Adaptive Difficulty

[Paul Jeffries]

I really don't see anything that adaptive difficulty would give you that simply providing different difficulty modes wouldn't do a lot better.  If people just want to play for the experience then let them play on easy mode, if they are there for the challenge let them play on hard.  Why have the game constantly try to second-guess what the player wants when you could simply have them tell you?

Lots of games have treasure rooms with a special challenge inside or gollems or skeletons which are frozen but only attack if you mess with them, or tombstones... There's usually something you, as a player, can do to make the game more difficult if you want to.

While these can offer additional challenge, they also tend to reward you quite well with items that can improve your odds of winning overall, so I think usually it's more of a strategic short-term-risk for potential long-term-benefit kind of decision rather than simply an elective difficulty thing.

[Nachtfischer]

I really don't see anything that adaptive difficulty would give you that simply providing different difficulty modes wouldn't do a lot better.

Well, it takes the "designer job" of picking the right difficulty away from the player. Anyone interested in how to do adaptive difficulty right should look into Auro (available for iOS and Android). It effectively features a "single-player matchmaking system" that lets you rank up (and down) between matches to keep the challenge level optimal (just as flow theory postulates). It's a great approach that gets rid of a lot of the typical roguelike problems, e.g. them being about "highscores" (and therefore runs taking longer and longer, results becoming more and more meaningless etc.) or you just never winning (and therefore missing out on an important side of the feedback).

[Vanguard]

Scoring systems and bad players never winning are both good things

[Paul Jeffries]

I really don't see anything that adaptive difficulty would give you that simply providing different difficulty modes wouldn't do a lot better.

Well, it takes the "designer job" of picking the right difficulty away from the player.

No, it does the exact opposite - it takes the 'designer job' of picking the right difficulty away from the designer and gives it to the player, but in an extremely obtuse and indirect way that means that it may not give the player what they want anyway.

I think declaring it a 'designer job' is a little arbitrary in the first place.  If I'm in a restaurant and the waiter asks how I want my stake cooked, I don't rebuke him with "that's a 'chef job'!".  I might get a bit cross, however, were he to hover around watching me eat, then decide that because I'm eating slowly it's obviously because it's too rare for me and grab it off my plate without asking to stick it back under the grill a bit longer.  That's basically how I see the difference between difficulty selection and adaptive difficulty.

Anyone interested in how to do adaptive difficulty right should look into Auro (available for iOS and Android). It effectively features a "single-player matchmaking system" that lets you rank up (and down) between matches to keep the challenge level optimal (just as flow theory postulates). It's a great approach that gets rid of a lot of the typical roguelike problems, e.g. them being about "highscores" (and therefore runs taking longer and longer, results becoming more and more meaningless etc.) or you just never winning (and therefore missing out on an important side of the feedback).

I haven't played Auro, so I can't speak with any authority and may well be wrong, but the way you describe it this sounds more like a progression system that allows incremental loss rather than the kind of adaptive difficulty that the OP was talking about.  Presumably going up in rank is a (meta?)goal of the game, so going down by one actually makes this more difficult (in the same way dying and starting the level/game again would).  It does sound like a good system, but I'm not sure it's what we're talking about.

[akeley]

When I hear the term "adaptive difficulty" all my alarm bells go off ringing and most likely I will steer clear from a game that employs it. For me, it`s a concept that somehow meddles with the integrity of the "gaming" itself and shouldn`t be used at all, especially in roguelikes. What is the point of a game that will roll over and try to play nice because I died 5 times? Where`s the challenge in that?

It seems to stem form the modern sensibility that all games should be for all people, nice and comfortable to play - (which in commercial gaming is of course $-driven) - something I disagree with.

I`m not even too keen on "difficulty levels" in RLs either - while these make a bit more sense in twitch games, I don`t think they should be used in strategy titles. Every time I encounter this choice I `m quite confused as to which one I should pick, since there`s usually no description explaining the mechanics in difficulty spike, just some vague "You`ll die!" flava. 

I don`t want to have to make this choice - I`d like to leave it to the designer to design his game, without trying to please everyone and staying honest to his vision and rules. And if I find it "too easy" - or "too hard" - I`ll try and find a title that`s  more suited to my mood or current experience level.

[Omnivore]

Most game designers are not psychics.*  Offering different levels of challenges makes your game available to a larger number of players across a spectrum of aptitude, experience, and even mood.  Many games that do not have an explicit difficulty selection option instead have different difficulty protagonist and antagonist choices.  This is especially true in strategic and tactical war games for instance where you can choose your side and a scenario.  It is also true of most CRPGs, there are easy classes and hard classes  even in old style real roguelikes.

Adaptive difficulty that makes the game harder is quite interesting.  Making it easier seems counterproductive, outside of explicit player choice via selecting easy difficulty mode, I expect a game to get progressively harder.  End game and late game play can easily become boring if the game doesn't become more difficult.  The trouble is, again most designers are not psychics.  It is a complete crapshoot as to whether or not the designer's initial choice in rate of progression is satisfactory. 

As to meta-gaming an adaptive difficulty implementation.. well if that floats your boat have fun.   Personally I don't see the point for stand alone crpgs or turn based tactical where you are only competing ultimately against yourself.  However, games are all about wasting time in amusing ways, so if it amuses someone, great!

As far as the designer having some holy vision and set in stone rule set, that's a recipe for failure.  A vision is a plan, great to have one, needed even, but as in the nature of all plans it will not survive contact with the enemy unscathed.  A vision and the ruleset which bounds its expression, must both be adaptable to the realities of the presentation environment and the audience if it is to have any lasting presence.  Adaptive difficulty seems to be a way to help automate that necessary adaption for the designer and let him or her get on with adding more depth or creating a new experience.

*Can't prove that none of them are, there are a few that may be, then again perhaps its another type of psych behavior: psychotic.

[akeley]

By "adaptive difficulty" I understand the process where game`s engine analyses your actions and adjusts parameters towards "easier" or "harder" on the fly - and by difficulty modes the ol`Beginner-Nightmare! selection at the start of the game, where parameters are changed according to choice and set for the duration of the game.

These definitions are what my opinion is tied to. I don`t consider classes to be deliberate difficulty changers, for me the easier/harder gameplay is just a function of of playstyle tied to one or another. Same goes for "sides" or "campaigns" in strategy games. Besides this is much more arbitrary - for one person it could be easier to play as a rogue, for another as a berserker - while base game rules stay the same. Of course, games have to get harder - or provide enough challenge - as your character gets stronger, but again, this has nothing to do with my - and I think OP`s too - definition of adaptive difficulty.

I`m not sure designer`s vision needs to be "holy" - for me "consistent" will do nicely.  But okay, I see where this went...of course I never argued that somebody should rigidly stick to what his initial design document contains, although I`m pretty sure in some point you have to make some concrete decisions otherwise your game will dwell in dev hell. As for the "competing against yourself" remark, I have no clue what does it allude to, so will pass on this one.

I`m just against hard coding adaptive difficulty mechanisms into a game, that`s all. Because doing so is in my opinion the true "recipe for failure". Perhaps supported by the fact that very few - if any - of the roguelikes I played use it, and yet many of them are considered classics with quite an audience and long lasting presence.

On the other hand, like I said in my first post, game that uses it risks stopping to offer any sort of real challenge and becoming a strange neutered monster that basically pleases by letting everyone win. It`s already a trend in big commercial games - "Skip the encounter?" pop up and such - and one of the reasons I avoid them.

[Tilded]

If you want variable difficulty without an overt difficulty selection, you could perhaps incorporate the difficulty selection into the game. For example, in a dungeon crawler, you could have several branches of the dungeon with different requirements for entry. Maybe each branch has a different guardian/boss, and the player chooses which boss, and this which difficulty, to face depending on their health.

[wire_hall_medic]

Scoring systems and bad players never winning are both good things

Word.

As far as an internal adaptive difficulty, I argue against it.  For me, permadeath adds tremendously to the fun of the game because it makes the experience more valuable; I can lose it.

If you have a need to put in such a system, logical within the world.  Maybe the potion merchant gives you a discount, or NPCs have less respect for you.  You know what'd be hysterical?  If you'd made so many trips into the dungeon that the villagers hired someone else, and now there's someone tagging along . . . splitting your experience and loot.

I really like Slash's Classical Roguelike definition, but if you feel your game will be more fun (and fun should be the ultimate goal of any game) by allowing them to to succeed despite not learning from their mistakes, go ahead.  Just don't let them believe they've mastered the game; give them save points