Tales of the Rampant Coyote

Adventures in Indie Gaming!

RPG Design: Boss Battles and the Positive Feedback Problem

Posted by Rampant Coyote on August 14, 2012

In my article last week about Frayed Knights‘ problem with boss battles, I neglected to mention one other factor that worked against my poor bosses which is peculiar to the RPG genre: One of my guiding principles of the overall game design is that the game should be fully playable and of reasonable difficulty for a player who chose “sub-optimal” progression for the party. In other words, a player should not be penalized for experimenting with character advancement and making some “poor” choices.

I stand by this choice, and unless something convinces me otherwise, I expect to make this approach with any and all RPGs I make in the future. The whole point of creating a rich character design / progression system is to allow players to experiment and customize characters as they would like to imagine their character, not as the game dictates.  More importantly, I would never want to create a situation where the player finds the end-game unwinnable (one way or another) because of choices made in ignorance they made earlier in the game. Should there be consequences for choices? Sure. But that’s not the same as penalization.

The problem with even a middle-of-the-road approach like this – where the game is balanced for middle-of-the-road ‘optimization’ – is a positive feedback loop. Yes, “positive” sounds like a good thing, but in games, it’s often a bad thing.  I’ve talked about this before, in fact exactly three years ago today. Must be the season or something. But I thought I’d revisit the topic with a bit more concrete examples, and relate it to boss design.

With a positive feedback loop, good players get an easier game, and poor players get a harder game. The “rich get richer, poor get poorer.” In coin-op arcade games, the positive feedback loop was  advantageous to getting players to pump in the quarters, as it led to a speedier conclusion to the game – and gave less-skilled observers some insight into the amazing things the expert players could do as long as they could “ride the wave” of their own success.

And then there are the boss battles. In general, boss battles are intended to test a player’s mastery of in-game skills. They should represent something of a break from the gameplay routine – an interesting challenge. Although, really, in an RPG as much as any other game, I don’t think you should have any uninteresting challenges. But anyway… a boss shouldn’t be just a “bigger, badder” opponent.  They should be a special, dramatic challenge for the player.

The problem – if you’d call it that – is that there is an additional layer between the player and the game, and that is his character (or characters). In a traditional game, the game can challenge the player directly – testing his precision and timing with jumping, or with jumping and shooting combined in an epic boss fight.  The potential capabilities of the player’s avatar are known entities at design time, and challenges can be designed to exactly test the player’s mastery of control necessary to get within a threshold of that potential.

But how do you do that in an RPG, when the player has invested their points into their characters ‘leet’ bread-baking skills?  Do the bosses now challenge the player to a bread-baking contest instead?

This runs into another contradiction of the purpose of a boss battle – to break up the routine. If the player has gotten to the point where he can win the game by simple, skillful application of bread-baking mastery, the player should not be able to simply bake his way out of  a boss fight! No, at the very least, he should be tested in the application of his skills in a cookie-baking contest, just to shake things up and get the player out of his comfort zone, right?

But what designer is going to create customized boss encounters for every area of character expertise – and non-expertise?

My poster-child for this problem is Vampire the Masquerade: Bloodlines, a game that’s failures were made so painful only because it tried so hard to push boundaries. So many encounters in the game could be resolved through means other than straight-up combat. They worked very hard to make the game very playable for a character that relied upon stealth or subtlety. But then – especially near the end-game – they dropped you into an arena with a straight-up combat against a powerful boss opponent, and if your stealthy or subtle character had neglected their combat skills – something remarkably easy to do to this point – then you, the player, were in for a very rough time.  There was no place to hide, and nobody to talk to. Just guns and fangs.  A conflict designed to challenge combat god builds was multiplied by three.

That’s analogous to having an entirely conversation-driven boss encounter, a la Vizzini’s “battle of wits” from The Princess Bride, and if your characters Wits and Conversation skills aren’t high enough, you’ll automatically die at the end and lose the game. What’s that really testing? Whether or not the player knew the designer’s intentions and demands in advance? Players would cry foul at this – and justifiably so – yet we let the combat bias slide all the time. We’re used to it, I guess.

There’s no easy answer to this. “Middle of the Road” bosses are certainly an easy approach – challenging an “average” player with an “average” build – but not a particularly satisfying encounter, and subject to the positive feedback problem. What are some other approaches?

1. Constrain player customization.  Class-based systems work well here, as the designer has limited the variations they must design for to a few well-defined options.

2. Force the player-characters to maintain a minimum level of combat skill. Level-based systems help here, too – regardless of how you’ve customized your character, your attack, defense, and hit points have still gone up a minimal amount. In this case, an overwhelming challenge may be made easier simply by delaying the encounter for a couple of levels.

3. Make the “optimal” tests optional. Here’s where you put the Vizzini and Master Baker and Combat God boss fights – somewhere off the main quest line of the game.

4. Provide alternative methods of winning encounters. You can fight the demon-lord head-on, sneak around him, or talk his minions into betraying him.

5. Provide alternative boss encounters: This is an expensive approach, but not unheard of.  Different “boss encounters” (or major challenges) may be chosen by the player, which test different character abilities.

6. Allow players to use custom skills to weaken the boss encounter: The “raw” encounter may be a great challenge for an optimal combat-build character, but the player can use other skills (in advance?) balance the encounter. This is really, IMO, the best approach, as it allows the player to “test” his skills at using the character’s other abilities.

7. Auto-Scaling the encounter: Yes, Oblivion handled auto-scaling poorly, giving it a bad rap. But IMO, this is potentially no different from a live “Game Master” building an encounter around his player’s abilities. It just needs to be handled much, much better, so it feels like less of a cheat.

8. Bosses should be able to counter the most-optimal strategies: This is generally good boss design anyway. I don’t think bosses should negate a player character’s abilities – this is simply the reverse of the original problem.  But a little bit of ‘negative feedback,’ not to penalize optimal character builds but to reduce their advantage – could make for more interesting encounters.

9. Fully Organic Bosses: There is no set “strategy” for defeating the bosses, but simply an interesting encounter and environment where the player is free to use any skill or strategy at their disposal to resolve it. I love this idea in theory, and lean this way in my own design, but in practice it can result in boss encounters feeling no different from any other.


While I didn’t always apply things perfectly in Frayed Knights: The Skull of S’makh-Daon, it tended towards the more old-school RPG approach of having the “bosses” be somewhat understated and less of an “event” anyway. My real desire is for all combat encounters to be interesting and present their own challenges, and I’m not completely abandoning that approach in the sequel, either.  But there will be a few “boss encounters” that will be “punched up” a bit more.

Filed Under: Design - Comments: 9 Comments to Read

  • Andy_Panthro said,

    This is generally the reason I don’t like boss-fights, because they are generally done so poorly. Do we even need boss-fights at all? In a more action-focused game, like Diablo, a boss is an interesting challenge (even if that challenge tends to become “how many potions can I drink per second”), but for other types of games it can seem really out of place (like Deus Ex: Human Revolution).

    Your Bloodlines example caught my eye in particular, since my only attempt to play that game came to an abrupt halt when I found myself in unwinnable combat areas after I had made a non-combat character.

    My final thought is one that I have considered for a long time: Do we rely on combat too much in RPGs? I’d suggest most of my favourite CRPG moments were non-combat ones.

  • Corwin said,

    There are 2 types of boss fights I detest; the one where the boss morphs into something even more powerful after you first defeat him while you are drained and badly wounded, and the one where you as a mage build have to fight in a suddenly ‘non-magic’ area where none of your spells work- Risen springs to mind here.

  • SniperHF said,

    One thing that always irked me with boss battles is when they don’t play by the same rules. If we’re talking CRPGs usually there is some sort of a rule set at play. But if the designer breaks all the game rules to make a challenging encounter I feel cheated.

    I do make exceptions to this though for story reasons. If there is a valid story reason a character in an AD&D game has a super spell no one else has that’s okay. But it’s not okay to give a level 10 boss level 12 spells just for balancing.

    As for your ideas:

    I highly agree with number 3,4,5,6 and 9 in almost any game.

    #2 could work in a combat oriented game and definitely in a single character game.

    #7 is interesting because I think it can be okay to some extent on normal encounters. But for bosses I much prefer they serve as benchmarks as to your characters progress. Thus I do not want them scaled.

    Not quite sure I understand #8. Can you give an example?

  • Rampant Coyote said,

    #8 would be very dependent upon the game rules. But we could say that if you min-maxed your character and really specialized, you could give him the nasty “Piercing Strike” skill by level 8, which is devastatingly powerful against single opponents.

    Well, at around level 8 you are supposed to encounter a lone boss who his balanced to be plenty tough for average player characters… but not for a character who is min-maxed with Piercing Strike, which might otherwise render him a trivial opponent.

    A bad thing to do would be to make him completely immune to Piercing Strike. That basically negates the advantage that the player sacrificed some “well-roundedness” to achieve. But maybe he does have a defense against the ability (“Flesh of Stone”) which reduces its effectiveness. So the highly-optimized character still enjoys some advantage over the boss, but not enough to end the fight in ten seconds.

  • Mike said,

    Perhaps the simplest solution is to separate combat skills and social skills (like crafting and persuasion). For example, when you level up, you get one point to spend on combat skills and one point to spend on social skills. This ensures that players are never too underpowered or overpowered for combat and that they will have the skills to explore the social aspects of the game as well.

    Ideally, games would have the full flexibility to avoid combat using social skills like the first half of Bloodlines, but the history of RPGs and the second half of Bloodlines itself seem to indicate that this is very hard to achieve. For most RPGs, the combat and social parts of the game like getting and resolving quests, crafting, and buying and selling goods are fairly independent, so separating the skills probably makes sense.

  • Rampant Coyote said,

    @Mike – Yeah, great suggestion. That would be another great implementation of suggestion 2.

  • Links for August 15, 2012 | Andrzej's Links said,

    […] RPG Design: Boss Battles and the Positive Feedback Problem […]

  • Galen said,

    That’s good analysis of the issue(s) (actually this is the first time I’ve been to your site but thus far all the blog entries have been really solid!).

    I’m not sure there’s a perfect solution to this problem, and this is by no means contained to game design! Take the original star wars movies (disclaimer: I’m not implying that George Lucas is a master story teller… but in the first series he at least had to live with editors 🙂 ), There was no believable way for young Luke to succeed against “the boss’ boss” (note: I’m also in the camp of thinking that if Vader could have simply tossed the old man when he was distracted, he’d have done it long before).

    Adding any kind of player mechanic that allows for specialization outside the “main” mechanics required to win consistently throughout the game, sets a player (or group of players) up for serious disappointment, and frankly many of the alternative approaches open up the “cheating” or weak-ending cans of worms.

    The best example of balancing I can think of, and admittedly this is still a fiction author, is R.A. Salvatore’s Dark Elf series. Ultra powerful characters, yes, but still mortal. Even more ultra powerful enemies, many immortal or danged near. But the solution was multi-character, no one character, or necessarily the “good guys” alone could account for their enemies, but combinations of characters achieved balance… I think that series is well written 🙂 Now, how to create that same model in a game.

    Some of it in my mind is believability. In a dungeon full of ever increasingly powerful enemies a couple things spring to mind that lead to maintaining immersion:

    1) Can any one creature ever maintain control over the current baddies group (barring gods or near gods)? This comes out of my “who turned on the lights?” theory of game design, namely, in a room/castle/dungeon/whatever full of zombies, vampires, and other undead, who the hell keeps lighting those torches? Thankfully, that and fast moving undead rotting corpses have been largely overlooked by the gaming public.

    But none-the-less, the boss spanking issue comes back to the same idea, if you in real life (ok, let’s role play a little here) actually grinded your way through all that carnage, why would you have any expectation that IF something was in-fact in complete control of that nightmare you just waded through that you’d survive? (let’s again overlook that a real “boss” is unlikely to wait until you’ve completely wiped out their army before coming out to play – I blame Tolkien for that story design approach… well, I have to blame someone 🙂 )

    2) If my “non-optimal” skills got me here, I must be exactly like the creatures I got past (and destined to be controlled by the current “boss”) or … why doesn’t my approach work now with the boss?

    There should be at least a lot of “mini-bosses”, where either I am encouraged (as in unable to progress) to build the necessary survival skills or slide down the food chain.

    Just some of my (rambling) thoughts.


  • GhanBuriGhan said,

    Late to the party but I wanted to comment anyway: I usually hate bosses in the classical sense in RPGs – the end boss in Risen was a vile example how grating that can be. I think in RPGs one should rather think of “major challenges” – these could be an unusual dungeon, a major battle (or series of battles), an incarceration and escape sequence, ar an encounter with an unusually strong foe (OK, yes, that’s a boss – but the emphasis here is on encounter, encompassing more than just battling the boss).
    Rather than design an unusual battle, try to design an unusual situation.