Saturday, 8 September 2012

Killing Me Softly, part 2

So last post I waffled a bit about non-damaging “soft attacks”. Since then, Dan H has mused a bit more focusedly on the role of actual chance and player beliefs in the use of soft attacks, and on the relationship between soft attacks and health-monitoring systems. They’re good posts, and if you’re at all interested in this topic you should read them, not least because I’ll be deliberately or unconsciously leaning on them from now on.

So what do I want to think about now, in my usual rambling way? Well, last time I said I’d consider whether there are alternative ways to model soft attacks, particularly if they can be usefully modelled as a range in the same way that hard attacks tend to be. Another thing, which has been prompted by Dan’s article, is to consider the point of soft attacks and how we would like them to play out in practice. Since this will affect my views on the first objective, let’s tackle that first. It also ties in to my approach to traps in the other series I sporadically do.

Broadly speaking, I’m inclined to say that soft attacks have mechanical, simulative and narrative origins, or alternatively, you can consider them from those angles.

Soft attacks can have a mechanical role in making battles easier, or sometimes in making them winnable at all. This increases the variety of battles available in the game. They can make it possible to take on very large groups of enemies (by disabling some), very powerful enemies (by reducing their deadliness), very tough enemies (by making them vulnerable) or otherwise deal with some problematic element of a battle. You can sneak into the guarded compound without raising the alarm if your sleeping-gas is successful. You can drug the kraken so it’s lethargic and possible to capture. You can keep up an EMP barrage on the enemy battleship so it can’t get a decent lock on you as you hightail it to the warp gate. From this perspective, we can view the soft attack as a tool that’s part of a tactical puzzle: “how do we succeed at this battle?”, and specifically as a reusable tool that’s part of a fairly standard problem-solving inventory, as opposed to one-off solutions that work for specific battles. As Dan points out in In Summary, point 3 this depends heavily on the genre in question, and the intended tacticality of the combat. Call of Cthulhu doesn’t need these kinds of soft attacks.

Soft attacks can also be seen as a way of modelling plausible options. Players and GMs are creative, and there’s a reasonable chance they’ll try something more complicated than hitting things with glaive-guisarmes. Here, the soft attack exists as a distinct entity from hard attacks because in the context of that game, it doesn’t make sense for it not to work differently. In a game of cunning tricksters, someone will try blinding or deafening or high-gravitating or drugging somebody, and the game needs to reflect the consequences in a way that makes sense to the players. Some soft attacks don’t really fit this view, because they are deliberately added to the game rather than growing organically from the actions of players – spells or unreal tech are common examples that add entirely new soft attacks. Following on from last paragraph, while Dan is right to say from a design point of view that Call of Cthulhu doesn’t need built-in soft attacks, situations can still arise in which it makes sense for an attack to inflict soft damage, partly because it takes a relatively realistic approach. However, for this type of game, it seems best to handle these rare events through ad-hoc judgements (which is how the designers left things) rather than complicate the rules.

Finally, soft attacks can be seen as cool options. It’s cinematic and exciting to dazzle the beast with your torch just as it rears up at you. It’s cool to modify the coffee machine and watch on the CCTV how the whole security team crumple unconscious ten minutes after their break. It’s nice to have options for disabling, restraining or rendering helpless your opponents without having to kill or cripple them, especially in games where you might be dealing with innocents or trying to minimise bloodshed. Similarly, a fight where a monster is blinding the heroes or rendering them delirious can feel tricky and exciting, and quite different from a straight-up hackfest. Here, the soft attack exists because the game is cooler or more interesting that way, regardless of whether the attack itself is realistic or plausible. These depend a lot on the genre and mood of the game in question. In games with a heavily tactical focus, having a variety of attack types that make different battles play out differently can make the game a lot more interesting and increase its longevity.

I’m going to mostly focus on the third aspect of soft attacks. After all, the prompt for writing these posts was that I thought it would be in-character and freakin’ cool for my awesome Space Marine to use exciting exotic grenades to overcome the wiles of vile xenos monstrosities. When I use a soft attack, I want it to have a reasonable chance of doing something, and to affect the capabilities of the target in an interesting way, indirectly contributing to our victory. I’d also like it to sometimes be pretty ineffective, and sometimes rewardingly devastating, just like an ordinary attack; and I think the chances of those should relate to the power of the target and the power of the soft attack. So if I hurl a flash grenade at a horde of gretchin, they should usually all stagger around blindly; if I use the same grenade on a carnifex it should flinch briefly, and I’ll need to turn to a Dreadnought’s actinic photon searchlights to have much effect. But just occasionally it’d be nice for the flash grenade to go off just right and send the carnifex reeling.


In the comments to my last post, I described my basic objectives for a soft attack system:

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

Dan offered a set of rules, which I think are also well worth shamelessly stealing taking into consideration.

If I had to provide rules for workable Soft Attacks, they would be these:
1. Soft Attacks should not simply provide a way to circumvent the rest of the combat system. An enemy with 250 Hit Points shouldn't go down to a single failed saving throw.
2. Soft Attacks should be of the same level of abstraction as the rest of the game. You shouldn't have rules for called shots in a game with highly abstracted hit points for example (a particularly perplexing feature of the 40K RPGs is the fact that it makes you determine the location of hits in combat, despite the fact that all damage is applied to a global Wounds score).
3. Soft Attacks are only appropriate in games in which combat is a tactical minigame. Call of Cthulhu, for example, doesn't need Soft Attacks.

His first point, I think, ties into mine: soft attacks shouldn’t circumvent the combat system, neither should they be neutered for that very reason; we want them to be a broadly useful tool in most situations. Let’s also add, drawing on his Chance and Credence argument, that players should understand soft attacks to be a generally useful tool. There is certainly a place for specialised niche tools, but I feel that common equipment and abilities shouldn’t be in it.

As to his second point, that also seems sound. A system with very abstract combat may be happy with a “generically impaired” result, or even treating soft damage as hit point loss. Maybe the largely-invulnerable superbeings are just generally ground down by attacks, be it bullets, poison, blinding flashes or sonic booms alike. Games that aren’t really combaty at all may also want to vague that sort of thing; if you’re not modelling physical injury, why model penalties? At the opposite end of the scale, if you have detailed wound models and healing rules, then a generic ‘stunned’ condition may seem too vague.

The third point I’ve already touched on above. Some games don’t need soft attack rules, even if soft damage may sometimes occur.

Let’s firm up those objectives a bit.

  • In a game of tactical combat, soft attacks should be (and be known to be) viable tactical options. In particular, if soft attacks require you to take specific equipment or abilities, players should be rewarded for that choice by having opportunities to use them effectively. Even disregarding questions of optimality, choices tend to reflect things people want to actually appear in the game.
  • Also, soft attacks should be appropriately cool. That doesn’t necessarily mean cooler than hard attacks. They shouldn’t fall flat in narrative terms, either by never working (particularly if that belies their description) or by making combat less satisfying. There’s satisfaction in being so powerful that you can simply freeze all your enemies solid every time, but it will make the combat itself less interesting, which is a problem if the game is supposed to focus on tactical combat.
  • Soft attacks should scale reasonably, so that they have proportionate effects when the power level of the attacker and defender are considered.
  • Soft attacks should have immediate effects that vary either in duration or severity, giving degrees of success similar to the variable hit point damage inflicted by a hard attack. A blow to the head or a dazzling light tends to affect you immediately, and slowly wear off. In this respect they contrast with hard damage, which tends to build up over time towards a state of ‘dead’. This immediacy is an existing feature of soft attacks that I’d like to retain.
  • I should also accept that no system will work for all games, given the issues mentioned above. Let’s assume that we are working with a hit-point-based system involving tactical combat between a wide variety of entities on a sliding power scale. So we’re talking about something that might work for games like D&D or Deathwatch.
  • In terms of monitoring, let’s say we want to model the effects of blinding, stunning and slowing attacks. These nominal effects cover impairment of senses, disorientation and impaired movement. A paralysing spell might stun and slow, a smoke grenade might blind and slow, and so on.
  • The system should readily handle recovery from soft damage. Whether an effect wears off with time or requires active treatment, it should be straightforward to calculate and record the recovery.
  • The system should readily handle creatures who are unusually resistant to soft attacks.

Okay, so now I have to think about the difficult part: actual mechanics. This isn’t entirely my forte, which centres more on tangential waffling and opining. As usual I will probably come up with random ideas and just see what happens.

One observation before I start: status effects can sometimes get very complicated (some iterations of D&D have loads of them and their definitions can be quite fiddly), and I’ve outlined some of the problems with binary soft attacks, but it’s worth noting that binary effects are easy to track. You’re either stunned or not. Keeping things easy to track makes play faster, which is important in tactical games – D&D 4E attracted a lot of criticism for its slow battles. It makes things easier for new players or those who aren’t really into heavy mechanics. It makes post-battle cleanup relatively easy, keeps character sheets fairly simple, and helps minimise confusion if there are breaks in play. It occurs to me that full-blown tactical miniatures games, like Necromunda (the most RPGish one I know) tend to do the same thing with hard damage: they have minimal hit point tracking and a very simple set of injury states to keep the battles flowing.

Oh, wait. Before I start this exercise, it may make things clearer if I have a specific game to model. Any excuse...

In a vast universe of squabbling galaxies, trouble can erupt at any moment: calamities, rebellions, coups, trade disputes, accidental hyperevolution, accidental necromancy, dimensional fluxes, awakening the sleeping armies of the lost Ghkrat, dreams becoming real, enthusiastic postgraduate researchers carelessly building an unstoppable army of invincible robot armchairs – the possibilities are endless. No conventional task force can be assembled in time.

In this unimaginably distant future, heroic teams of spacefaring reptiles maintain the tenuous peace between squabbling mammalian empires. No drifting wreck is too sinister, no jungle world too unexplored, no asteroid belt too pirate-infested for these fearless trouble-shooters. With modern technology and ancient wizardry, they preserve the fragile web of intergalactic civilisation. Only their hardy poikilothermic bodies can survive nanotech implants, arcane infusions and the harshness of intergalactic deepjumps. They are the Monitors.

Our intrepid bands of anthropomorphic armoured cyborg warlock lizards will confront a wide range of troubles, many of them violent. They may have to battle anything from a million-strong sea of rock-grubs, through entire ships full of parasite-possessed parrotfolk, down to a gang of vicious pirate jellyfish or a single giant robot.

Arbitrarily, I’m going to call Monitors a skills-based d20 system, with special abilities and gear tied fairly directly to skill levels, giving substantial scope for scaling. Armour is represented by damage reduction, so sometimes a hit will inflict no damage.

Idea: Soft Hit Points

I’m going to start by looking at this idea, not because I think it’ll work very well (I don’t) but because it’s very obvious, and deliberately ignoring it seems to be passing up a learning opportunity. If a game has hit points as a way of modelling physical health and endurance against hard attacks, why not have an entirely parallel system for modelling soft damage?

In our straw system, let’s assume that a monitor has 5hp per level, and so on for other damage types. At fifth level, the monitor has 25 hit points, 25 blind points, 25 stun points and 25 slow points. A typical attack inflicts 1d6 points of whichever damage type it inflicts.

The first problem is that tracking hit points is a significant amount of book-keeping. I’m not sure that people would welcome having soft points to track as well. Marking that your character is stunned is easy; marking off stun points requires more effort. This problem is directly proportional to the number of different status effects you want to track: if you have ‘blinded’, ‘slowed’ and ‘stunned’ to monitor as well as ‘injured’, you are effectively quadrupling your damage-related book-keeping. On top of that, there’s space to consider – you’re potentially looking at a much bigger character sheet to track all those points. These problems are particularly acute for the GM, who already has tons of information to keep track of. Ten hostile foxfolk marines with different hit point boxes is bad enough, without introducing three types of soft damage to track as well.

You’d also have to decide what the points actually indicate. In most abstract combat systems, it’s only when hit points run out that characters actually suffer any real consequences. So running out of blind points leaves you completely blind. In that case, as blinding attacks are usually much rarer than hard attacks, having blind points equal to your hit points would probably mean never being blinded. You would basically soak up blind points happily until you suddenly went blind. A model like that runs counter to my vision of soft attacks as immediate and temporary.

On the plus side, the system can easily handle resilience by reducing the damage inflicted. A thick-skulled rhinoborg takes half stunning damage, for example.

It can also handle recovery fairly easily, treating it as a parallel to normal healing. You simply recover the appropriate number of soft points in response to treatment or rest. If you want effects to be temporary, the creature may heal a certain number of points every turn. While it’s straightforward, you might well end up losing and regaining soft points constantly, which could get fiddly.

I think I’m fairly happy that this copy-paste system would be unsatisfactory. It requires too much book-keeping, and makes it quite difficult to actually affect somebody with soft attacks.

Lessons learned:

  • In a cumulative system where soft attacks are relatively rare, each attack needs to inflict a relatively higher amount of damage than a comparable hard attack if it’s to be mechanically worthwhile.
  • Point pools building towards a total are not a good way to model immediate effects that wear off gradually.
  • The less information that has to be tracked, the better.
  • Quantitative models can handle both damage mitigation and recovery in a straightforward way.

Idea: Penalty Stacking

The next idea that occurs to me is a simple system of stackable states (ooh, alliteration). This is basically the cumulative inverse of the whittle-away hit point model. Rather than just switching your blind/nonblind status, or deducting 1d6 blind points from a pool, a flare in this system might inflict 1d6 levels of blind on the target. These levels would stack in a simple way and inflict straightforward penalties. For example, blind levels might give a -1 penalty to any skill roll requiring vision, including searching and watching, attacks, first aid, computing, reading and so on and so forth.

This system automatically allows some scaling. A more powerful blinding attack would inflict more blind on the target, giving a higher penalty, so it makes sense for these to be rarer or require a higher level. Similarly, any given attack will be more effective against weak opponents (who have low skill levels) than against strong ones (who have higher skills and can more easily tolerate the penalty). At the same time, it means than any successful attack will have some effect on an enemy regardless of how powerful it is; a -1 penalty might not mean much against a skill of 20, but it’s still a penalty, and a number of such weak attacks can build up to something substantial, just as many weak hard attacks can slowly whittle away hit points.

While it requires some book-keeping, this also seems more straightforward than soft hit points, simply because there are going to be smaller numbers involved.

On the downside, I think a quantitative approach to blinding is less narratively satisfying than a qualitative one. For example, impaired vision should create new difficulties rather than just exacerbating existing ones. Perhaps you should need a roll just to read some text, or see where another person is? Also, some tasks rely on vision more than others – you might be able to cook blindfolded, but guaging an attack on sound alone is very difficult, and tracking a wild animal more or less impossible. We’ll have to accept some arbitrariness for the sake of a working system, but it’d be good to keep that under control.

Just as with the last model, penalty stacking can handle resilience simply by reducing the number of levels inflicted by an attack. A crocodilian spy’s extra eyelids give her blind resistance 3, so she only takes 1d6-3 levels of blinding from a flare.

Recovery is also very simple, as we simply have to alter the level of the penalty. Our crocodilian spy has been dazzled by a laser misfire, giving her blind 8. At the start of her turn, she blinks and recovers to blind 7. The amount of book-keeping needed is likely to be similar to the soft points model, as the numbers still go up and down regularly. On the upside, because the numbers are likely to be smaller, it may be possible to track them using counters or dice. Back on the downside, it’s still much fiddlier than tracking a blind/nonblind binary status.

Lessons learned:

  • Stackable penalties allow immediate effects
  • Stackable penalties allow scalable soft attacks
  • Stackable penalties allow weak attacks to be meaningful against strong opponents
  • Stackable penalties can be fairly easy to handle
  • Numeric penalties may not be enough to make soft damage convincing; qualitative levels may be needed
  • Stackable penalties work well with damage mitigation
  • Stackable penalties are easier to track than hit points, but more difficult than binary statuses

What next?

At this point, it seems to me that stackable penalties offer a lot of advantages, so I think it’s a good jumping-off point for whatever model I try next. However, I do think the skill penalty model is too simplistic and likely to feel unconvincing. If there’s any point in having attacks that blind or deafen or paralyse, even though we’re trying to introduce scaling, at some point they need to feel like the target is unable to see, or to hear, or to move.

The next idea that occurs to me is vaguely inspired by Necromunda, slightly by Hellcats and Hockeysticks, and partly by some other game I genuinely cannot remember. Possibly war machine damage in Epic, bizarrely.

Anyway, the idea I’m considering now is having a stackable range of qualitative conditions that interact with quantitative damage. The D&D 3.5 system of fear does something a bit similar: it ranges from shaken to frightened to panicked.

Idea: Status Thresholds

One option would be to use the stacking model, together with thresholds of effect that are crossed as penalties stack up. For example:

Xerxes the gecko detective is attacked by a pirate, and pulls out his flash pistol to get some breathing space. His first shot inflicts 4 blind on the villainous squirrel, lightly dazzling him. A second shot inflicts a further 3 blind, to a total of 7. Once the blind level reaches 5, a threshold is crossed: ‘dazzled’. Long Tall Nutkin now can’t make out fine detail, including reading, and has to spend pass a perception roll to identify individuals, which makes combat extra difficult, so he fails to hit Xerxes.

A third shot takes Nutkin past the 10 threshold to ‘half-blind’; he can now only make out strong light-dark differences and large motions. It’s very much like operating at night. In this condition, as well as the mechanical -10 penalties, he can’t identify individuals visually, see colour, distinguish targets under normal lighting conditions, read, or make out details of his surroundings, but he can just about manage to stagger away down an alleyway and avoid an oncoming hoverlorry.

Xerxes decides to end matters with a well-aimed magnesium bomb, inflicting another 8 blind and passing the 15 threshold to ‘blind’. Nutkin is now entirely unable to see. He can’t identify anyone or anything visually, orient himself, read, observe anything that’s happening nearby, spot open manholes, or avoid walking into walls. He can attempt to drive, but unless he has the windows open or some other way of getting non-visual signals, he’ll automatically fail. On top of that, he has a -18 penalty to any activity that involves vision, including cooking, defending himself, most conversations, juggling and sorting change. Where there’s no particular failure condition, like eating or navigating through an empty building, he’ll simply be very slow. Unable to work out which direction he's heading, Nutkin stumbles over a dustbin, and he doesn't have the time he'd need to fumble his way to safety by trial and error. Xerxes has him in pawcuffs within moments.

This system calls for the designer to determine thresholds and the effects of crossing them, which may become quite detailed. There’s also the question of static versus relative thresholds.

Static thresholds are simple to track and remember, but don’t scale very well. A low threshold (say 5) allows low-level flash pistols with 1d6 damage to actually affect a target reasonably often, and makes powerful blind attacks very potent; however, as characters increase in level and get more powerful attacks, it becomes trivial to blind equal-level opponents. If the threshold is high (say 20), the flash pistol can only affect low-level opponents with repeated hits, making it basically useless. This system would also make it more or less equally difficult to blind low and high-level targets.

So we’re looking at some kind of relative threshold that reflects the level and resilience of the target to that attack, in the same way that hit points increase with level and Toughness. The details will depend on how exactly the system works. In D&D, with its formal levels and stat increases, a threshold for blinding might be “level + Dexterity modifier”. In Deathwatch, which relies on stats and skills, the threshold might simply be “Perception modifier”. Sadly for tracking, these vary between creatures and characters; on the plus side, they aren’t that difficult to calculate.

Let’s decree arbitrarily that Monitors has no formal levels: characters increase their skills and stats through accumulated XP. Therefore, blind damage is modelled with thresholds based on multiples of the creature’s Reaction stat, since quick-reacting creatures can blind or look away quickly to avoid the worst of the light. A slow-thinking terrapin accountant with Reaction 4 is dazzled at 4, half-blind at 8 and blind at 12, while a hyperactive shrew pickpocket with Reaction 8 needs a hefty 24 damage to blind it – that’ll take repeated flare blasts or some more powerful effect. Similarly, stun damage might come from Toughness and slow damage from Might (stronger creatures can work harder to reduce the effect).

It does occur to me, though, that having soft damage types tied to particular stats for their thresholds could get unconvincing. A psychic stunning attack should perhaps be compared to Mind rather than Toughness, as strong willpower and clear head make it easier to stay focused. Might is fine for overcoming slowness due to entangling vines or increased gravity, but shouldn’t a paralysing poison depend on the creature’s Toughness? Many systems reflect this by having the soft attack target a suitable defensive stat, which works well. Our system can still do this for actual attack rolls, but when it comes to considering the effects of a hit, it’s tied to a specific stat.

Lessons learned:

  • Any threshold-based damage system needs to reflect the target’s attributes in order to preserve scaling, just as physical hit points tend to
  • Defining thresholds and their effects might get quite fiddly; there’s always more you can say
  • If a threshold is based on stats, it probably won’t convincingly model all the attacks that could inflict that kind of damage

Coming up next on Killing Me Softly: soft damage boxes, healing soft damage, natural recovery and modelling resilience. Additional ridiculous reptile antics are also likely.


  1. I do worry that adding a hit-point-esque incrementing scale of "soft damage" would add unnecessary bookkeeping, and I worry that it removes the "decisiveness" of soft attacks. One of the things that's nice about a stun, say, is that you can stun one enemy while you fight another, for example, and that's rather lost if you have to ramp up Stunning Points over several turns.

    1. I hadn't specifically thought about that case, but I'm trying to bear in mind "the way SA should work" as I go. Not sure how well-articulated this article is, there was a fair bit of chopping about. This is very much an exploratory sort of project rather than one where I've got any firm idea what I'm doing. Problems like book-keeping and faff are definitely things I'd rather avoid.

      Exactly what it is I want out of soft attacks I'm not yet sure, but it would not usually involve building up points. Xerxes aside, I wouldn't really expect people to use soft attacks exclusively on a regular basis.

      Taking your example, where you're fighting a group of level-appropriate enemies, my feeling is that a "successful" stunning attack should stun the target in the traditional can't-act sense, allowing you to focus on the others. A poor damage roll would leave them able to act, but at a penalty.

      Where stacking up Stunning Points (if we used them) might be necessary would be either a) allowing a series of unlucky damage rolls not to be entirely worthless as they currently are, by adding up to a meaningful impact over two or three attacks; or b) making it possible to affect even powerful opponents with sustained soft attacks, in a non-swingy manner.

      Obviously one possibility of the series is that I'll end up deciding there isn't a viable alternative to the binary system.