blubberquark:

When VVVVVV and Super Meat Boy came out in 2010, they were radical departures from established concepts of game difficulty. Games used to have autosave or save slots, but also lives and health, so that when you overwrote your save in a low-health state you could paint yourself into a corner. Other games had regenerating health, or let you restart over and over from the last checkpoint.

Instead, VVVVVV and Super Meat Boy both had a binary state of life or death. Every hazard was lethal. In VVVVVV, the landscape was riddled with autosave checkpoints. Super Meat Boy was difficult, but levels were short. Restarting was instantaneous. There was no “Game Over” screen, no death animation that could not be skipped, just getting back into the action instantly with a single button press. Most importantly, unlike Super Mario World, neither game has a penalty for dying 50 times in a row.

Super Meat Boy could have been different, with lives and mushrooms and coins, and when you are low on of lives, you might be tempted to go back to the equivalent of “Donut Plains” to farm some 1UP mushrooms. But Super Meat Boy was innovating on Super Mario World by being easier in one way and much, much more difficult in another. There is a very different concept of difficulty, failure, and losing.

Instead of being “easier”, these games took the sting out of failure states, and got the player back into the game quickly.

When I am making a game, I try to focus the difficulty into the game’s core gameplay, and in the core loop. If it’s a puzzle game, I want the puzzles to be difficult and engaging, not the stuff around it, and I don’t want the player to be able to grind his way through the meat of the game by spending effort on all the other parts. That means I’d rather make a boss fight just a little more forgiving by default than give the player the option to grind. I’d rather make platforming easier in order to put emphasis on puzzles.

In my experience, about nine times out of ten, when you identify a difficult section in the game during playtesting, it needs to be dealt with for players of all skill levels. The solution is rarely to just give the player “more health” to tank the hits. It’s usually better to more clearly telegraph what is required to overcome the challenge, to put the required tutorialisation earlier in the game, or to re-design the level/boss/timings/attacks for all skill levels. During playtests, I have often observed players struggle and fail at a certain point because they tried the wrong strategy over and over convinced that they could win by executing their flawed strategy perfectly rather than thinking of an alternative.

Some games can’t have difficulty settings. I can think of some persuasive games, for example Depression Quest, Dys4ia, and You Have To Burn The Rope, whose central idea would be undermined by difficulty settings. I can also think of un-gamey interactive experiences such as Proteus, The Stanley Parable, Mountain, or Windosill, where difficulty is just not applicable.

In cases in which difficulty is applicable, there is rarely only one way to implement it. The Curse of Monkey Island had a “regular” mode and a “harder mode” with more puzzles, but no “easy mode”. The different difficulty settings in Mobility change the game’s goals and platforming mechanics. The difficulty settings in VVVVVV slow down the game by up to 40%. You could give characters more health, more lives, drop more loot, change the sizes of hitboxes, remove obstacles from levels, let the player jump farther, or make the enemy AI stupid.

More drastic interventions like increased jump distance can turn easy mode in a completely different game, such that getting better at easy mode won’t help with normal mode. With increased jump distance in a platformer, solutions that work in hard mode might not even work in easy mode either.

Other games have a different focus. Nuclear Throne is a coffebreak action roguelike-like about dying and retrying. If there was an “easy mode” and a save/reload function, players might be tempted to crank the difficulty so low they can beat the game in one run. This is clearly not what the designers intended. Vlambeer’s next game Luftrausers was even more explicitly focused on dying and retrying.

Reasoning like “so that players of all skill levels can complete the game like the designer intended“ implies that the designer intended the player to beat the game in the first place. Saying “all games should have difficulty settings“ is an expression of a certain expectation of what games “should“ be, and definitely incompatible with moral objections to the term “walking simulator“.

Ending a run of Papers, Please in starvation and poverty is just as “valid“ as doing your duty for Arstozka, just as valid as getting caught helping dissidents, just as valid as buying fake documents and fleeing. Still, it’s conceivable that such a game could have difficulty settings - but it doesn’t need to.

1. Balance your game so that the difficulty lies in the core gameplay, not in minigames, movement, or ancillary mechanics. Limit the downsides of failing minigames.
   
2. Re-work situations that are too difficult, rather than giving the player “more health“. Make it easier to dodge or seek cover rather than tank hits. Make sure the player knows which parts are supposed to be difficult, and what is impossible. Tell the player the information necessary to beat the next challenge - unless the challenge is a riddle. These changes often solve the same problem that difficulty levels would solve. Corollary: Remove grinding and resource hoarding when possible.
3. If applicable, rethink your failure states, implied failure states, save system, penalties for losing, ease of restarting, distance from save points, and permanent progression.

When your players lose too often, make losing fun!