Making a Puzzle Boss

The steps for building a puzzle boss are pretty much the same as building any kind of encounter, or any kind of story, ever. That is:

  • Goal: The players want something
  • Conflict: The DM puts an obstacle in their way
  • Resolution Mechanic: The players and/or characters overcome the obstacle (or fail)

In most combat encounters this is incredibly simple:

  • Goal: The players want to go to a place or get a thing
  • Conflict: Monsters want to stop the players from going to the place or getting the thing
  • Resolution Mechanic: The party uses violence until there are no more monsters

The party could use diplomacy or stealth or magic or whatever tools they want, but the general structure of an encounter is the same. An encounter even uses the same steps from the perspective of the DM, but that’s another post.

What’s important about this structure is that it’s nested. As you drill down into it, you’ll find the same steps repeated over and over until we get down to a game mechanic. In D&D, that’s usually a d20 roll:

  • Goal: The players want to kill the orcs in the forest
  • Conflict: The orcs do not want to die
  • Resolution Mechanic: Combat
    • Goal: The ranger wants to use her bow
    • Conflict: Bow attacks provoke attacks of opportunity
    • Resolution Mechanic: The ranger goes to a place where the orcs cannot reach her
      • Goal: The ranger wants the high ground
      • Conflict: The high ground is very high
      • Resolution Mechanic: Climb check vs. DC
    • Goal: The fighter wants to hit an orc with his sword
    • Conflict: The orc does not want to be hit
    • Resolution Mechanic: Attack roll vs. AC

Most conflicts have fairly straightforward goals, conflicts and resolutions, and that’s important. The key to a puzzle boss is that it doesn’t:

  • Goal: The fighter wants to hit the orc with a sword
  • Conflict: The orc has a magic barrier around him
  • Resolution Mechanic: Disable the barrier
    • Goal: The party must stop the barrier
    • Conflict: The party does not know how
    • Resolution Mechanic: The party makes skill checks, or tests theories, or thinks really hard, etc.

The puzzle is in figuring out how to resolve the conflict. It’s not a matter of what the best answer is (for example, swords versus clubs versus fire) but how to answer it at all. The simplest way to do that is to make direct, obvious combat either impossible or just a very bad idea, but there are several ways to go about it.

Puzzles are too broad to have a clear set of creation steps. They’re not like a monster or a trap where you can say “this type of puzzle is this difficult, and it takes this long, so it has this Challenge Rating”. It’s more of an art, where you have to balance the combat, the story, the players, the schedule, and the mood at the table as it shifts. But like art, while there aren’t rules there are strongly-worded suggestions:

The party has to discover the puzzle quickly. Combat has certain expectations in D&D, and one of those is that it can be resolved via combat mechanics. Breaking from that is fine, but breaking from it in secret and expecting the players to read your mind is not. You don’t want your players to miss the runes on the floor and spend four rounds trying to hit the invincible necromancer with a sword. They’ll rightly consider it a waste of character resources and real-world time.

The answer can’t be too obvious. Things like “kill the healer first, or he’ll heal his allies”, or “the wizard has terrain advantage” aren’t puzzles, they’re complications. They’re a good way to make fights more interesting and dynamic, but that’s not what we’re talking about here. You have to escalate the difficulty until the players don’t have a evident answer. Consider “kill the healer first, or he’ll heal his allies, but everybody is acting similar so we need to find out who the healer is”, or “the wizard has terrain advantage because she’s on a ledge with no visible access point.”

There should be more than one answer. If the puzzle relies on an arcane leap of logic or an absurdly specific action, there’s too strong a chance the players won’t solve it at all. Continuing the above examples, if the party can only find out who the healer is by remembering that all clerics of the healing god have a small brand on the back of their neck, or if they can only make it to the wizard by throwing paint at the wall to reveal the hidden ladder to the ledge, that’s a bad puzzle. The key word is “only”. Both of those are fine answers, but when you reject other plausible answers because they’re not the one you had in mind, the puzzle you’re actually giving the players is “guess what the DM is thinking”.

You need to accept that the players are going to try something else, like using detect magic to find the telltale signs of a healing or illusion spell. Reward them for ideas that fit with the puzzle you’ve presented. I usually require that an answer have at least two of the following traits: it’s clever, and it shows the players are really thinking about their situation; it’s entertaining, and it gives us joy to see the characters try it; and it’s successful, because the players rolled high or set things up well enough that a roll doesn’t matter. A few examples:

  • A player suspects the enemies all worship the same god, but asks if their holy symbols are the same. With a good Perception check she notices (or declares) that they’re subtly different, and asks to use Knowledge (religion) to recall whether those symbols indicate different religious disciplines. This is clever because she’s considering and manipulating the backstory of the enemies and she rolled well enough to be successful, so she learns that only one enemy has access to healing magic.
  • The fighter can’t get to the wizard’s ledge, but he knows it’s attached to a wall. He starts attacking the wall, trying to bring it down. The party asks him to stop because it might destabilize the architecture of the rest of the room, but he persists and starts rolling damage. This is entertaining because it can add a dynamic element to the combat, and with good damage rolls he can actually break through a wall. It also helps that I have a soft spot for players who hurt themselves to fit their character, so if the fighter is a berserker who loves destruction despite the party’s comedic exasperation, I’d consider that entertaining.
  • The party decides to stack themselves on each other’s shoulders so the rogue can climb them and reach the ledge. That’s a clever use of party resources (time) and mechanics (carrying capacity, height, and weight), and it’s entertaining enough that I want to see it happen. If I’d ask them to make rolls at all the DC would be very low. I might give myself a bonus to attack players at the bottom of the pole, but that’s unlikely to impact the success of their maneuver.

Per the above examples, “entertaining” usually means “funny”. But it can also mean “dramatic”. A player who uses a forbidden spell, sets aside their code of honor to do something underhanded, or sacrifices themselves so their allies have a chance, and who’s willing to accept the narrative fallout of their decision, deserves success as much as somebody who’s trying to pick up an enemy and throw them into another one.

Have a purpose for the puzzle. The puzzle needs to fit in with the pacing of the campaign and the story the campaign is telling. Having a puzzle boss because “it’s time for a puzzle” isn’t a sufficient reason. It should happen during a battle where it makes sense that just hitting the opponent isn’t a complete solution. Often the best way to do that is to telegraph it in some way; if the players find out this group of fighters has never had a single casualty, they won’t be surprised when they see it’s because they have a hidden healer. The party goes in expecting something out of the ordinary and they’re mentally prepared for it.

In addition, the puzzle needs to work with the boss itself. The powerful orc warlord with a reputation for ripping his opponents limb from limb is not going to hide behind his defenders and take potshots with a crossbow. He’s going to wade into battle with supernatural fury, and it’s up to the players to find out why their blows are bouncing off his skin. This is a good place to toss in a plus or a minus; the warlord is all but immune to the barbarian’s axe and it falls to the sorcerer to notice the warlord’s necromantic aura and realize he’s actually a zombie powered by a magic item. A clever player might even swing their role as a minus into a plus, like the barbarian who realizes her axe isn’t working, drops it, and grapples the warlord to mitigate his attacks and buy time for everybody else. That’s a good kind of clever, even if it does mean I have to find my grappling flowchart.

Make it fun. This suggestion overrides everything else. A puzzle can be well-balanced, creative, and narratively meaningful but still be a tedious slog. If the party just has to pull a lever every round to keep poison gas from filling the room, all you’ve done is give them an action tax. There’s a difference between “I can’t believe we did it!” relief and “I’m so glad it’s over!” relief, and that line can move based on your campaign, your players, what happened last week, how tired the players are from work, and so on. One of your roles as a DM is to focus on the right thing: puzzle bosses, like everything else in the game, are a means for having fun. If yours isn’t fun, it’s wrong.

If you’re looking for a detailed breakdown on types of puzzle bosses, you’re out of luck. I don’t have one either. But I can recommend the following works that have some examples I’ve used myself, as further reading:

  • Jojo’s Bizarre Adventure – Starting in the third arc they get into the concept of “Stands”, which allow the main characters to fight with supernatural, usually external entities. Most battles boil down to “Who is my enemy, what is his power, and how do I survive it and punch him in the face?”
  • World of Warcraft – I’ve gotten a lot of ideas from browsing strategies for raid and instance bosses. The numbers don’t really mean anything (I don’t know if “3000 Nature damage” was meaninglessly low or an insta-kill in the relevant expansion) but the general ideas are good for keeping fights dynamic and adding an element the unknown.
  • Final Fantasy – Each game has several bosses that don’t fit the normal “grind and smash” mold. I’m told the single best game for this is the remake of Final Fantasy IV for some recent portable system, where almost every boss is an actual challenge because they each need a different strategy. Games that let you change your party, like Final Fantasy Tactics, may also work for general ideas, but I’m legendarily terrible at strategy games so I don’t have a ton of experience here.
Posted in DMing | Leave a comment

Puzzle Bosses

In my career I’ve seen the designers of D&D try to approach boss monsters in several ways. In 3E a boss is a monster with a CR higher than the party level, which usually means it has higher defenses, bigger attacks, and powers beyond the party’s scale. In 4E a boss has largely the same combat numbers, but has quadruple hit points and gets more than one offensive action per turn. In 5E it gets even more actions and some ability to ignore player attacks. Pathfinder uses 3E rules, but the conventional wisdom is to make a boss a standard-CR creature with enough allies or minions to ramp up the difficulty and keep the boss from being swarmed.

All of these share a common trait: a boss is a normal monster, but scarier. For the most part they use the same rules as any other creature, sometimes with a template or bonus stacked on to give them different abilities (“solo” in 4E, “legendary” in 5E). An aggressive way of saying it is “a boss is a normal monster who breaks the unwritten social rules of the game by being numerically out of bounds.”

That’s my big issue with them: not the phrase “out of bounds”, but the word “numerically”. A boss is defined by their numbers, and higher numbers challenge the characters rather than challenging the players. A boss who’s a normal creature, except it deals 50% more damage and you have to roll two higher to hit it, isn’t actually a boss. It’s just a longer fight. That can be dramatic if the numbers push it just out of the players’ comfort or competence zones, but it can just as easily be a boring slog or frustrating TPK.

In the Zelda campaign I hit the players with more bosses than I have in any other campaign, and I settled into a rhythm that I think made them more interesting than just being mathematically superior. In most Zelda games, the players gets an item in the middle of a dungeon. They use that item to beat the dungeon, and it’s usually required or helpful for the dungeon boss as well. The campaign worked the same way; every dungeon boss was in some way vulnerable to the item in its dungeon, and the players had to figure out how to leverage that. This included not just learning how the item affected the boss, but also when to use it and which player was best suited to it, in the middle of a standard pitched combat, and the end of an adventure where the players were running on depleted resources. Beating each one felt like more of an actual achievement than rolling slightly higher than normal for significantly longer than normal.

This is a puzzle boss, a boss fight solved by a strategy other than “hit it really hard until it falls down”. Often a puzzle boss works via some weak point the player has to discover and exploit, like an opening in the boss’ defenses after a big attack or a nearby object that can damage the boss when attacks cannot. I think my favorite version of this is the one in Zelda games or Jojo’s Bizarre Adventure: the boss seems unbeatable, but once you figure out their trick you can open them up to one-sided, incredibly satisfying melee combat.

That’s where I find myself as I start to think about my next campaign, which will be in Pathfinder, the land of “if you ain’t doing a hundred damage a round, you ain’t trying”. I spent a good while trying to find published monsters immune to Pathfinder’s rocket-tag gameplay, and when that failed I tried to find monsters resistant to it, then gave up on that and just tried to find monsters that didn’t actively contribute to it. But instead of using the narrow subsection of the game that works with my style of play, I’m probably going to do what I normally do and make something new.

Part of my reluctance is that Pathfinder is still very simulationist. Monster design in 4E is like the Wild West, where you can’t assume that one monster’s swallow whole power works in any way like it does with another monster. Pathfinder doesn’t do that; “swallow whole” is a specific monster ability with a glossary entry and a programmed set of steps and results. The monster creation rules in the Bestiary actively discourage DMs from making something out of whole cloth:

Monsters should use abilities from the Universal Monster Rules whenever possible, instead of creating new yet similar abilities—when you do create new abilities, use the Universal Monster Rules as a template for how to present and create the new abilities.

The rules work on a very “like = like” system. For example, a character can break through nearly any barrier given enough time, and with the right tools they can usually make it trivially easy. Now consider a boss who is physically weak and controls minions or machines from behind a transparent wall. That wall can only be made of two things: a physical material like glass or invisible steel, which a strong character can destroy rapidly; or a magical material like force, which is impossible to break manually but falls immediately to specific spells. There’s no room for a surface that can only be broken by filing minions at it or tricking the machines into hitting it. That material would work unlike everything else in the world, and the rules don’t like that.

So I find myself in a situation where I’m planning encounters that work in a gaming system while deliberately subverting the intent of that system, skirting the line of “it works this way because the DM wants it to and for no other reason” without crossing it. It’s a strange kind of perpendicular design, making encounters with the feel I want by working despite the system rather than within it. That’s not bad per se, and I’m finding it a lot of fun, but it means I’m trying to come up with combats that marry both the existing mechanics and the gameplay I want without just telling the players that a given monster breaks the rules and leave it at that.

I don’t want to just hand-wave away Pathfinder’s simulationism, either. Say I’m committed to that magic wall that can only be destroyed by flinging a minion at it. I have to explain it somehow, so I’m going to say the minions are also magic. They could be magic in a similar way, which disrupts the wall because they overloads it or because they interact with it in a way other objects can’t, or in an opposite way, which damages the walls or exploits its weak points. I have to give the players some way to find out about the minion’s special properties—a previous encounter, a book of notes from their creator, symbols on the wall that match the symbols on the minion’s faces—and trust them to draw appropriate conclusions.

This trust highlights one of the big issues with designing puzzle bosses: they have the downsides of both a puzzle and a boss. If the players don’t figure out that the minions affect the wall, it’s over. If they figure it out but draw an incorrect conclusion, like thinking the minions enhance the wall and trying to drive them away, it’s over. If the get the right answer but fail on their rolls to enforce that answer, it’s over. If the players ignore or circumvent the puzzle entirely, like digging their way through the floor under the magic wall, the battle works but it’s deeply unsatisfying. And the whole time I’m trying to hit them with an exciting, dangerous encounter, so I’m taxing their resources even harder than normal. It’s not easy.

But the results are worth it. If the players succeed in the fight, and I succeed in designing and running it, we get an encounter that stands out from the rest of the campaign. It feels different because it is different, and the party has to adjust its tactics to compensate. If the players solve the puzzle in a way I didn’t expect, like tricking the minions into attacking the wall themselves, that’s even better. The players accomplished something, not the characters, and that feeling of legitimately deserving success lasts a lot longer than winning because your math outpaced the game’s.

I’m still trying to figure out if I have enough to say about making puzzle bosses to fill an entire blog post. I guess we’ll find out soon.

Posted in DMing, Pathfinder | Leave a comment

More Achievements Unlocked

It was a pretty rough night.

Posted in Pathfinder | Leave a comment

Changing 4E Monster Hit Points

Our house roles are the kinds of things that pop up organically. We don’t sit down at the beginning of the campaign and say “rolling a 20 on your critical confirmation roll has a chance to kill an enemy outright, because I want that to be a thing and the game doesn’t do it by default”. Instead we see something happen, decide it would be neat if the rules reacted to it, and over time reinforce that expectation. Most of the time these fall into two categories: rewarding the players for doing something neat (my players know rolling all sevens on something is a big deal, and we’ve used the charge rules in several ways that were clearly not intended) or adding something to fill a perceived gap (like our secondary skills). Very rarely do we create a house rule to take something away. Normally we prefer to resolve that by talking with the players, instead of outright banning something overpowered, or reskinning, instead of banning something that violates the feel of the campaign.

So it’s interesting that we’ve been removing something essential from our monsters in 4E: hit points. Starting in the One Piece campaign our group has cut three hit points plus three per level from every standard (non-elite, non-solo) monster. That is:

Standard Modified
Skirmisher 8 + Con + (level x 8) 5 + Con + (level x 5)
Brute 10 + Con + (level x 10) 7 + Con + (level x 7)
Soldier 8 + Con + (level x 8) 5 + Con + (level x 5)
Lurker 6 + Con + (level x 6) 3 + Con + (level x 3)
Controller 8 + Con + (level x 8) 5 + Con + (level x 5)
Artillery 6 + Con + (level x 6) 3 + Con + (level x 3)

There are several reasons for this, applicable both to our specific meta and to the meta of 4E in general. The biggest reason is how it changes the pacing of a day. Major arcs in the One Piece campaign involved several fights without long rests in a single location, usually punctuated by one or more boss fights. 4E has a pretty hard limit on how much punishment a party can take in a day: all other expendable resources aside, when a party is out of healing surges they largely can’t fight any more. Encounters are expected to consume slightly less than one-fifth of a party’s resources, so a party can fight five times a day but probably not six. That meant we could have three standard encounters, a miniboss, and a boss. It’s fine for a normal dungeon, but not for an arc-ending, four-session-long climax.

The books don’t really cover how to deal with long days because they rarely discuss and barely acknowledge a world outside the game’s design and testing scope. By extrapolating from their advice on long fights, we can say they expect long days to work via intermediate rests, longer than short but shorter than long. Maybe the party can hole up in a storage closet, catch their breath for an hour or two, and gain back two healing surges and one daily power but no more. That’s fine, but it encourages the player to take several of these rests, engineering them whenever possible, instead of splitting them up narratively. It also only works for certain situations. The Zelda campaign’s long days were in the middle of dungeons where “sit and wait for a while” wasn’t an acceptable solution.

We went about it the other way; instead of making the players’ resources deeper, we made encounters easier. The simplest way to do this is by lowering monster hit points. We could have lowered attack, damage, or defenses, but that leads to DM frustration and makes individual creatures feel less dangerous. A lower hit point value makes the monster just as scary, but for less time. The players can kill it quickly so it doesn’t spend eight rounds assaulting them, then move on. It also let us keep the same number of monsters in an encounter so our fight design wasn’t limited by an artificial head count, and it let us be more clever with those monsters because we knew the players could focus-fire truly perilous creatures more effectively.

Consider the end of this campaign, when the players and most monsters were L19-20. That meant each creature had about 60 fewer hit points than typical monsters. That’s two hits from the party’s blackguard, or three to four from the swordmage, and anywhere in that range from the ranger with multiple attacks, assuming all attacks hit. That’s two fewer rounds the blackguard had to spend in pitched solo combat against the enemy brute, which is two fewer rounds of attacks she had to take, which is two fewer rounds of healing she needs after the battle, and two more rounds crossing the battlefield and assisting with other monsters. In making the fights faster and easier we accidentally made them more dynamic; like players, monsters spent more time using their big guns and less on at-wills, and the battlefield changed more often as they disappeared.

This had a pronounced effect on table feel. Fights went so fast we frequently had two per session and occasionally three on top of our typical talking, exploration, and puzzles. Players used encounter powers more often, so they got more value out of those frequent, signature attacks and spent less time in the “nothing left but at-wills” state. They used daily powers less, often only once in three or four sessions, which made them feel more significant without changing a thing about them and let the players have bigger spikes by changing the average around them. And since we left elites and solos alone, they felt even more significant than before, closer to the climactic battles we wanted.

One reason this worked is because we already have a healthy disrespect for D&D as written. I haven’t awarded a point of experience in almost eight years. Instead we usually award level gains at the ends of arcs (or, in this campaign, whenever the party beat a boss or collected four Heart Pieces). Because of this we don’t know or care how our changes affected the XP budget. It’s why there’s no XP amount in a monster’s header in my Monster Manual.

In the Zelda campaign I also often gave the party fewer enemies in a combat, which made them faster still and gave them that feeling of “always fighting” you can in the games but can’t manage in D&D. Like in a video game the monsters’ job wasn’t to be a serious, life-or-death struggle in every room. They were there to give the feel of a Zelda game, to explain the theme of the dungeon, and to whittle away at party resources so the players didn’t notice they were running low until the boss music started. Further, it meant I could have monsters as simple as their source material warranted. Players notice when a monster has spent six rounds using just one attack. They’re a lot more forgiving of a monster with only two powers when that monster isn’t around for very long, especially if those powers are both instantly recognizable.

All told this change has been great for our sessions, our DMing, and our story pacing. It is, however, a 4E-specific change. In order to make things a little easier on myself in my next campaign, I’m going to have to approach monster design in a different way, which I’ll talk about soon.

Posted in D&D 4th Edition, DMing, House Rules | Leave a comment

Campaign Report – Monsters

The monsters for this campaign were weird. When I started in 4E, I wanted to make as many creatures as possible on my own. I didn’t trust the official monsters to be able to do all the things I needed, and I wanted to take advantage of how free-form monster design was in the new system. But as the campaign went on, I found myself with less time to build monsters and less enthusiasm for working in a space that was at once numerically restrictive and narratively overly permissible. Instead I got into reskinning, using published monsters in ways their designers clearly did not intend.

I expected to do much the same thing for the Zelda campaign, because monsters in video games are all about taking narrow sets of actions (move in four directions, damage the player on collision, sometimes shoot a laser) and applying them in several ways. But this time I found published monsters didn’t work at all because they didn’t accurately mimic the originals. Instead of looking for monsters based on their function, I was looking for them based on their feel and how closely they matched an existing target. I could reskin basically nothing, and I had to design almost every monster in the campaign from scratch.

So now that I’m basically sitting on a fifty-page custom Monster Manual, I figure I might as well make it available to anybody who wants it. Everything in here is a Zelda creature except for the dungeon bosses, but I don’t have pictures for obvious copyright reasons. You may need to take a trip to the Zelda Wiki, which has basically been my home page for two years, to find out what the more esoteric monsters are.

In going through this, I found that I actually lost the entire stat block for one creature, the boss of the Skytree Tower. It’s not even in the backup file I made one month after the fight. Still, if I designed 152 monsters and I have notes for 151 of them, that’s a pretty good record.

If I were to do this again, I would have put more weaknesses on monsters corresponding to the item in each dungeon. I did this a few times early, but not often enough as the campaign went on. I spent most of that energy making sure each item was required for the boss battle and the dungeon puzzles. I also should have done more in the way of slashing, piercing, and bludgeoning damage, something we carried over from 3E. Zelda games have several monsters resistant or vulnerable to swords, bows, or blunt objects like shields and pots. I think I fell out of it around the time we got a character whose primary damage type was “chicken” without further clarification; once things got that wacky, inventing ways to leverage it wasn’t rewarding any more.

But the biggest reason was completely cosmetic: new powers and resistances would have taken up more lines on the page, and I really wanted to squeeze in as many monsters as possible without flipping back and forth through the document during combat. That may not come through because converting the file from OpenOffice to a Word document ruined my margins and section breaks, but in the original file it makes total sense.

The mathematically observant will notice that monster hit points are usually wrong. That’s intentional, and I’ll talk about it next time.

Posted in Campaign Writeups, Campaigns, D&D 4th Edition, DMing | Leave a comment