Making RPG Combats fit Dramatic Structure

There are multiple methods for analyzing dramatic structure.  Most of them don’t really apply to RPGs that well.  This post assumes some familiarity with dramatic structure.  I want to compare combat in RPGs to dramatic structure, and find out where the combat should fit in.

In a combat, as opponents are killed off the players continue to survive.   The intensity of the conflict diminishes the longer the combat goes on.  The climax of a battle is the very first round, and everything afterwards is just denouement. Therefore, if a combat is to be a climax the rising action needs to build up to the very first round.

Creating this rising action is the key difficulty.  Many people enjoy a series of combats that escalate in difficulty, but I find that too time consuming.  Skill checks and roleplaying scenes are nice and quick, in comparison.  However, often skill checks and roleplaying scenes don’t have very much at stake.

Remembering that the goal of this is create a suitable rising action prior to combat-scene based climax, there is no actually need for more at stake in terms of game mechanics.  It is better if the players feel more invested in the story and the outcome.  A good way to do this is to create open-ended problems that the players can solve creatively.  Then, when a player makes a skill check, it’s not just success or failure on the line.  The player has invested some of their own creativity and effort into that roll, and they will care a bit more.  In that situation, failure is a loss of control over the narrative.  The control would then be in the hands of the GM, who gets to determine the consequences failure has on the story.  Instead of increasing penalties for failure, the sense of intensity form the rising action can be created by more complex and more open-ended problems for the players to solve.

Using open-ended problem solving will help get the players invested, but players will eventually notice if nothing bad happens to their characters when they fail.  There’s three ways to make mechanics part of rising action: steadily increasing the penalties for failure, increasing the rewards for success, or both.  Which one is best suited really depends on the plot.  Penalties for failure will usually be the easiest for GMs to use, but that doesn’t make them the best.  If the GM takes away the rewards the players earned with all their creative problem solving, it will feel like more of a loss than if the players get a penalty on future rolls.

After the combat, the conflict that drove the plot needs to be resolved very quickly.  It can be safely assumed that the conflict was caused by the thing the PCs just killed, because otherwise the climax would probably not be a combat at all.  Part of resolving the conflict is resolving associated loose ends.  Going to one’s employer to get paid comes to mind as a very common loose end in DnD.  These scenes ought to be relaxing and complication free; they make up the denouement.  They must also be short.  I think this is actually a good place to put in certain types of structured roleplaying scenes, perhaps with associated XP awards.  A denouement is often  used to show in a very unambiguous way how a character or the setting has changed, and this is best done with a roleplaying scene.

To make combats in RPGs climatic actually relies on having a very large amount of time dedicated to things other than combat.  Of course, there are lots of reasons to enjoy combat in an RPG that have nothing to do with drama, and there’s nothing wrong with having lots of battle.   Just remember that having lots of battles actually makes having climatic battles more difficult.

Visual Supports and Creative Thinking

Consider the players visit a town.  In that town there are three things: a market, a castle, and an inn.  If your players are like mine, they will visit places in the following order: the market, the castle, and then the inn.  How dull.  This is a danger of presenting information in the form of a list.  Of course, instead of a list the information could be presented with lots of flavour text and exposition.  This runs the risk of boring some players, even though it will entrancing the others.

Recently I’ve been running a game for children.  They love pictures, and I’ve noticed a big difference in how they play hinging on when present the pictures and the amount of detail in the picture.  The players interact in a much less linear and a much more creative fashion when the information is presented in a picture, and that picture is imperfect and imprecise.

Imagine, instead of the list, I say “you go to the town and it looks like this.” and draw a doodle that looks like this:Image

Players will say things like “what are those tents?” and I’ll say “that’s a market.”  They might ask why the people at the market are in tents, and I’ll say something like “the merchants in this kingdom are mostly members of a nomadic ethnic group” or some such thing.  They might notice that this seems pretty sparse for a town, and ask what else is around, and I’ll say something like “Like most parts of this kingdom, this area is surrounded by lush swamp and impoverished dirt farms wherever the land is dry enough”.  This kind of logical engagement with the setting is a good thing to foster in an RPG.

Note how this replaces GM exposition with players asking the GM reasonable questions.  Generally speaking, listening to exposition is not fun for players and writing an exposition is not fun for the GM.  Exposition is a necessary evil, but there are many tricks to minimize it.  Instead of preparing a lengthy exposition about a location, a GM just needs a doodle and to be ready to improvise reasonable answers to what the players ask.  It will seem like the GM had a really well prepared setting all along, but it’s mostly made up on the spot by logically extrapolating from a few simple points.  The players, in turn, get to feel like they’re investigating to get the information, instead of just listening to a lecture.  It is much more engaging.

Players are also less likely to just visit every place in a single particular order, the way they are if locations are presented in a list.  That being said, I know one guy who insists on examining each place on a map in a clockwise order.  I just make sure my maps aren’t round when playing with him.

The key features of using visual supports in this is that the pictures are abstract, not precise.  Pictures that are detailed and precise do not invite the same amount of interpretation.  A stick figure with a viking hat and tusks leaves much more for interpretation than the picture of an Orc in the bestiary.  This makes pictures for communicating information very different from the pictures that are used because they are cool.

This doesn’t need to be just for maps.  It can be used for objects or monsters also.  In the case of monsters, their most threatening features can be exaggerated so that players know what to be careful of.  This is pretty obvious, but it’s still more engaging than if the GM says “it’s claws look dangerous.”  Look at the following monster, and guess what it’s abilities might be.

example doodle mindflayer

Assuming the players can tell that those are tentacles coming off it’s face, and the GM clarifies that they are “waves of psychic power” and not “stink lines,” the players will realize that they are fighting a sort of humanoid opponent with dangerous face tentacles and psychic powers.  Perhaps you agree with me that it was more fun figuring that out for yourself than if I just told you it has “dangerous looking tentacles and big brain that indicates psychic powers.”  This monster is intended to be a mindflayer.  The picture would work especially well if their characters shouldn’t know what a mindflayer is, for one reason or another.  Unfortunately, most D&D players are experienced enough to identify a mind flayer instantly.  If you used this technique in a game like GURPS, where the GM could make almost anything, the open ended interpretation of the visual will be especially beneficial.  In either, though, if the players think of something novel that it seems reasonable for the creature to be able to do based on the drawing, it’s a very good idea for the GM to add that to the abilities of the monster.  When my wife saw the picture she thought it’s body looked like a kangaroo, and said she thought it would be  able to jump really far.  I would  completely add that to the list of mindflayer abilities without a second thought if a player said that during a session.

It is possible to use visual support for “room-escape” style puzzles.  The key difference between this in an RPG and a room-escape puzzle game is that in a roleplaying game there ought to be many correct answers.  Just remember that anything logical ought to have a chance to succeed, and anything logical that includes the kind of details hinted at by a picture ought to be very nearly foolproof.

.example doodle room escape

This room was intended to have 4 features: a pile of garbage, a window to the outside, a puddle, and a locked door.  Players might look at this and infer that the garbage would include a big branch because it’s under a window, and use the branch as an improvised lock pick.  The wall with water pooling against it could be structurally unsound, either due to water damage or the puddle indicating a sinking floor.  Maybe they think the water will freeze at night, and want to try and drip water into the lock MacGyver style.  When I drew the bars, some of them were accidentally kind of crooked.  Maybe the players will notice and just say they can break the bars with their incredible brute strength.  Any of these would be acceptable ways to escape the cell.

Using these visual supports can make what is normally a routine or dull task be an open-ended and creative experience.  It also saves time, and prevents boredom, by replacing exposition.  It can also be used to communicate important information (such as “watch out for that monster’s mouth tentacles and psychic attacks”) without outright stating it.

Improvisation in RPGs: Building the Platform

In improvisational theater, the beginning of a new scene normally needs to introduce a setting, the characters, and any important props or objects.  This is referred to as the platform, as it is what the scene is “built on top of.”  In most RPGs, the GM has control over the platform, and the players have control over how the scene plays out.  With this post, I will discuss how everyone at the table can participate in building the platform instead.

Adding to a platform is done by “offering.”  When a player offers, they take an action or say something that implicitely creates something in the setting.  Consider the following exchange:

GM: you enter the tavern.  there’s a shady character in the corner.

Bilbosh the Baritone Bard: I push my way through the swarm of patrons, climb on top of the room-long banquet tables, and use my impeccable singing voice to get everyone’s attention.

Fred the Flirty Fighter: I meander through the bar looking for chicks (preferably gnome chicks) but I cannot overcome the mystery around the shady figure in the corner.  I watch her closely (while being careful to not look like I’m watching her closely), and am glad to find that she is only about half my size and carrying books.

The GM offered that there is a tavern.  Bilbosh accepted the offer, and offered more of his own: that there is now a crowd, and that the tavern a classy kind of tavern that has banquets.  Fred acknowledged and took part in Bilbosh’s more detailed bar, and then added some details about the shady figure in the corner.  Now they have a much more detailed platform to build a scene off of.

For players, platform building can make it so that the various abilities that are only useful under certain circumstances come up more often.  I any game that uses a grid, the shape of the room and what fills it can be altered during roleplaying, before the GM has to draw what a room looks like.  This can adjust what kind of area of effects are useful, the benefits of mobility, and the benefits of range.  It can also adjust the kind of skills or powers that increase mobility.  By being able to add details to NPCs also, situations where the players benefit from things like smite evil or favoured enemy are partially under the player’s control.  This introduces a new balance concern: a player who is skilled at improvisation will benefit more from playing characters whose abilities are balanced by their usefulness being limited by the frequency of how often they can be used.  For example, a player who is a very strong improviser would benefit more from being a ranger than a fighter, due to favoured enemy and favoured terrain.  Before, the responsibility to balance these was entirely on the GM.  In GURPS, this problem is made much more severe by Frequency of Appearance rules for advantages, and the similarity in effects between skilled improvisation and the serendipity advantage.  In GURPS games where I allowed players to build the platform, I used the guidelines in the GURPS: Action series of books for how to use serendipity-like effects.

For GMs the burden of creativity and scene construction is actually made much more difficult.  However, the amount of time spent preparing a session is greatly reduced.  Only a rough sketch of the most important details of every place and person needs to be created.  Any more will cut into the player’s contribution, and any less will make developing plots and settings impossible.  How much is the “bare minimum” is a matter of preference; I personally tend to use lots of minor details for settings, and have only a small number of very heavy handed details about NPCs.  Keep in mind that the players are limited to details that build the platform for the current scene only, so that the plot for the campaign as a whole and the sweeping, setting-wide themes are still under the GMs control.

Playing out the scene after the platform has been made is actually where most of the fun is.  Players are likely to be more invested in each scene, since they’ve added details that interest them.  It is not relevant to improvising a platform whether the scene becomes a roleplaying scene, a combat scene, some kind of skill-related obstacle, or a puzzle.  However, the platform that everyone builds together will often imply a particular way to play it out.

Platform building needs to end at some point, and adding some formal structure makes this much easier.  If players add platform details late in a scene, it will normally seem arbitrary or like a deus ex machina.  GMs tend to be a bit more aware of what constitutes an unfair surprise as opposed to a fun surprise, just because they are used to taking on the responsibility that comes with authority.  As such, limiting introducing more details is normally limited to the players.

I suggest that the platform phase ends after a period of time equal to 1 minute/player + 1 minute.  Since players are usually used to having time to think, this will often be a substantial time crunch for them.  To encourage full participation, I suggest using a reward mechanic where the reward is larger the more people participate.  For example, if 1 person participates the XP Pool could be 100, but if two people participate they get an XP pool of 220 XP (110 XP each), and if 3 people participate it’s 360 XP (120 XP each).  This means that players are more likely to try and include the other players, both with out of character banter and by doing things in-game to prompt a response (such as in character dialogue).  Alternatives to XP for rewards exist, but I’m not going to talk about them here.  There are issues with XP based reward systems, and the set of alternatives warrants lots of discussion.

Of course, participation can be made mandatory by using a turn taking system, much like combat rounds.  In such a case, I suggest the person who goes first gets to go twice, so that they have to do something that acknowledges the contribution of other players.  Personally, I don’t like turn taking here because it discourages spontaneity.

A common problem with platform building is that players can become very silly.  There is nothing wrong with a funny scene, but when the platform of a scene contains a joke the result is sketch comedy.  Sketch comedy does not lend itself to longer plots or character development.  Some players are fine with this, some are not.  It really interferes with my enjoyment of the game as a GM or as a player, since I appreciate very detailed settings and plotlines.  In my games, I make it a rule that any platform detail added because it’s mere existence is funny is “erased.”  Players interested in being funny ought to use the platform to set up jokes and humorous situations when building a platform, but not deliver any punchlines until the scene is being played out.  This is consistent with guidelines for improvisational comedy.  The kind of comedy to aim for is more like in Cheers or Arrested Development, less like Monty Python or Benny Hill.  That being said, I think I may want to switch to some kind of reward system for players who are funny in the way I want them to be, instead of punishing the players who are funny in a way that I find disruptive.  I’ll need to think about that more to flesh something out.

Integrating platform building into game mechanics is relatively easy.  Quite implicitly, the control it gives players helps them control what mechanical abilities will be relevant when playing out a scene.  If a GM wants, however, they can require checks prior to adding details.  The way I do this is to make it so that successful knowledge, perception, or sense motive checks allow a person to add details about the subject of their knowledge or observation.  This works better in GURPS, in my opinion, just because of the much more expansive skill list.  Since doing this introduces a barrier to full participation from all players, I don’t like it very much, but if players are used to platform building it can help integrate the details they add with their character concept.  This is particularly good for any semi-realistic science fiction roleplaying, as it forces players to narrate platform building through the eyes of their character’s own professional expertise.

When I next get a chance, I’m going to use my ideas from earlier post “Round Replacement” as a way to fill in what happens after building a platform.  As a personal challenge, I want to have a game that has everything that’s not combat be so much fun that players avoid combat solely to increase how much fun they are having.  Even the players who thrive on imposed objectives, structured gameplay, and “winning” the RPG.

The strengths and limitations of reward systems in RPGs

Reward systems can be used to adjust behaviour.  My real life profession is actually all about doing just that for therapeutic purposes, so I know a thing or two about this.  The key part of creating a reward system are deciding what behaviours you want to increase.  When that behaviour occurs, a specific reward is given to the person who performed the desired behaviour.  For the purpose of this post, lets assume that the reward is XP.

Here is a short list of behaviours that are often rewarded with XP in RPGS:  fighting, overcoming obstacles, roleplaying, being funny, cool stunts.  Of these, fighting is usually worth the most xp.  It also usually has the most structured gameplay, and also has the clearest objective.  Overcoming obstacles is a broad category, and normally the way to overcome any given obstacle is structured within an RPGs rules.  They normally are worth less xp than fighting.  Finally, rewards for roleplaying, being funny, or doing cool stunts are usually the smallest and the least structured.  They tend to require substantial judgement calls by the GM.

First of all, assuming the players are motivated by the reward, they will be motivated to engage in behaviours that will result in the reward.  This tends to make PCs into incredibly violent and dangerous individuals who’d just as soon kill someone as talk to them, and this makes perfect sense given the reward system.  Not only is combat usually rewarded more than any other way to spend time at the game table, but if a player manages to roleplay or otherwise be creative in a way to get out of a potential combat they only might get a reward.  It depends on the GM’s discretion.  Combat has the most rigorous and strongest reward mechanic, so of course players engage in combat very frequently.  You may find even players who claim to prefer roleplaying, storytelling, and acting to combat will often use violent solutions when clear alternatives are present.  Reward systems have that impact on people, often subconsciously.

So, if you want players to act differently, alter the reward system.  This requires picking the specific thing you want the players to do, and tying it to a reward mechanic.  What the players do needs to be a behaviour, not the adoption of an attitude.  However, the adoption of certain attitudes can often be a pleasant side effect.

For roleplaying XP, one thing I did in the past was have the players pick a specific way they can show their character’s traits during play.  It wasn’t enough for a character to be “lawful good.”  They had to pick a character trait they could act out in game that was representative of “lawful good.”  A character had to pick a trait they could act out that was “lawful” or “good,” such as honourable, honest, kind, etc.  If they did that during gameplay, they would get rewarded.

Another part of a good reward system is timing the reward and how it is presented.  It can be presented immediately or after the behaviour.  In theory, it can be presented before the behaviour, but that is strange and very rarely a good idea.  In the above case about roleplaying rewards and character traits, I waited until the end of the session and people could “present their case.”  Waiting until the end of the session and allowing people to pick what in their play showed the trait allowed them to highlight when the trait was implied by subtext or context, but was not overt.  Since often character traits will not come up directly in play, this was important.

The behaviours that a GM wants to encourage with a reward system can be almost anything.  It can be used to encourage certain table top social dynamics, or to encourage creative solutions to a problem.  It can be used to encourage certain uncommon uses of existing rules in games, to encourage intrigue between the players, or to encourage certain themes.  The trick is defining the behaviour the players have to exhibit.

The reward motivates the players to take the initiative, which can really change a game’s dynamic.  The GM does not need to be responsible for presenting opportunities to players for everything the players want to do.  They will take initiative to earn the reward.  The GM also needs to be prepared to improvise, and allow the players to take the lead.  This more reactive approach makes the game very different.  It becomes more of a team effort.

Unfortunately, my trait roleplaying system worked too well.  You see, there was a list of 4 traits in total that players were able to roleplay.  By the end of the first month, the players were trying to hit all four of their traits as efficiently as possible.  By the third month of the game, some of the players started yelling out “CHECK!” each time they did one.  It became a roleplaying grocery list.  It was pretty funny, and a lot can be learned from how the system broke down.

Sometimes a reward system can encourage behaviours until they are routine, predictable, and boring.   This will interfere with why the reward system was implemented in the first place, because fun is a much more important reward than XP.  When this happens, either a more complex, more difficult, or more abstract system can be used to replace the previous one.  This breaks up the routine.

A thing to watch for is accidentally rewarding other behaviours.  At the end of every session, when it came time to convince me that they acted in a way that implied the character traits, some of the players were very convincing with their argument but in fact did not act much at all during the session.  I didn’t reward these people (which was the right call), but there was another problem I missed.  I didn’t withhold the reward from players who artificially created stupid opportunities to roleplay all 4 of their traits.  In so doing, I reinforced the mentalities that led to the “grocery list problem.”  When to withhold the reward requires some difficult judgement calls.

Another potential problem is reliance on a reward system.  With the grocery list problem, the players were very good at roleplaying the traits they identified at the start of the campaign.  They developed the skills to roleplay and better define their characters, but they did not use this to make their characters change over the course of the campaign.  They only took it as far as the system required, even though they probably would have had more fun if they took the initiative to go further.  Reward systems often have this problem, to the point where it is unrealistic to expect people to behave otherwise.  In behaviour therapy, there are ways to work around this reliance.  Most of them do not transfer to a game.

The last thing to remember about reward systems is good advice in general: keep it elegant.  If a game has too much going on at once, it gets very confusing.  Further, there’s only so much time to spend at the table, and it cannot be spent simultaneously emphasizing every single fun thing that people can do during an RPG.  There’s just not enough time in a game session to do that.  A game can’t be about everything at once, no matter how clever the reward system is.

Round Replacement

I want to create a replacement for rounds that accomplishes the same things, but is less restrictive.  I mostly intend to use this outside of combat for the various kinds of scenes that are important enough to warrant structured game time, but thrive on open-ended participation in a way that combat usually doesn’t.  Diplomatic encounter, strategic preparations, and breaking and entering scenes come to mind as good examples of where this will be useful.  First, some background, then my idea.

Most RPGs have game time divided into structured and unstructured time.  Structured time is normally divided into “rounds” that allow everyone to take a turn.  In D&D and derivatives, this has two notable effects: ensuring participation from all players and budgeting time between players.  The former is seen when a player only participates in the game on their turn.  The latter is seen when a player insists on being the limelight, and only willingly steps aside to let other people take their turns.

If you’ve played GURPs, you know that the system has a wide variety of skill challenges (I think they’re called Complex Skill Checks) that are done by one person.  The key feature is a certain number of required successes using a single skill, and only one check is allowed within a certain time frame.  This is structured, and additional participation from other players is optional and usually provides bonuses.  This can be contrasted with 4e Skill Challenges, which us rounds exactly the same as combat uses rounds.  This makes it effectively mandatory for all players to participate, and it is less open ended in how the players contribute as it is limited to the skill list.  GURPS has lots of rules for complex skills but not a lot of rules for dividing up time and attention at the game table.  4e has very minimal skill rules, and uses rounds.  GURPS doesn’t encourage full participation from all the players.  4e doesn’t encourage any creativity or complexity from the players.

So, I want to encourage full participation, without diminishing the players’ or GM’s creativity.  To accomplish the prior, I think I’ll use a reward method.  To accomplish the latter, I want to break “complex” activities into discrete objectives.

The reward method is actually very simple: XP will always be divided amongst all players, regardless of participation.  If all players participate, the total XP is greater.  For example, for overcoming an obstacle players may get XP as though they defeated a CR 2 opponent.  If everyone did at least one thing, they all get XP as though they defeated a CR 3 opponent.  Players need to at least ask if anyone else would like to go after the completion of each turn, to provide other people the opportunity to take a turn.

Complex activities will be broken down.  Imagine a skill challenge where the players needs a target number of successes, in the vein of  ” Accrue 5 successes before 2 failures, using diplomacy, to convince the king to send in the army.”   Instead of having a target number, create a number of discrete problems to overcome.  In the case of the King and his army, create 5 reasons the king will object to the request.  Of course, the players might come up with some very persuasive arguments of their own, and reasons for the King to send in the army (instead of reasons why the objections to sending in the army are insignificant).  This makes skill related challenges a lot of work for the GM.  Especially if the players start getting creative (as they will tend to do if given the opportunity) the GM will have to improvise a lot.  However, this means the GM and the players are all being creative, using in-universe logic.

A negative side effect of this is that skill challenges will be shorter (in terms of number of checks), but they will also probably take up more time at the table.  This additional time will be spent on imagining, problem solving, and roleplaying, so I consider it a good trade.  It ought to enable more story to happen in less time, while ensuring the story developments are part of the game play (not just tacked on fluff).

For pathfinder or GURPS, the main challenge to including this is to make sure that all the PCs are actually are able to participate.  A fighter with 8 charisma might not be able to contribute to a diplomatic encounter very well.  The same applies to an average IQ barbarian helping out a superscientist from the future in GURPS.  Overcoming this will likely require house rules or some very creative scene construction.  For 4e the difficulty is overcoming the incredibly specific skills and power rules, to enable open-ended contributions.  I’m not sure how to begin with accomplishing that.

Analysis of Mechanics in Pathfinder

There are three broad mechanics in pathfinder that allow players to interact with the world.  How they intersect and balance between each other is part of what makes different classes “feel” different to play.  Also, a good understanding of how these different mechanics impact “feel” really broadens the kind of gameplay (and thus genres) that Pathfinder can be used for.

The three mechanics:
1.  Rolls that Use Ability Score.  This is especially skill checks, but I suppose also includes attack rolls.  I do not include saving throws in this group for most practical purposes, but you could.
2.  Spells.   Since spellcasters only have so many spells per day, a “minigame” of ressource conservation comes into play with spells.  Since many spells can accomplish things without any roll of any kind, the limiting factor is spells known.  This makes gameplay (even for sorcerers and other spontaneous casters) a much more planned out, less spontaneous affair than for classes that rely on ability score related checks.
3.  Hit Points and Damage.  These point tracking systems are how success and failure can be measured.  Once HP reaches 0 (maybe -10), that side has lost.  It also impacts objects.  I’m of the opinion that since Saving Throws are normally used passively (ie. the GM does something that forces a character to use them, instead of the character initiating their use), a character who has really high saving throws ends up feeling like a character who has high HP.

The ways to balance these three mechanics between each other are difficult.  They are very well balanced for combat, but the rules for any other kind of scene that incorporates all three are scant.  Here’s an example for stealth obstacles.

An issue with stealth obstacles in pathfinder is that often only the classes that have lots of skills can participate.  To address this, I will find ways of incorporating spells and damage into stealth obstacles.

Since gameplay with spells incorporate resource management, there has to be a reason to conserve spells.  As such, the obstacle the players are overcoming must be long enough and complex enough that the players don’t want to use all their magic at the very beginning.  It also should include checks of varying difficulty, so that there can be optimal and suboptimal times to use a spell.

Classes that have a lot of hit points also tend to be able to do the most damage.  Playing those classes is often about dependability.  As such, if a player attempts to break down a door, bust down a wall, break the floor to go down a level, or any other incredibly destructive act during a stealth scene, it’s okay if it takes a longer amount of time than if someone else used a skill or spell.  Object’s hardness will make this occur often.

In order for dependability and ressource management to be relevant, the stealth obstacle will require structured time.  Something similar to combat rounds would be acceptable, although there might be more fun options out there.

Finally, there’s the problem of failure in stealth scenes.  In most games, a failure to be sneaky results in combat.  In most structured scenes, failure results in loss of hit points.  Hit points are not a very sensible choice for failing a skill check to sneak past someone.  What makes sense is being seen.  This has narrative consequences, not mechanical consequences.  Bridging this gap is difficult, and worth much more thought than it would get in the last paragraph of a post about something else.