'Chasm' Wants to be an Endless Metroidvania, But the Randomness Only Hurts


#1

On paper, Chasm's pitch is compelling: a never-ending Metroid, where every experience is different. It's not totally random, as the game is pulling from a pool of hand-crafted parts to build the world, but the way it links them changes every time. Chasm's problems are myriad, extending far beyond its approach to level design, but even brushing them aside, it never manages to sell you on its most basic pitch. Rather than creating a sense of mystery, Chasm's randomness leaves a giant, frustrating question mark over your playthrough. Is this level actually boring, or did I get a bad roll of the dice?


This is a companion discussion topic for the original entry at https://waypoint.vice.com/en_us/article/j5ndd3/chasm-wants-to-be-an-endless-metroidvania-but-the-randomness-only-hurts

#2

So it’s randomized but there’s no permadeath to force you to reroll and runs are many hours long?

I don’t recall any game with such a structure having a chance at a boring roll, the randomization may vary the difficulty but over the course of several hours you should encounter enough dice rolls that it somewhat averages out. If it’s boring for that long then the chance of it being boring in general, as opposed to only your one seed, is extremely high (and if there are seeds that are fun but most are not then I’d call that a major issue in the game anyway).

The basic principle of a procedural Metroidy game is doable as a fun game, A Robot Named Fight makes for a good one IMO but that doesn’t take much more than an hour for a run either.


#3

Could this same type of criticism apply to 4x games like Civilization, where the map is randomized and individual games may take a dozen hours? If so, it only lends credence to Chasm just not being a well designed game, randomized world or no. It’s a shame, because all the footage of it looks so good. I had high hopes.


#4

Dang, I was just about to post about Civilization. Also, this can be a problem with survival games like Minecraft, Terraria, ARK, etc.

This seems like a tough problem to solve for any long-play-session game with a procedurally-generated world. If you get a bad start on a deserted island with minimal resources, you might not have much fun in the first few hours. Would “Desert Island Spawn” be a good name for this issue?

Over time, the Civilizations series has made a lot of effort to mitigate this problem. The map generator in modern Civ games is much less random than earlier games. It seems heavily biased towards generating large Pangea-like megacontinets, and tends to avoid generating disconnected islands. The starting position algorithm also seems more advanced, and it carefully picks locations that will give access to good resources, and places each player within a reasonable distance of other players to facilitate early-game interactions. It isn’t perfect, and it might have less variety than earlier Civ games, but it seems very consistent and works great for most matches.


#5

I think the important difference between a 4X strategy game (or most crafting survival games) and a Metroidvania is that in the latter the map is a place you explore, while in the former the map is a place you create. It’s less important to have an interesting starting condition when the whole point is how the world is reshaped by expansion and conflict.


#6

4x’s are about putting the players on more or less equal footing (mechanically anyway, map generation can give advantages and disadvantages) where single player games put the uniquely powerful player against ai’s. At least when a game has an infinite world like Minecraft or Terraria you can always just keep traveling until you find something you like. I’m not sure what the benefit is from generating a world that is finite, relatively linear in that you’re going to explore most of it eventually, and also time consuming enough that you can’t just re-roll though a bunch of variations.


#7

I played a Rogue-lite game a while back called A Robot Named FIGHT! (Steam link) that tried to make a randomly generated Metroid-style map every run. It generated an upgrade in each section that allowed you to open a door into the next area, and so on, much like a classic Metroid game.

Here’s the thing: it totally succeeded in doing that… but it turns out the things that make Metroid great, like exploration, secrets, and “gating”, become asinine and irritating in a Rogue-lite. And without a human touch, all of it felt just… empty. Nothing was exciting or interesting.

I remember seeing trailers for Chasm what feels like ages ago, and it’s been on my radar since then. But the second I heard it was both a Metroidvania and procedurally generated, I was concerned. Both of these qualities are becoming cliches in the indie sphere, and both of them are quite easy to screw up. (Also, it’s not like Metroidvanias have ever really suffered from replayability problems, like… the speedrunning community basically started because of Super Metroid.) From other reviews I’ve seen, some people actually like the game feel and combat design unlike Patrick, so actually playing the game might be more fun for some people. But it seems to me that a problem with A Robot Named FIGHT! shows up in Chasm, too. Without a real human being there to breathe life into the structure, the design of Metroid-style maps usually become vapid, depressing slogs through identical tunnels. It’s a lot like the original Metroid on the NES. The caves seem to go on endlessly, showing no sign of variation or surprise. Everything feels… empty.

I actually want to point out a game that’s been in the news recently: Dead Cells. (website) I returned to it after reading the Kotaku article about it’s (honestly awesome) internal structure, and found myself enjoying it a lot more than I originally was. Regardless, I’d like to point to why I think Dead Cells gets it’s Metroidvania structure so right. To me, one of the keys is, rather than having players traverse space to backtrack, the game has players simply “backtrack” by returning to routes on successive runs. With games like Dead Cells, I am convinced that there is some way to synthesize these two genres, but it’s a dangerous and fickle combination.

I’m genuinely bummed Chasm, after six years of hard work, has such basic structural flaws. But at the same time, I can’t be too disappointed, because I sensed this coming a long, long time ago.


#8

metroidvanias mostly work for me as avenues through which to perform neat environmental storytelling, often using a lot of foreshadowing. They therefore - at least at this point, technologically - necessitate bespoke design for them to interest me. I want to explore a world not for the sake of exploration, but for the sake of (directed) discovery. I don’t know that these types of games reward that sort of discovery when they lack cohesion or intent.