> Roguelike is a self-descriptive classifier -- something like Rogue. The defining characteristic of roguelikes is exploring game mechanics amidst permadeath.

The problem is no one agrees on what those defining characteristics are. You've given your opinion, but it's just an opinion. It's highly subjective. I hear people all the time say the same thing "it's self evident!" and then go on to list "turn based/grid based" as the self evident requirements.

The author and I disagree on this topic (I wrote the linked article "The Roguelike War Is Over"), but.... I do feel for fans of traditional roguelikes. They had a well understood genre of game for about 30 years that meant specific things: turn based, grid based combat with random levels and real permadeath. Many of these fans just can't get into action games and they have a hard time finding what they're looking for these days.

The real "problem" here is that we're getting more lines fuzzing right? Instead of just having Rogue, Nethack and Caves of Qud we're getting some games that dip their toes in the water to grab a few features those designers appreciate (i.e. permadeath a mechanic lots of devs have used to make their games less stressful by keeping expected lives rather short - a usage of a mechanic for a purpose far different from original rogue-like's intentions)...

I think this is just like music genres - nobody owns a genre and games will come out that do different things with that genre... the market is open and new games just add to the collection of choice - it might be that rogue-lites end up stealing some community support and income from more classic rogue-likes but nothing has destroyed Rogue or any of the other games people have fallen in love with along the way.

I have my own classifications for what roguelike would be and Diablo (even in permadeath mode) wouldn't fit into it - it's far too fixed and predictable of a game and quite easy to initially get into - I'd consider a lack of accessibility to basically be a rogue-like requirement.

> I'd consider a lack of accessibility to basically be a rogue-like requirement.

I just have to say that's a strange take. Lots of newer traditional RLs are quite accessible and accessibility was a big goal for my game. There's no real reason to have a game use 95 separate key commands. Then again this classification is no less arbitrary than the low value factor "ASCII display" in the Berlin Interpretation.

Oh it's definitely arbitrary, the reason why dwarf fortress made the "sorta" list but other strategy games that are technically just as perma-death (i.e. every RTS ever from Starcraft to AoE2) don't... in fact Dwarf Fortress is notably _less_ roguelike than those other games - it's just more "brutal" almost certainly because there is no ability to tune the difficulty and some gameplay requirements are not clearly explained as necessary. In DF the presence of your prior fortress can potentially make your life a lot easier (or more difficult) by tackling overworld threats and providing security (or becoming a hive of scum and villainy)... And there's even a "recover my save" sort of game more in launching an expedition to reclaim your fortress.

Everyone who has ever played a dwarf fortress game for longer than four seasons will tell you "Oh yea - I realized you're going to need plump helmet" - compare that to starcraft where often times players will realize they'll eventually need a military of some form in their first game.

> there is no ability to tune the difficulty

FWIW, this isn't actually true; you just need to manually apply the gameplay changes that a difficulty setting would normally affect. The two main one's I'd recommend are increasing the number of embark points (in init/world_gen.txt#EMBARK_POINTS) so you can buy more starting resources, and increasing dwarves' speed (in raw/objects/creature_standard.txt#CREATURE:DWARF#APPLY_CREATURE_VARIATION:STANDARD_<x>_GAITS), which just generally makes everything much easier, including increasing your fort's economic productivity to give you better ability to absorb any disaster that do happen.

Your overall point stands though; we proudly call it a learning curve^Wcliff^Woverhang for good reason.

The absolute core gameplay mechanics of rogue like games historically was a combination of being strategic rather than twitch, resource management, permanent death, risk with randomized rewards, and progression. A game like Slay the Spire fits that mold much more closely than something like dwarf fortress.

Being brutal is just an outgrowth of those basic mechanics. Meaningful risks require meaningful costs. Using an unidentified scroll consumes the scroll. Fighting monsters consumes an unknown amount of resources to receive some benefit. But, if you tone the game down to near safety then without twitch based tests it’s just wandering around freely.

Similarly, if you can memorize the layout or the meaning of a scroll for the next life, then that’s not a risk you’re taking.

So would playing Civ6 on ironman mode (assuming it was enforced and/or you did it by honor) and with a random map and a randomized tech/civics tree count as a rogue-like?

When you move that scout you don't know if you're going to encounter a barbarian or a tribal village - do you hold off on building a warrior to escort your settler hoping that you can settle without it being captured?

I have a clear picture in my mind of what roguelike is, and that picture is nethack because someone introduced me to nethack and said "This is a roguelike" and I still basically believe them - trying to abstract the specific mechanics that make something a roguelike is hard... Here are some questions:

Do you have to just be one person - can a roguelike be a party of adventurers?

Does a roguelike need to have an rpg-style leveling system - can strength be gained by item acquisition alone? Does a leveling system make it not a roguelike?

Do you need to use pixel graphics for display?

Does the PC need to be precisely on coordinate pixel big?

Dwarf fortress, for instance, is turn based but the only real difference between turn based and real time is that twitch factor - if you made a version of starcraft where the game paused every five seconds to let you revise as many orders as you wished would that suddenly be really close to a roguelike?

My core thing here is that rogue was a collection of game mechanics, and games can be judged to be similar to that in an entirely subjective manner - you can objectively call out different features that differ or are the same, but those features only really matter if the person you're talking about strongly believed (subjectively) that that feature was "core" to the gameplay. A young kid playing rogue today is probably just going to walk away saying that rogue-likes use ascii art and I don't think it's wrong to accept that definition - for that kid.

Where Civ6 in that game mode breaks the mold I just described is the game eventually stops having that random risk vs reward element as a major gameplay mechanic. Again StarCraft is effectively deterministic, there isn’t any kind of clicking an ability and wondering what will happen phase.

As to being twitch or not, again that’s from a gameplay standpoint. Slow down Starcraft dramatically and it’s still real-time, but eventuality APM stops being a meaningful factor.

> Many of these fans just can't get into action games and they have a hard time finding what they're looking for these days

Welcome to the "RPG" club. This is just the nature of games. No genre can remain pure, at least not once it becomes popular. Nor should they.

I mean, RPG describes a lot of games. Baldurs gate and Diablo are both RPGs but are pretty far from each other. Even Team Fortress could be classified as a RPG with the most pedantic reading.

That's why we have additional classifiers. Diablo is an A(ction)RPG, Baldurs gate is tagged as D&D and Classic on Steam, and TF2 is a multiplayer FPS (without the RPG tag because most people are not pedantic af).

As far as Roguelikes go, I subscribe to the belief that they pretty much have to be permadeath, tile based, procedural generation for levels and items. Most games don't fit the bill, but so what, it's good enough as a general guideline. Recently we also got the "Roguelite" term to describe Roguelikes that are not very Roguelike-y :)

Bottom line, why are people so up in arms about a tag that doesn't even mean that much?

"Roguelike is a self-descriptive classifier -- something like Rogue" Well play Nethack, Sil, DCSS, POWDER, Cogmind, Frozen Depths, DoomRL. Then play Spelunky, BoI, FTL, Nuclear Throne, Risk of Rain, Ziggurat, Hades. You'll very quickly notice the first group is a lot like.... Rogue. And the second group is a random mishmash of unrelated genres - Platformers, FPS, Twinstick shooters, slasher/looters, and whatever FTL is.

If someone liked Nethack, I would happily recommend any game from the first list. If someone liked Ziggurat, I wouldn't recommend any of the games from the second list because they're so different. Anyone doing the experiment in my first paragraph would clearly come to the same conclusion.

I think the second group of games is great. And so do a lot of people judging by their popularity. However, I prefer the first set but now searching for games like the first set is much harder. You can't just search for the term that was always used to describe the set. And I've already seen people describe Spelunky as a "classic" or "traditional" roguelike because it's 12 years old now.

I think the point the author is trying to make is that "roguelike" is emphatically not actually a genre but describes both the development process and the end result, at least in the "traditional" sense.

In the same sense that some strongly distinguish "FOSS" from software which is just "open source" or just "free as in beer", I think Roguelikes to many people indicates not only the "style" of the game but both the development style and shared history of the game.

The "lowest common denominator" roguelikes like Angband, ToME, Nethack etc. are pretty dissimilar games, apart from some shared procedural generation and dungeon crawling aspects, but they all share either a literal or spiritual lineage to the original Rogue game, itself written as a kind of demo for the curses library in BSD UNIX.

Hack began life as a Rogue-clone written by a highschooler on the school's PDP-11, was rewritten a few times, eventually ended up becoming a collaborative effort on the "net". Angband is based on the gameplay of UMoria, a rewrite of Moria in C for Unix, itself originally a rogue clone for the VAX written in BASIC/Pascal/assembler because UNIX didn't exist for the VAX then and the developer missed playing Rogue from the PDP-11.

To some Roguelikes aren't just a "style of game" but a community that encompasses both playing, developing, distributing, and modifying the games. Many people have learned coding from making small modifications to Roguelikes , which are in turn modified by others, similar to game mods today.

Unlike most game mods, however, because of the open-source nature of "traditional" Roguelikes, they tend to be "forked" and released under new names as large monoliths, which can themselves be forked. Instead of the modern phenomena of installing 100+ mods to your favorite RPG, a developer would throw 100+ patches into Angband and call it YAngband, and then some other person would throw another 100+ patches and call it NeoYAngband.

I agree that at the end of the day being pedantic is pretty silly and people shouldn't lose sleep over something as silly as this, but I also think it'd be sad to forget about the community of "original" roguelikes and abandon the practice of kids picking up the last generation's game and forking it.

Thing is, that hasn't happened -- it's just that "games that mechanically resemble roguelikes" have become sort of mainstream, and the group of people who just want to play games and be done with it has grown much faster than the sort interested in the old way.

As new games appear the lines will continue to get a bit blurry. Dungeon Crawl Stone Soup and Hades are both roguelike by my estimation, but runs affect future runs in both games.

Hades is typically called a roguelite, and the progression elements are rather strong for the genre. You get significantly more powerful with some permanent upgrades. This does make a lot of sense for the game, especially with the strong focus on the story. You want most players to be able to complete the game after a certain amount of tries, but not too early to shortcut the story. And the real difficulty slider is in the endgame anyway.

There's no metaprogression (runs affecting future runs) in DCSS. You must be confusing it with a different game. It's a classical "Berlin Interpretation" roguelike in every regard.

Right. The only thing that can potentially carry over across DCSS runs is a "player ghost", which actually makes the game much harder rather than easier. In any case that's a very small part of the game.

DCSS ghosts also guard random loot, not a bones pile like in NetHack. And in recent versions the game has a set of built-in ghosts that can appear even on a fresh install, so there's not even a meaningful difference between your first playthrough and later ones.

> It's a classical "Berlin Interpretation" roguelike in every regard.

This is no surprise. David Ploog and his sister Johanna helped write the Berlin Interpretation. David served as de facto PMs of crawl for years.

I think they're likely referring to ghosts.

If that's the case, the pairing with Hades (a game with strong metaprogression) muddles the point. As I mentioned in another comment, DCSS doesn't have "bones" to loot and it will generate ghosts regardless of whether you've actually played before, so the impact of previous runs on future runs is essentially cosmetic.

> runs affect future runs

That is more or less a rogue-lite in my opinion. Games like The Binding of Isaac, FTL, Rogue Legacy, Crypt of the NecroDancer etc. It's a fuzzy definition, though.

Many of the core examples of rogue-likes have the concept of bones files, where you can encounter the level a previous character died on. So in those cases some runs affect some future runs.

I assume that what's being talked about in DCSS and Hades is a stronger effect, though?

I have not played Hades, but I think referring to these types of games on a spectrum is more apt.

In The Binding of Isaac, each run starts the same. The difference is playthroughs allow new items to be found, but the main character is not enhanced. FTL unlocks different starting ships, but there are no "upgrades" as far as I'm aware, just different ships with pros and cons.

In Rogue Legacy and Crypt of the NecroDancer, successful playthroughs enhance aspects of the main character. This to me is a hallmark of roguelites. The Binding of Isaac technically has iterations but since only the random item pool is modified, I would keep it in the roguelike camp.

> I assume that what's being talked about in DCSS and Hades is a stronger effect, though?

Very much so. In some games, past runs only unlock additional content (more possible items, monsters, etc.). In Hades, for example, you actually get stronger with each run.

Hades is pretty far on the 'progression' spectrum, yes. Both the story and the character's capabilities evolve with each run. There's also resource gathering. The game has additional mechanics so you can speed up progression if you want.

> Many of the core examples of rogue-likes have the concept of bones files, where you can encounter the level a previous character died on. So in those cases some runs affect some future runs.

They affect one another, but there's no meta-resource and meta-progression as in an Enter the Gungeon, Hades or Dead Cells.

"Rogue-lites" usually refer not just to persistent state, but to persistent progression across runs.

Unlocking ship types in FTL is technically progression but is of a purely horizontal nature. When you unlock a new ship it has zero affect on runs with any of the other ships. This is markedly different from something like Binding of Isaac where you unlock new things to find that affect all future runs.

FTL has unlocking of the ship types but that doesn’t affect future runs of a given ship at all. That’s like having unlocks for different character classes. In no way does that make FTL “lite”.

You're right, I wasn't thinking when I lumped in FTL there.

This is also true in Nethack. Possessions lost in previous runs may be found again in later runs in bones levels.

Yeah. And there exists a very scummy form of meta-gaming exploit known as bones stuffing where you deliberately create well-stocked bones levels for future characters to find.

If you don't scum though bones are far from guaranteed and often a bones level just means getting murdered by a summoned demon lord or something. I wouldn't call it a progress mechanic like we see in roguelites.

That type of mechanic is often referred to as rogue-lite

Here's the important bit for me.

Roguelike, as much as the term refers to a game style rather than a development process/community thing, describes a game that is turn based and tile based. I find the turn-based nature of roguelikes core to the genre, because they define so much about how you interact with the game. "Real" or "traditional" roguelikes allow you as much time as you need to decide your next move.

What this means is the genre is completely removed from any sort of mechanical or reflex skill. If reaction times are at all involved, to me, it's not a roguelike. Under this rubric I find FTL is mostly a roguelike due to its pausability, whereas Spelunky is definitely not, given the required mechanical skill.

(I may even consider being turn-based to be more core to the genre than having strictly no between-run progress; FTL gates off some ships behind multiple runs but this barely registers for me. Some people take this 'requirement' a little too strictly for my tastes.)

Why this matters, beyond pedantry, is that blurring the genre lines has made searching for new games in the genre almost impossible. Some great roguelikes have found their way to Steam (Caves of Qud, for example), but finding them amongst the 100s of games that get the "roguelike" tag is difficult. Not a huge issue, of course, but it's why it matters to me.

Beyond that, the term "roguelike" is really not very descriptive unless you're already part of the RL community and are up on how the community defines the genre, so I'm just clueless as to why it has become such a marketing buzzword. "Run based game" has only one extra syllable and provides such a cleaner description of what the game is actually like. And can we really say that a game like Slay the Spire (a game I love) is anything at all like Rogue?

>Some great roguelikes have found their way to Steam (Caves of Qud, for example), but finding them amongst the 100s of games that get the "roguelike" tag is difficult. Not a huge issue, of course, but it's why it matters to me.

Try the "traditional roguelike" tag, it's exactly what it's there for and carries the meaning well. The author of the article argues against its usage but uses it himself many times throughout.

Why does it have to be tile-based? Enter the Gungeon seems very Rouge-like to me.

FWIW, monetization strategies do influence how a game plays - at least for people who don’t pay extra.

Energy limits how much you can play the game in a particular day.

XP Bonuses ensure that there’s a greater-than-normal (probably dull) grind to progress through the game.

Cosmetics mean that there will be few, if any, cosmetic rewards for playing the game. (An aside: yes, this matters. If cosmetics weren’t an important part of a game, why would they sell them?)

Lootboxes (gacha, etc) mean that the ability to make substantial progress through the game will rely on luck (and grinds to get the chance to test that luck).

And, something new coming to roguelikes/lites, “Undo” consumables mean that you’ll be starting over more than you’d expect to.

Otherwise, I agree with your overall definition of Roguelikes.

I would not consider FTL to be a roguelike because (a) you can't go back to previous levels, you can only explore the current map, and (b) you're forced to progress to the next level on a timer.

Rogue also doesn't let you go back to previous levels (except during the ending sequence when you escape with the amulet) and forces you to progress on a timer (you starve if you go too long without eating).

Dang, you're right. I've played so much nethack after rogue that it overwrote that.

> (a) you can't go back to previous levels, you can only explore the current map

a. While this is common in roguelikes I'd never pick it out of a list of defining traits of a roguelike. It's like defining first person shooters by whether they have keys that open doors. Sure it's common but I'm not playing Doom to find red keycards, I'm playing to shoot things. The keys just help direct the experience and there are plenty of mechanics that could fill that same gap.

> (b) you're forced to progress to the next level on a timer

b. There is a timer that prevents you from leaving before it's up (engine charging), and there is escalation of force the longer you stay in a sector (rebel controlled beacons) but neither of those design decisions strike me as incompatible with the roguelike formula.