Saturday 15 September 2012

Killing Me Softly, part 3

Last time in KMS, enthused by a moderately interesting start to a new project and a flicker of interest elsewhere, and also by cyborg warlock space lizards, I continued to stumble aimlessly into the tangled morass that is “soft attacks”. By which I mean, of course, attacks that don’t directly deal damage but inflict status conditions or penalties on an enemy.

So far, I’ve mainly discovered that trying to work out alternative systems for this stuff is really freaking hard, that it’s quite difficult to plan mechanics in a general way without a pretty concrete system in mind, and that if I deliberately invent silly RPGs for demonstrative purposes I will be seized with a desire to write and play them, at least if they involve reptiles.

Mostly I’ve been investigating ways to implement a range of effectiveness for successful soft attacks in place of the commonly-implemented binary succeed/fail options, in the hope of making them less swingy and therefore more comfortable for everyone. This project is mostly preliminary research, by which I mean that I come up with ideas and then poke them to see what breaks. Ideas so far have included:

  • Pools of ‘soft hit points’ to monitor variable damage, heading towards a tipping point
  • Stackable penalties to actions, so a better damage roll inflicts a higher penalty
  • Stackable penalties combined wih fixed thresholds at which defined status effects kick in

Common findings seem to include:

  • All systems are fiddlier to track than the binary one
  • Pool-type systems are not very good for modelling immediate and temporary effects
  • Quantitative models can handle damage mitigation and healing easily, and scale easily

The ever-helpful Dan has also pointed out that accumulative soft attacks tend to be less immediately game-changing, which is unwelcome if we’re expecting them to be occasional spice in the soup of hard attacks.

There are a few things I wanted to think about this time, including more possible models, and ways of modelling recovery and resilience.

In which further models are expounded

I think it’s best to get the models out of the way first so the rest of the discussion can include all the options. This is a sort of tick-box model of escalating status effects, which is the one I think might relate to Epic.

Idea: Damage Boxes

In this system, each soft attack type has several thresholds which impose fixed penalties. Let’s stick with our earlier ‘dazzled’, ‘half-blind’ and ‘blind’ example. Each is represented by a box. A successful blinding hit marks off the ‘dazzled’ box. Each subsequent hit marks off another box.

The main problem with this simple approach is that there’s no discrimination between weak and powerful attacks or creatures, so the same scaling problems apply and the same binary problem applies to some extent – three flare pistol hits will totally blind a Robosaur, so GMs will end up taking steps to prevent this working. If, on the other hand, the blindness isn’t that effective against a powerful creature like a Robosaur, then at high levels soft attacks become useless, which is another no-no.

One thing we could do is add damage ranges to each box. Roll damage and compare to the boxes to see how damaging the hit was. Once the hit is resolved, the fixed penalties for the box apply and the exact damage done is no longer relevant. This allows for power differences between weapons, and helps preserve scaling, particularly if the ranges are linked to the target’s resilience.

I decided last article that in Monitors, stats tend to increase through XP accumulation, so it’s reasonable to assume that more powerful creatures (both veteran fighters and dangerous beasts) will have higher stats. Let’s say that an average 1st-level stat is 4 (yes, yes, this contradicts an example in the last article). A fairly simple system would be to have the ‘half-blind’ threshold equal to Reaction and the ‘blind’ threshold equal to 2 x Reaction. Anything lower than the threshold inflicts a ‘dazzled’ status.

Xerxes has got himself into trouble again. After digging too successfully into a corruption case, he runs into a couple of besuited raccoon heavies in a dark multistorey skimmer park. Luckily, his flare pistol is always to hand. Raccoons have Reaction 4. A snap shot inflicts 5 blind on the first raccoon, leaving it staggering in the dark ‘half-blinded’ and out of the fight for now. A shot at the second only inflicts 1 blind, so it’s merely ‘dazzled’. If Xerxes shoots it again, we don’t care what the current penalty is; we just roll to see the result of the next shot.

In a system like this, the GM doesn’t need to track numbers, only statuses. Basically each status is broken down into several levels of severity, which are inflicted according to the damage rolled. It calls for more tracking than binary models, but less than quantitative models. The amount of tracking depends largely on the number of statuses implemented.

What this model doesn’t do is provide any way to stack up soft attacks, which I’d like to investigate.

Idea: Accumulating Damage Boxes

In this adjusted model, each damage box has a damage range as above. However, each subsequent hit becomes more deadly because of the existing damage, so a sustained barrage of soft attacks can be more devastating than the sum of their parts. This makes sense to me: being ensnared in four nets is more problematic than one, repeated loud noises are more likely to leave your ears ringing, and several doses of KO drugs will affect you more severely than one.

We could simply to say that every hit marks off at least one box, regardless of the damage rolled. However, this suffers from the familiar problem of overpower against tough opponents, and I suspect isn’t really worth implementing unless the system involves lots of boxes, so a very large number of soft attacks would be needed to overwhelm tough opponents. For example, a system with ten boxes might be viable, because it would then take a typical PC group of 4 characters more than two rounds of only using soft attacks to completely blind the Robosaur, during which it would have ample opportunity to chomp them, and might already be recovering from the first few hits. However, they might be able to hire ten mercenaries with flare pistols to help out. Basically, the extent of the problem depends on the willingness and ability of players to exploit this loophole, and whether is fits the tone of the game. If you feel that fourteen soldiers with flare pistols should be able to blind a Robosaur with concerted fire, it’s not a problem at all.

A more scalable option is for each ticked box to grant a bonus to future damage rolls; in our example, it might be a bonus equal to the Reaction threshold of the ticked box (so +4), leaving Xerxes rolling 1d6+4 for subsequent attacks on the first raccoon and with a reasonable chance of hitting that 8 blind threshold to completely blind it.

There’s one last model I’d like to look at. Honest. I can stop any time I want.

Idea: Damage Charts

So this is another model that I don’t think will work, but I’m going to look at it in passing. In this system, each soft damage type has a damage chart available, with escalating penalties for higher results. The successful soft attack simply rolls on the chart. More powerful attacks roll a larger die, just like hard attacks.

Here's a sample blinding chart:

  1. Lightly dazzled: -1 penalty to rolls involving vision
  2. Dazzled: -2 penalty to rolls involving vision
  3. Badly dazzled: -4 penalty ro rolls involving vision
  4. Half-blind: -5 penalty to rolls requiring vision; must roll Perception for actions based on vision that normally require no roll, such as reading or identifying creatures
  5. Near-blind: -10 penalty; roll half Perception for vision-based actions
  6. Blind: fail all actions or rolls requiring vision; halve all rolls involving vision

The difficulties here are that in many systems it’s relatively difficult to predict the damage that an attack will roll, and that the variation between low- and high-level damage would make for some very weird charts. For example, a 1st-level Dungeons and Dragons 3rd edition fighter is probably rolling something like 1d8+3 for damage from a decent weapon and a +3 Strength bonus. A critical 20 with a spear might allow her to inflict 3d8+9 damage in a single hit, with perhaps another +2 from magical buffs for a maximum total of 35. However, it’s incredibly rare (about 1 in 10240 attacks) and the average is going to be more like 7, with about one hit in twenty doing around 20 damage. If soft attacks had a similar range, the damage chart would have to handle ordinary blindness within at most a range of 1-20 while allowing space for amazing results for 35. Bonuses and multipliers stack up as levels increase, making the problem worse: a 10th-level rogue may easily end up rolling 1d6+5 damage normally but 8d6+10 on a critical sneak attack, for a result anywhere from 6 to 58, and that’s without any complex multipliers from spells or magic items.

The scaling problem can be overcome to some extent if we used enemy stats to modify the result.

Captain Ussami is investigating an SOS from a mining factory. She soon finds out the reason when erratic robots begin to attack her. Her disruptor glove inflicts 1d8 stun damage on robots, giving her a roll of 1d8-4 on the damage chart against the standard workers, with a minimum result of 1 (inflicting a -1 penalty to actions) and she manages to battle her way through largely unscathed.

A few months later, the Captain finds herself confronting a complex full of elite security bots. Their improved cogitator units give them Mind 6 and her odds look a lot worse with a 1d8-6 roll. She’ll need to pick up a more powerful weapon to stun these bots reliably.

This model would probably work best with established limits on the damage soft attacks can inflict, as otherwise swinginess will become a problem again. This is likely to be a system where hard attacks also have a fairly limited range of damages, without too many stacking modifiers and multipliers to complicate matters.

Basically I think this idea might be fun in a straightforward system where (again) creatures are fairly similar in power and likely to be making attacks in a fairly similar range, without much in the way of power scaling. As randomness and scaling enter the model, it will become increasingly difficult to create a table that both a) produces satisfactory results for all possible damage rolls; and b) isn’t incredibly bland. Since Monitors is going to involve wildly different creatures across a significant power scale, this isn’t going to be an appropriate system. There’s also the problem that a damage chart with many different results will get fiddly to track. Oh! And as the chart above demonstrates, it's not easy to create a chart that's actually interesting for such a specific damage type, which to me suggests that it's not an improvement on the simpler stacking penalties model (with or without thresholds).

Okay, enough playing. Back to theory. At the start of the article I was wondering about the impact of recovery and resilience modelling on the way soft attacks worked. In the process of the last few examples another point has come to mind, which is the way that severity of effect is modelled. These – like everything else – interact with one another.

But I Get Up Again

As I mentioned previously, the way I envision soft attacks is generally as a sudden and immediate effect that changes the way the target can act, but wears off over time. A flash of light blinds you until your eyes adjust. A drug knocks you out until it works its way out of your system. A layer of gloopy webbing hinders your actions until it’s rubbed off.

I’m aware of three main ways that binary systems handle this recovery, sometimes using more than one in the same game:

  • The effect has a fixed duration or wears off at a specific point, such as the end of combat
  • The effect wears off after a length of time based on the attack (which may be fixed or random)
  • The effect allows a saving throw each round to recover

If we’re using a non-binary system, the options are a bit different. Some I can think of include:

  • *You heal to a random extent each round
  • *You heal a fixed amount each round until you recover
  • *You attempt a recovery roll each round to mitigate your situation, possibly recovering entirely (a slightly more nuanced version of the saving throw).

The other main consideration is what recovery should be based on, which might include ‘nothing’ (it’s invariant), ‘power’ (level) or ‘resilience’ (as compared to equivalent-level critters).

In a very simple model, recovery works exactly the same for every single creature. Fine for some games, but for Monitors? The hyper-metabolism of a Centurian velocibadger allows it to adjust to environmental conditions within seconds, and recover almost instantly from overstimulation or chemical attack. Obviously, that one example can be handled with a special rule; but in a system with wildly variable entities, having their capabilities reflected smoothly by the models seems like a good goal.

The two major options seem to me to boil down to how you want resilience represented in recovery. Should it make you recover damage more rapidly on average (a larger fixed amount or a larger die roll), or should it give you a better chance of recovering each round (a better chance of success on a roll to recover a standard amount)? These choices will tie in to the model you choose. Rapid recovery makes sense with quantitative models that allow fine control. Probable recovery makes sense with models using discrete categories of penalty.

Example: Recovery speeds

Our beta for Monitors is testing the stacking penalties model. Xerxes’ trusty magnesium bomb lands squarely in a cluster of Tetulan porcupinoids and their war-squid, inflicting 20 blind on each (thanks to a set of fluke rolls). The porcupinoids recover 5 blind at the end of each round, so it will be the fifth round before they can act normally and for at least three they will be basically useless. However, the hardy squid recovers 10 blind each round, so it has only two rounds of penalties to endure.

Example: Recovery probabilities

Another session tests a probability model. The ‘stun’ status has been allocated levels of ‘distracted’, ‘dazed’, ‘stupefied’ and ‘stunned’. Doctor Clymestra’s devastating Invocation of Primal Nightmares wracks the minds of an angry owl mob, leaving them all stunned. Each subsequent round, the owls can roll a d12 against their Mind of 5, and a roll equal to or less than the stat allows them to recover by one level. Amongst the owls, though, is a chessmistress, whose acuity and willpower is represented by a Mind of 9, leaving her far more likely to recover each round. Of course, this model means a ‘stunned’ status always takes at least four rounds to recover from, but then it’s relatively difficult to inflict that status.

A possible tweak to the probabilities model would be to allow a second roll in certain circumstances, from the simple “after any successful roll” to “a natural 1”. This would give a small possibility for a creature to recover totally in any round, especially a very tough one. On the downside, it strikes me as reintroducing swinginess by making the soft attack occasionally very ineffective (though assuming rolls happen at the end of the creature’s turn, it would still be stunned for one round), which could be annoying when it’s very difficult to stun creatures with a high Mind in the first place. While there’s room for specific creatures in a game that are both very tough and recover rapidly, such as those with high Armour and regeneration or healing powers, I’m not that comfortable with having that combination an intrinsic part of the ruleset.

Let’s go back a couple of steps. In that simpler model, with no variation in recovery rates, resilience is only taken into account when establishing the initial damage. This might actually fit better with the way hard damage resilience works: high Toughness-equivalent tends to increase the number of hit points you have, but not how much damage you heal compared to the next lizard. If we used that kind of model, everyone would recover at the same rate; in a probabilities model, this would be a fixed roll, such as 4+ on a d6.

Example: Static recovery

A veteran bionic mole commando and a sheep politician are attacked by newt kidnappers en route to a crucial trade meeting, starting with a stun grenade. The neural scrambling leaves the politician stunned, but only dazes the hardy commando. Regardless what method is used to track recovery, both recover at the same rate. This means the politician will be out of action for about twice as many rounds as the commando.

I think the decision between systems mostly boils down to whether you want resilience to model how badly you are affected, how quickly you recover from the effects, or both. We’ll also have to consider how we want the severity of a bit to be modelled. But this seems a good point to talk about resilience.

You’re Indestructible

Okay, definitions again. I’ve talked a bit about resilience here before. What I’m talking about here is how difficult a creature is to affect when compared to other creatures. I usually mean in comparison with other creatures of a similar overall power level, but that won’t always hold true.

Sometimes resilience reflects the natural attributes of a creature: thick-skulled rams are more resistant to knockouts than kangaroos, even if both are humble peasants. At other times, it reflects training: a veteran combat mage has a focused mind and is better able to keep mental coherence than a mere apprentice. Often the distinction is less clear-cut: a hulking mammoth marine is both innately hard-headed and trained to stay focused in combat.

There are several ways that resilience gets modelled in games, which relate to the rest of the combat system and often co-exist. I’m going to include physical damage here because I think it’ll be useful. Some examples that spring to mind:

  • Penalties to attack rolls that take opposing stats into account
  • Saving throws against attacks that are increased by a high stat
  • Hit points that are increased by a high Toughness
  • Ability to reroll failed saving throws to avoid an effect
  • Reduction in damage taken from an attack (as in Deathwatch)
  • Mitigation of specific status effects to a lesser version
  • Regeneration of damage each turn
  • A bonus to fixed saving throws taken to recover from status effects (as in 4E D&D).
  • Ability to reroll failed recovery rolls to recover from an effect

The first one models resilience through a change in the attack roll. The next five make the attack less immediately effective. The last three speed up recovery from an attack.

Dungeons and Dragons, for example, uses several of these systems in parallel. High Toughness gives extra hit points to absorb physical damage. Stats modify either (depending on edition) attack rolls or saving throws that determine whether a soft attack is effective. Many creatures have damage reduction that reduces damage taken from attacks. Regeneration is a fairly common ability. More rarely, several abilities can mitigate status effects to a lesser version, or grant bonuses or rerolls to saving throws. In 4E, hardy enemies gain bonuses to the fixed saving throws used to recover from many persistent effects.

There’s always the possibility of special rules for special critters, but we’re interested in the base mechanics. The main way resilience comes in to play in binary soft attack systems is that it increases the die roll needed for an attack to take effect. If we’re looking at other systems, we have other options.

Last time I talked about stacking penalty models, where the damage from a soft attack roll becomes a penalty to the enemy’s actions. Here damage reduction is an obvious way to handle resilience. On the downside, it offers the opportunity for some attacks to inflict no damage; however, the same is true if damage reduction applies to hard attacks. Deathwatch is a prime example of this, with both Toughness and Armour reducing the damage taken. In a system like that, I don’t know that it would bother players too much.

In damage box models, we could apply a similar rule. If we’re rolling for damage, the reduction would help determine what level of penalty applied. If we’re just ticking box after box, it would reduce the number of boxes ticked.

We could also say that resilience affects not how severely the attack hits, but how quickly you recover from it. In this situation, we’d use a stat-based recovery model, whether faster recovery or a better chance of recovery.

You know, as I go on here I’m finding more and more factors that play into this modelling business and interact with each other in complex ways. I’m going to do some thinking about before I go any further. In any case, I’m not really sure whether all this is actually going anywhere.

No comments:

Post a Comment