Save Game Systems

I once heard Peter Molyneux say that during the development of Populous he didn’t want the player to be able to pause the game. His reasoning was that Populous is a world that goes on with or without the player. Luckily, his friends talked him out of it, pointing out that sometimes the doorbell rings, the phone rings, or the baby cries.

Players should have some basic rights. They have lives outside of our games and we should respect those lives and design our games accordingly, rather than expect our players to design their lives around us. Players should be able to save anytime they want, or more precisely, they should be able to stop playing your game anytime without losing their meaningful progress.

This is an old argument where one side talks about the convenience of saving anytime and the other talks about the need to make games challenging, but this is a false dichotomy. We can allow the player to stop playing without excessive penalty and make a challenging game. It’s just a matter of defining what “saving” actually means.

As an example, Mario 64 doesn’t literally allow the player to save anywhere they want, but it still meets this requirement in spirit. The point of the game is to collect all 120 stars, and every time you collect a star, you “save and continue.” You cannot save your exact position in a level, but such a feature isn’t needed anyway. The geography of the game is designed such that a player can reach the entrance to any level in just a few seconds by navigating Mario’s castle and getting back to any specific goal in a level doesn’t take long either. This preserves the game’s difficulty (players can’t save and load to get the stars more easily) and it also means the player can turn the game off at any time, knowing that the only important progress (collecting stars) has been saved.

Save Point vs. Checkpoint

God of War 1 and 2 and Resident Evil 4 all use the same save system, which is also common in many other games. They all have save points and checkpoints. Save points let players save their progress and load it later. Checkpoints are sprinkled invisibly between save points and if they die, they go back to the last checkpoint rather than all the way back to the last save point. This system isn’t too bad, but it doesn’t do a good job of letting the player save and quit at any time, either. It would make more sense if the player could pause the game at any time and save progress up to the last checkpoint. I’m not suggesting that the player should be able to take a step, save, fire a shot, save—just that they should be able to stop playing the game and resume from the last checkpoint. After all, that would happen anyway through dying.

Why separate save points from check points in the first place? I think the answer is for technical reasons rather than design reasons. God of War was designed for the PlayStation 2 and Resident Evil 4 originally appeared on the GameCube (and later on PlayStation 2 and Wii). These consoles take a few seconds to write a save to the memory card, so doing this every time the designers wanted a checkpoint would probably have been too annoying to the player. This lead to spread out save points and the addition of check points for convenience’s sake. In the future, we won’t have these technical restrictions.

Gears of War was designed for the Xbox 360, a system capable of writing a save file quickly. Gears of War’s save system is a definite improvement over God of War’s and Resident Evil’s: The player can play through the entire game without having worry about finding save points, but can also quit playing at any time and automatically start at the most recent checkpoint. Gears of War does this by having many checkpoints, all of which automatically save progress without any action required from the player. This example well-illustrates the false dichotomy I mentioned earlier. The save system is both very convenient and does not interfere with the difficulty of the game. In fact, Gears of War could be tuned to be arbitrarily difficult without sacrificing any convenience in its save system.

Multiplayer

Save systems get a little trickier in cooperative multiplayer games. Players expect to be able to join a friend’s game and leave at any time, and to save and continue their progress later without the game’s save system getting in the way. Gears of War does a great job here too, allowing a friend to join an in-progress game at any time (taking over the AI for the character named Dom). The player can get through a couple of chapters alone, then have a friend join who can leave at any time and pick it up again later. Even if the friend is new to the game, they’re still allowed to join someone who’s playing the last level, because Gears of War is trying to be as convenient to the player as possible.

One hitch is that when the friend leaves, the player must briefly quit the game then restart it from the same checkpoint. On this matter, Lego Star Wars has Gears of War beat because it allows a friend to seamlessly join or leave a game without ever quitting out to a menu screen.

Playing Gears of War with a friend is easier than playing alone (there are no AI adjustments between coop and single player), but it could have been incredibly difficult had the designers wanted it to be. The save system’s flexibility doesn’t prohibit difficulty. That said, if you were really serious as a designer about creating a meaningful leaderboard for single player and co-op play (Gears of War doesn’t do this), then you’d need a single player mode where no one can ever join in, and a co-op mode where the two players are set from the start and can never switch out. This would be highly annoying, so it should only be used as a hardcore leaderboard mode inside a game that also offers a more forgiving system.

Massive Saves

In massively multiplayer online games things get even trickier still. On the plus side, players can log out at almost any moment they want in these games, and their character’s progress (such as items or experience points) will be saved. In World of Warcraft, players can’t log out while “in combat,” and must wait 20 seconds when they do want to log out, but it’s pretty player-friendly overall. There’s even a hearthstone that lets players teleport back a city (once per hour) so they can end their play session at almost any time with character progress saved. What’s much harder to save is progress on a quest or in a dungeon. If a group of four friends is halfway through a three-hour dungeon, one could log out, but it’s socially unacceptable and that player won’t be able to continue their progress in that dungeon later. This is a worse problem during raids where 25 people must coordinate their real-life schedules and the ability to log off at any time is basically gone.

Blizzard has taken some steps to simulate the kind of save points seen in offline games, though. The Scarlet Monastery dungeon starts in an ante-room with four separate portals leading to four different wings. This allows players to play just one fourth of the total experience, stop, and come back later. Also, the Mauradon dungeon gives players an item half way through that allows them to teleport back to the half way point, so they can continue their journey later.

Blizzard added even more winged dungeons and pseudo-save points half way through dungeons in the Burning Crusade expansion to World of Warcraft. Players welcomed these changes as they make the game much more convenient, though they still fall somewhat short. A single player game with save points more than an hour apart would be considered lacking, but at least Blizzard is moving in the right direction here. There is opportunity in the MMO genre to be even more friendly to players’ real life schedules.

Diablo 3

Blizzard's Diablo 3 is notable for having such different save game systems on different platforms, at least for a while. In the Mac and PC versions, the save system was similar to an MMO. That means it saved everything you do at every moment. Also like an MMO, you couldn't pause the game. If you brought up a menu screen, the game was still running underneath it and monsters could still attack you.

In the Xbox and PlayStation versions, you can pause though. Bringing up those very same menus in these versions pauses the action and monsters can't attack you. The console versions of Diablo 3 have a lot of good features going for them, but the ability to pause turned out to be one of the best "innovations" to me over the Mac/PC versions. (I'm busy and often get interrupted, so this comes up constantly.) Thankfully, after the first expansion to the game, the Mac/PC versions were updated to also allow pausing.

Outliers

Let’s return to single player games and look at two unusual examples: Dead Rising (Xbox 360) and Castlevania: Dawn of Sorrow (Nintendo DS). Dead Rising has save points, but no checkpoints. The open-ended nature of the game makes it very easy to forget to save at all, especially considering that the save points are off the beaten path inside the various bathrooms of the shopping mall where the game takes place.

When players die in Dead Rising, they are given a confusing choice: they can restart from their last save point, losing all character progress since they last saved, or keep their character’s progress, but lose all save points. Yes, you read that right. If a player wants to keep their character’s progress since the last save (such as experience points gained and moves learned) then they must restart the entire game from the opening cut-scene. Even stranger, Dead Rising only allows a single save slot per Xbox 360 profile, per storage device.

That means the game is trying its hardest to restrict people into playing the game only the way the designer wants, while still remaining easily defeatable if one makes a new profile or uses another memory card. By “defeatable,” I mean this grants users the ability to create two save files, a feature common to almost all games.

The reasoning behind these decisions in Dead Rising was probably to create a very specific experience for the player. They are supposed to care about finding those save points, and care that they are in constant danger from zombies and that if they die, the last save point was a really long time ago so it’s going to be a big deal. The world is against the player—as it almost always is in the horror genre—and so the game’s difficulty is intentionally very hard. If the player keeps playing through the game and dying and starting over, they’ll start each time with a stronger character and with more knowledge of how to navigate the game correctly and save the various victims from the zombies. Incidentally, this same save system was used in the game Breath of Fire: Dragon Quarter, which was also by Capcom and is rumored to share some team members with Dead Rising.

I understand why a designer might create a save system like this that reinforces the concepts of the horror genre, but that's making a game that's too much for the designer and not enough for the player. I was personally annoyed by this system to the point of quitting because I could not play it the way I wanted. Dead Rising is an amazing technological showcase and combines the design concepts of a sandbox game (go wherever you want, do whatever you want) with the horror theme of a mall overrun by zombies. And yet, I’m not allowed even two save slots, I’m bullied into playing the same parts over and over because I feel obligated to restart all the time, and the save points require me to actively seek them out, which means it’s very easy to play for an hour or so and forget to save, then die. That type of save system may work for hardcore players (who border on sadomasochism anyway), but the fictional Little Jimmy from Idaho (the person I often design for) is just going to quit playing out of frustration. I know I did.

On the other hand, Castlevania: Dawn of Sorrow has an unusual save feature that is intended specifically for the player’s convenience, rather than for the designer’s vision. This game has standard fixed-location save points (with no check points) and it also has a second method of saving called a save marker.    

Players can pause the game at any time and create a save marker, and then the game quits to the title screen. When they want to play again, they can either load a game that was saved at a save point or they can resume from their last save marker. The tricky part is that if they resume play through either method, then the save marker is destroyed. That means if the player is in the middle of a boss fight, they can save, stop playing, play something else, then later resume from the exact moment they saved. But players cannot reduce the game’s difficulty with this feature because it does not give them a second chance of any kind. This is another example where the game can remain very challenging, and yet still allow the player to save and quit at any time. This same save system was also used by Fire Emblem (Game Boy Advance) except you didn’t even need to pause and create a save marker. It was automatically created for you any time you turned the Game Boy off during gameplay.

New Mario, Old Trick

One of the most surprisingly bad save systems of recent times comes from an otherwise wonderful game: New Super Mario Bros. (Nintendo DS). It would have been very natural in this game to allow the player to save at any time on the map screen between levels. Instead, the player must beat either the castle at the end of a world or the tower halfway through the world in order to save. For example, in World 2 this means beating a minimum of five levels before reaching a save point. Players can also spend their hard-earned star coins to buy a powerup from various mushroom houses which also lets them save, but they very well might not want to spend their coins.

The need to keep the player at arm’s length from the ability to save is conspicuous here given the traditions of the genre (Mario 64 did much better) and doubly-so considering this is a handheld game. Surely the concern wasn’t about keeping the game challenging, because NSMB lavishes the player with extra lives the whole way through. My girlfriend once asked if she could play Nintendogs on our DS, and I had to explain to her that no, she couldn’t, because I just spent almost an hour collecting nine star coins and didn’t reach a save point yet so I had to leave the DS in sleep mode until I could save. I’m not sure which game designer sensibility this restriction on saving serves, or why it would ever be more important than allowing my girlfriend to play with her virtual dog.

NSMB really stands alone here. The most incredible part is that when you beat the game, you unlock the ability to save anytime you want on the map screen! This proves that no technical limitation made the save system the way it was. The convenience of saving anytime was deliberately withheld from the player, and given as a reward at the end. As designers, we can’t do this, and must instead put the real lives of our players ahead of our game designery ideals.

Saving For The Player

A save system should allow the player to stop playing at any time, allow the player to pick up where they left off with as close to zero replaying as possible, and save as automatically and seamlessly as possible, so the player will not forget to do it.

Saving should be treated as one of the player’s natural rights, not an earned privilege or a game mechanic around which to make strategic decisions. The design space we have to create new games is so unthinkably large that we lose virtually nothing by restricting ourselves to designs with friendly save game systems that don’t presume to override the real-life needs of players. As I have shown, this does not even require a tradeoff with game difficulty; even difficult games can have convenient save systems.

We should always try to design a save system that simply serves its purpose and fades into the background, otherwise we might end up like New Super Mario Bros.—a game with sales of over 10 million units worldwide, and with ten million girlfriends unable to play Nintendogs.

Pacing for Impact

Is it better to save your best content until the end of a game so you have a strong finish, or is it better to make the first few minutes of gameplay as good as they can possibly be? If your best stuff only shows up after the player has invested 20 hours, reviewers and some players might not even know it’s there. But if you "give away the farm" on the first level, the game has nowhere to go but down.

The general trend I see in successful games is that they tend to show a great deal of their coolness (but not all of it) in the first few minutes to half hour of gameplay. Let’s look at some case studies.

Metroid Prime

As of this 2006, Metroid Prime is the 4th highest rated game of all time on gamerankings.com (10th highest as of 2014), receiving a 9.7 from Gamespot, a 9.8 from IGN, and a perfect 10 from EGM. It’s sold 1.3 million units on GameCube, according to TRSTS data.

The first few minutes of Metroid Prime show off an amazing amount of the game. We learn basic movement (R button shifts to freelook, L button shifts to strafe/lock-on). A button shoots and B button jumps. If you hold the A button, you get a fancy charge-up shot, while the Y button fires missiles. The X button turns you into a ball (with 3rd person camera) equipped with bombs (that make you bounce), and everyone loves rolling around as that ball. Your visor lets you scan the world to get info and tips and even open some doors. Most doors you just shoot to open, and your charged shot can be used to clear rubble. You also learn how to operate elevators and use the save stations. After only a few minutes, you fight a boss where you learn how to circle strafe while locked on and dash sideways during a lock-on. A few seconds after that, you get to use your grapple gun. (It was probably a mistake that they had you use the grapple gun for the first time during a timed sequence, but oh well.) You also get a taste of Metroid Prime’s map and mini-map, which are probably the best in-game maps of 3D levels the industry has seen yet.

That’s an incredible number of cool features revealed in the first few minutes of the game. It makes you realize right away that Metroid Prime is a class act that deserves your time. Incidentally, after the intro sequence, your character gets damaged and loses access to the morph ball, charged shot, missiles, and grapple gun. The game designers need to give you these items slowly over time to reward you, but they wanted to make sure your first few minutes were packed with coolness, so they gave you a great taste of what’s to come.

Grand Theft Auto 3

This infamous breakthrough title conveys its core ideas in the first few minutes. The game starts with a very short series of three missions: First, get in the car and drive your buddy to Point A, then drive to Point B, and finally pick up a certain passenger at the hospital and take her to Point C. This sequence teaches you how to get in and out of cars, basic driving (gas, break, turning), how to change the radio station in the car, how to pick up passengers, how to get a new car if your current car gets too damaged, and how to use the mini-map to find mission objectives.

After those first three missions, the game turns you loose into the world to do whatever you want. Doing whatever you want is the core concept of Grand Theft Auto 3, and the player realizes it right away. You can drive anywhere. You can fight people on the street and take their money. You can crash cars, wreck stuff, and steal cars. You can totally ignore the story and mission structure and make up your own story and missions. In doing so, you quickly learn about the police "star" system where committing worse and worse crimes increases force of police who are sent after you. You have to hide out or find secret police stars hidden in the world to reduce your infamy rating and get the cops off your back.

I’ve watched several people play GTA3 for the first time, and all of them abandoned the game’s mission structure within 5 minutes to explore the world and create their own goals. No wonder it sold over 5.6 million units on PlayStation 2 alone (and many times more than that once you factor in other platforms and expansions).

Castlevania and God of War

Castlevania: Symphony of the Night (PlayStation 1) and God of War (PlayStation 2) are both examples of a near ideal distribution of “good stuff.” Both games start by showing you a large portion of the game mechanics. Castlevania uses the same trick as Metroid Prime where the player gets to start with a bunch of cool moves and weapons that they won’t get to use again until much later. God of War introduces basic fighting, ground throws, air throws, opening hatches, walking tightropes, a boss fight, special finishing moves, and use of magic all within the first few minutes. Note that best boss is the first one (the Hydra) and the most fun and effective magic power is the first one you get, (Poseidon's Rage, the 360 degree lightning attack). Each of these games is putting its best foot forward to get your attention from the start.

The interesting thing is that these games feel great right off the bat, but they don’t feel like the 9/10 or 10/10 games that they are. In each case, something later in the game takes the quality from 7 or 8 up to 9 or 10. In God of War’s case, it’s the emotional content of the excellent story that builds to a very satisfying conclusion. Even though the game is a “fight a bunch of guys game,” the story and presentation elevate it to the status of “memorable experience,” rather than just “brawler game.”

Castlevania: Symphony of the Night has one of the biggest surprises in games, and if you haven't played it, I’m about to ruin it. The game leads you to believe you've reached the end when you find Dracula and kill him. There’s a map that keeps track of what percentage of the game you’ve visited, and it approaches 100% by the time you reach the big boss. The surprise is that this is only the halfway point! The boss causes the castle (the entire game world) to flip upside down, and you must now play through it all over again, this time walking on what used to be the roof. Chandeliers stick up from the ground, you walk on the undersides of stair cases, and you begin to realize that the entire game was planned from the start to support an entirely different upside down game! Wow! All the enemies are replaced with harder enemies, and the various keys are hidden in new places. The awesome design of the upside down world elevates Castlevania to a very memorable experience.

So if starting out strong, but ending stronger is the key to victory, then what are some examples of games that break this trend? Games that start out weak and end up weak aren’t very informative here, but games that start weak and end strong would be great examples. We’d expect those games not to sell very well.

Psychonauts

I hate to pick on Psychonauts because Tim Schaefer’s great writing in previous games is one of the reasons I joined the game industry in the first place. That said, the first 12 minutes of Psychonauts are, from a gameplay perspective, a very poor experience. The only interactive things I did in those 12 minutes were enter my name, move the camera to the right and then up one time each in a tutorial, and walk two steps to an NPC that triggered even more movies. The rest of the 12 minutes was all movies. I just wanted to play the game. What’s just as bad is that after two hours, I didn’t get even a single Psi-power. Only after about three hours did I get to see anything that set this platformer apart from any other platformer, and the real interesting stuff isn’t until much later in the game. Even though many people told me that the game has wonderful ideas and cool gameplay as you get into it more, the first time I played it, I never made it past the first 12 minutes.

Traditional RPGs

RPGs in general also suffer from this phenomenon. Most (Final Fantasy-like) RPGs start you out with a wooden sword, no spells, and have you fight a few rats or something. Over time, your arsenal of spells and attacks increase and you usually get the ability to do combos of spells (or use your party members together in combos) that are pretty interesting and fun. This fun tends to come later in the game though, at hour five rather than minute five. This is perhaps why the single-player RPG genre isn’t selling as well anymore, except for games called Final Fantasy or games that have the Star Wars license. [Post publication note: Oblivion and Skyrim are RPGs that did well, but note that they use the GTA3 sandbox gameplay model.]

Pace for Impact

It should be no surprise that you need to start out strong, or at least strong enough to grab the player’s attention. Burying the best content at the end is generally not a good idea, but it’s a question of degree. If your final boss is a 9/10 experience, but your first level is a 2/10, you have a major problem because no one is going to see that final boss. On the other hand, if you can get that first level up to a respectable 8/10 experience, then ending on a 9/10 boss is great, and perhaps nearly the ideal scenario. In any case, don’t be tempted to save all the fun until after the 20 hour mark because your first level is going to be your most played and most judged level. Go the extra mile to make it stand out, even if it means giving the player a preview of a few fun mechanics you planned on saving for the end.

One last note: if you really want to start in a way that no one ever does, then get rid of all that junk that players are tired of waiting through when they turn on a game. Get rid of the intro movie with the publisher’s logo, the intro movie with the developer’s logo, the legal screens (put them as an option on the main menu), and any other non-content cruft you can find. It’s getting totally out of hand how many screens of garbage games start out with before getting to the main menu. You’re much better off building brand awareness by actually making a good game than forcing everyone to see your logo every time the game boots. Put your logo in the main menu itself if it matters so much. When I pay $50 for a game, I expect to be exempt from even 5 seconds of this stuff. (Technical disclaimer: sometimes those logos are shown during loading time, and removing them wouldn't actually save any time.)

Almost no games follow the above advice. Yours could be one of the few. Oops, I saved my most interesting idea for the end of the article where no one will really see it.

Designing Defensively: Guilty Gear

The more variety there is an asymmetric game, the harder it is to balance. When each character (or faction / race / deck / etc.) is extremely different from the rest, there's more potential for some characters to do things that other characters just can't handle. But having radically different characters is really interesting, so is there some way we can keep things balanced anyway? The fighting game series Guilty Gear shows us how.

How They Did It

They handled offense and defense differently.

Universal Defense: all characters have equal access to an unusually large number of safeguards and defensive abilities. This includes failsafes that solve problems the designers didn't have to specifically know about ahead of time.

Unique Offense: each character has unique mechanics that stray further from the standard template than they would in most other fighting games.

Because the designers can count on all characters having so many ways to get out of trouble, they can give each character radically different offensive tools. I think of features shared by all characters as the "design skeleton," and the different options each character has as the "meat on the bones." Even though all characters share many defensive options (and a few offensive options), they feel extremely different. That's because your offense options define your gameplan—what you're actually trying to do when you play.

How different are we talking? The characters in Guilty Gear are more different from each other than the characters of just about any other asymmetric game that I know of, yet the series is still reasonably well balanced and definitely suitable for tournament play. The characters have such different mechanics that sometimes they feel like they're from different games. Here's a quick sample: 

  • Zato-1 lets the player control a second character at the same time as the main character.
  • Venom can place pool balls on the screen and change their arrangement and trajectories in complex ways.
  • Faust can throw random items onto the screen and he has to take advantage of whatever he happens to get.
  • Jam has to summon "cards" in order to power up her moves
  • Johnny has a limited number of coins per round, and hitting the opponent with coins charges up his "mist finer" attacks. He can also throw a mist onto opponents that makes them unable to block "mist finer" attacks.
  • Robo-Ky has his own unique super meter as well as a heat gauge that fills as he does moves, and a way to let off steam so he doesn't overheat.
  • Bridget can throw a yo-yo around the screen and then return it, turn it into a floating razor-bear, and many other tricks.

The Design Skeleton

The reason Guilty Gear is able to have such diverse characters (that are still well-balanced) comes from the common set of things all characters share. Let's look at some of those things, staring with the most basic.

The Basics

With only a few exceptions, all characters have access to this suite of movement abilities and basic attacks:

  • Double Jump
  • Ground and Air Dash
  • Sweep Attack
  • Overhead Attack that Launches
  • Ground Throw and Air Throw
  • Air Recovery (aka "tech recovery")
  • The ability to block high and low

There aren't any characters who can't block low, for example. There also aren't any characters who can't jump, or who can't walk forward. Characters lacking such basic things would inevitably cause major balance problems, so Guilty Gear wisely avoids doing that, but then so does just about every other fighting game.

Getting Out of Trouble

Some character is attacking you and doing whatever crazy and unique thing that character has. What do you do? No matter which character you are, you always at least have these options.

"f+p" invulnerability. Every character has a move performed by holding forward on the joystick and pressing the punch button. For every character (except Testament), this move grants some invulnerability to the upper body. This means that if an enemy jumps in at you, a f+p move is very good "anti-air." That is, it works well against attacks coming from above (unless the enemy expected it, and double jumped).

Green Blocking (aka "faultless defense"). While you are blocking, if you hold two buttons down, your character becomes surrounded by green rings. During this time, your super meter depletes, but you will take no "block damage" as you normally would from projectiles and other special attacks. Also, when you green block an attack, you will be pushed farther away than you normally would. This allows you to loosen up your opponent's traps by pushing them too far away. You can also use green blocking while you are in the air to block attacks from an opponent on the ground (you can't block those types of attacks without green blocking). Green blocking allows you to get out many situations that might be potentially brokenly oppressive, but it does cost you in-game resource (super meter) to do.

White blocking (aka instant blocking). If you block an attack at the very last moment before it hits you, your character flashes white. This reduces your character's blockstun, meaning that you recover from blocking slightly faster than you otherwise would. It's difficult to time a white block correctly, but it can help you get out of a sticky situation. If there's some sequence of repeated moves that you have to block, white blocking might let you recover just fast enough that you can interrupt the next incoming move.

Alpha Counter (aka "dead angle" attacks). While you are blocking an attack, you can perform this maneuver to cancel your blockstun with a pre-set attack of your own. It costs a lot of super meter to do, so you can't do it that often. If you're being overwhelmed by attacks though, blocking and then alpha countering gets the opponent off you. It knocks them down and gives you a chance to start up your own momentum.

Infinite Combo Safeguards

These next several features contribute to preventing "infinite combos." An infinite combo is a situation where once the opponent lands the first hit of the combo, they can continue the combo forever until they win. The more complicated a fighting game is, the more likely it is to have unwanted infinite combos. The designers of Guilty Gear put several systems in place to reduce the likelihood that such combos could exist.

Burst. This is the most blatant failsafe to stop infinite combos: a move that lets you break out of a combo. The game Killer Instinct pioneered this concept, but implemented it badly. Guilty Gear perfected it. While you're getting hit by a combo, you can "burst" out of it to knock the opponent away and avoid being hit further. It's your "get out of jail free card," and every character has it.

You can only burst about once per round. The designers surely knew that this mechanic was so powerful that it had to be closely regulated. If it cost super meter to use, there would be some character who's good at generating super meter and would be able to burst too much. Instead, burst has its own meter. It starts full at the beginning of the fight, and using your burst empties the meter entirely. The meter fills up automatically over time, but the only thing that can affect the speed it fills up is how much you're getting hit. If you get hit a lot, you get it back a bit faster. There's no trick to get it back instantly.

Your burst meter does not automatically refill each round, which was a clever decision. If you're at the end of a round and getting hit by a combo, you might very well choose NOT to burst so that you'll still have that ability at the start of next round. You should only burst if you think it will actually help you win the current round. The overall effect is that you don't see bursts every round because often players save it for the next round. The failsafe is there, but it doesn't get used excessively.

Also, a clever opponent will expect the moment you'll burst and they'll voluntarily stop attacking right before that. That makes your burst whiff, so they can punish the recovery of it with an even bigger combo. Even the failsafe itself has a counter.

Guard Meter. Right under your health meter is a little red meter called the guard meter. It starts at 50% full, and naturally tends to wander back to 50% if it gets higher or lower. The more attacks you block, the higher that meter gets. The more attacks you get hit by, the lower that meter gets. The higher the meter is (the more attacks you recently blocked) the less you benefit from the game's normal system of damage scaling. Usually, when you get hit by a combo, each successive hit is "scaled" down in damage more and more. But when your guard meter is high, even an ordinary combo can do massive damage to you because you are not being protected by the usual damage scaling. This is meant to punish overly defensive players.

On the flipside, the lower your guard meter is (meaning you got hit by a lot of attacks in a short period of time), the more damage scaling you benefit from. A very, very long combo will eventually do only one pixel of damage per hit because of this feature. So even if an infinite combo did exist, it would take an incredibly large number of hits to actually kill you and you could probably burst out of it before then.

Furthermore, you receive another even more important protection when your guard meter is low: reduced hitstun. Every time you get hit by a move, you are briefly stuck in a reeling animation where you can't do anything (except burst). This is the basic concept that allows combos to exist at all, since the opponent can often hit you again before your hitstun ends. But in Guilty Gear, the more you get hit, the shorter your guard meter becomes, which then makes your hitstun shorter. So if there exists a combo that is a "loop" of repeated moves, it may be possible to do 3 or 4 repetitions of the loop, but eventually the opponent's hitstun becomes so short that the combo simply stops working. (They'll be able to block at some point.)

Progressive Gravity. The longer your character is being juggled by a combo in the air, the greater the force of gravity on your character becomes. Many infinite combos in fighting games involve "juggling" a character in the air with attacks. Much like with actual juggling, it is a fight against gravity to keep it all going. While it might be possible to do 3 or 4 repetitions of a juggle loop, eventually the victim's body falls so fast to the ground that the juggle is no longer possible.

Summary of Mechanics Shared by All Characters

Every character has upper body invulnerability with their f+p move (except Testament). That alone is a great help in stopping attackers. White blocking can get you out of some situations without spending any super meter, but it requires precise timing and has only a small effect. If you're willing to spend super meter as a defender, you can green block or alpha counter. Green blocking will protect you from all damage while you block (except throws) and it will push the attacker away from you. Alpha countering is another method to get the attacker off of you, and it can be performed while you are in blockstun. 

That's a lot right there, but there's plenty more. If you actually get hit by an attacker, you have all sorts of things going for you. Your guard meter will eventually reduce the damage you take by the combo and shorten your hitstun allowing you to escape. Increased gravity will also eventually thwart their combo. You can nip it all in the bud, though, by simply bursting right at the very start of their combo, avoiding almost all damage. And don't forget that a great way to nullify attacks is simply not to be in the way of them. Every character can double jump, and every character but one can air dash.

The Freedom to Design Crazy Attacks

Every Guilty Gear character has a lot of options and system features that help them escape or avoid degenerate situations. The beauty of it is that the designers didn't need to know what those situations even were. They don't need to know which combo could juggle forever, because no matter which one it was, progressive gravity probably stopped it. They didn't need to know which non-juggle combo went on forever either: the guard meter's reduced hitstun solves that.

This gave the them the confidence and the ability to arm each character with pretty extreme and wildly different mechanics while maintaining reasonable balance. So that's the recipe: a robust, shared system of defense and failsafes with diverse and unique attacks for each character.