In the previous segment, Never Tell Me the Odds 1 and Never Tell Me the Odds 2, the basic combat mechanics for five RPG systems were examined with the idea that if you don’t know the rough probabilities of doing what you want, you can’t really evaluate what you can do. Using the power of tautology for good rather than evil, perhaps.
Probabilities are one thing, but the basic decision process to resolve the success or failure of any course of action, the actions you can take productively, are strongly influenced by how the system decides to resolve tasks.
The basic steps of resolving any effort break down into three concepts: Action, Opposition, and Effect. Given the violent topics covered here, that’s usually something like “I bash him in the face,” “he tries to duck but fails,” and “I break his nose” if you come at it from a descriptive point of view. From a more gamist sense, it’s often something like “I roll to hit,” “he tries to dodge and fails,” and “I roll 1d6-1 for damage.”
Some systems merge Action, Opposition, and Effect together in various ways; others let the tests be basically independent. How this is done drives the decisions available to you, and dictates when it’s critical to make them.
Agent of Action
The term “player agency” will be thrown about quite a lot in this column. There’s a bit of definitional pornography involved (“I know it when I see it”), but basically I’ll steal a good concise definition inspired by Papers and Pencils: Agency describes the ability of a player to make a choice that will meaningfully impact the outcome of the event in question.
In this particular case, combat mechanics, the attacker nearly always has agency. He can choose his target, perhaps employ one of a variety of weapons, and decide what special perks, powers, or abilities to bring to bear depending on his character.
But to the “I know it when I see it” comment: Agency is provided to the one who makes meaningful choices. If as a defender, you just sit there and take it, you have no agency for that action at that time. You will likely face the opposite side of that coin if you live to your next turn: you get agency, while your opponent gets to take one for the team.
In each section highlighting one of the options presented below, there will be a bit of analysis titled Agency Enabled. This will discuss the kinds of choices for which the mechanic provides support. Those choices, when they happen, and how, will strongly influence – if not dictate – how the system provides flavor to combat in the game.
Looking at the possibilities of Action, Opposition, and Effect, we can create options from combinations of the three. Some of them are easy to see as primary combat resolution flavors, some less so. Ultimately, though, either all are combined, all are independent, or two are combined together, with the third independent. There are some variations in what that means, and I’ll take the interpretation that’s the most pertinent.
I’ll use some cheesy symbolic representation to show my work: A+O represents a combined Action and Opposition roll, while A, O shows two rolls done separately. An illustrative example, where possible pulled from D&D as a common reference, is provided for flavor. In many cases, other options are available (my current D&D5 GM has us roll damage against doors, for example).
- A+O+E: Action, Opposition, and Effect rolls are all done as a combination subsumed into one trial. Many binary tasks are this way. In Swords and Wizardry, with STR 17, you open doors by rolling 1-4 on 1d6.
- A, O, E: Action, Opposition, and Effect rolls are totally independent, and executed in three trials. Make an attack roll, if that succeeds, the defender must roll to avoid the attack. If that fails, roll damage. GURPS standard combat resolution mechanic.
- A+O, E: Action and Opposition are combined into one roll, and the Effect roll is separate. The standard D&D to-hit and damage roll: Roll vs. Armor Class to hit, but damage is a separate trial.
- A, O; E=A+O: Action and Opposition rolls are separate, and Effect is determined from the balance between the two. Another way to look at this is that Effect is combined with both Action and Opposition. A grappling attempt in D&D5, or any Exchange in Fate.
- A+E, O: Action and Effect are determined in one trial, and Opposition in a second. Soaking a damage roll, or a “Save or Die!” trial after being bitten by a Giant Centipede.
- A, O+E: Action is an independent trial, but Opposition and Effect are combined into one roll. Damage reduction saves, where once an attack is made, the victim must pass a trial or take damage.
One Roll (A+O+E)
The simplest method, and one that reaches back into the wargaming roots of RPGs, is the simple binary test. If you’re playing a game of tank battles, you might target an opposing APC, roll 1d6, and expect to hit and destroy the opposing unit on (say) a roll of 4-6. The miniatures boardgame Ogre, from SJG, takes another simple take on this, with a single die roll (again 1d6) allowing either no effect (for example, roll 1-3), a “disable,” (roll 4-5) or a destroyed unit (only on a 6). That is, the Action roll (a 1d6) provides a spectrum of effects directly. There is no opposition roll in this case; the agency for opposition is to maneuver units to not be in range.
The most common all-in-one test in roleplaying is likely a test of skill. Whether it’s picking a lock in GURPS (roll vs. Lockpicking at -3), or kicking down a door in old-school D&D (Open doors on 1-4 on 1d6), the action is the die roll, any opposition is simply a modifier to the roll, and the effect is going to be binary success or failure.
“From a certain point of view” the basic Fate success test for combat, which rolls 4dF for both attack and defense, is really just splitting up an 8dF randomization pool into even chunks. (And remember when Obi-Wan says ‘from a certain point of view,’ it’s in your best interests to find out what the other point of view is before you start passionately kissing your sister.) The probability distribution does not change even if you give the attacker 7dF or 2dF, and the defender the balance (1dF or 6dF). The agency that the players can bring to bear is whether or not to spend Fate points to invoke an aspect, or choose to go full defensive, or other options that change the mean effect, but not the randomization of the rolls.
That being said, this is likely a poor way to look at it, because while the basis of Fate at its Core might devolve to this, the infinite number of ways that the GM and players are encouraged to muck about under the hood with Stunts, alternate dice, and interesting mechanics means that very quickly it can be meaningful that the dice are divided up between attacker and defender, and vital that they be equal.
Night’s Black Agents invokes an unusual form of One Roll to Rule Them All in its Full Contest mechanic. The participants all take turns making a test of the ability in question, spending from their General Skill pools as they like. The first person to fail a test loses the contest. Each individual trial is Action, the Opposition is a target number, and the effect is “you lose,” but only if you fail. If you succeed, you narrate what happened, but in a way that holds out the scene for someone else to pick up. If your foe also succeeds, she tells what happens next, but again, it’s not fully resolved. This passes around until someone biffs a roll. The target number is typically 4 on 1d6, so it’s 50-50 if you don’t spend from your skill pool. Your foe is making the same calculus. The only agency denied you is the ability to narrate the end of the contest until someone fails.
This particular mechanic finds a home in combat where the only real agency is “where do I point my weapon?” More directly, even if the agency and choices are mostly in the hands of the attacker at the moment of decision, there are plenty of ways for the defending or assisting characters to get in on the game. The first, though a bit subversive, is by pre-empting an attack by a judicious use of deferred initiative or something similar to GURPS’ “Wait” maneuver.
“If X happens, I act!” at a time of the defender’s choosing allows for ambushes, despite being in a (theoretical) turn-based game with a one-roll mechanic. Another way to drive agency on the defender’s side is to declare actions on his own turn that raise the effective difficulty of the attack roll. Taking cover, declaring a “full defense,” or a similar option does provide a meaningful choice that can influence the outcome. It’s deferred agency (or predictive, in some cases), but it’s agency nonetheless.
If limited defensive agency is a disadvantage (and that’s an eye of the beholder thing), one advantage that this mechanic has going for it is that it can be satisfyingly fast. The attacker declares her action, she rolls her dice, and that roll also dictates the effects. Boom, next turn. In a game where it takes many such effects to win, either because you get partial victory conditions such as ablative damage or because there are just that many orcs on the board, this kind of alacrity in resolution can ensure that a combat isn’t a priori the only thing going on that evening.
Of course, in the world of unopposed skill checks, in combat or no (“I fast-draw my sword!”), the basic One Roll to Rule Them All test is the go-to mechanic in most cases.
Contested Outcomes (A, O; E=A+O)
If a single roll that determines everything is one mechanic that appears frequently in unopposed skill tests, another that will come up more often in combat is the opposed skill check. In this type of trial, two characters roll against each other, and the victor is determined by a hierarchy of result, but based solely on the rolls of each of the trials for Action and Opposition.
The results can be based on the degree of success, or can be a yes/no matrix, but in this kind of trial, no independent effect roll is made.
GURPS has a similar mechanic – the Quick Contest of Skill – which is used for many things, not the least of which is to resolve certain combat moves, such as grappling takedowns. Each participant rolls a test of her skill, and the margin of success or failure for both is used to determine the effect.
Frequently, this effect is binary – you either take the guy down or you don’t. Sometimes, though – such as for determining the amount of damage inflicted by applying pressure during a joint lock – your effect is equal to the margin of victory on the contest. If you succeed by a lot, and your foe fails, you collect the Chewbacca Prize for impressive dismemberment.
There are two more versions of Contests in GURPS. The Regular Contest is a matrix effect. You only win the Contest if you make your skill trial, but your foe actively fails his. If the reverse happens (you fail, your foe succeeds), you lose. If both succeed or both fail, there’s no effect and the Contest continues. The archetype for this is trying to Pin a foe following a grapple (at least in the Basic Set. Some hack went and wrote an expansion.)
The final type is a more qualified type of Quick Contest, where not only must you win by more than your foe, you must also successfully make the underlying implied skill roll. The implication here is that not only must you do something right, it must also overcome your foe’s resistance. This is seldom used in vanilla GURPS, but is frequently found in the aforementioned grappling rules.
Because the dice are so easy to combine without changing the outcome, it’s easy to lump these two tests together and just call it a generic 8dF. That’s not how it’s going to feel to the players, though, and that’s important. Each is rolling their own set of dice, and the effect is determined by the combination.
Mathematics aside, the Fate resolution system is basically a contest of skill as well. The attacker chooses what Aspects, Stunts, or Boosts he will leverage (providing a flat bonus), the defender does likewise (including having declared a Full Defense on her own turn, for another flat bonus), and each rolls their allotment of 4dF, and the shift delta between the two provides the effect.
The contest mechanic hands agency to both actors in a struggle, and often pits identical – or at least highly related – skills against each other. A warrior wishes to grapple her foe, and must beat her foe’s Dexterity or Athletics with her own Athletics test (D&D5) in order to successfully grapple and incapacitate him. As mentioned, Fate Exchanges are an Action Roll as an Attack, a Defense roll as Opposition, and when the die rolling and point spending is done, the number of shifts between the result dictate the amount of stress or degree of wounding applied.
This works particularly well in games with a narrative or dramatic focus, especially where the turn order and length are somewhat nebulous. It’s not necessarily the right fit for an “I go, you go” turn mechanic in combat, and the finer the turn resolution, the less appropriate it can feel.
In a game like Fate, grappling someone and throwing them to the ground might be Creating an Advantage, and it would be opposed with a defensive action in the same way an attack would. The narrative might be that if you succeed, your foe is thrown to the ground or put into a hold (and the situation aspects granted might give +2 to something you want to do or -2 to a foe’s ability), while if you lose, you tried something and your foe struggles against you in a way that they can choose to have it not happen, or they get a benefit instead (at least, that’s how I’d play it). Continuing the same example, a failed grapple or takedown attempt might just be a no-result, or the defender, who won, might slap his erstwhile attacker with the “Off Balance” aspect.
Even in a detailed game, any time skill can be met directly with skill (as opposed to, say, the half-skill Parry and Block of GURPS), this mechanic is sensible. It’s also very intuitive, and lends itself to fast play and easy interpretation.
To-Hit Roll, Damage Roll (A+O, E)
The archetype for RPG mechanics for applied violence has to be the version of combat trials that has existed since at least I started playing D&D. It was present in the Moldvay Basic Set, and AD&D, it’s definitely still there in both Pathfinder and D&D5. Even where you might have to squint a bit to make it work, such as firearms combat in d20 Modern, the classic method of rolling to hit against a fixed target number, and if successful, rolling an independent damage roll is the standard by which mechanics are judged. One of the complaints lodged, for example, against games that allow active opposition is that you can throw an attack “good enough to hit,” but have that attack negated either by defenses or by armor on later rolls.
As noted, when someone says “roll to hit,” the first question out of most gamers’ mouths is “OK, what’s my foe’s Armor Class?” Most editions of D&D are largely designed such that the highest unmodified (through magic or Feats) Armor Class (AC) is in the neighborhood of 20. In the most recent D&D5, this is nearly exactly true: full plate and a shield will get you AC 20. Lighter armors have lower AC, but this is usually balanced by the ability to receive DEX bonuses. So with the right combination of DX and armor, AC 17 without a shield is within reach (though you may need DEX 20 to claim it!), and AC 18 with full plate. As magical bonuses are more restricted than in prior editions, AC is likely to stay lower than 30 regardless of the level of the character.
Pathfinder is more lavish with bonuses, and full plate and a tower shield, combined with at least DEX 12 for a +1 bonus, can rack up an impressive AC 24 – if you can pay for it. If you can find them, stacking armor, shield, and special protection enhancements (usually on a ring) could potentially boost by another +15, putting the upper range of AC near 40. That’s OK, though, because a high-level Fighter (say, 15th level) will get his first attack at +15 for his level, likely +5 to +10 for Strength bonuses, and if you’re facing a foe with +5 armor, you likely have a +5 sword. So AC 40 can be met with 1d20+25 or 1d20+30, giving effective target numbers of 10-15 on an unmodified d20 roll: 25-50% chance of success when titans clash
Savage Worlds has the defensive actions subsumed into the target number, for the most part as well. You roll an attack against your foe’s Parry score (2+half the Fighting die), and the degree to which you exceed the total gives some benefits. However, there is a separate effect roll (see the next section) and the Raises you get from the attack roll influence, but do not replace, that roll. So that one’s related but not truly combined – and it’s more properly an example of combined Action-Opposition rolls, like D&D. The enhancement is that margin of success on the hit roll matters, with Raises boosting the effect, but not dictating it.
In keeping with its narrative base and simple flair, Night’s Black Agents resolves attacks with a single 1d6 roll vs a target number, while damage is likewise 1d6 plus a (sometimes surprisingly small) modifier – a .50 BMG is 1d6+2, while a 9mm pistol or sword is 1d6+1, and a kick or punch 1d6-2.
GURPS usually uses a three-roll mechanic (see below). However, in battles between low-skill foes, the chances of hitting can be so poor that one must take the “All-Out Attack (Determined)” option to even have a chance, and maybe even stack the “Telegraphic Attack” option on top of that, in order to have a decent chance of landing a blow. These two options give up your own defenses completely, and give your foe a better chance to defend because of how obvious your strike it, respectively. In exchange, you get a monstrously large hit bonus. If both foes are exercising this option (or at least your target is), then the usual sequence of Action-Opposition-Effect degenerates to Action-Effect, much like D&D.
This classic mechanic is a simple step away from its wargaming roots – the addition of a damage roll. The decisions that matter must be made by characters on their turn, be they decisions about fighting defensively, taking cover, movement, or what-have-you. Defensive agency is minimal, and by the time someone’s shooting or stabbing at you, all that matters at that point is the final decision.
Further decision-making power is added, however, by the separation of the hit and damage roll. This encourages a level of detail in the creation and employment of weapons, since the damage statistic is usually a combination of the character’s ability (often a Strength or fighting skill, or both) and a weapon’s properties. As the archetype, D&D takes a weapon-based damage (say, 1d8 for a sword) and adds to it a modifier based on the user’s Strength score. Savage Worlds combines, for melee damage, the attacker’s Strength die and a bonus for a weapon – axes and shortswords add a d6, while great swords and greataxes add a d10.
Night’s Black Agent’s compression of effect rolls into a range of 1d6-2 to 1d6+2 is almost certainly a design choice driven at keeping the awesome power of the spotlight on the players and their characters, not the fictional gear they tote around. Thus, a 9mm with just shy of 600J in its projectile is only slightly less likely to injure a character as a .50BMG sniper bullet, with something like 20x more energy (and a documented history of blowing targets into two pieces in sniper incidents in Iraq and Afghanistan). The hardware is flavor and chrome, not the central point – and a mook is taken out with narrative description and a point spend anyway, so why fuss over energy-to-dice-modifier conversions? It’s not that kind of game.
Set Dice Rolls to 11! (A, O, E)
The most agency in both attack, defense, and damage is obtained by separating the Action, Opposition, and Effect rolls into entirely separate trials. GURPS is the archetype for this, with a hit roll only showing an attack is “good enough” to possibly have hit, but then provides a great number of on-the-spot options for the defender to attempt to negate the attack.
GURPS, Master of Options
Of the five systems used as examples in this column, GURPS is the only one that uses the three-roll mechanic. Each of Action and Opposition has a continuum of options that can be as simple or as detailed as the group would like, but the upper end of the detailed spectrum can push right on past “varied” and clear into “option overload,” especially for new players. There are four attack maneuvers (including those in Martial Arts) ranging from All-Out and Committed Attack, to a plain-vanilla attack, to a Defensive Attack. Each has benefits and drawbacks in movement, hit bonus, and defensive options. There are also options that can be stacked on any maneuver, the most important of which is Deceptive Attack – a voluntary penalty to lower your own skill in exchange for lowering your foe’s defense rolls. This is very important (and all-too frequently overlooked) when high-skill opponents face off.
On the defense, you get full selection of dodging out of the way, parrying a blow with a weapon, or blocking with a shield, if you have one. You may retreat in nearly any direction, or not, and you may also employ the same sort of option on the defense as offense – a Riposte lets you lower your defensive score in order to lower your foe’s defenses on your own next attack – opening him up through fancy bladework of your own. Shields and some weapons (notably the quarterstaff) can provide boosts to the defense roll – armor, in a revision to prior editions, does not.
Once you throw an attack, and your foe fails his defense roll, you roll for damage, but your foe’s armor, if any, subtracts from that. It’s entirely possible to fight a foe encased head-to-toe in steel or sci-fi armor with so much protection he can simply stand there and let himself be hit. Some of the thicker plate armor can behave this way, especially with magical protection thrown in, and if a player is willing to have his character encumbered, plate over mail can provide double-digit Damage Resistance.
While this column will focus on the five demo systems I’ve selected, it would be a mistake to think that only GURPS is the oddity here.
Warhammer Fantasy Roleplay: You can declare a parry against a successful attack. This uses up one of your attacks for next turn, and if successful, subtracts damage from the incoming blow, rather than warding it completely.
Shadowrun (Fourth Edition): The attacker rolls a trial based on his combat ability, scoring a number of hits, while the defender exercises his own options, which subtract potential hits. Any remaining hits get added to a damage pool, which are then further reduced by the defender rolling a trial against the combined total of toughness and body armor.
Runequest (Sixth Edition): The attacker spends an action point to attack, and if the defender chooses to spend an action point, she may parry. If this is not successful, damage is done. RuneQuest makes use of pretty extensive special effect options when superior die roll results are scored. Other editions allowed one Reaction per combat action, so you could parry as many times as you could attack.
There are others, as well, but most place a keen emphasis to tactical and deadly combat as a common thread.
The benefit to all of this is that it allows very fine-grained “verisimilitude,” the seeming of reality. Moves can be broken down almost step by step, and for those that wish it, the fights can be almost perfectly choreographed in reverse.
The down side is option overload. If allowed, each choice can take far too long, and seemingly endless optional detail can be mistaken for required choices.
For the defender, the paradigm that once a turn is taken, she just has to sit there and let blows rain down on her is no longer operational. She can parry, dance backwards, do a grabbing parry to set up a judo throw or knee to the unmentionables, or even launch an Aggressive Parry that if successful injures the foe on his own turn. A “defensive fighter” is now fully possible in this scenario, via positive action at the moment of decision.
A bountiful supply of attack methods, skills, and locations, matched with defensive options and weapon stats to match provides unending flavor to fights, and every choice can matter. The down side is that if every choice matters, the game can bog down as one or more players (or the GM, running twenty foes) hit the wall of decision paralysis.
The last two combinations mostly occur when an attack or other move has a predetermined effect. Either the primary effect is part of the casting of the spell (such as the Fireball), where the declaration of the casting is automatically successful, and the only thing that the attacker does (in Fifth Edition) is to roll the appropriate damage type. The defender can mitigate this damage with a roll. In many cases it’s somewhat arbitrary to define the A, O+E versus its opposite number, since the binary effect is inherent to the attack. Some sort of attack reflection or reversal might qualify, but for the purposes of this discussion, I’m not going to work that hard to separate these fairly closely-linked mechanics.
The saving throw from D&D provides a good example of how muddled the two can be. For attacks, spells, or traps that call for a saving throw, the action on the attacker (or just the inanimate trap) is often just to declare the event, possibly rolling some sort of base damage as the Action. Casting or triggering the trap is done without a skill roll (though this need not be the case), and the defender gets to mitigate usually-known effects with a die roll. The always-popular Save vs. Poison counts here – if an attack succeeds, the defender gets a Saving Throw to avoid instant death from poison. Whether that’s O+E or A+E could be argued either way.
Another common representation of this mechanic, and often used in the same manner for the same things. You see the combination of Opposition and Effect when resisting a poison or some other persistent effect in GURPS. After a poison is administered, the victim must make a series of HT rolls, sometimes at a penalty. Failure brings a known, set consequence. Similar mechanics are used (nearly always a HT roll, but it might be DX as well) if navigating an environmental hazard. Choking, noxious gas (roll vs HT-2 or become Nauseated) or navigating across a slippery surface (roll vs. DX or fall prone) would both qualify as O+E.
The agency is mostly still with the attacker on this one, and the defender’s actions (often providing binary effect results of on/off or full/half) provide an ability to resist an offensive action, but no real meaningful choices. “No, I don’t resist the poison” and “I don’t think I wish to roll my saving throw against that fireball” aren’t really in it – this type of trial is usually representing some sort of autonomic response.
In most cases, while the player may not have true meaningful choices, these mechanics are activated in order to provide some ability for the character abilities to influence the outcome. Whether it’s that high-level characters are less susceptible to certain attacks, or that a certain combatant has purchased an advantage like Resistant to Fire and is being bathed in flame, this mechanic provides something as important as agency: differentiation.
Wrapping it all up, each method of resolving violent conflict could potentially be used within any given game system, and frequently some or all of these are employed in different situations. In D&D5, the standard attack method combines the Action and Opposition rolls, but many other tests are a Contest of skills. There are plenty of binary skill tests in the game as well, if not directly in combat.
Combat Agent Man
Of the types, the most active agency is provided when there are meaningful choices for each trial. Of my five systems, GURPS uses this the most actively in combat, as it provides meaningful and detailed choices for each roll. Combined with the one-second timescale (a topic for another time!), one can get choices in sub-second intervals. Looking at an example longsword exchange in tournament competition, this resolution is needed – each blow and defense (often leading into other attacks) can happen so quickly that the time mesh of one second is just about right for the decisive moment of a fight. “Micro-resolution” of each choice is appropriate and dramatic for this case.
Contests provide the next most visceral feeling of agency. Again, this is related to the fact that the maximum agency is almost always provided to the attacker – it’s “her turn,” after all – while the defender will get a chance to react. Fate provides agency here by allowing more or less the same latitude to the defender as the attacker – if the defender wishes to spend a Fate Point to invoke an aspect, he can. And those spends are important, mechanically and probabilistically. The same applies to Night’s Black Agents, as spending from the General Skills pool is the primary agency you have in the game. Both GURPS and D&D allow choices in what skill is being used to resist some contests, especially in combat.
The most common mechanic – at least by market share – is the rolling of attack against a target number, while effects (usually damage of some sort) are separate. Agency is provided to both the attacker and defender through choices made on their own turn, but if you didn’t choose to fight defensively, there are no takebacks. It would be relatively easy to hack such games, of course. You can declare some sort of last-ditch all-out defense in exchange for loss of agency on your next turn by skipping it or taking some huge penalty to attack (-10 in D&D5, perhaps?). But the existing rules put the choices in your hands only on your own turn. Savage Worlds has the roll against a target directly influence the effect roll, too, through the mechanic of Raises. A particularly good action roll can give bonus dice to the effect roll.
The last three mechanics, the all-in-one skill test (such as GURPS’s use of unopposed binary skill rolls to try and Ready a weapon as a free action) and the combination of opposition and effect into a combined trial are of the least agency. Sure, the combined all-in-one test allows for the actor to choose what he wants, but the amount of variation can be fairly limited. The opposition/effect roll seems to be mostly used for binary yes/no avoidance of something that’s already happening. Poison resistance, saves versus spells, and the like. The player may be rolling dice, but there are no meaningful choices in the heat of the moment – your choices ran out when you got into the situation to begin with (or when the GM cast bolts from the blue at you, if he’s that sort). Hey, it was your choice to delve too deeply, where the fouler things dwell.
So, more agency is good, right?
Not always. The more meaningful choices that can be made in any given exchange (to borrow Fate’s term), the longer each action can take. The longer each action can take in real time, the more the players will feel compelled to get the most out of each one. If analysis paralysis sets in, this can go from tedious to interminable to wake me when the pizza arrives.
That’s no fun, and so a balance is often struck between agency and rapidity of action – especially when introducing new players to the game.
Ultimately, matching the right degree of agency and play speed to what is going on within a given scene, fight, or action sequence doesn’t have to be a fixed thing. Night’s Black Agents notes this explicitly with its Player-Facing Combat mechanic. When the plot dictates it, hapless guards can be taken out with a quick die roll and a small point expenditure (if needed), with basically no opposition required. When things get more serious, either point expenditures and target numbers can be raised, or you can break out the full combat rules. Night’s Black Agents was designed for this sort of narrative-based telescoping; not all systems handle it gracefully, but most can be made to do so.
Removing agency from NPCs usually doesn’t bother anyone, and can be a big help to a potentially overwhelmed GM. Even GURPS, with all its provided options, recommends in the chock-full-of-detail GURPS Martial Arts to remove agency from untrained or low-morale fighters, and recommends a simple 1d6 roll: on 1-3, the “fighter” will All-Out Attack the nearest threat, casting his defenses aside; on a 4-6, he will turtle up, selecting All-Out Defense and using any movement allowance to back off from battle.
Ultimately though, the mechanics selected for a game matter. They provide the default degree of agency to be expected, and increasing or decreasing this from what the players have grown to expect needs to be understood beforehand, either via conversation or (even better) by running trial combats.
Failure to account for this may lead to strife. It could come as easily from loss of agency (“what do you mean the opposing fighter closes thirty feet and brains me with his axe? I have a freaking bow-and-arrow ready to go here!”) as from the perception that the game is only moving at a snails pace (“will you please just make up your mind? Parry the thing and move on!”). As always, the golden rule is to have fun and meet the expectations of the entire group (including the GM).