[TL:DR] I am currently working on an untitled project. It's a dungeon crawling top-down shooter with procedural weapons. At first I dismissed the idea of making the game infinite when my co-developer brought it up. There were so many bugs already, if we made it infinite, well the number would increase significantly. Buuut, curiosity got the better of me, and I realized the step into infinity was a few hundred lines of code, and nothing more. I now have a build you could technically play forever, though when you get to a few thousand rooms, your frame-rate will steadily decline to unplayable levels. It got me thinking: "Is this a good game design choice? Does it add anything of merit?".

I haven't answered those questions just yet so feel free to discuss them, but I have come to a few realisations. In writing, there is a rule, never take more than you give. If a reader is super invested in a character, and you kill that character, you must provide some reward, or catharsis to the reader that is equal to or greater than that loss. In games, our catharsis doesn't have to be a single moment, it could be, but it could also be spread throughout the entire game. In my game, there are a few rewarding things in the current build. (Quick definition), for something to be rewarding, it must make the player feel better, even if slightly, than they felt the moment before. The simplest reward in the game is surprisingly cosmetic. Blood. There's lots of blood. It sprays bloody particles across the scene when you kill an enemy, leaving a deep crimson splatter where it died, sticking to walls and columns, it's fun blood. The second reward is impact. When you fire a high-level shotgun, it throws you back five feet, sending a column of fire and black smoke in front of you as the massive sound rips through your headphones. The third reward is something oddly specific, trickshots. The sniper rifle in the game fires a bullet that ricochets like a pool ball. Many a time have I lined up a shot through a door on a lone rocket-bot, only to have the bullet ricochet around the room, killing every single enemy it contained. Or when I deliberately bounced a bullet off a wall, it striking a spider, knocking it down to half health, killing three space crabs and two rocket-bots afterward- you get my point. I think it's also safe to say that when something is difficult in a game, surmounting that difficulty can be a catharsis in and of itself. But there is a delicate balance, some players play Volgarr The Viking for hundreds of hours, failing thousands of times until is is beaten, and feel great about it. Other players simply die ten times and get their refund. The difference is partly in the game, and partly in the player. If I made my game impossible, with enemies that do ten times your damage and have fifty times more health than you, at the start well that's unfair. But what about player generated difficulty? In my game, you find weapons that are progressively better the deeper in the dungeon you are, but if you decide to keep your weapon and lose to a high-level enemy, well, you're back at the last save point. Speaking of points, where's mine? That's right: how does infinity effect the relationship between reward and difficulty? This game gets progressively difficult as time goes on, it balances out eventually, but the curve is long. The reward has no such mechanical curve, but it just might have a similar curve dependent on the player. If our player is aggressive, doesn't mind dying a lot, plays with risk, runs after a high level shotgun into a room swarming with enemies, and knows they could probably die doing it, then this game will become increasingly rewarding. The risk gets higher, the reward gets better, fun fun, right? However, if the player is cautious, hates dying, and can't stand the idea of running through the same thirty rooms to grab that burst rifle to fight that impossible mini-boss once again, they might not like it. All that being said, I think I'll have an option in the start of the game, limited, or infinite, to hopefully mitigate this issue. If you've read this far, thank you for listening to me ramble. That's all folks!

From a technical perspective, the fact you think that as the rooms increase the framerate will inevitably drop shows a lack of understanding in the design.

No matter how large the generated content, you only need to process and draw that which you can currently see on screen, so you should be able to have an "infinite" design without this having any technical issues like that.

Ah yes, I knew this would come up. I was referencing the current build. Right now everything but enemies are processed only while onscreen. I just need to make the enemies children of the empty parent of the room they spawn in, so they get de-activated like everything else when too far from the player, and that will increase performance dramatically.

[EDIT] It's now fixed. I just didn't want to misrepresent the current state of the game.

The largest problem with the infinite design is that because of the inconsistency it's difficult for players to care about the world.

An example is the loot system you see in a lot of MMO-RPGs.

Each weapon is generated using a code, names are based on the random stats of the weapons, because of this there is no unique items or any reason to care about the weapons, players just equip the strongest weapon and move on.

To counter this, item sets are made, these items allow players to match their gear with their playing style, it also gets players to care about the loot that is dropped.

In dungeon crawlers levels have a theme that changes every now and again to give players a sense of progress and to serve as a kind of land mark. They will also often include set pieces that can be found only on the designated levels; Diablo is a good example.

Games like Don't starve do the same thing in there own way, the Pig king is a constant on most maps that players can use as a landmark.

Making a infinite game just because you can is a bad idea, it takes a lot of planing to make these games good.

I don't know if infinity will have the meaning you expect. If you will never conquer this world, what good does making progress? I'm not sure that challenge *in and of itself* is a strong draw without the context of mastery, and if there is infinity, then there is no mastery. I would probably suggest going a more Dark Souls-esque route where there may be 1000 different levels, and the player is tasked with beating them all. Something perhaps too hard for any mere mortal, but something that is at least theoretically possible to do. The Catharsis for something like that would be... just incredible. If 1K is too few, go for 10K, but the point is... give the people a chance to win the game. I suspect that will enhance the challenge sensation rather than diminish it.

Edit: Unless, of course, your game is about the futility of life, in which case, GENIUS.

I don't know if infinity will have the meaning you expect. If you will never conquer this world, what good does making progress? I'm not sure that challenge *in and of itself* is a strong draw without the context of mastery, and if there is infinity, then there is no mastery. I would probably suggest going a more Dark Souls-esque route where there may be 1000 different levels, and the player is tasked with beating them all. Something perhaps too hard for any mere mortal, but something that is at least theoretically possible to do. The Catharsis for something like that would be... just incredible. If 1K is too few, go for 10K, but the point is... give the people a chance to win the game. I suspect that will enhance the challenge sensation rather than diminish it.

Edit: Unless, of course, your game is about the futility of life, in which case, GENIUS.

You've caught me, I'm making a philosophy game. I like the idea of letting the player beat the game. I could have a final boss at the 10,000th room, but have that boss be slightly different for every player, so there's always a mystery. I think I'll give the player a win condition, but endless difficulty levels, once you beat the game. You go back to the start, but this time there are different bosses, more zones and events, perhaps even different basic enemies.

I could have a final boss at the 10,000th room

If you do this make it clear that the boss is there.

Most roguelikes stick with 100 because it's what player know and they are so difficult that most players never reach level 100. Recently I played a new roguelike and it was much easier to reach level 100 than most games however I had no idea how near the ending I was and stopped playing at around level 150, because with no clear goal there was no reason to progress.

After I stopped playing I learned that the final floor was 500, if I had know at the time I would have played it as a kind of endurance game.