Dying in the game for the 30th time, you involuntarily wonder: did the game designer think everything through and did he mess up with the balance? It is not always possible to accommodate unexpected changes, especially when they are generated by procedural generation.
Further material, which understands the role of randomness in roguelike games and the genre as a whole - what is fraught with ill-conceived systems of randomness and what, according to the author, is wrong with "roguelikes".
I don't usually play roguelike or rogue-lite. But some look really worthwhile - it seems that the developers were able to get around all the shortcomings of the genre. And every time I regret that I started the game.
What is a rogue like?
Arrogant
Rogue is a 1980 computer game. This fantasy title is largely known for its use of ASCII encoding for graphics reproduction and random map generation. The game proved to be very successful and spawned many roguelike imitators such as Angband and Nethack.
In earlier versions of Rogue, you couldn't save. Saves were added later as the game got longer and more difficult. They allowed you to go through the game in several approaches, restart it from the moment of the last save after death, or if the random did something that you would not like.
Users began to abuse this, so the developers created a system in which saves were deleted after a reboot. That is, it was possible to save when exiting the game, but after the start of a new session, the saved data was erased - without the ability to reboot in case of death or undesirable development of events.
Death became a permanent outcome and was called "permades" (from the English permadeath - permanent death). The one-death mode has become a key mechanic in roguelike games. In 1993, Chunsoft released Fushigi No Dungeon for the Super Famicom, and in 1995 an even more popular sequel, Shiren the Wanderer, was released.
These games not only paid tribute to the predecessors of the roguelike genre, but also applied interesting decisions: what to develop and what to leave in the past. They have beautiful 16-bit graphics and animated characters. At the same time, random level generation, turn-based movement and attack systems, hunger mechanics, random attack values and permades, characteristic of 80s bagels, were preserved.
Thanks to its graphics, sound, and bizarre, intricate maps, Shiren has become a cult title in Japan and among American fans of Japanese games. And in 2008 it was released in the United States under the Nintendo DS.
The resurgence of the roguelike genre
There are hundreds of roguelikes on the market right now, most of which are indie titles that scream loudly that they are exactly roguelike. Canonical projects have all the hallmarks of the genre: randomized levels, random attack values, turn-based movement, hunger, and, of course, permades. Some titles are classified as rogue-lite because they do not borrow all the elements that are characteristic of real roguelikes. Usually these are random levels and permades, but sometimes some others.
Why such popularity? There are two main reasons:
- Procedural level generation is a blessing for beginner developers. If you are an indie developer and make a game with certain levels, then you will have to manually collect at least 20 of these. But you can create a system that will generate an infinite number of them. That is, for X investments you will receive 20 units of profit, and for X + Y investments - infinite profit. What is Y and how balanced and good these procedurally generated levels are compared to manually collected ones is another question. We will return to it later.
- The roguelike genre has a certain prestige. This is because non-game designers confuse the feeling of “this is what I hate about the roguelike genre” with “it takes some work, but everything is fair.” The second really happens: in games like Dark Souls or in PvP modes against very skilled players.
So what's the problem?
In early arcade and console games, death was permanent and forced the player to start from scratch each time. But the game sessions back then were very short, and the goal was to achieve a record in an endless (unless the game crashed due to a bug) series of repetitive levels. And all because of memory limitations.
Modern home computers are equipped with hard drives that allow you to not only bypass the ROM restrictions of arcade and console games, but also save data. Developers can create longer and deeper titles, and users can save their progress, complete games in multiple approaches, and not return to the very beginning in case of character death. The ability to respawn works great in games with set content and titles that need to be completed through trial and error. But in games with random elements, this approach does not fit so well, especially when random elements are generated on the fly and players can reload an infinite number of times until they get the desired result.
When the ability to save was introduced to Rogue, the developers were quick to add permanent death to it so that players would not try to outsmart the system by gaining an unfair advantage. But "permades" also means the almost complete depreciation of the knowledge gained, since the player starts from scratch and the levels are generated anew. It's not that bad and can even be fun if done right, but in many cases the combination of permanent death with the infinite randomness of roguelikes unfairly puts the player at a disadvantage.
A bit about labyrinths
This is a labyrinth. Take a couple of seconds and go through it.
Happened? How long did it take you to realize that it was impossible to pass?
Here are three more mazes. In the third, you need to take the key and open the door to get out.
Here you can immediately see that the first labyrinth can be passed, but the second cannot. But you will have to think a little to understand that the third cannot be passed if you start from the top, and you can if you start from the bottom.
Here is another maze. Here you need to eat an apple every five cells, otherwise you will die of hunger. Can it be passed?
The larger the maze and the more complex the rules, the longer it will take you to go through it or evaluate whether it is possible. Even if you study a hundred examples and determine the passability of the labyrinth at a glance, you only need to limit the field of view to torment you.
Now you have to explore at least part of the maze.
Now I have to explore even more. Will you be able to pass it? Maybe you just haven't found the way with the optimal number of apples?
That's why I hate roguelike games: most of the time they can't be won, because random elements add up to one of millions of different options, and it's impossible to win. It's like a book with a hundred labyrinths, 99 of which are dead ends, but they are large and complex, and it takes several hours to understand that they cannot be passed. And then you need to start all over again in the next maze, having not received anything for the time spent in the previous ones.
Too many directions
Of course, you will say that this is nonsense! Developers are not sadists to make titles that cannot be won, and even develop a bunch of systems that hide the fact that they cannot be won.
And you are right. Developers are unlikely to intentionally create impenetrable games. But many of them use procedural level generation. Not everyone understands that you need to constantly analyze and balance randomness so that the game can be won.
The apple maze came about thanks to the latest roguelike title I tried to play. He looked cool and I wanted to support the indie developer. There was the usual procedural level generation and permanent death, as well as four more health parameters: damage, hunger, thirst and temperature. If even one of them reaches zero, you die and start from scratch in a new generated world. Unlike most roguelikes, the world in this game is linear. You move from one location to another in linear ways, while the order of these locations and their content are determined by randomness. I died a few times but thought I just needed to get used to the system. Then I starved to death because I had no food on the way. No matter how skillfully I played, I would still die for lack of food.
The developers put a lot of effort into creating this game, thinking through all the systems and elements of randomness, but did not realize that a couple of “unfortunate antics” of the randomizer would break everything. Perhaps I was even lucky: I abandoned it pretty quickly, but I could spend hours on careful planning, only to lose by chance and lose all progress.
I hate roguelikes because they don't care what you do, how well you think things through or understand the game. You can accidentally lose and have to start again without any compensation for your efforts.
Of course, the opposite is also true. When Shiren the Wanderer became popular, I played it a little too. Tired of losing to random, I used a save scamming emulator and bypassed random. When I was hungry, I saved, opened the chest, and if there was no food in it, I rebooted until I found it. When I couldn't deal damage, I reloaded until everything worked out the way it should. I did this until the very end of the game, which made my friends very unhappy. They spent hours relying on luck and losing, thinking they were improving their skill in the game. My way of "relying on luck" had the same right to life as theirs, only the result was always in my favor.
In "bagels" is not just one or two random elements: the very essence of the genre involves a dozen of such parameters. Balancing all the randomness is not easy. At the same time, some developers do not understand that something needs to be balanced at all. With all these layers of uncertainty, it's hard to tell when something has gone wrong. It is not clear whether the random system works correctly. Especially if there are several.
Some people like randomness because slot machines continue to exist. It seems to me that roguelike fans believe that skill is more important in these games, not luck. Ignorance in matters of game design and the confusion of these titles make players think that defeats are the result of wrong actions, and victories are the right ones, and this is not a blind accident. People are used to games that can be played and do not think that "bagels" are arranged differently.
Pass the impassable
The two main problems of the roguelike genre are permanent death and the ubiquitous randomness that makes games impossible to complete. How to fix it?
Permanent death "must give"
I lied to you a little. "Permades" does not mean complete loss of progress. So it was in Rogue and the first games of this genre. But since Shiren (or earlier), roguelikes have introduced small bonuses to mitigate the effects of permanent death. In Shiren, you meet characters that can be sent to the first city - even after you die, they can be found in the tavern. They give small buns that help to pass the game. Spelunky compensates for permanent death in its own way - it has the Tunnel Man. It asks for a huge amount, which you can pay in installments over multiple playthroughs. Having received all the money, he will build a tunnel that allows you to skip several levels on subsequent playthroughs of the game.
These bonuses do not remove the disadvantages of permanent death, but rather are taken as an acknowledgment that it is a bad decision and an apology for using this mechanic.
The opposite of this is a quick save (quicksave), that is, the ability to instantly save at any time and restart if things go wrong. This system has been used in PC games since the 90s, and in console games since the advent of the Xbox 360 and PS3. Built-in hard drives made it possible to save quickly and without problems.
Quicksave has its drawbacks. The ability to save at any time significantly reduces the risk and the game becomes less exciting. This is where the derisive term “savescamming” that I used earlier came from. It means that the player saves at every turn and rolls back to the last quick save. And not only in case of death, but in any situation where the game does not go the way he would like. Savescam has a particularly negative effect on the passage of games in which the element of chance is important. The player can save before the random item is determined and then reload until they get what they want. That's what I did when I used the save emulator in Shiren. Compared to savescam, permanent death seemed like a better option.
An intermediate option - although it is closer to quick saves - is save points. You can only save at special checkpoints. Sometimes it has to be done manually, sometimes everything happens automatically. In case of death, all the progress that you had at the time of passing the checkpoint is loaded. There is still a risk of losing progress from the last save point, while there is no longer a risk of losing all progress altogether and starting all over again. The developer maintains the balance in the game by adjusting the number of checkpoints. Games where savepoints are far apart support a higher level of risk than games where they occur at every turn.
There are two types of savepoints. The first are manual, when you need to make a conscious decision in order to survive. Usually in these places there is a special landmark. The second ones are automatic, in which the game is saved on its own after certain conditions are met. Usually this is due to some kind of story or quest event. Automatic save points are often paired with quick saves to make sure the player can exit the game and return at any time without losing progress.
Chulip, the cheeky PS2 game, can be manually saved using the toilet
Salt & Sanctuary politely reminds you not to turn off your computer while autosaving is in progress
Over the past ten years, a middle ground has emerged between permanent death and save points. These are the so-called Souls Death, which gained popularity thanks to the Dark Souls series. These games have the usual checkpoints, upon death you return to the last point, saving your progress and equipment, including what you discovered after saving. In this case, all the currency remains at the point where you died - you can go back and find it. But if you die earlier, they will be lost forever, as death creates a new save point for the accumulated funds, which you can find after resurrection.
Despite the confusion, the system was well received. Instead of permanent death with the loss of everything, users were offered checkpoints with progress saved, new risks were added to the game and a chance to return lost resources.
Tame your random
In the picture above, from left to right, "permades" is complemented by more frequent saves in roguelike games. The user has more and more opportunities to explore the frenetic randomness, a key element of this genre. The curtain rises, and it becomes obvious that permanent death is necessary for the functioning of the random house. Without it, players can use saves to some extent: this tactic can be resorted to even if the checkpoints are located far from each other. You can spin the random as you want, and not fight it and mistakenly perceive it as a fair challenge.
That is why permanent death is the most important element of the roguelike genre. All other bagel mechanics are tied to it. If a game has everything except permanent death, then it is usually not classified as a roguelike.
All roguelike games are created exclusively in the Permades region of France. Otherwise it's just a "procedural dungeon"
Even without removing permanent death, you can make the "bagel" passable, but you will have to reduce the amount of randomness. The first thing that is abolished is the random outcome of battles. Instead of relying on luck, players need to develop skills. This is much more fun, and the game becomes more fair. Most modern roguelike titles have already taken this approach.
Further, everything becomes much more complicated and requires significant effort without visible reward. There are three options:
- Make it so that the random house system analyzes itself. Then the player will not end up in a deadlock due to randomness.
- Make sure that any random outcome is favorable. That is, even if the player does not get what he wanted, and he has to change his strategy, the outcome will still not be negative.
- Make it so that the randomness is less defining. Then the player will be able to compensate for any negative outcome with his skill.
Option 1: Random introspection
It is difficult to find suitable examples, since such processes take place behind the scenes. This approach was definitely used in roguelike and games of other genres with elements of randomness. For example, we have implemented a sophisticated analysis system to ensure that the semi-random distribution of planets is "fair" in . But without looking at the game's code, it's hard to tell for sure whether such a scheme was used.
A hypothetical example would be an improvement to a game where I didn't find food and starved to death. An introspective system would be able to provide food at the first X locations, and then food would appear at every Y±Z location. Then the player would not die of hunger by chance. Further, a logical design decision would be to create situations in which the player knows that he will soon find a source of food, but he does not know exactly when this will happen. You will have to choose: play safely and stock up on food, or take risks and take less food with you, but more things that allow you to craft rare supplies.
Option 2: always a positive outcome
Examples include Let It Die and Sundered procedural dungeons. In Let It Die, the dungeon is divided into zones. Each of them has several rooms with several exits. Their location is random, and the exits randomly lead to other rooms, forming a specific area.
Here is one of the possible room configurations in the Tagahara zone in Let It Die. The circular area towards the middle is a wide vertical shaft with one entrance at the bottom and two at the top. Such a map element creates different navigational challenges (as well as situations in which you need to fight off enemies), depending on which entrance you use.
The Sundered map has both static areas that are always the same, and large dynamic areas that contain small rooms with a random configuration.
On this map from Sundered, the areas with dark vignettes are static, they are always the same. Light gray, on the contrary - randomly generated. These rooms fill the entire space, all the doors to them can be opened. But to find the best path, you have to explore the map.
Regardless of the location of the premises, there is always a passage. Different configurations need to be played in their own way, but there are no “dishonest” cards in these titles.
Another plus is that procedural generation is applied in certain areas, while the rest of the areas remain static. Players are given landmarks instead of being thrown into an ever-changing ocean of randomness with no clear signs.
Option 3: reduce the influence of randomness
This option ensures that failure (either a single random event or a chain of them) does not lead to the death of the character, leaving the player no chance to overcome these events or adjust to them. Although Fortnite is a multiplayer game with short play sessions, the chests in it are a great example of this approach. Each of them contains a random set of weapons and other items. Good random items give the user an advantage, but a skillful player can win even if he finds only the worst items in every chest he opens.
Conversely, this approach ensures that the player does not win purely by luck without any skill. Again, in Fortnite, even the coolest weapons from the chests do not provide much advantage if the player does not know how to use them effectively.
An example would be the Oriental Medicine sticker in Let It Die. It is found randomly and allows you to restore the health of the character. Without it, you need to constantly monitor the level of health and have food in stock in case of recovery. With this sticker, you don't have to think about healing - you are invulnerable for the first third of the game, until you get to enemies that deal damage faster than you regenerate health.
Conclusions for developers
If you're making a video game, be it a roguelike or not, a board game, a tabletop RPG, or anything else, try playing around with the settings.
Set everything up so that the player is totally unlucky with randomness. Perhaps the game is now impossible to pass, regardless of the skill of the user. If there is a playable build, try to generate every random element in the worst possible way. If your game only exists on paper so far, run a similar simulation in your head.
Then do the opposite: turn up all the settings for the most fortunate set of circumstances, and see if it is possible to lose the game. Alternatively, you can imagine the consequences of a positive event or a chain of events that offset all risks. The game will turn into a simple dummy where the user misses out on huge chunks of story and mechanics because they can be bypassed if you're lucky enough.
Note that I'm talking specifically about random elements, and not the level and progress of the player. A billion-level player who easily passes all obstacles is very different from a player who accidentally received an invisibility ring and also easily passes all obstacles. The first put in a lot of effort, and the second was just lucky.
Conclusions for the players
I hope this crazy tirade has helped to sort out a bit the element of randomness in games, and you will become more critical and conscious about what depends on your actions, and what depends only on the will of chance.
Of course, a certain amount of randomness can be interesting, and everyone has their own threshold of sensitivity to it. But in my opinion, the better a person understands what depends on his choice and skills, and what depends on luck, the more pleasure he gets from video games.
Source: habr.com