Difficulty, Yoshi's Island

Yoshi's Island is considered by a lot of people to be 'surprisingly hard'. I am of the suspicion that this is because it has an extremely forgiving 'health' mechanic. (I suspect, by contrast, that part of why people view its difficulty as surprising is precisely because of the extremely forgiving health mechanic) Yoshi's Island allows you to take as long as you like and get hit as many times as you like -so long as you never let Baby Mario stay separated from Yoshi for too long of a stretch. This means the game can be pretty generous with danger. So long as it isn't lava, spikes, or bottomless pits -all of which instantly kill Yoshi- any hazard is one a player can run headlong into, experience consequences, but still keep playing the level.

Most games immediately and severely punish mistakes, which means they have to carefully construct themselves so that a 'typical' player will only rarely actually experience mistakes. Games in which there are health pickups or other ways to 'undo' mistakes to some extent or another can extend the number of mistakes allowable within a level theoretically infinitely, but it will remain the case that the number of cases where the designers expect mistakes have to be kept within the bounds of the innate resistance+such power-ups.

To place this in concrete terms, let's look at the Donkey Kong Country games, which came out in a similar timeframe and are both platformers where beating the game and 100%-ing the game are different things.

In DKC games, you get two Kongs.

When you take a hit, you lose a Kong. If you're down to one Kong, that kicks you out of the level entirely and it's time to restart from the beginning, or maybe from the halfway mark if you reached the Star Barrel before you died.

There are DK Barrels that the player can use to get back to two Kongs when they're down to one Kong. If you already have both Kongs, DK Barrels are no more useful than a generic Barrel -you can't get a third Kong out of them. You can, to a limited extent, carry around and set down DK Barrels to give you more room for error, but this is always impractical and often is impossible, as sequences are often impossible to traverse while carrying an object. (Also note that taking damage while carrying a DK Barrel means you lose a Kong and the DK Barrel is dropped/destroyed without replacing the Kong you just lost) Returning to a DK Barrel to use it later is also usually impractical and sometimes outright impossible, depending on level design.

Taken all together, the Donkey Kong Country player has room for exactly one meaningful error (That is: an error the game actually punishes, as opposed to an error like not jumping quite high enough in a situation where you can just re-do the jump until you make it) at any given moment, as a second error will kick them out of the level. Each DK Barrel the player finds is one extra error over the course of the level that the player may make, though it will never allow them to make two or more errors in close succession -you can't find 3 DK Barrels, hold onto them, and make four mistakes close to the end of a level without dying as a result. The absolute best case scenario is that the player can return to previously discovered DK Barrels after making a mistake, which is not always possible in a level and even when it is possible is usually not reasonable as a course of action. More typically the player can simply hope that they find the DK Barrels with good timing -right after they've made a mistake.

In turn the level designers need to place the DK Barrels in a manner appropriate to this, and indeed design the level around this. Levels must be constructed such that a player new to the level can be expected to make a mistake about once between each DK Barrel, as anything else causes the level to become too punishing, limiting enjoyment and increasing the likelihood that the player simply hurls the controller through a window. (I should hope it's obvious that this is a Bad Outcome)

DKC's design represents a fairly typical design in many games, and I don't simply include action platformers here -for instance, many first-person-shooters (Particularly old-school ones) have a health mechanic that, though more complicated to accurately describe, amounts to the same principle. You have health, it goes down when enemies hurt you, and there are "medkits" that you can pick up to undo the damage, but only up to whatever your maximum health is. This is essentially the same thing as Donkey Kong Country, just with bigger numbers, more variation in the details, and more ambiguity. As such, an FPS using this system also conforms to the same principles of level design -the game must construct itself so that a new player of average competency will expect to take somewhere under their maximum health in damage from any given stretch of the game, with medkits giving the game more room to inflict damage on the player, but never to the point that it is extremely likely or outright certain that a player will lose health equivalent to their maximum in a single fight. (Unless it is expected that they will be able to pick up medkits within the single fight, but the principle still holds, it just makes a precisely accurate description more of a pain to construct))

This contrasts sharply with Yoshi's Island, where a player can commit an infinite number of mistakes in a level as long as the player never loses Baby Mario for too long in a row. The result most people would expect is that this would make Yoshi's Island easier than other games, and this isn't actually wrong, but it overlooks that in turn Yoshi's Island can place enormous demands of skill on the player and still expect even quite bad players to make it through a given level in a single play session eventually -by virtue of actually mastering the skills the level demands of the player, because they have room to make mistakes while trying to come to an understanding of what the level is actually demanding of them and how they can make that happen.

A game like Donkey Kong Country, however good it is, simply does not have that kind of room in its design to place extremely high skill demands on a very consistent basis because the player cannot be reasonably expected to flail their way to the solution before they give up in frustration entirely.

Another point to consider is how exploration-friendly each system is -the DKC-esque model is very exploration unfriendly, as it discourages players from taking unnecessary risks, only taking risks that are either necessary or essentially "free". (Such as when a health pickup is on hand and cannot be taken with them anyway and they cannot come back for it -might as well try then) This isn't necessarily obvious looking at the DKC games themselves, which love to hide secrets and encourage the player to go looking for them, but these games have to employ secondary measures like using bananas as signals that a seemingly dangerous decision (eg 'jump into pit') is actually a safe one. And even then, I speak from experience when I say that a player is still going to often hesitate to actually go looking for a secret when in the middle of a difficult level, for fear that it will cost them success at the level itself.

Yoshi's Island's approach allows players to open themselves up to considerably more danger on the off-chance there's some kind of payoff. Why not go checking out that side passage riddled with obnoxious enemies? So long as there's not any of the major hazards -and it's worth commentary that spikes and lava almost never show up outside of the castle levels- exploring is safe.

For those who might feel like Yoshi's Island is too easy and boring for their palate, have no fear: the game is designed with you in mind, too! It's possible to flail your way to the end of most levels and complete them, but your score at the end of a level is determined in part by your star total ie your health equivalent. While Yoshi will always 'regenerate' up to 10 if Baby Mario is on his saddle, to actually get a perfect score you need to end the level with 30. (Plus have found all the Red Coins and Flowers) If all you want to do is to get to the end? That's fine, flail away. If you want to master the game, though, you'll need to get good at minimizing mistakes. Indeed, the game rewards good scores by unlocking even harder optional levels! Which is a nice reward for someone who is already choosing to seek out a more challenging form of play. It's very clever.

In summary: the greater the punishment for mistakes, the more the game has to work to to make it nearly impossible to make mistakes. Conversely, the more forgiving of mistakes a game is, the more room the game has to place high demands on the player with a reasonable expectation that they can meet those demands.


  1. Did Yoshi's Island actually pioneer the hybrid regenerating health model that's a staple of modern action/adventure type games?

    We had a SNES back in the day, however I only had a chance to play Yoshi's Island when we bought a SNES Classic. By the time it came out, the PS1 was all the rage, and because it used custom hardware on the cart the game never appeared on any of the pirate multicarts that were ubiquitous in my part of the world. I found it very enjoyable and thought that Yoshi's Island alone justified the SNES Classic's price!

    1. That's distinctly possible. I've yet to hear about an earlier game with a similar model, at least, and Yoshi's Island is (justifiably) plenty well-known. I'm personally not strongly familiar with the range of more modern action-adventure games, so I wouldn't know -I'm honestly a bit surprised to hear that hybrid regenerating health has become more ubiquitous- but I find it plausible.

    2. I'm not that up-to-date with games either, and I'm not sure it's as ubiquitous as I made it sound. I think most games still go one way or the other (no regen with healing items vs. pure regeneration). The more well known examples that I've played recently are the 2010s Deus Ex games, Human Revolution and Mankind Divided. It works quite similarly to the Yoshi's Island model; your character regenerates up to a certain amount of health, and there are also healing items which depending on how much health you had either made you heal faster or bumped it above the regen max up to a certain absolute maximum. In-universe this was justified by cybernetic enhancements, and you can get upgraded for faster healing and more health. That said, this really wasn't too important in the modern Deus Ex games because a) your character was always a glass cannon unless you maxed out all the health and armor upgrades, so an additional 20 HP here or a few seconds faster healing there didn't matter that much b) the series is less about open guns-blazing combat and more about having multiple ways of solving (or bypassing) your immediate problem. The developers justified the regen mechanic by saying they didn't want to derail players' experience such that if they ended up with low health the immediate thing to do is to look for healing opportunities.

      The original Deus Ex didn't have the hybrid mechanic, it had the usual food/health packs/health stations thing that contemporary games did. DX1 does have an (optional) regeneration enhancement but it needed to be triggered manually and consumed energy units when activated, so it's not quite the same as the Yoshi's Island model.

      Then there are also games like Halo which has the regenerating shields on top of non-regenerating health, the key difference to the YI style being that it's the regenerating part that (usually) got depleted first.

    3. Ah. I could buy convergent evolution just as readily explaining this, but yeah, I can see the similarities. I swear I'm familiar with some other 2000-onward games that had you regenerate if you dropped below a minimum, like an FPS or something, but it's not coming to me.

      Halo's approach is mechanically similar but solving an entirely different problem -that FPSes struggle a lot with hitscan enemies demanding either stealth-heavy gameplay (Which most FPSes don't want to do at all) or that the player is something of a bullet sponge. A regenerating shield is basically perfect at resolving that issue by making it so getting hit at all isn't player failure, but there's still a clearly-defined 'you are taking too much damage too fast' cut-off point, which health-pack hitscan FPS end up producing anyway, just in a nebulous way a learning player can't predict. (ie if a fight has a full heal placed right after it, losing 99% of your HP in the fight isn't a serious problem, especially if you won't be able to come back for it later. On the other hand, if there's only a half-heal followed by a similarly difficult fight, losing 99% of your health in the first fight is a lethal problem...)

      And as a bonus, saying the shield is an in-universe technological widget helps with immersion; yes, Master Chief *is* supposed to shrug off a certain amount of small-arms fire over and over, in-universe and everything. Where a lot of FPS protagonists are, what, fantasy trolls who can literally regenerate lost limbs in minutes?

    4. Hmmm, as a counterpoint, the Freespace space-combat games come to mind. That series had shields (a standard sci-fi trope, at any rate) _before_ it had hitscan weapons.

      At the beginning of the first game, you don't have any shields. Your weapons are the standard spacefighter lasers-that-behave-like-slow-projectiles, and various types of missiles. There are also slow-moving torpedos for blowing up capital ships. Some way into the campaign, your fighters get upgraded with shields, but capital ships (with one exception) generally still don't have them and simply rely on having tons of HP for durability. This made them sitting ducks, however; the other fighters can chase you around and be shooting at you till you run out of shields, while the capital ships were slow so you can just afterburner out of its weapon range, shut off your engines so you can divert full power to shields and wait for them to come back, rinse and repeat until the target is dead. Kinda made them boring to fight, but it wasn't too important for the first game because technical limitations meant you can't have many large, very detailed ships present in the battlefield.

      When Freespace 2 rolled around, the devs clearly wanted more screen time and direct contribution from the capital ships, so two capship-type weapons were introduced: flak guns, which were area-of-effect weapons for stripping off attackers' shields, and _hitscan_ beam weapons which _ignored_ shields altogether. The shielding system didn't otherwise change, but this made the battles against caps very different. The beam weapons also have the side effect of ragdolling your fighter, which is really annoying but was actually a mercy because that meant getting hit immediately tosses you away from the path of the beam and any follow-up shots. I've played with mods that modified the beam weapon behavior, and without other changes the game gets severely imbalanced; turning off the ragdolling meant that a single hit was pretty much an instakill (because hitscan is undodegeable and staying in the beam's path meant you took instantaneous damage continuously), but making it not bypass shields made them ineffectual.

      The way FS2 dealt with enemy ships having beam weapons is mostly with Command yelling at you to stay outside the range of them when you got anywhere near. That meant wingman management was a huge pain, because they'd always end up inside enemy weapon range and end up however, because turn-of-the-century AI was really stupid and you can only give them very basic commands. Later in the game, you're given some long-range weapons for taking care of point defenses, and they are actually quite overpowered and seem to only be there to get you through the latter missions. Although that might just be the usual video game thing that later missions tend to be much poorer tuned than earlier ones (development schedules and all).

      Tl;dr, in FS shields came first and was so effective that the series introduced a hitscan weapon that bypassed them, effectively flipping Halo's problem _and_ solution!

    5. Not sure I'd call that a counterpoint, but certainly interesting. I'm especially surprised to hear of a game avoiding passing out shields to capital ships -scifi usually runs the other way, where if shields exist and aren't completely universal its some of the fighters (Or all of them) that miss out.

    6. There's some in-game justification about the shielding units not being able to scale to the size of capital ships. However, there is actually one capital ship in the story that canonically has shielding, but for the most part its shields are (literal, justified) plot armor; that particular ship does not meaningfully figure into the gameplay and mostly only shows up in cutscenes or heavily scripted sequences. There are a couple of times you get close enough to take shots at it, and if you do it doesn't produce any shielding visual or sound effects, it's just a regular ship that's flagged as being invincible.

      I suspect not giving large ships shielding had more to do with contemporary technological limitations than anything else. This was back in 1998, where "3D acceleration" was still a buzzword, and it was probably impractical to render the shielding effects for large objects (in a way that didn't look super weird, at least). There are later mods and total conversions that do give caps shields, but obviously they had bigger gameplay and balance tweaks to compensate.

    7. Ah, yeah, that makes sense. It's surprising how often major design decisions get made due to technological limitations on the graphical end, where something gets done because it's new ground being broken and thus shinycool (The Total War series' core design has been heavily defined by the fact that they had a breakthrough that let them make non-flat 3D terrain that didn't look awful), or does not get done because the team couldn't make it happen without being unhappy with how ugly it looked.


Post a Comment

Popular Posts