Thursday, 23 August 2012

Killing Me Softly, part one

I've written and re-written this post about ten times now, rearranging and rephrasing all kinds of things, to the point where I can't even evaluate it any more. So I'm just going to shove it out and hope.

I was planning to work on the next part of Trappery, but a recent game of Deathwatch inspired me with another musable thought and I want to strike while the iron’s hot.

Assault Marine by me, based on this picture.

I played a calculating Assault Marine, tooled up with kit to keep tactical options open, including photon flash grenades, which are used to blind enemies. Those sorts of soft attacks have always appealed to me; I like esoteric bits of kit and options that aren’t just “cause damage” but create an advantage in other ways. As things worked out, I readied a flash grenade several times, but never ended up using it. Quite often, it was because I didn’t think they’d be mechanically effective, rather than because I didn’t want to, and that’s what I want to muse about today.

Soft attacks

Most games with combat feature some kind of non-damaging attack. In my case it was flash grenades, but you might find sneezing powder, hallucinogens, paralysing charms, freeze rays, spiderwebs or slow spells. Instead of causing damage, they penalise the targets, making it difficult for them to harm you or defend themselves, and so making the rest of the fight easier. I’ll refer to these sorts of attacks as soft attacks in the rest of this post, and I’ll refer to stunning, blinding and other such non-damage penalties as status effects. In contrast, hard attacks are those that deal actual damage. Also, I’ll use level to indicate any kind of progressive scale of power from ‘puny inexperienced peasant’ to ‘godlike veteran adventurer’, and the corresponding NPCs and monsters.

Sweeping generalisations ahoy!

Obviously, there’s massive differences between games and I can only make very broad statements based on the arbitrary subset of games I’m thinking about. Some don’t have soft attacks, some may not play out in the way I’m discussing here. They’re not really germane right now.

Soft attacks are a useful and effective addition to simple hack-and-slash combat. They can lock down dangerous targets, disrupt enemy tactics, or help eliminate key threats much faster than brute force alone would allow. In practice, they tend to be either highly effective or completely ineffective, without much middle ground. That’s because soft attacks are mostly a binary option: either they affect the target fully, or they have no effect at all. Soft attacks are quite swingy to use compared with the variable-but-reliable damage inflicted by hard attacks. One is the flashy, dashing and photogenic cavalier; the other is a solid, unassuming middle-aged guard who doesn’t even get a mention in the credits.

From Complete Adventurer and Players' Handbook 3.5 art galleries (C) Wizards of the Coast.

This is mostly down to differences in the mechanics for the two attack types, which I’ll get round to looking at later on. One feature you tend to find is that the more powerful an enemy is, the swingier soft attacks are. This mostly comes down to a combination of the way resilience is modelled, the invariant nature of status effects, and the way powerful enemies are used in games. Against very powerful enemies, soft attacks can become mechanically useless, or deeply suboptimal, but this mechanical reality doesn’t always match up to what logic or drama would suggest. Sometimes the cause is flat-out immunity, and sometimes it’s just very high resilience to soft attacks.

Immunities in general can be interesting and characterful. Eyeless cave-drakes can’t be blinded, robots don’t need to breathe, and water elementals can flow through entangling vines. These make narrative sense, shake up tactics a bit and avoid things getting too predictable. In other cases, immunities seem arbitrary, and largely designed to stop key battles falling flat through skilled or lucky use of soft attacks. ‘Bosses’ in many games (especially computer games) are immune to stunning, sleep, paralysis, nausea, and a whole list of other effects, either through special traits, equipment or magic, and this often seems designed to make the fight pan out as the designers intended.

While immunity to soft attacks is usually deliberate, near-immunity can be a side-effect of creating powerful opponents, an artefact of the way the game models creatures. Resistance to soft attacks often scales with the level of the creature, while above-average statistics also bump up resistance. In Pathfinder, for example, Will saves are based on the creature’s level and its Wisdom statistic; since powerful monsters tend to have high levels and high stats, their Will saves tend to be higher than those of weaker monsters, even when both are intended to challenge the same adventurers. This is part of being ‘more powerful’, in the same way that hit points tend to increase. Resistances based on physical toughness are usually the most strongly affected, since lone powerful opponents tend to be extremely tough (so they last long enough in a fight) even if their other scores are not particularly high.

Monster Power!

It might be useful to clarify what I’m talking about here. Most games (excepting things like Call of Cthulhu) have levels of monster power that correspond roughly to the level of the PCs, so that it’s easy to work out what monsters are suitable opponents at any stage of the PCs’ careers. However, within that band of ‘level-appropriate opponents’ there’s a range of monsters of varying power. Generally speaking, this means there are weak monsters who appear in hordes, middling monsters who appear in groups about equal to the PCs, and powerful monsters who appear individually. Though the overall difficulty of the fight is about the same, one powerful monster is a much more dangerous opponent than one weak monster. This power typically comes from being relatively high-level, but may just be down to higher stats or better equipment. Often it’s a mixture of these three.


As I see it, powerful enemies tend to be very resistant to soft attacks, and this can make the game less convincing or satisfying. More specifically, powerful enemies tend to be disproportionately resistant to soft attacks, if not completely immune. This is sometimes a deliberate design decision to stop a game falling flat when the powerful central villain of the piece fails a saving throw and is frozen solid before it gets a single shot off. At other times, it’s just a consequence of the way the system models resilience.

Case Study: Battling Broodlords

Let’s returning to our Deathwatch game for a minute. Towards the end of the game, we faced off against a genestealer broodlord and its kin. I had a flash grenade ready to throw as soon as any enemies appeared, but by the time it burst out of cover, everything else had been mown down in a hail of bullets and flame, leaving only the broodlord. Quick calculation time. Assuming a successful hit, I’d have to overcome the broodlord’s Toughness, which would be high. Even worse, while I didn’t know the Deathwatch system at all, general experience suggested such a key enemy might be entirely immune to the attack. I also vaguely remember from older editions of tabletop 40K that Tyranids were highly resistant to many soft attacks. So realistically, there was maybe a 10% chance of a successful hit actually affecting the Broodlord and hampering it for a short time. Using a grenade, I’d also only be able to make a single attack.

Alternatively, I could just charge it; I’d get two or three chainsword attacks at 60%, so at least one should hit and inflict some damage even through its armour. There wasn’t really much of a choice. I’d have to spend five rounds grenading the broodlord to have even odds of blinding it, and there’s a real chance it might be unblindable anyway. In five rounds of conventional attacks, I could expect to land at least five hits and inflict maybe 30 wounds, helping my battle-brothers to finish it off faster and with less injury to ourselves. There was really no comparison.

It might just be me, but this is not how I imagine special attacks working. Heroes don’t stand around methodically lobbing grenade after grenade at the beast ripping through their comrades, hoping to get lucky and blind it; they hurl one to get a quick advantage as they charge in, sending it reeling long enough for someone to land a crucial blow.

Because powerful enemies are so difficult to affect with soft attacks, you end up not even trying, because you expect conventional attacks to be more reliable and often more effective. Many soft attacks provide a relatively small benefit to the attacker, such as a small penalty on the target’s dice rolls. This is a worthwhile tradeoff against a couple of hard attacks, but rarely more. Trying out an unreliable tactic against ordinary opponents is one thing, because you’re not taking too much risk; against something very powerful, you want to aim for maximum efficiency because every round they fight is a serious threat, so unless the unreliable tactic will be devastating if it succeeds, it’s not worth the trade-off. It’s particularly not worth trying any tactic that is both unreliable and less effective than hitting things with your sword.

The logical outcome here is that against a broodlord, you don’t bother with realistic use of soft attacks. You either rely on hard attacks, or you spam very powerful soft attacks; although the chances to succeed are slim, if you can beat that 10% chance to paralyse it, the effect will be devastating.

For one thing, this is a bit boring. For another, it's slightly disheartening if you've envisioned your character as a tactical or tricksy sort, and you're either ineffective in major fights, or have to resort to fighting on someone else’s terms. For a third, it goes against the grain of realism, because fights against dangerous opponents are exactly the ones where you should be turning to tactical combat and indirect options to gain every possible advantage. You might go toe-to-toe with a wolf, but if you’re hunting rogue elephants nobody says “no point trying tranquilisers or poison on that tough old brute... I’ll just have to face it mano a mano, pass me a knife”.

Analysing the problem

So as I see it, the problem is that it’s mechanically difficult, and often suboptimal, to use soft attacks against relatively powerful creatures; and that this reduces tactical options and makes the game less interesting. I’m interested in seeing whether this is true, why, and (if so) whether anything could be done to redress the balance, without allowing soft attacks to become too powerful: as I’ll explain later, this is a real danger.

The Sword and the Stun: hard and soft attacks

In many cases, there are significant differences between the ways soft and hard attacks work, which affect the way they’re used. I think the main ones are the mechanics for making the attack, the way the effects of an attack are modelled, the way progression is modelled, and the way resilience is modelled. I’ll be generalising hugely from now on, and your pet system may work differently; for example, not all games involve variable damage from attacks.

Dice (typical role playing game dice)


Both hard and soft attacks usually call for an attack roll to see if the attack is on target, and some systems allow the defender to try and dodge or parry a successful hit. At this point, the mechanics diverge.

Hard attacks tend to go like this:

  1. attack roll
  2. (optional) dodge or parry roll
  3. roll for damage (damage roll)
  4. if the damage inflicted exceeds any damage mitigation*, the defender suffers the difference as damage

*things like Deathwatch Armour or Dungeons and Dragons Damage Reduction or Energy Resistance.

Soft attacks often go like this:

  1. attack roll
  2. (optional) dodge or parry roll
  3. roll to see if the attack is effective (resistance roll)
  4. if the attacker won, the defender suffers a fixed penalty

Unsurprisingly, these different systems produce different results. I’ve knocked out a diagram below, with the width of columns broadly corresponding to probability.

Fifty percent extra unlikely

In the hard attack, there are two pass/fail rolls (attack and dodge) that can negate the attack. The third roll establishes the effect of the attack, which varies from low damage to high damage. In the soft attack, there are three pass/fail rolls (attack, dodge, and resistance) that can negate the attack. The effect of the attack does not vary: it’s a fixed status effect (though see “Tangent: duration rolls” below).

If hard and soft attacks use the same attack and dodge rolls, which is typical, the extra pass/fail roll for resistance means that soft attacks are statistically more likely to be negated. Assuming a simple system and two identical opponents, each of the rolls may be 50%, which means a hard attack has a 25% chance of inflicting a variable amount of harm, while a soft attack has a 12.5% chance of inflicting a fixed penalty. In reality things are rarely so simple, but you get the idea.

This statistical disparity isn’t necessarily a problem. It depends on the relative effectiveness of hard and soft attacks, and the interaction between this and other factors.

Effects: there is no ‘try’ in ‘soft’

While both attacks involve the same number of rolls, the hard attack resolves effectiveness through a variable damage roll. In contrast, the soft attack resolves it through a binary resistance roll. A soft attack either works or it doesn’t, whereas a hard attack can be more or less damaging to the target. The lasrifle inflicts between 1 and 10 points of damage to the target, while the taser stuns it for two rounds if it fails its resilience roll.

Tangent: duration rolls

In reality, soft attacks quite often add a variable dimension by having a roll for (typically) the duration of an effect. The strength of the effect doesn’t vary, though. In these cases, the soft attacks require more rolls than a hard one, and are still more likely to fail outright, as well as allowing the possibility of a low duration roll.

This point attracts my attention most, because it runs contrary to reality. Things like being dazed, blinded, tranquilised or weighed down don’t have an arbitrary threshold. A punch that sends most people reeling may only make a boxer a little dizzy, and give a world champion momentary pause. A tranquiliser that knocks out a dog will leave a horse weak and lethargic, even if it doesn’t bother an elephant. Thick underbrush is a major obstacle to a toddler, a tiring inconvenience to you, and irrelevant to a gorilla. If I’m going to look at changing anything, this right here seems like a good place to start.

Progression: Bigger, better, faster, stronger

Where games include progression, it usually involves a system for increasing the effectiveness of attacks.

Attack rolls tend to scale with your level, but so do dodge rolls (or Armour Class, or whatever else determines the difficulty of landing a blow). Broadly speaking, hitting enemies of your own level is usually about the same difficulty, regardless of the actual level involved. This progression affects soft and hard attacks equally, and so is mostly irrelevant here.

Damage from hard attacks also tends to scale: low-level attacks do less damage than high-level ones. This usually balances with the hit points of level-appropriate enemies, which also scale, so it takes about the same number of combat rounds to kill an opponent regardless of your level. Its main significance is that any given hard attack becomes less effective as the level of the enemy increases, because the damage inflicted is a smaller proportion of their total. However, a successful hit still generally inflicts some damage and contributes to your victory. Thus, the value of a successful hit depends on the relative level of attacker and defender. Stabbing a rat with a dagger is a lot more effective than stabbing a blue whale, but even a whale will drop eventually.

In contrast, the effect of soft attacks is rarely level-dependent (though duration may be influenced by level). Instead, the attacker’s and defender’s levels are reflected in the resistance roll, with a low-level attacker being less likely to win. This usually balances with the resilience of level-appropriate enemies, so the threshold for a resistance roll against an appropriate opponent is about the same regardless of your level. Its main significance is that any given soft attack becomes less effective as the level of the enemy increases, because it will probably be negated by the resistance roll. However, an attack that does overcome resistance is exactly as effective against all enemies, regardless of level. Asleep is asleep, frozen is frozen, blind is blind, be you goblin or titan.

This is very important, because it means that the effectiveness of soft attacks doesn’t depend on the level of the attack. All that matters is winning the resistance roll.

Tangent: high-level enemies

In practice, high-level enemies in some systems have abilities that mitigate the effect of soft attacks, most notably dispel-type powers that help them to remove status effects, which are not generally available to low-level enemies. They may have special senses that limit the effect of blindness, or teleporting powers that make having your shoes tied together less of a problem. So it’s not necessarily “exactly as effective”, but it’s close. High-level enemies may also have healing abilities that mitigate the effects of hard attacks, so hey.

Resilience: Was that supposed to hurt?

Because hard attacks affects hit points while soft attacks inflict penalties, and also because one is variable and the other is binary, they have different ways of accounting for the toughness of the target. Essentially, hard resilience is modelled by reducing the relative effectiveness of a successful attack. In contrast, soft resilience is modelled by reducing the chance of an attack succeeding at all.

Creatures that are entirely immune to hit point damage are vanishingly rare, while effects that mitigate or heal damage are fairly common. Creatures that are immune to specific or all soft attacks are fairly common, while effects that mitigate status effects without entirely negating them are rare.

Again, systems vary. Traveller, Deathwatch and D&D all have damage-reducing defences that can reduce damage down to zero. What this tends to mean is that relatively low-level attacks are ineffective, level-appropriate attacks almost always inflict some damage, and high-level attacks always inflict some damage. It renders creatures immune to the least powerful attacks, while otherwise acting more or less like extra hit points. This does not, in my view, significantly alter the balance between soft and hard attacks.

So what?

Taken together, these things mean that soft and normal attacks work in significantly different ways.

  • A hard attack causes a variable amount of damage. Unusually resilient enemies take relatively less damage. The damage is less effective against higher-level enemies. It is more effective against several weak enemies than one strong enemy, even if their total hit points are equal, because killing enemies prevents them from attacking.
  • A soft attack either works or it doesn’t, and against unusually resilient opponents it is less likely to work. The status effects it imposes affect enemies of any level about equally. It is more effective against one strong enemy than several weak ones, even if their total hit points are equal, because a higher proportion of the enemies are penalised.

Why are those designers so mean to soft attacks?

At this point, you may be thinking: “But Shimmin, that’s not true. Low-level soft attacks aren’t effective against high-level creatures. You already discussed that!” Absolutely. Designers understandably limit the power of soft attacks to deal with these problems. You can’t just stun whatever you like; you have to see if it’s tough enough to resist you, typically with a resistance roll. If the defender wins, then it shrugs the attack off completely. It’s not stunned, just a little disoriented. It’s not asleep, just drowsy. It’s not blind, its eyes are just watering a bit.

Very occasionally, soft attacks impose a lesser version of the status effect if the resistance roll is successful; being dazed instead of stunned in Pathfinder, for example. This seems like a promising line of investigation.

Soft attacks complicate games in various ways, which I suspect is one of the reasons that immunities appear, and also why it’s okay if soft attacks are less likely to succeed. Here are some of them.

  • Soft attacks are particularly dangerous to powerful enemies, as they tend to work alone or in smaller groups. A lone giant that’s frozen in place can quickly be hacked to death, tied up, or showered with other soft attacks until it’s out of the fight for good. In contrast, freezing one of fifty goblins hasn’t helped much.
  • Soft attacks also make power-scaling awkward. Typically, you start off with rusty spoons fighting diseased goblins, and work your way up the hero ladder until you’re battling greater demons with a flaming omniversal railgun that fires angels’ tears and was carved from the fang of the Dragon Queen. As you progress, your early weapons become less and less effective against the enemies you’re facing; there’s no point using that rusty spoon that causes 1d2 damage any more, because it’ll take you a hundred rounds and four real-life hours to kill one opponent. However, as I said above, an attack that stuns one opponent for one round is just as useful against one greater demon as it is against one kobold. If there was no resistance roll, this would allow low-level soft attacks to wreak havoc on high-level enemies, which disrupts the game balance system. Low-level characters who could never defeat a demon by main strength are able to take it down through soft attacks. The balance of the game becomes skewed towards soft attacks.
  • In any game with major physical differences between opponents, realism is also affected: thick-skinned Greater Xrgax Beasts are much more resistant to laser fire and bullets than Lesser Xrgax Pups, but just as easy to knock out with an uppercut..?
  • Low-level attacks are typically more readily available or cheaper to buy than high-level ones. This works for hit-point weapons, whose damage scales readily with level, but not for soft weapons. Without scaling effectiveness, there’d be no reason to buy an expensive 100th-level stun grenade instead of a hundred cheap 1st-level ones. The balance becomes skewed in favour of buying a hundred stun grenades instead of a railgun.

What to do about it?

As I see it, the main problem is that most systems treat status effects as binary. You’re either blind, or you’re not. This makes the difference between “effective” and “ineffective” much greater than for a weapon that rolls low or high damage. A poor sword attack against a powerful enemy might only inflict 1 point of damage, but that's still chipping away at them. In contrast, a poor stun attack achieves nothing, while a good one grants a substantial advantage.

The resistance roll allows creatures to soak up soft damage, just as hit points allow them to soak up hard damage. The difference is that hit points are slowly whittled away by even the puniest attack, while soft attacks have no cumulative effect. It takes twenty hits to whittle down the ogre’s health and force it to surrender, and those hits build on each other in a predictable way. Any individual sleeping charm has a 5% chance of dropping it immediately, and it’s entirely random whether that’s the first, the twentieth or the one-thousandth charm you use.

What I’m wondering is if you could usefully model status effects as a range, which would allow for more flexibility in their impact on targets, making them less of an all-or-nothing proposition. I’d like to make soft attacks, not better overall, but easier to use in a fun and useful way. This might well require a system built around that aim, rather than a superficial tweak. In a future post, I’ll try and see what possibilities there are, and whether it would really be feasible at all.


  1. To be fair to /Deathwatch/ I think we might have been overestimating the game's unwillingness to let Soft Attacks work - I've just checked and Tyranids don't get any special protection from Blinding attacks, so the grenade might have had about a 50% chance of working, although that's still a 50% chance of doing nothing, which isn't great.

    1. Personally I might have been inclined to reduce the Broodlord's blindness penalties if there were other 'stealers present to act as its eyes and years. (Hive mind's gotta count for something, right?) But you had in fact killed every one of its underlings before you got to it.

      With respect to soft attacks in general, I think a) you overestimate how much blinding a subset of a heap of lesser enemies can help (a quarter fewer attack rolls against the PCs next round can be the difference between life and death). I also think you're not necessarily taking into account party dynamics; in some systems it might be worth 1 player giving up a hard attack in a round in order to do a soft attack, so that the next round all the PCs have a substantially higher chance of hitting with hard attacks. (Likewise, if you have 4 people doing actions then having one person occasionally do a soft attack still means your foes have 3 hard attacks coming their way this round, whereas if you go with a 1-on-1 model where the choice is either 100% hard or 100% soft then it's different.

      But I do agree that lots of systems sell them a bit short.

    2. (Above reply was to Shim, not Dan.)

    3. With respect to soft attacks in general, I think a) you overestimate how much blinding a subset of a heap of lesser enemies can help (a quarter fewer attack rolls against the PCs next round can be the difference between life and death).

      I think that depends a lot on the way your soft and hard attacks actually work. If I'm fighting six genestealers, and my options are to lob a Flash Grenade or to melee one of them, the flash grenade is probably a better option because it should take at least half of the 'stealers out of action. On the other hand if my choice is to lob a flash grenade or a *frag* grenade the choice gets a lot trickier, because I might outright kill as many Genestealers as I would have blinded previously, and the other ones will be injured instead of totally unharmed.

      I also think you're not necessarily taking into account party dynamics; in some systems it might be worth 1 player giving up a hard attack in a round in order to do a soft attack, so that the next round all the PCs have a substantially higher chance of hitting with hard attacks.

      This is very true but I think the problem, as Shim implies in his post, is that uncertainties stack exponentially. If I want to make a Soft Attack then not only is there a chance I'll just miss, but there's also a chance that the enemy will save and there's *also* a chance that the GM or the scenario designer will decide to make the enemy immune.

      If we simplify everything into 50/50 chances that makes the average Soft Attack have a 50% chance of missing, a 50% chance of being saved against if it hits, and a 50% chance of being hosed by GM/designer fiat (or if we want to be really paranoid a 50% chance of being hosed by designer fiat and a 50% chance of being hosed by GM fiat even if it isn't hosed by designer fiat) all of which adds up to an attack with has about a 90% chance of doing nothing.

    4. Yes, step one in alleviating this is for GMs to commit to not fiating away soft attacks (and to work behind the screen to scrape away designer fiat).

      Then again, to take Deathwatch as an example you can run up against foes where under normal circumstances your weapons have a 90%-100% chance of not actually doing anything in hard attacks, in which case the best thing really is to select an appropriate soft attack and then coup de grace the thing whilst it's helpless. But both designers and GMs don't really think about that much so I agree that it is a problem.

    5. Wow, debate...

      @Dan: I think your credence argument is pretty much on the money. I'm a bit surprised that 'stealers have lost that protection but it seems sensible.

      @Arthur: did you mean "underestimate" in your first reply? You're entirely right as far as blind grenades go, because they have a blast radius so affect more targets than a chainsword. However, for simplicity I was assuming a like-for-like comparison with a single-target attack. Blast attacks get complicated very quickly - on photon vs. frag, blinding two of eight 'stealers is arguably better than wounding eight of eight, but potentially killing a few termagants and wounding others is probably better than blinding.

      Similarly, I think the party dynamics angle is very sound, but depends somewhat on credence. A soft attack that's very likely to help the rest of the party is worth it, whereas it's harder to justify taking a long-shot if you could reliably inflict some damage against a dangerous opponent. If, for example, we'd snuck up on the Broodlord, I'd say an unexpected flash grenade is absolutely worth it over a bolter shell, for the chance of giving you an early advantage, but once it's charging at you you want "certainty". Does that make sense?

      Basically my take on soft attacks is that they're sometimes really valuable (particularly blasts against large numbers of fairly susceptible enemies) and sometimes hard to justify the time (single tough enemies). I don't think they're underpowered as such, in fact in some cases - D&D CRPGs spring to mind - you can cakewalk large sections of the game if you want.

    6. Roughly what I'm hoping to look at is whether there's a way to model soft damage so that:
      * the distinction on when to use soft attacks is less "you're a fool not to" vs. "don't waste your time" and they are a generally useful tool in most situations
      * soft damage does not disproportionately affect more powerful creatures and cause game balance problems that designers feel obliged to defend against
      * the effect of soft attacks is less swingy, so less inclined to make them fall flat in either direction