remember Stellaris? No? 4X space game that kinda flopped a little?

Time to learn some game design. Let's examine why Stellaris is flawed and how to fix it.

Oh, no, we're not talking theory. We're talking mods.

This mod.

All the mod does is make your starting "warp" drive travel at light speed. In normal Stellaris, a journey takes around ten days. With the mod, it takes a few years to get anywhere. A year charging up, a year charging down - it represents light speed travel.

And the result?

Stellaris is suddenly fascinating. Absolutely amazing.

There's a lot of moving parts to examine, so let's get to it.

1) The stars are not our friends
Stellaris' normal play feels... anemic. One big reason is that space itself feels anemic. The number of stars you can touch is immense, only bounded by random enemy empires and beef gates. Similarly, the number of stars that can touch you is immense. Enemies are always less than a month away, and most star empires seem to want to wage war across those stars.

There's no sense of mass, or slowness. You can reach out and touch ten different empires. At an instant, your fragile empire can be taken away by an enemy so far away that their stars are tiny points of light in the sky.

By limiting things to light speed, all of those problems go away. Every star you explore is a multi-year expedition. Every star is a treasure, and the shape of your empire is stamped deep in your brain because of it. Empires are easy to talk to, but wars take years to unfold - easily time to mobilize your defenses.

This gives your empire a sense of mass and weight. It also makes every jump matter, and you quickly learn to schedule predictive jumps. It's a radical change that, yes, slows the game down.

2) Planets are our friends
Now, in most 4X space games, this would make things unbearably slow. But Stellaris is fundamentally well-made and powerful. There are other systems which pick up the slack and absorb the pacing change - and also end up making your empire feel more real and weighty.

The on-planet options you have for managing your empire are deep and interesting. Constructing buildings, managing populations, installing space station modules, expensive building upgrades. Add in some of the additional mods allowing for more complex populations, more complex resources, and buildings with adjacency rules, you have a lot of really interesting literal world-building to delve into.

That said, this is an area that could be improved a lot. Not to blame Stellaris: it was never intended to carry this much of the gameplay load. But there are a few tweaks that could be made to make managing your empire even more interesting and rewarding. I might go with more variations within the population and a lot more on-world events, along with actual environmental evolution based on industry/geo-engineering projects. Interplanetary stuff might also be a lot more interesting and fun, maybe with "micro-bases" on moons and such, managed on the same screen as your main base's population.

All well and good, but all that stuff is theory. What's actually in the game works reasonably well.

3) Look at your hands
The basic result here is that Stellaris goes from being a mediocre game about endless conquest to a really interesting empire-management game. The change in focus arises from a simple change in pacing:

Stop giving the player stuff. Instead, make the player struggle to get anything new. Make them value what they are already holding.

This makes the player value what they have quite highly. As long as the play is dense enough to make that interesting, it makes everything meatier and more important.

Yes, you can make a billion random stars and a trillion random planets. But they only matter if you make them matter, and that means making the player value each one. Easiest way to do that is to make each one very expensive.

4) Moral of the story
The worst part about this story is that these mods are really unpopular.

Almost nobody plays Stellaris with the mods that make Stellaris fun, because most people think "more power = more fun". Millions of mods that make interstellar travel faster, or give you a higher colony cap, or add "sexy space babes"... but the mod that actually improves the game goes unnoticed.

As game designers, this is the heart of the problem. Every part of a player's feedback is going to be about how they want more, faster, stronger. When we take this to heart, we end up with vanilla Stellaris: a game that gives players far too much, far too fast.

Instead, consider that impulse as a source of power. If a player wants more stars, make the stars twice as expensive. If they want more ammo, make the ammo half as common. At least experimentally, for play-test purposes.

Because their urge for more is what makes them value what they have. If they want more stars, that means they think the stars are valuable. Condense that. Put that energy into fewer stars, and they'll treat each one like a priceless treasure.

That's my theory, anyway.

Directed Sandbox (Planet Coaster)

Let's talk about Planet Coaster's gameplay. Let's really, deeply examine it, and analyze its place within the overall genre of "games where you build stuff".

First things first: Frontier Developments is a company that has been pumping out literal theme park games for decades. Most of their dozens of games are about either zoos or roller coasters. Because of that, I'm going to assume they crafted Planet Coaster exactly as they wished: no accidental elements.

The core gameplay of Planet Coaster is quite straightforward. In the first phase, you build a park out of canned rides and shops. This involves laying things out, hiring and training people, balancing the budget. There is a fair amount to learn: it took me five hours before I realized you could train people.

This "early game" of Planet Coaster is a logistical challenge, but it's largely pass/fail. Once you understand enough of how the park works to get a reasonably good layout, balance of canned rides, and sufficient employees, you're done with the early game and move into the late game.

In the late game, the park runs itself. You have a steady stream of income and no distractions. This is typically the phase where you start to build coasters and track rides, although you might build one or two small ones earlier on. Since you have no particular constraints, you are free to slowly work out your coasters and express yourself.

Moreover, the progress you make on your coasters doesn't much change the dynamics of your park. You don't need to change the number of shops or the number of trash cans or any of that stuff. Minor tweaking, at most. There's no distractions such as fires breaking out or Godzilla attacking. You can do as you like.

I call this a "directed sandbox".

The directed sandbox lets you do as you like, but also comes with some useful time-based constraints. This means you're likely to build low-tech coasters first, then grow into more advanced coasters. I think this is a good progression. Even when I'm familiar with the range of options and how the high-performance coasters work, I still like "warming up" by pushing the constraints of the starter coasters and the initial layout constraints.

Planet Coaster has caught a little bit of flak because of its easy late game. This contrasts with classics like Sim City, where as the game progresses more concerns are introduced. Requirements such as police, pollution management, and outright disaster response are phased in to keep the challenge alive.

It is true, Planet Coaster does not do that. Fundamentally, the game is not about managing a park: it's about creating a park. Expressing yourself.

However, if you simply hand someone a toy like that, most people aren't going to be inspired. They'll feel lost, like what they're doing is pointless.

So Planet Coaster gives you some friction up front. Plan the park. Tend to the employees. Make the park feel like it exists, even though it's basically busywork. This will get you interested in the nature of your park, deciding on a theme, building a virtual experience you can feel in your bones.

Then, when the time comes to express yourself freely, you'll have a foundation to build on.

You won't be faced with a blank page. You already have a sketch.

Planet Coaster does have a pure sandbox mode. I don't really like it, the unstructured construction ends up with an untextured park whenever I try. Nothing wrong with that, but the directed sandbox is actually more freeing for me.

I think we can learn a lot from this. Games like Kerbal and Space Engineers are directed from the other side - they have goals you work towards instead of foundations you build on. I think both approaches are viable.

But I would like to talk about one of the big weaknesses in Planet Coaster - one shared by many base-building games. Decoration.

Decoration is a big part of making things feel lush and lived-in, but left on their own, players will focus on mechanical elements and leave decoration aside. Some games try to embed decoration into the mechanical elements to insure the end result is decorated. For example, in Evil Genius not only are the various rooms automatically decorated, but adding decorations creates a stat boost in a radius around them.

Planet Coaster does this by simply making everything have a decoration rating. The rides are the most obvious example, since they have a rating stapled on their summary, but under the surface the characters constantly think about it everywhere they go. Their thoughts reflect this if you click on them.

My issue with Planet Coaster is that this short-circuits. The now-famous "surround everything with flamethrowers" approach is the fastest and easiest way to make this max out, and so many players simply have a theme park that looks like hell, both literally and figuratively.

Clearly they made this choice on purpose. I'm sure they tested other methods. The fact that flamethrowers are cheaper than a ratty old cart shows they clearly wanted to make them dominant, for some reason. But let's talk about some other methods.

One of the things that goes into actual theme park engineering is theming. Much like a roller coaster, the experience of moving through a theme park involves a variety of experiences paced to pull you in. The actual beats are different, and it's not as linear, but it's a well-understood arc. They build casinos using the same methodology.

It would be possible to allow the player to construct a theme park experience in a more explicit manner, where the themes in each area and the experience of walking through it are combined into a final "feel". Rather than theming the whole park as "pirate", there's a lot of ways you can refine that. This area is about high-seas battles. This one's about shipwrecks. This one's about desert islands and searching for buried gold. This one's about the scurvy port town. This one's about being in a prison. This one's about pirate ghosts.

It would be relatively easy to spam through these as well. "Oh, I want to get the points for a pirate sub-theme, so I'll just stick ten thousand cannons here in a huge pile." This is why I might recommend that the experience unfold in much the same way as a roller coaster's arcs do. People are moving through the area. Things that are close have a greater weight, things that are far away matter based on their height and visibility. This can be simplified or made complex as you like, but the point is that the actual walking along a path has an emotional pattern to it.

It sounds nitty-gritty and annoyingly obsessive-compulsive, but that's what designing a real theme park is like. Yes, they agonize over whether to use gray brick cladding or red brick cladding. They agonize over whether to use real-world branding for their food services, or just put up a sign saying "grog". This is what it's about.

There are a lot of different things you can do with this, things that are every bit as deep as a roller coaster. For example, you can use a monorail to move people from one area to another, but you can also use the monorail to reset their expectations and rest them up, so they're ready to tackle the new area. Correspondingly, you can put the energetic and terrifying rides beneath the monorail, in the area the monorail skips. Excited teens will ignore the monorail and move into the high-energy area, while tired families will hop on the monorail and get rested up.

Anyway, Planet Coaster's pretty good if you understand that it's a directed sandbox and not a game with a beginning, middle, and end.

What do you think about their approach?