‘Tharsis’ Reveals the Luck Inherent in the Roguelike

Tharsis can't fudge probability with a random number generator and that upsets people.

Tharsis is a game centered on the luck of a dice roll, but there’s also a considerable amount of strategy revolving around those dice. It’s a game about managing risk, knowing how to manipulate the many systems in play, so as to increase the odds in your favor. The odds will never be completely in your favor, but that’s fine. That’s part of the challenge. How do you make the best of a bad situation?

This is the question at the core of every RPG roguelike. This is a genre that regularly makes us the underdog position, an underdog pitted against an incredible force, and then asks us to beat it. In Tharsis, you have to survive on a ship seemingly held together by string and prayer. In FTL, you’re a single little ship fleeing an entire galactic armada. In Darkest Dungeon, you have to beat back the cosmic horror of an evil god. In Out There you’re battling the uncaring universe itself. In Skyshine’s Bedlam you have to cross a No Man’s Land desert hellscape while also trying to kill its king. This is hard work, hard but always doable.

There’s always an element of luck in these games, and the challenge lies in knowing how to properly manage all the unforeseen risks that you might face. In every one of these games, you can be screwed over by an unlucky random encounter, and depending on how you play, in the early game you can encounter a mid/late game fight that’s literally impossible. That’s the way that these games are designed. Early mistakes have lasting consequences.

Tharsis is no different, but based on its mixed critical reception, it seems to be taking way more heat for this balance between randomness and strategy than its peers. This critical contrast is starkest when you realize it was released around the same time as Darkest Dungeon, which has received pretty unanimous praise. That praise is deserved, and I’m not here to bash Darkest Dungeon, but the difference in reception is worth analyzing, considering how similar the two games are. Why does one punishing game with a large element of randomness receive praise while another punishing game with a large amount of randomness receive hate? How do we like our punishment?

We like our punishment to feel fair. That’s kind of an obvious statement to anyone who has enjoyed a difficult game (and likely anyone reading this post), but that’s also a tricky statement because it’s both ironclad and vague. Punishments must feel fair, but what is a “punishing” game and what constitutes “fair”? Fairness, specifically, is really tricky because it’s more a matter of perception than balance (see context), and we tend to perceive the world (or at least a virtual world) as more ordered than it really is.

We naturally assume that a game world is based on a set of logical and predictable systems. We assume this order, and so we apply that sense of order to the world as we experience it. This tends to result in us viewing the world as more favorable to us than it really is, since we’re the player and we know this world was made for us by a developer: We’re the center of this world so naturally the rules favor us. This isn’t true for all games or gamers of course, but this is an insidious belief because it manifests in subtle ways, even when we know that we’re playing games that don’t put us at the center of things. We still inflate our importance even when we know that we’re not important.

The other night my roommate was playing XCOM 2 and complained when a soldier missed an attack that had an 87% chance of success. Now 87% is high, but it still leaves a 13% chance of failure, and that’s not an insignificant number. This is something that I’ve found myself doing as well in other games — inflating 80-90% to 100%.

One of the developers of Tharsis, Zach Gage, talked about this in a recent interview at Polygon:

“It turns out that if you tell a player they have a 90 percent chance to succeed and then they fail, that player will feel ripped off and confused,” Gage explained. “So in Civ games they pull some trickery to weight odds so a 90% chance actually wins a lot more often than 90 percent.” Gage once saw a talk discussing this odd bit of player behavior and psychology by the eminent designer Sid Meier. But what if you want to make a game that doesn’t fudge the numbers?

Tharsis doesn’t fudge the numbers. It can’t. Resources for each turn are determined by die rolls and those dice are actual 3D models, so how they roll is based on physics, not on a random number generator. The dice thus become a symbol of true randomness, an embodiment of luck and gambling that seems to reject order. That’s what makes people uncomfortable about Tharsis. We can’t inflate the numbers. We can’t impose our sense of order on the dice. The inherent randomness of the roguelike is exposed to us, and we’re forced to accept it.

Darkest Dungeon is different in that it hides its randomness within several layers of combat mechanics. The randomness is still there in the hit percentages and in determining what enemies we face each fight, as well as in the size and shape of the dungeon itself, but those bits of luck are forgotten as we strategize during a fight. We’re more concerned with the positions of our soldiers, their growing stress, and their shrinking health. Randomness still has to be taken into account since many attacks aren’t measured in absolute numbers (we’re given a range of damage, anywhere from three to six points, instead of a consistent five points), but that randomness is overshadowed by all the other combat systems. It’s just not something that we focus on.

But the funny thing is that dice aren’t actually that random. As Gage says:

“Dice very quickly fall into a curve of results depending on how many you roll,” Gage told Polygon. Catan players and anyone familiar with Vegas odds know this. With two dice the most likely outcome is seven. Anyone familiar with Vegas odds will also be very much aware of this reality. The next likely roll is a six or an eight. “

Dice are governed by probability, and probability can be factored into a plan. Granted, that plan has to be rather fluid and open to change, but that’s part of the fun — watching your best laid plans go to hell and having to make a new plan as a result. I’ve written before that the key to Tharsis is knowing how to manage risk, but it may be more accurate to say that the key to it is managing probability, knowing how to manipulate the mechanics around the dice so as to give you a better chance of rolling those good numbers.

That’s what we do in any roguelike. We build our ship or convoy or civilization or expedition in such a way that we feel like we’re prepared for most outcomes. We manipulate the mechanics so that the odds are slightly more in our favor. The difference between the two games is ultimately one of perception and preference. One only seems more random than the other, and gamers don’t like randomness.

Gage also notes this in his interview: “We got a really amazing positive response from streamers, board game players, developers, some journalists, and non-video game-focused publications, and a really negative response across the board from video game-focused publications.”

Video gamers crave control, even if it’s just an illusion. We still like to think that we can “beat” the game but we can’t “beat” Tharsis (or Darkest Dungeon for that matter) any more than we can “beat” a game of Blackjack. Our best hope is to learn and manage all the probabilities at play. We win by learning to play with the game, not against it.