Difficulty and Punishment


Everyone likes a different challenge. Some people like to play an FPS and die at least 20 times on the same level. Some people play that same FPS and get frustrated when they die twice. Compounding this problem, people have drastically different skill levels. There are people who will breeze through a Halo game on Legendary and people who will die a thousand times on Easy. What’s more, it’s usually the most hardcore gamers who are the best, and yet also want the most challenge, while people still learning the controls, or who don’t play video games a lot often want to be challenged less. What this means is that to adequately challenge both of these players we need massively different difficulty levels.

A big part of my disagreement with people arguing against objective review of games is that I think that you actually can review games 95% objectively, whilst also going in depth. Difficulty would seem to contradict that, because different people have such different requirements, that two people can have massively differing views on the appropriateness of the difficulty and both be telling the truth from their standpoint. The best example is the Souls series, which have one extremely hard difficulty level (although it’s hard mostly due to poor resumeability which I’ll get to later). For people with the right blend of skill and desire for difficulty, it’s a perfect wonderful game. For people who don’t have the right mix of skill and desire for difficulty, it’s not a fun game at all. It’s very easy to cop out here and argue that this proves that it’s all just opinion and there is no such thing as a universally loved difficulty level.

Which is a total strawman. A game with great difficulty is a game that has a “just right” level of difficulty for a maximum amount of players. It’s why a game that has three difficulty levels that different people play at will always be better than that same game with just one difficulty level. Of course, that wouldn’t affect the players who play at the difficulty level that we keep, but it would negatively affect every other player, either by making them play a game that is too hard or too easy. Now some people would point out that we can’t say that the game is better, because it’s only better for some people, and some people are totally unaffected. While that’s possibly true from a critical perspective, it’s definitely not true from a game development perspective. Besides, if, as a critic, your argument is that a change in a games design that makes a massive positive difference in a huge amount of people who play that game doesn’t actually make that game objectively better, then who do you think your criticisms could possibly help?

But we can actually do better than having discrete difficulty levels. Having 3 or 4 difficulty levels just means that, assuming they’re well designed, you have 3 or 4 difficulty levels that come pretty close to truly satisfying a lot of different people. There is a huge risk though in having the highest difficulty be too easy for some players, and having the lowest difficulty be too hard for some players. What that means is that we either need to miss some players at the high and low end, or have massive gaps between our difficulty levels.

Imagine a FPS game where the hardest difficulty, called Insane, has the player die in one hit, and the easiest difficulty, called Cupcake, has the enemies do such little damage that, when combined with constantly recharging health, the player can just kill one or two enemies from each group before they can’t mathematically kill him anymore. In such a game we have 1% of players who love the Insane difficulty, and 1% of players who love the Cupcake difficulty. That means that, since we only have 4 difficulty levels total, we really only have 2 difficulty levels for 98% of our players, which seems hopelessly inadequate. This is probably better than having Insane be too easy for the best players, and Cupcake being too hard for the worst players, but it would be much better if we could more finely tune the difficulty to suit each individual player.

The stopgap solution is just to have an extra difficulty mode. So we have Cupcake, Easy, Medium, Hard, Insane as our difficulty levels, and this will be better than without Medium. I want to stress that this may in fact be good enough to satisfy 99% of players 99% of the time, and difficulty is just one part of the game. However, if we want to knock it out of the park then we need less discrete difficulty levels. Actually, having pre-set difficulty levels for gamers to quickly pick from is a great idea, just like pre-set graphical quality options are a great idea. What we ought to do is allow the player to individually adjust certain parameters as they wish. These parameters will vary from game to game, but some that are probably relevant for most games are “Damage by Enemies”, “Amount of Health”, “Damage by Player”, with others being game specific such as “Average amount of enemies per encounter”, or “Speed of Sniper chargeup”.

This can be criticized by pointing out that players have to tinker with the game in order to get exactly the difficulty they want. I would agree, but also point out that, provided there is still pre-set difficulty, that players don’t have to do any of this unless they feel like it. There is a better criticism though, which is that the game itself ought to be adjusting itself on the fly to provide the right difficulty to the player. The reason for this is very simple, if the player knows that Easy is too easy, but Medium is too hard, then they can tinker with the difficulty, but what they produce, although probably being better for them then the presets, might still be significantly off. What’s more, they will only learn this after playing for a while and testing it all out.

With that in mind it’s easy to argue that we should go for on-the-fly changing of difficulty, but there are three problems with this. The first is that every player has both different levels of skill, but also different desire for difficulty. A simple algorithm that notices that the player has died 5 times in a room and slowly starts to lower the difficulty might be too slow for one player who has already quit, but far too quick for one player who’s loving the challenge. Likewise, players don’t always play to win, and someone might just be screwing around having fun dying time and time again in a room, only to be disappointed when the game starts spawning less enemies. Or maybe someone isn’t very good at the game but goes through a small part of a level playing extremely well, and so the game turns up the difficulty higher and the player starts dying over and over again for a while.

We could solve that by asking players how challenging they want the game to be for them, and then balancing the game around their skill level. That has the problem that we don’t know what their skill level is to start the game, and thus we don’t really know what difficulty to start the game at, so we’ll probably get it wrong. Additionally, if we’re giving them the Easy, Medium, Hard, Very Hard type difficulty selection, only for challenge rather then pre-set difficulty, then we have all the same problems as before, Easy being too hard, player wants Medium-Hard instead of either, Very Hard too easy, etcetera. This can be somewhat solved by having difficulty on a slider, which TES:Oblivion does fairly well, so I guess it’s a minor negative.

The second problem with having the difficulty being adjusted on the fly is simply that there is something satisfying as a player about overcoming a set difficulty. I think some of my favourite gaming moments are starting out being killed a lot at a certain level of difficulty, before eventually having that become too easy, and going up to the next level of difficulty. If the game were to constantly be changing the real difficulty behind the scenes I wouldn’t have such a sense of accomplishment. In other words, it’s kind of fun to have the difficulty of something organically become easier as we do it, as opposed to stay the same relative to us.

The third problem is that I just really don’t trust any algorithm to do a good job of adapting difficulty to the precision necessary to be truly better than just giving the player a bunch of sliders to tweak along with some pre-set difficulties. Even if it was possible, I don’t think it would be nearly better enough to justify the time that would go into doing that.

Long story short, a games should have various levels of difficulty. The easiest difficulty and the hardest difficulty should be too easy and too hard for 99.9% of people, or at least the people who actually play your game. Additionally, the closer the game can come to any individuals preference for difficulty the better. Practically that means that games need to offer pre-set difficulty levels covering a wide range as well as at least one slider that players can adjust to tailor the game’s difficulty to them.


There is another complicating factor that people lump in with difficulty which is actually not difficulty, or at least not directly. When the player succeeds in a game, they move on. If it’s a linear game they move on to the next part of the game and if it’s a more open game then they move on to whatever they feel like. If the player fails then typically they are set back some ways in the game. The amount that a player is set back by can be thought of as the punishment for failure, and varies from game to game.

Most linear games just have the player re-do the section of the game that they failed. Some games, usually open-world games, get more creative. World of Warcraft has the player weakened until they can find their body at which point they are back to normal. Bioshock, not quite an open world game, had the player revived at a chamber someways back, while keeping the world the same. I’m sure there are some games where death is totally impossible and you just lose a bit of score when you die/fail in some way. Commenting on the specific ways that games punish their players for failure, while interesting, also doesn’t really concern me right now. The severity with which games punish players for failing is what concerns me.

Punishment for failure sounds really bad but actually it is a necessary part of the game. If you design a game where no matter what the player does the same winning outcome occurs then you haven’t designed a game. The closest I can think of is a game with “god mode” enabled, or invincibility. If the player can’t be killed then they can fail constantly by getting hit by all the fire in the world and they’ll still live. I’m far from condemning this, I think that games should give players the tools to break them much more often then they do. That being said, if we want to actually create a video game, unless we’re doing something unlike 99.9% of video games we have failure modes. This could be dying in just about any game, not finishing a lap fast enough in a racing game, playing a song so poorly you get booed of stage, and lots of others. When the player fails there must be some amount of punishment here in terms of something, or else they have no reason to win.

This is totally seperate from difficulty. Imagine that game from earlier on Insane mode, versus on Cupcake mode. We could have the game on Insane mode with auto-saving every 10 seconds, ensuring that players never lose more than 10 seconds of their time. We could also have the Cupcake mode have no saving, so that if the player dies they have to start over from the beginning. Is it really accurate to say that these two modes are of the same difficulty? I’d argue that the Insane mode is much more difficult, it simply has drastically less punishment for death than Cupcake mode.

I mentioned earlier that the Souls games aren’t actually very hard for people who play games a lot, they just punish the player massively for dying. I recently played Tom Francis’ Gunpoint, which autosaves every 3, and 7 seconds and keeps multiple saves from which you can pick from when you die. I’d say that difficulty wise the two games are the same, but Gunpoint doesn’t waste nearly as much of the players time when they die, nor does it weaken the player at all.

I said previously that people have different skill levels as well as different desire for difficulty, and that those two things together meant that a game ought to offer a wide variety of difficulty. People also have a differing amount of desire for how much they should be punished for failure in a video game. Therefore a game that offers different levels of punishment for failure is simply better than a game that only offers one level of punishment for failure, as that level is probably going to be too high or too low for most people.

The solution is quite simple, just give players more options. I hate the current developer-placed checkpoint system that is in most linear games nowadays. I designed an open world RPG that had a system where the player could pick from “no saves allowed” (therefore perma-death) to “auto-saves every 10 seconds” or “only saves allowed in a safe place, not the wilderness”, or any combination that actually made sense. Because why make a game any more or less punishing than some people want when you don’t need to?


Make your game have varying, user selected, levels of difficulty, and make your game have varying, user selected, levels of punishment for failure. If you do that your game will be objectively better than it was with a single difficulty and punishment for failure, ignore anyone who disagrees.

This entry was posted in Game Design and tagged , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s