The best kittens, technology, and video games blog in the world.

Friday, May 31, 2019

What is a Roguelike?

Mini Garfield by Neticola from flickr (CC-ND)

Every gaming blog is required by Laws of the Internet to attempt answering this question, so here I am fulfilling my duty.

The Wrongest Answer

The totally wrongest answer is called "Berlin Interpretation", and it's just a completely unfocused list of features early Roguelikes shared. Half of that list is just limitations of computers from early days, nothing to do with game design. It's the worst. Let's just get it out of the way.

The Wrong Answer

So my first idea was that Roguelike is Procedural Generation plus Permadeath. This definition is a trap. Minecraft on Hardcore mode is not a Roguelike.

Procedural Generation

Procedural Generation is a spectrum, and there are very few games with fully fixed content. Even Tetris gives you random tiles. And no game is completely random, whatever that might mean - there's generally some fixed outline, which is filled by some combination of fixed content (often initial or final parts), random selection of handmade elements, actual procedurally generated content, and purely random events during the game.

For games where "map" is a meaningful concept, a reasonable dividing line is having procedurally generated map.

So Skyrim (completely fixed map, mostly fixed quests, some random encounters, random quest variants) is definitely out. Minecraft (random map) is definitely in, at least by this criterion.

So what about a game like XCOM: Enemy Within? It's a series of battles, each on a randomly selected handmade map, and with procedurally chosen aliens. If we see the whole XCOM campaign as a map, and each battle as like a dungeon level, it feels like it's mostly in. XCOM 2 went with even more procedural generation, and so it feels even more Roguelike than XCOM 1.

This criterion gets a bit fuzzy, with most games being in between, but it's a fine part of any definition of a Roguelike.


So what about Permadeath? It has absolutely nothing to do with being a Roguelike or not, other than very indirectly.

Minecraft definitely has procedural generation. It has Hardcore mode with permadeath. Is it Roguelike? What if Civ5 had ironman mode like Paradox games? Nowadays all kinds of games have optional permadeath mode, since it appeals to a subset of players, and takes very little effort to add.

It's pointless to argue that permadeath only counts if it's not optional, or that these games were not "designed" to be played in permadeath mode, like author's intent can be meaningfully known. Saying Minecraft on Hardcore or Civ5 on Ironman counts as a Roguelike, while Minecraft or Civ5 on Normal don't just shows how badly this attempt at a definition fail.

Permadeath is a dead end.

Tactics Based

Before I get to what to consider instead of permadeath, we need one extra criterion. Roguelike must be primarily tactics based, not dexterity based.

There's no amount of procedural generation that's going to make a game like Modern Warfare, or Super Mario a "Roguelike".

So games like Rogue Legacy are not even remotely Roguelike.

I'm not being arbitrary here. Tactics-based games usually have partial or full procedural generation, and dexterity based games often have extremely fixed content. Different genres have different requirements.

I'm using very broad phrasing here. Turn based systems totally work, but "Real Time with Pause" systems like Total War or FTL are just as fine.

This is of course also a spectrum, as games like Factorio are overwhelmingly tactics based, but they still have minor dexterity based combat system. Which by the way should be removed from the game completely, and it can when starting a new game.

Short Playthrough Time

So now that we killed permadeath as a possible criterion, let's see what we can use instead.

Why would Civ 5 on Ironman mode not be considered a Roguelike? Mostly because it's intended to be played in long campaigns.

The key characteristic of a Roguelike is that it's meant to be replayed over and over. Short playthrough time is key to this. You can't just "start over" a 20h XCOM campaign or a 100h EU4 campaign.

Because playthrough time is short, permadeath by default is reasonable. I find Paradox idea of forcing ironman mode on games with very long campaigns like EU4 and CK2 totally idiotic. Bad RNG, misclick, or game bug destroying a run that takes 30 minutes anyway is totally reasonable. Bad RNG, misclick, or game bug destroying a campaign that takes 100h to completely is just shitty game design - especially since longer campaigns give far more opportunity to game breaking bugs to happen.

And yes, if you took a Roguelike game, lowered difficulty (so permadeath can happen but it wouldn't be a common occurrence), and extended its dungeons to 10x current size, at some point it would stop being a Roguelike.

On the other hand, if you took a not-really-Roguelike game, and just made a short mode out of it, it'd probably become Roguelike. Like let's say if XCOM had a mode where you do just 5 random battles on procedurally generated maps, it would be basically a Roguelike.

And if you took a Roguelike, and added a save game system, it would still be a Roguelike.

Into the Breach is another interesting case. It's as Roguelike as Roguelikes get, but it doesn't have total permadeath. You get one time reversal per battle (two per battle with one of unlocked pilots). You're playing time travelers, so it makes perfect sense, it's a highly controlled feature, and removing it would just make game more frustrating pointlessly.

Links Between Runs

This also answers the mystery why pretty much every Roguelike nowadays have some ways to keep a bit of stuff from previous runs in your new runs.

This definitely goes against the idea of permadeath, but it makes so much sense when you realize that short playthrough time is the key. Letting you keep an unlock or a small bonus from previous run simply incentivizes starting over.


Of games mentioned:
  • DoomRL - Roguelike
  • FTL - Roguelike
  • Into the Breach - Roguelike
  • XCOM - not a Roguelike (but could maybe have a Roguelike mode)
  • Civ5 - not a Roguelike
  • Diablo - not a Roguelike
  • Factorio - not a Roguelike
  • Rogue Legacy - not a Roguelike
  • Skyrim - not a Roguelike

Full Definition

Roguelike is a tactics-based game with mostly procedurally generated content, meant to be replayed over and over thanks to its short playthrough time.

It frequently features permadeath, and there's often some links between runs, but these aren't essential features.

Why any of that matters?

You might be puzzled why so much has been written on this subject. Game design is still in very early stages, and even when we make a game that works, it's mostly by accident, and even best games are still full of design fails.

Having more clarity on which features go or don't go together and why can only improve this sorry situation.

And of course there will be games that break any established rules, but it's really helpful to understand why those rules exist in the first place.

Having no clue and just blindly borrowing features from mismatching genres just got us to design failures like ironman mode in EU4, and dexterity based combat in Factorio - otherwise great games.

Saturday, May 04, 2019

Total War: Rome II Review

An embarrassment to all feral kind... Tom cats all over the world are shaking their head in disgust.... by praline3001 from flickr (CC-NC-ND)

I'm really late to the party, but this game had such awful launch, I delayed it a bit, and then I was busy, so here I am, reviewing a 6 year old game.

It is not Rome I

At first I tried to play it like it's basically Rome I with better graphics and weird settlement system, and that really doesn't work.

Battles work more or less like in previous Total War games, but on campaign level it's definitely not so. The biggest difference is that you literally can't have any troops not attached to a general, and limit on number of armies you're allowed to have is quite low.

This breaks a lot of common patterns like recruiting more units and sending them to frontlines, leaving some units behind as extra garrison, or to protect settlement from rebels, splitting big army in half to clean up multiple leftover AI troops, or separating a single unit to send it ahead to scout.

There's a new system of provinces made out of (usually) 3-4 settlements. This really cuts on micromanagement. A very interesting thing they did is that province's capital settlement has walls, but none of its minor settlements do. This completely avoids the problem Medieval 2 had, where 80% of battles were assaults on walled settlements, which might have been historically accurate, but not terribly exciting.

Autoresolve all the things

A surprising and very welcome change is autoresolve not being ridiculously biased against the player, like it was in all previous Total War games. In Empire I'd sometimes try to autoresolve a trivial fight where I had 10:1 advantage and I'd likely wipe out the enemy completely without a single casualty, only to be told that my army lost. None of that here.

Even if win was guaranteed, manually fighting all trivial battles was necessary because reinforcing required getting back to high level settlement (in Rome I and Medieval 2), or paying ridiculous amounts of money (Empire).

Here, army losses are surprisingly inconsequential. Armies reinforce for free, and quite fast, so as long as none of the units in your army get completely wiped out. This was introduced back in Napoleon, but together with reasonable autoresolve, it means there's no point in fighting most one-sided battles. You'd mostly fight close battles, which are far more interesting.

Female generals drama

For my first campaign I took the obvious choice of Ptolemaic Egypt. I don't really give much shit about Western Rome, Constantinople the only true Rome etc. Back in Rome I, Egypt was infamous for being ridiculously ahistorical, with bronze age armies thousand years out of their time. In fact "Egypt" by that time was a Greek kingdom, with the usual Hellenistic armies of heavy phalanx supported by skirmishers and light cavalry.

Now Egypt has a reasonable unit roster. Except all the generals I can hire are female. I was quite baffled by that, and thought that maybe they're some family members, which would maybe be excusable, but nope, they're just some total unrelated randos. WTF?

So it turns out there was this big drama, about 5 years after release, Rome II silently pushed a patch which added female generals, at ridiculously high spawn rates, to all factions including those which had absolutely no business having them. And then instead of toning it down to reasonable levels, and just to factions where it would make some sense, or at least making this silliness optional, devs went full "fuck you all, don't like it, don't play the game" mode. They got very well deserved Steam review bombing for it, but did not learn their lesson.

It's shockingly different from how well a game like Crusader Kings 2 handles gender. Playing a king is quite different from playing a queen, different cultures and religions handle status of women differently, and when you start a new game you can choose a few options to expand state of women. Or if you reform a religion.

Immersion failures continue

One really annoying thing about historical Total War games is that they start by hiding the whole map except your country and its immediate neighbours. It's good that I remember what the map looked like, so I can play based on that. And it turns out Rome II map has very little to do with actual history. Seleucids are really tiny!

In 272 BCE Seleucids were basically half the map, a mega-Persia stretching from Western Anatolia through Mesopotamia, Persia, all the way to a chunk of Central Asia and Afghanistan and Western India. Instead they have 6 settlements on Syrian coast and a bunch of vassals. Wat?

Let's talk immersion. People play historical games for the same reason they watch 22nd Avengers movie. They have connection with historical countries or established characters. I've heard Shogun 2 is a good game, but I've never played it because I don't give a fuck about all the Shimazus, Takedas, and Uesugis. Who is that even? EU4 has about 400 countries, but it turns out over 60% of games are just top 15 nations. Only half of these are even that strong.

It's just so much more fun to immerse yourself in all those historical conflicts. Playing Byzantium in EU4 is borderline masochistic, and yet 1 in 40 of all games is someone trying to stop the kebab menace and restore the glory of Constantinople. People even make mods for restoring Byzantium in HoI4. It's great to also have Hins Kayfa and Tannu Tuva for people who are looking for their 100th campaign, but even these campaign feature mostly well known countries as key antagonists and NPCs.

What does it have to do with it all? When you start with historical setting, or established fictional setting for that matter, you have a budget for how much you can change before people go "fuck this shit, it's not the Harry Potter I love". And many things already demand a chunk of this budget. Better gameplay or technical issues will require breaks with history. Tiny Seleucids might very well be good for the game. Being historically inaccurate to increase the coolness factor is good use for the budget. Being true to people's perception of history rather than actual history (like Rome I Egypt) is fine use of the budget. Every MCU movie needs to spend some time introducing new characters viewers don't care for yet, that uses up part of that budget.

Go too far, and break immersion stupidly, and you get backlash. Even most beloved universes like Star Wars and Harry Potter have breaking point. For historical games this budget is a lot lower. Blowing up a big chunk of immersion budget on something as stupid as forcing female generals on Greek and Roman factions is just so fucking dumb. Not listening to the players is even dumber.

Interface prioritizing minimalistic style over functionality

Anyway, back to the game. Older Total War games had big interfaces where all relevant information was always easily accessible. Rome II instead uses minimalistic highly stylized interface, with completely meaningless icons without text, and where information is hidden behind multiple layers of tooltips, or requires alt tabbing to a wiki. Like, how do I know which buildings I can build in a settlement once I expand it? As far as I can tell, there's no in-game way at all.

From UX point of view that's just atrocious. If you play a lot and don't mind alt tabbing to wiki, you'll get over it eventually, but your first few campaigns it will be a constant pain.

In older games it was really easy to understand what's going on. When game gave me the choice which building to construct, or which unit to recruit, all relevant information was there. In Rome II it's just not there. What's the difference between those two units? Here's 10 sliders, have fun figuring out what they mean. How much money will this building generate compared to that other building? Can anyone even figure this out without alt tabbing to Excel?

This complexity doesn't make game deeper. On the contrary, without any clear information what which choice does, players will either pick at random, or just read somewhere what's the optimal choice, and in either case they'll make no meaningful choices during gameplay.

One interface issue that is highly problematic every single time is that routing units are basically invisible during battles, and chasing them is very important. Giving them white flags like in previous games would be such an obvious improvement.

Bugs 6 years after release

It really did not help my first impression of the game that during the first tutorial siege, AI army hit some invisible wall in the settlement, and just stood there stuck. I tried to attack them, but my armies were also staring at invisible wall in the middle of some street. I finally figured out that if I take my troops out of the settlement and walk in from same direction AI took, I can fight them. It was the only bug I encountered so far (not counting Greek female generals, which are arguably a bug), but wow, those were not good first impressions.

Politics stuff

Rome II has whole extra layer of managing politics of your faction, with other families, something like 40 interactions, civil wars, senate, and so on. It's not clear what all of that does, and so far I've been mostly ignoring it, and it seems to be fine to ignore it.

One baffling thing is that I can't find any options for getting my children married. Maybe all the women joined the army, so there's nobody left to marry?


Battles are great. Maybe comparing your best Medieval 2 battle to best Rome II battle, Medieval II still wins. But thanks to autoresolver and reinforcement changes getting rid of most one-sided battles, much better mix of walled sieges / unwalled sieges / field battles, and how well battles play, I'd say that a median Rome II battle is more fun than median battle in any previous Total War game.

Campaign changes reduce micromanagement, but consequences of the choices are much less clear, so it's a bit mixed.

Interface is just plain bad. It prioritizes style over functionality far more than is reasonable.

Immersion is mostly fine. I'm reasonably tolerant, but I'll get a mod to fix the biggest silliness for my next campaign.

Game performance is so far totally great. It runs better than Rome I on my hardware.

I had horrible first impressions of the game, but mostly positive second impressions.