Making a ROGUELIKE? Watch This First...

"monke is monke" solid life advice

Where do I watch your design crit?? This is insanely good advice

You could have the same discussion about people thinking that roleplaying is only about leveling up.

the most important part of the rogulike, imo, is building a character/playstyle every run, and the procedural generation works around that in order to learn in depth the sistems and mechanics in order to adapt to the ever changing levels.

Not mentioned in the video, but essential to getting the most out of this focus on engine building, is to avoid limiting the range of build options at the start of the game. This is because experimentation is the key driving force behind the replayability of rogue-likes and by drip feeding unlockables you are impeding enjoyment. That can seem contrary to normal game design philosophy where you need to avoid overwhelming new players with too many options but keep in mind that the kind of gamers who are drawn to rogue-likes are the ones with prior RPG experience.

Balance in rogue-likes is also approached very differently. There is a misconception that the genre should have a punishing difficulty, but this is wrong. The difficulty should be a consequence of the build the player has gone with. This is where you have those god-like runs where players are enjoying a power fantasy because the proc gen and build synergies just worked in your favour but also those times where the odds are against you - there is a lot of skinner box psychology at work in rogue-likes.

Furthermore, due to the enjoyment coming from optimising the engine, your game needs to avoid a common problem with RPGs where a build is "complete". To some extent this means character classes should be mutable or non-existent so that players can be forced to switch play styles mid-run. Consequently you don't want two extremes as suggested in this video but rather a variety of distinct styles which have no overlap so that hybrid builds can exist between these points.

On my 3D roguelite I am building a procedural environment that allows for different situations based off your build, for example places that are higher to reach to encourage picking up a jump upgrade, underwater and frost biomes that require certain upgrades to be confortable playing into and an overall player progression that focuses of fewer, deeper gameplay mechanics that I basically sum up with: every upgrade should allow for new actions to be made and new gameplay scenarios to be played.

Otherwise I often feel it's really just variation and stats progression, or bullet-hell hell.

I agree with the "develop 2 character types" sentiment. I've been on a couple projects where the player controller was pretty tied into the core code of a certain character type. Expanding out from that, weeks down the line, wasting a lot of dev time. Seems less important for an MVP, but thinking ahead is helpful!

Great insights, I’d like to see more of this kind of genre breakdown videos with actionable advice for designers.

TBH, I feel like procedural generation should be one of the last things a roguelike works on, like Dr Joe said about making the heart first.
Heck, Balatro is a roguelike, despite it not really having anything to "procedurally" generate, it's just a bunch of random outcome tables; you get this boss condition from the list of condition. The shop has these jokers from the list of jokers. Etc Etc

This has definitely given me a think though because I've got a roguelike idea rattling in my head and I'm realizing that the idea I have revolves around making your own card, which isn't very randomized and that if I wanted to make it and make it a good game, I'm gonna have to make some revisions to how the progression works so that players don't just beeline to the thing they want to do.

I wanted to send you a demo of my roguelite action platformer (yeah, yeah :D) with a list of my current concerns for a full version and what I think to do about them, but this video covers one of my most important concerns. I think I'll first sit down and work out how to better give a player more meaningful choices between runs :)

Some deep insights in this video, thank you! I personally think of roguelike games in terms of their capacity to present a constant drip of meaningful decicisions to the player. Making builds or "engines" as you refer to them in the video is just one aspect of this decision-making, which is the actual generalization in the genre together with other aspects related to progression, etc. But in my mind the fundamental characteristic of the genre is decision making over engine-building (which is just one aspect of the decision making). Anyway great video.

Did you change the thumbnail of this video? I was trying to find it and couldn't recognize any of the thumbs xD
Great video btw ^^

can you get deep with the "engine building" ?
I knew some things but surely it's far from enough to say that I have a solid understanding.
This will help me in my future projects and I'm sure many others are in need to know this too, though many people don't even know if they need to know.

This video hit me like an hammer

hahaha, I have no comment on this topic because they aren't my thing but you know, gotta write something to keep you know who happy ;)

I believe the main issue is "how to find the correct definition of a genre". I said "find" not "define" cause we're playing a genre and try to find any elements which constitute that genre and which elements which aren't. This is not an easy task cause it takes time to do it properly and most players don't bother. That's why we many genre misconceptions from famous genres such as metroidvania, soulslike and linklike.
Maybe you could make a video describing how to do the above properly next time.
That being said, devs with genre misconception could accidentally make new genres when trying to replicate the genre with their games. The result could be good or bad.

Oh, wow, this makes me also understand Tower Defense games on a new level. Setting up the defense engine and seeing how it goes.

This reminded me that i want to make a space exploration game similar to early no man sky, and my first priority isn't the planet, but how to make the game fun on barren planet, thus the name of the game, ie no crutches of civilizations, ruins, flore and fauna, only geology, mineralogy, and weather lol

One thing I'm surprised you didn't mention, especially with the focus on engine-building, is player modification synergy.
THAT, I think, is one of the two biggest motivators for players to explore, not only your game, but the depths and extremes of your game which keeps them coming back.
The other is emergent gameplay factors, but that could mean a dozen different things depending on how your game engine is built and what cross-effectiveness it allows between otherwise-segmented factors.

I did some of the things you talked about. The thing i've learned about engine building with a healer game is that it's extremely easy to reach an upper limit to where your choices stop mattering. It's a lot easier to overheal than it is to overkill.

Quickly becoming one of my favourite channels. You're doing great work!

There are a lot of 'Cargo Cult' ideas in game dev where we imitate what we see in games without necessarily understanding what makes them compelling in their contexts.

Would love to see an engine design video

Really enjoyed this talk as I just built my procedural level generator this morning. This definitely helped me get a better understanding of the design issues i am working on.

I discovered this the hard way while working in my roguelike, I spent too much time focusing in gameplay and other details that don't really matter, thinking that then I can sprinkle it with random power ups and it's done.
The power ups or rather decisions you can make to alter the run is the core of the game, other aspects of the game could be mediocre and I think the game still could work.

Release date is less than a month away and now I gotta go and re-kajigger the upgrades so that things have a tweak more interplay

Thank you for the annoying problem to solve lol

The term roguelike deckbuilder feels very narrow and unweildy, almost interchangeable with Slay the Spire clone. But I think the way it's used has made it synonymous with engine builder. Games like Peglin have been labelled with it, even.

Also I agree with procedural generation not being the most important thing. I'd also argue permadeath is not what makes roguelikes engaging. Brotato added an option to retry waves, completely removing permadeath. And I've actually had more fun experimenting with builds that way.

To me permadeath is often implemented in a way that's like a built-in stress and dissapointment mechanic, and when you defensively react with "oh I'm probably going to die soon, who cares" it can encourage you to treat individual runs as segments of meta progression, not trying too hard or caring much about what's happening in the present. That's why Donkey Kong Country 2 is still objectively the best game.

I just wanted to say I really appreciate your channel. Your content is always so engaging and informative. I've learned so much from watching your videos. Keep up the amazing work! I'm excited to see what you create next.

Yeah. I think a similar angle to judge your mechanics by is asking whether your random choices are providing difficult decisions to the player. Like which card to choose, which boon to get, which loot to take. It will be a difficult decision if multiple strategies are available.

My example for this is Brotato! I'm not the biggest survivorlike player by any means but I connected with brotato a lot more than other games of the genre because the emphasis is on utilizing your unique character trait and key items to build economy, damage, survivability etc. I find a lot of action focused roguelikes focus on presentability and polish rather than having a large interaction space. Which is good for the first 40-50 hours but hurts the game in the long run. Brotato doesnt even have proper proc gen, because engine building is the main point rather than moment to moment gameplay

This is exactly the sort of thing I’ve been giving a lot of thought to recently! Great timing

Your advice is all great but I hope no one takes away that they don't have to test their procedural generation very well or can just add it on at the end a week before release. I have played games where:
a) clearly testing did not happen enough and I am stuck in a dead end where I can only, in the best case, kill my character if not hard reset.
b) have procedural generation which creates different variations of the same boring layout making procedural generation look like a lazy way of not having to do level design.
c) poorly optimised generation that takes ages to load. I don't want to see a loading screen in your 2D pixel game because your algorithm is calculating every single possible permutation of layout especially if it ends up suffering from the problems in a) and b).

So sickk fo the term rogue like

I think one aspect of Roguelikes that is undertalked (also because it doesn't fit all roguelikes), is the probability to build an "overpowered" or almost "broken" build in your run, when making the right choices and have the stars aligned. This pushes player to try again and look for that super rare and powerfull build that feels good, without breaking the actual game because the build is only valid for the duration of the run

Looking good with the Steve Jobs sweater style.

Writing this before watching through just to see what I consider important in roguelikes personally

- Runs that are different in some meaningful way
- Element of randomness which is cause for decisionmaking often by limiting resources in some kind or having environments showing up only in some runs
- Player skill progression over multiple runs by having some sort of mechanical or decisionmaking learning
- Power progression within a run
- It should give a feeling of discovery. Be that through finding secrets, unique build or any other means. The best roguelikes have enough interesting things going that you feel like you're able to find new small things even after you've spent hundreds of hours in them

Excited for this one!

New term suggestion: Rogue-ish

This definition is quite limited. A game like Spelunky does not really have the concepts of engine building or classes as seen in modern roguelikes but it is still undoubtedly a rougelike. I fear advice like this is encouraging people to stick closely to current market trends, which is unhealthy for a number of reasons. I'm not trying to be clever in the comments, but I don't think this is how we should be encouraging independent creators to approach their work...

I clicked so fast