Game Review: Underhill

Note: This review contains no screenshots, because this game doesn’t actually exist.

The dwarves are mining, the bugbears are lumbering through the mushroom farms, the imps are scurrying to and fro, and all the traps at the entrance to the dungeon are armed and ready. From a bird’s eye view, it seems as though everything in the underhill is humming along, but that’s only because the problems are invisible from far away. The dwarves have found a new vein of moonmetal, which they’re taking to their infernum foundries to melt down and make better tools and weapons. Unfortunately, the moonmetal has a waste product, and the imps have been transferring that to the midden rooms. Normally there’s a garbage troll that loves to feast on all kinds of scraps, but the moonmetal byproduct is toxic to him, and in another two days he’s going to wind up in the infirmary, which will cause the middens to overflow. That, in turn, will result in general disorder and work stoppages, and the dungeon will find itself on the verge of collapse.

This is Underhill, the newest and most ambitious game from Kyle Mormont. He describes it as ‘a left turn from Rimworld’ and claims heavy inspiration from Dungeon Keeper, though anyone who goes in expecting their experience with those games to help might find themselves frustrated.

Take the case of the poisoned garbage troll. In a Rimworld knockoff, you might expect that you would designate a midden zone and assign a garbage troll to it, then check a box to make sure that the moonmetal byproducts are sent somewhere else. In Underhill, there’s nothing like that level of information or control, and trying to avoid the poisoning, if you even know that it’s coming, takes a much different sort of work.

Excavation

The game starts with a single ability, ‘Dig’, and gives you a side of a hill to dig into. The UI is essentially non-existent, with only two small Diabloesque orbs in the lower left and lower right to show your mana and power, and a very small selection of powers that accumulate over time, but which stay very small throughout the entire runtime of the game. While the UI is minimal, it’s clearly had a lot of work put into it, particularly in the way that new buttons are added: when there’s a new power, the UI shifts to accommodate it, making it feel slightly uncomfortable, like a mole that you weren’t sure was there the day before. At every stage, it looks as though that will be its final form, but it’s always difficult to be sure if you’ve seen everything there is to offer.

The game is also cagey with the information, especially at the start. You’re not told what to dig, though the hill is the only thing in front of you, and you’re not told why you should dig. But unless you’re particularly stubborn, you’ll dig, and once you have a tunnel dug out, you’ll get your first visitor poking their head in, almost always a goblin, the most basic and least specialized of the dungeon inhabitants. If he finds things to his liking, he’ll make it a home, and the first trickle of power will start coming in.

“The game is meant to be played blind,” explained Mormont, who flagged me down when I joined the game’s small but vibrant Discord. “You’re meant to come in knowing nothing and experiment to figure things out. That’s supposed to be the joy of it. The goblins come in, and you don’t know what they want, so you have to watch them and figure it out. It’s a watching game, a learning game.”

Goblins like tunnels that aren’t much bigger than they are and rooms that are about three goblin heights across. There are no rulers or grids, at least not when you start the game, but the initial option to dig has a standard width, and that’s just about as wide as a goblin tunnel should be. Nothing in the game communicates that this is what a goblin likes except watching what the goblins do. On my second playthrough after an untimely collapse of my dungeon ecosystem, I understood the game a little bit better and did some of the research work that it seems to want, which meant creating eight different rooms of various sizes to figure out which ones that goblins would go to. They prefer to be close to the things they need, which in the beginning means being close to the dungeon entrance, but my notes eventually filled up with details on the proper height, size, and shape of a goblin room.

Mad Scribblings

Underhill loves that sort of thing. There’s an in-game book that’s unlike any I’ve ever seen before, a blank journal that fills in with drawings and details, especially with regards to the ecosystem components, but is completely idiosyncratic. One of the design goals for the journal was that it look like a real journal, something that someone was slowly filling with their own observations, rather than being an encyclopedia with lost pages. For that reason, the journal is dynamic, filling up as you go, the notes stretching across pages. I’ve only had a few hiccups with it when the unseen writer wrote down a detail or two that I hadn’t figured out on my own.

“It’s messy, it’s organic,” says Mormont. “That the aesthetic. There are numbers in the background, but you should never see them, and they should be very difficult to intuit. I want people to be thinking on the non-number level.”

This is one of the reasons given for using a system without a grid, though the trade-off is that it gets difficult to get anything looking nice and ordered. Digging out a goblin home in a square that’s three goblin lengths across is an exercise in frustration. In theory this encourages messiness and a ‘let it be’ approach, but in practice it can be hard to embrace the organic mess that the game is trying to encourage.

Once the goblins have settled in, you can watch them go about their lives and see what they need. The process of discovery is one of the game’s main selling points, and as you watch, you’ll see that goblins form themselves into families, which form into clans. Goblins have biological needs in the form of water, food, and waste, and also seek shelter, which is why they move into your dungeon in the first place. If the dungeon is cramped with goblin families, they’ll throw their trash just outside the dungeon, but if there are enough rooms, the goblins will designate one of them as a ‘midden’ and start throwing their food scraps, broken tools, and other waste there.

The game doesn’t tell you that the room is a midden, just as it doesn’t tell you most things. The midden is one of the things you’ll learn about over time by watching. And it’s from one of those early middens that I got my second dungeon denizen: the garbage troll.

The Age of Discovery

I don’t want to spoil everything in the game, because it’s a game of discovery, but it would suffice to say that the garbage troll took care of one problem and created another. The garbage troll has his own needs and wants, and if those needs can’t be satisfied within the dungeon itself, he’ll either go out into the wider world where he might create all kinds of problems, or conversely, create problems within the dungeon by eating things that aren’t trash — an example being goblin possessions.

The game rolls on like this, with more monsters slotting themselves into place as it goes on. The ability to dig is your only tool for what seems like slightly too long to me, but as more creatures come to occupy your dungeon, your power slowly grows, and new abilities do eventually make themselves manifest. Water is one of the early ones, and comes up more than I had expected from the start, being one of the primary tools you have to shape the dungeon and its inhabitants. There’s a dungeon species that can’t cross running water, which means that it can be kept to one side of the dungeon and out of trouble by having a small stream trickling through. Similarly, water is one of the main ways to keep a dungeon clean, and helps to automate the movement of sewage down into somewhere a colony of garbage trolls are living. When the dwarves move in, they use the water for their own fastidious cleaning.

Part of the joy is in watching all the elements interact with each other. Even right at the start, there’s joy in seeing the goblins go out hunting beyond the range of your vision and come back with food, which they clean and prepare before eating. The animations are crude but evocative, done procedurally, and the game has a lot of clutter even when the dungeon is still developing, whether that’s fast-growing moss that creeps over the rock walls or the tiny mushrooms that grow in the midden (and can be cultivated by a druid later into a permanent food source). Surprisingly, everything is procedurally driven, even when it doesn’t feel like that would be necessary, and this is used to full effect to allow different varieties of creatures to have different motions to them. The goblins come in different sizes and body types, and can even grow from children to adults.

Obsessing Over the Depths

Sometimes, all this work leaves me scratching my head. One of the later game creatures, the nibbler (named after pen nibs, not a Futurama reference), goes around your dungeon and counts things, which are recorded in its notebook and exposed to you through a special button in the UI. In a different game, this would just show you the internal count of everything that the game knows the dungeon contains, but in Underhill, the creature has his own modeled understanding of the dungeon, and will only report on things that he can directly count. If you want to know how many goblins there are, and don’t want to count yourself, you have to wait until the accountant goes to take a peek into the goblin warrens. If the number of goblins changes, you’ll have to wait until he checks again to get the updated number.

I was watching the nibbler take stock of one of the dungeon storerooms, and noticed that he was using his finger to count the boxes, which was a fascinating detail. What was more fascinating was that he apparently lost count and had to start over while in the middle of this. It was such an immersive detail that it seemed like few people would ever notice, and had to have taken a lot of time. But as I watched more, I saw that he was losing his place while counting far more often than I thought he should, sometimes twice a room.

When I asked on the discord whether this was a bug, Mormont responded within a few minutes asking me whether I had dwarves in my dungeon. When I replied that I did, Mormont had an answer ready to go. “The dwarves like to brew alcohol, and if you have nibbler, you’ll see him drink some ale when he stops by there doing his count. If he’s drunk, he has a harder time counting. There aren’t that many mitigation strategies for that yet, since it’s hard to restrict the nibbler’s movement.” When I suggested that the behavior could be triggered a little less often, Mormont had a rant ready to go.

“That’s not how it works,” Mormont wrote. “There aren’t triggered actions. There’s not some variable in the game that passes a certain threshold and says to play a confusion animation and restart the count. The nibbler is actually counting. I had wanted to do a full vision system for all the creatures, but there’s too much overhead, so it’s just simulated instead. It counts with its finger because that makes the process go faster. It gets lost in the counting when it’s a bit tipsy because it can’t see its finger as well and its internal count of how many objects there are is more likely to be wrong.”

The obvious question was why you’d choose to do it that way when you could just have the nibbler report the actual numbers.

“Because it’s funny,” said Mormont.

There was a long pause where I think I was supposed to agree that it was funny, and then Mormont started typing and posted a wall of text five minutes later.

“One of my formative memories in gaming was when I was playing Oblivion,” he wrote. “I was trying to steal from this woman, and she saw me, and that was fair play, but then she started attacking me, so I thought to myself ‘wait, I can just kill her’, and so I did. I went out of the house and into the countryside, then to a major city, where a guard stopped me and asked me to answer for the crime. He had no way of knowing that it was me, and I found it really frustrating, because it didn’t make any sense. Obviously what was going on behind the scenes was that there was some kind of hasMurderedSomeone flag that was triggered, and it instantly went to every guard in the whole world the moment the murder happened. As a game designer, why do you implement things that way? Because it’s easy. But it has an impact on how the game plays, and I think you either have to make that a part of the story the game is trying to tell — psychic guards — or work to make sure that all the little moving parts work together. This is a game of moving parts.”

These are the kinds of rants that Mormont likes to go on. He’s more of a preacher than a game developer sometimes, and it’s the small things that seem to get him going.

Does this make for good gameplay? I think it does, with the right mindset. There’s a risk with the opaque approach to information that a player might not be able to tell quite why something is happening or how to stop it. If you view your job as being that of an investigator and scientist, the oddities are engaging rather than frustrating. However, if you’re trying to build the perfect dungeon that has all the creatures working in concert with each other, it can hurt to have it all spiral out of control and not be able to diagnose the problem after the fact.

The Secrets, Cataloged

After I had put in twenty hours, I opened up a channel on the Discord for veterans of the game, which turned out to be a mistake. I won’t spoil it, but there were entire aspects to the game that I had been missing out on simply because there were some conditions for attracting certain dungeon denizens that I had never thought to try. From reading through the different comments people have, that’s not an uncommon experience, and “there are witches in this game?” is a common sentiment. Much work has been put into cataloging all the game’s secrets, and there are three different spreadsheets that seek to track the interrelations of the different elements.

“I don’t like the spreadsheets,” Mormont says in a post below each of the pinned spreadsheets. “Making your own notes and discoveries is the game. Understanding and watching is the game. The game isn’t about making a perfect dungeon from instructions that someone else left you, it’s about being surprised and seeing what happens, using the scientific method to get an intuitive understanding of what’s actually going on. As soon as it’s all numbers and figures it becomes dead, like a butterfly nailed to a corkboard. This isn’t meant to be a team game. It’s not meant to be a game that you watch someone else play on Twitch. It’s a personal journey of growth and discovery. It’s balanced around a regularity of discoveries, so the average person keeps on hitting them. It’s digging in the science mines and continually hitting new veins.”

I didn’t delve into those documents. Instead, I did as was suggested and added to my notebook, both the one in game and the one that I kept beside my mouse. When Underhill hits, it really hits, and there’s something immensely satisfying about understanding these little creatures that move around in your dungeon, going about their business. By itself, that might almost be enough, but aside from the note-taking and investigation, there are the fresh injections of newness that come with new denizens, deeper depths, and new materials.

(Never) Ending

The dwarves were a turning point in my game, but apparently they come much later for most people. Their habitats need to be square, and they’ll spend a lot of time with chisels making sure there are as many right angles as possible. Dwarves will take over if you let them, because unlike goblins, they can dig on their own and see to all their own needs. They want to live in the dungeon and seal themselves off from the outside world, and so long as you don’t get in their way, they’ll develop their own city that meets its own needs.

My first reaction was that this defeated the whole point of the game, but after some time sitting there watching them work, I realized that it was just another way of underscoring what the game had wanted me to get from it all along: I was supposed to be learning from the dwarves, learning about the dwarves. Eventually, I was learning all the things that dwarves won’t do for themselves, all the ways that they would naturally make a society that was worse than the one I could help them make. It was a variation on a theme, in a way. As it turns out, the game is full of those.

I’m fifty hours in now, and still seeing all the ways that the game is developing its core ideas, stumbling through different lessons and trying to figure out the inner workings of all the creatures, materials, and substances. But if I were a goblin, this would be a hole that was just the right size for me.