Saturday, January 2, 2021

A Note on Deceptive NPCs and More Crystal Frontier


December has been a productive month, though projects keep moving off sideways rather then plodding on properly to their finish. I’ve put out a second ‘Mini-Adventure’ for Tombrobbers of the Crystal Frontier, while the original project still needs some art, layout and editing to finalize.

The Beast

At 1900 words, this second Crystal Frontier adventure isn’t quite a One Page Dungeon, but it’s only eight keyed locations, a lair, and it certainly lacks all the elements of a full scale dungeon adventure. Most importantly there are no random encounters or other mechanisms that put time pressure on the party. Instead it revolves around negotiation with two NPCs: an untrustworthy exiled sorcerer and the undead, demon tainted assassin that hunts him on behalf of a powerful, but distant regional faction: The Warlock King. The Brujia, The Beast and The Barrow is less of a single one session adventure (though it’s also that, and I think it contains a couple of interesting puzzles) and more of an introduction to one possible source of intrigue and a useful NPC who can remove curses(curses and magical disease are a prominent feature in the setting).

The adventure itself is something I’m happy with, the layout and art are properly brooding while maintaining the colorful, slightly psychedelic look that I’ve picked as the overall visual theme for Crystal Frontier. The keys are short and while the barrow isn’t expansive, it holds a few puzzle style traps and dungeon furniture that tells a simple story about if the players want to seek it out.


It’s available as a PDF here on DriveThru RPG.

I’m currently at work on another of these mini-adventures, though this one is 14 keyed locations and revolves around a dangerous “hunting monster” that stalks the characters as they search the location for treasure. It’s an interesting variation on the dungeon crawl, a style of threat that’s hard to do well but seems like it should offer possibilities -- I’ll see how well I can manage it. “Broken Bastion” should be out next month, also as a $1 PDF on DriveThru.

Maunderings about play style, 5E, and theory follow. They are entirely absent from The Brujia, The Beast and The Barrow.  


Deceptive Non-Player Characters: liars, betrayers and “questgivers” that go back on their word can be a useful addition to your game, there’s a great joy in catching them at their schemes or better in thwarting them. However, even for players familiar with the concept of lying NPCs the require delicate GMing.

First, lying NPCs must speak in their own voice, not that of the GM - their lies have to be in character and separate from narration and description of the setting.

Second, players still only have the detail and description that the GM provides, and exaggerated indications of untrustworthiness are often useful. Conversely, because the GM is a trustworthy narrator the GM should never declare an NPC lie as truth.

Third, Players hate deceptive NPCs and will often try to avenge themselves even for minor deceptions. A deceptive NPC should never be necessary to the setting and their deceptions should never conceal information that’s needful to understand a location or event.

For players unfamiliar with classic play and deceptive NPCs greater effort might be necessary, don’t be afraid to indicate that an NPC is not telling the truth, but avoid revealing the nature of the lie. This is especially true for unimportant NPCs who lack robust description or whose lies are unimportant. For these NPCs it’s also useful to save time by confirming that their truthful statements are that.

A few of my readers and editors for the adventure, one of the more thoughtful designers and most experienced GMs I know as well as a long time player -- one of the most astute and engaged I’ve ever had the privilege of running a game for, pointed out that The Barrow offers a lesson in classic play. Specifically, the adventure revolves largely around how to judge NPC offers, how to gauge NPC truthfulness and deception. That this isn’t simply an obvious and constant part of play hadn’t crossed my mind, but looking at newer adventures for contemporary traditional play it’s true that NPCs simply don’t engage in deception.

Thanks Anne, Ava, Eric and Nick!


 I’ve recently had a read through of Wizard’s of the Coast’s newest 5th Edition adventure, Rime of the Frostmaiden, and while it’s an improvement in some ways it follows the Wizard’s house style fairly closely. Rime is still largely a linear epic, built more like a computer roleplaying game then an adventure for an in person game. Part of the set of play style expectations and design principles is a sort of simplification and minimization of NPC and faction personality. NPCs in Frostmaiden don’t really have their own goals or complex relations with other NPCs/factions. At least they don’t have a variety of goals, some of which may conflict with player goals. NPC interaction is designed in such a way that it occurs in simple scenes that offer at most one or two choices, but are largely static - at the end the players will have a ‘quest’, that once completed results in the NPC sort of fading into the background. It’s almost as if NPCs in this contemporary traditional style design appear at first highlighted by a ray of light to indicate that the players can ‘talk’ to them to receive a quest, or will appear and start a ‘cutscene’ that advances the story. A lot of this may stem from other design choices or simply aesthetic differences. Reliance on boxed text for example leads naturally to NPC scenes and monologues. However, at some point elevating a predesignated narrative path over a toybox setting of factions has an effect. Frostmaiden is better then many other Wizard’s adventures in that it sometimes hints that the players may make choices outside of or even contrary to the path -- they may murder NPCs or depart form their plans, and it strikes me that this is a sign of improvement in the Wizard’s house style.

This isn’t to say Frostmaiden is a fully linear adventure or a bad one. More than most other WotC 5E adventures I’ve read it does make decent gestures in the direction of a more open world, especially in its opening “chapter” of 100 pages. Not only does it provide some variety in the 12 or so “quests” that make up this section, but it offers alternate hooks for most of the locations described to help a GM improvise a less stilted narrative flow then the expected “new town, meet Questgiver, undertake quest”. However, at some level the mechanics of 5E force a certain style of design and faction relations -- specifically the steep power curve of 5th edition and its reliance on either combat or milestone leveling mean that encounters/adventures must be more closely level gauged. This is something that’s hard to get away from when running 5th edition Dungeons & Dragons both because the ethics and mechanics of contemporary traditional play promote tactical combat as a solution to most obstacles (Frostmaiden struggles mightily to subvert this - with mixed success) and because the steep power curve means that threats must be balanced against level with a fairly high degree of specificity. To put it another way -- in a low level adventure following the classical design tradition, such as The Brujia, The Beast and The Barrow one can include a powerful monster that will demolish a 1st level party in open combat because (A) It’s not a sure thing - the power curve isn’t as steep so a clever or lucky party may win such a fight, (B) There’s little expectation among players that combat will be ‘symmetrical’, and (C) Combat doesn’t provide the means for player level advancement either explicitly as an XP source, or implicitly as the clear marker of quest completion i.e. there’s no expectation of “boss” monsters.

Still, I can’t say that my own mini-adventure departs too greatly from this ‘questgiver’ formula -- it can’t in three pages. The sorcerer Marble-Eye can do something that most Crystal Frontier (or really any setting) characters will want and he has a price for it either monetary or in the form of a quest. Yet a couple differences are notable between a toybox questgiver and a path questgiver that I think stem from the overall way the design tradition treats NPCs. Marble-eye and the Barrow are designed as a place within the setting that’s independent of but still part of a larger player created story. The barrow and the sorcerer’s problems exist in a fairly static state of equilibrium that the players can disrupt, either for the short term benefit of curse removal (leaving the demon-bull assassin trapped by the barrow’s magic), long term benefit of curse removal, treasure and creating an ally (somehow destroying the demon-bull) or the lesser benefit of treasure and ultimately disadvantage (freeing the demon-bull to kill Marble-Eye, so losing someone who can remove curses/diseases for a reasonable price). These outcomes all change the overall setting in some minor way, but they aren’t part of a predetermined narrative structure, even one as loose as the more open chapter based structure of Frostmaiden.

In a more linear, story based design structure, especially with milestone leveling, NPCs paradoxically don’t do a huge amount of work to create a story, the ‘chapters’ and overall flow is set in advance. In Frostmaiden the party will do chores around the region’s towns, then when they have collected enough milestones/levels they will move on to investigating and confronting a grey dwarf threat to the region and ultimately exploring a lost city. It’s quite a lot really, and mostly it’s fairly open to player choice (though the Duregar portions are a bit more structured), but because of the ethics of balanced encounters and mechanics supporting a milestone based level structure with a steep power curve, the design needs to gate these larger chapters off. It does it well, but there’s less space for factions and individual NPCs to remain relevant then in a more open toybox.

However, it’s not really useful to compare a 200 plus page adventure path/campaign to a three page adventure, but there is a distinct difference between the way the NPCs in The Brujia, The Beast, and The Barrow function and those in most Wizard’s products work that’s worth highlighting. Specifically the problem of how to run “lying NPCS”.


One of the interesting aspects of Rime of the Forstmaiden is that despite a large number of NPC interactions there’s little deception by the NPCs - except for the most obviously evil ones. I may be reading too much into it, but this tendency seems like an aspect of the Wizard’s house style. The only instance of an NPC lying about their intentions that I could discover in the first several mini-adventures in Frostmaiden is a pair of talking winter wolves that attempt to trick the characters back to the lair of their mammoth boss. Notable about this obvious set up is that the characters get a skill check to detect if the wolves are lying. Another interesting element of this encounter (and one that’s likely worth a bit of analysis) is that the lie doesn’t use a text box. It’s an NPC interaction that offers the GM the goals and a few possible tactics/statements by the wolves without a full text box of “dialogue”.

Maybe this odd approach to deception is a useful comment on the role of skill checks in 5th edition design (and who players use the meta-knowledge of a roll’s existence even when they fail it), but it’s also interesting in that most of the classic games I’ve played in or run are packed with deceptive NPCs. It’s almost always worth doubting what an NPC has to say, and analyzing it based on their situation or needs - not the players. I personally avoid untrue rumors because they often occur during the “haven turn” or outside the detailed description and danger of the dungeon or adventure itself and rumor mongers are often lacking sufficient description to give players the tools to judge their veracity.

Still, NPC deception just doesn’t seem like it’s a big part of the Wizard’s design principles for 5th edition. NPCs tend to relay truths, and while they don’t always know everything about the situation, they also aren’t often deceptive about it. Whatever the source of this difference in design, it’s something that could present difficulties for GMs used to the Wizard’s style and making a transition to more classic play. There’s tricks to running lying NPCs.

One needs to accept that maintaining the players’ trust is a key aspect of classic play. There’s several reason for this, but most importantly the players need to be able to feel that the GM is a neutral arbitrator or a distant ally. GM description is simply too important to the problem solving style of play and the consequences of character death too present for play to work well if the players have doubts about the GM’s veracity and trustworthiness. Yet, at the same time as the GM needs to be unimpeachably honest, the setting needs to be able to deceive and trick the players.

The necessity of some deception in play and a way to make it work might be best thought of in comparison to word play or riddle. Riddles traditionally depend on deceptive word use, somewhat like a pun or double entendre so that when the solution is revealed, what was confusing or tricky is revealed as obvious - a different way of looking at a mundane thing. For example, there’s this riddle from the Hobbit:

“Alive without breath,
As cold as death;
Never thirsty, ever drinking,
All in mail never clinking.”

As a child, hearing the Hobbit’s riddles for the first time they seemed enormously clever, and very hard - I don’t think I figured a single one out. Bilbo figures it out, because he’s the book’s hero, but also because he thinks about the things that his riddler Gollum knows about - mountains, darkness, and the blind fish he catches and eats in his subterranean pool. Yet the answer is so obvious once provided, that even as a child confused by the riddle it seemed eminently fair. Of course the riddle describes a fish. There’s no way to view a Hobbit style riddle as deceptive, or cheating -- withholding key and necessary information to make it impossible to answer. One wants deceptions and tricks in RPGs to be similar -- possessing a clear answer that was obviously available from the clues the players had on hand. This assures the players that they can trust the GM, and like rolling dice openly puts negative results back into the player’s hands as the result of their failures to put clues together correctly or choice to engage with a risk. It’s not always easy, but that’s the goal - offer the fun of unpuzzling the riddle’s tricks and the sense that even in failure there was a clear answer available from the clues and tools provided by the GM.

Lying NPCs don’t exactly work like this, but if approached wrong they can break up the sense of fairness, creating a situation where the players feel like the NPCs falsehoods aren’t part of the fiction of the setting, but part of the GM’s narration or metafiction of play itself. This happens when the GM’s voice and that of the NPC get confused - and the lies of the NPC suddenly become a betrayal by the GM. The goal then, especially with players familiar with WotC’s style of truthful NPCs (and the veiled warning of perception rolls) is to make a clear distinction between what NPCs are saying and what the GM is saying and to offer sufficient information about NPCs that players can make judgments about their character and truthfulness that doesn’t later feel like a deception.

This may be a time to be pretty heavy handed as a GM. It might even be a time to use funny voices. A different voice or set of acted mannerisms can help make clear that the unreliable voice of the NPC is talking rather than the oracular voice of the GM. However, perhaps better (or at least better for those of us who don’t like using funny voices) is to remember to be very clear when the NPC is talking, and to offer NPC description.

Description works well with a few memorable details. They don’t need to be much, with Marble Eye the Sorcerer all I’ve included is a physical description “filthy”, “thin”, “eyes of polished stone” and his sartorial choices: “the blood robes fashionable at the Warlock King’s court”, “a coin tiara” and “cinnabar necklace”. This should be enough to make him memorable, at least in the context of a three page adventure. Hopefully it also gives the players enough of an initial impression to make judgments about Marble-Eye’s situation and decide if they like or trust him. A picture can help, even if it’s just an image pulled from online of a now forgotten movie star.

Most important though is clarifying the separation of GM voice and NPC voice. Never simplify or gloss over deceptive NPC (or any NPC really) statements.

“You learn from the sorcerer that there’s an evil in the barrow that he keeps trapped and you need to go offer it a pig”

Instead, try to provide the exact wording of the NPC.

“Marble Eye the Sorcerer says:
“”I’m protecting us all from the ancient evil I trapped in the barrow long ago and only the sacrifice of this hog will renew my ward

For players with limited experience in classic games it may even be useful to always reveal that a lying NPC is “not telling you everything” or “seems to be concealing something”. There’s little downside to this unless the GM reveals exactly the nature of the NPCs lie, and especially for regular people (drunken bar patrons peddling rumors for example) it provides the players with both a warning that NPCs can lie and a sense that their characters are competent sorts who know enough about the game world to detect bullshit.

Following from this idea is the second way of distinguishing NPC mendacity from GM antagonism -- remembering that and playing as if the PCs are competent. The GM can provide the players with the characters' sense impressions. The characters might be able to smell fear or spot an eye twitch that proves an NPC is a liar, but here the GM can’t themselves lie. If an NPC is telling the truth, and a competent judge of human nature might be able to tell, especially if giving the players the information saves play time in a basically narrative or rumor finding interaction, it’s fine to say that the NPC “seems too scared to lie” or “Sounds like they are telling the truth”. However, unless the NPC is the victim of powerful mind control or an accomplished liar (the “Spymaster” might sound truthful even while lying and it’d be unfair for a player to believe that the GM had tricked when a literal professional liar had deceives them) the GM should never say that a lying NPC is telling the truth. Even with the ‘Spymaster’ a few hedges or uncertainties are in order because it’s almost always true that the GM’s statements of ‘fact’ are more powerful than a player’s impressions. When the player has doubts about an NPC’s truthfulness, these doubts will (and should) fade if the GM says that the NPC is telling the truth, and this is how GM’s transform NPC lies into GM deception.

These tricks may not seem like much, but formalistic distinctions between subjective NPC voice and always trustworthy GM voice are important.

In Barrow, Marble Eye might conceal a bit of information, but he’s fundamentally honest and keeps to the deal. The undead assassin “Rolling Calf” (a name pulled from Caribbean folklore) doesn’t - and hopefully a trapped undead demon-minotaur with burning eyes seems untrustworthy. Beyond his obviously demonic and horrific nature there’s a situational reason that he’d lie - he’s trapped, anything to escape! Both the unsavory nature of Marble-Eye (he uses human skins to keep his hut windproof for example) and the obvious horror that is Rolling Calf are designed to imply an untrustworthy NPC. Even for players used to the transactional style of computer RPG interaction they will hopefully have doubts, and if they don’t they should at least realize that real world badges of untrustworthiness apply even in Dungeons & Dragons. Both of these NPCs are excessively, obviously untrustworthy -- and that’s intentional, the GM cannot provide the level of nuance and detail that one experiences in the real world, and larger than life characterization is one way to make up for that. Genre is another, and the “Spaghetti Western Fantasy” of Crystal Frontier should imply schemes, grey v. grey morality and greed mad, double crossing, dry-gulching varmints at every turn.

While the majority of Game Mastery involved in deceptive NPCs is presenting their deceptions in a way that makes the lies distinct from the GM’s authority so they don’t erode players trust as a designer there’re a few other issues at play. First, NPC liars have short life expectancies and second they need to create story rather than cut off content from play.

Players tend to develop strong feelings about NPCs fairly quickly based seemingly on random details or if the NPC crosses or spites them in even the most meaningless of ways. They tend to murder these NPCs. One should never create an annoying NPC whose death limits outcomes, and one should never create an annoying NPC whose danger isn’t entirely evident.

In Barrow Marble Eye lies, though not in a way that endangers the PCs - if they follow his instructions, and get past a significant trap, they’ll be fine and receive the reward he promised. They will also know that Marble Eye deceived them - though given that he’s a deeply sketchy sorcerer this shouldn’t surprise any but the most trusting of players. Moreover, marble-eye is clearly dangerous - a stone eyed, unarmed man living atop a cursed barrow in a hut roofed with human skins and wearing the robes of a court sorcerer shouldn’t seem like ready victim. THis of course isn’t to say that Marble Eye would be especially hard to kill - if one didn’t mind casualties. He’s an annoying NPC but 1. Offers useful help 2. Is clearly dangerous 3. His lies shouldn’t actively harm the characters. There’s a significant chance that he will survive.

A Trustworthy Face be Sure!

Second NPCs that lie should never be essential to the plot, despite Marble Eye’s advantages I’d guess that about ½ of the parties playing the Barrow will at least try to murder him or betray him to the Rolling Calf - not because the promises of the Calf are enticing, but because a chance to anhillate any NPC that wronged any player in even the slightest way is a core joy for many players. I’m not saying people are bad, but I will say that D&D reveals the ever present sadism that most people have in their hearts in ways that are only matched in my experience by divorce litigation. Yet Marble Eye is not essential to any larger plot, and Rolling Calf less so, a positive menace. It’s fine if one or the other both die, and letting Rolling Calf kill Marble Eye is a perfectly reasonable ending for the adventure - one where the characters may even do well, gaining a magic spear and Marble-Eye’s treasure if they are clever.
Creating these sorts of NPCs, and recognizing their transitory nature, but also that they may endure means giving them more then just a set of actions. It’s enough to describe a simple questgiver and say what they want and how they will reward it, but for a deceptive NPC the GM will need more because the players are likely to engage in more complex conversation. Like a faction a brief set of goals, fears and plans are in order while the addition of personality and vocal or visual tics can help GMs who need to run the NPC.

This is the nature of Toybox setting design, putting a lot of elements into place and letting player choice determine which ones become important and which ones are a one session obstacle. Untruthful and deceptive NPCs are one way to encourage this because liars are hard to see as friends, and deception disrupts the transactional ‘questgiver’ structure by demonstrating that NPCS have their own agendas and potentially may make their own changes to the setting.

I can't say the The Brujia, The Beast and The Barrow embodies all of these ideas perfectly - it’s limited by its length and by a few mistakes, but I do think it offers a good moral quandary and I’m happy with the aesthetics. It’s true that Marble Eye could use a bit more detail about his motivations -- he lies because he both wants the PCs to die - a better sacrifice to the barrow then a pig and because he’s one of those secretive evil wizard types who simply deceives out of habit. Yet, this sort of detail might go too far, because it’s most likely that players will make the choice simply to let Rolling Calf kill him




  1. I appreciate this robust answer!

  2. I ran this for some friends this week! Had a great time.

    I wrote up a short play report if you're interested:

    1. I certainly do! Feedback and reviews are the best sort of thing people can do for an adventure writer. Even the harsh ones help me learn how to improve things and what doesn't work.


Old Games

Let’s talk about old tabletop roleplaying games - specifically the kind of games played in the 1980’s and recently depicted in the nostalgia...