[spellbound] alpha playtest post mortem

[merxa]

I was able to run a small playtest for my usual group a few months ago and it had a number of useful takeaways.

Character creations and a rules summary took about 90 minutes and the play session was 5 hours. The setup was they were all from the same remote farming village and were tasked with finding a missing merchant. They explored the road and forest leading out of town, avoided some wolves, encountered a single bandit, then a bandit camp of four, and the final combat encounter was against a large drider-like creature armed with a spear.

The most important takeaway is that damage needs to be reworked. This won’t be a surprise to anyone who has followed along with spellbound development, but validating this with a playtest was what I wanted. The basic issue was damage either did too much or too little, with it rarely falling into the expected middle.

After some consideration, since I am using a wound system it seems like a dice pool would be a good fit; this should make calculating average damage easier, smooth out damage variance, and provide more space for damage to ramp.

~

How much damage? Lets test some numbers using some simplified progressions. As an aside, by damage pool, I mean to roll d6s and a 4,5,6 is a success or 1 damage, so when i write '4d' that would be roll 4d6 and count successes.

Lets say level 1 starts at 6d and damage increases… by 2d? Per level, that would translate to an average of 3 then +1 per level. Damage reduction would scale similarity, in scenario 1 lets have damage reduction start at 1 and increase 1 per level. So the average damage would be 2 and that would stay consistent across level to level opponents. And average damage of hits would go up or down by 1 per level difference.

Comparing level 1 to higher levels, this does mean that once a two level gap opens up, average damage for the lower level becomes 0, although ~33% of the time the lower level would still do 1 or more, at a 3 level difference this becomes 11% and at 4 levels it drops to a very negligible 1%.

All else being equal, as levels increase, the average damage doesn’t change but the variance increases and somewhat dramatically.

For example, for 6d, the average damage is 3 with variance 1.22

For 10d it is average 5 with 1.58 variance. Going way up, to say 20, we see an average of 10 with variance of 2.24. To give those numbers a bit more context, in practice the vast majority of results will be within two standard deviations (ie 5.52 - 14.48), and practically never see results outside 3 deviations.

Since one of the systems underlying goals is to be in person playable, I likely will cap damage dice and begin adding a static modifier, 10d and 11d seem like ok caps to me – so in practice for every 2d above 10 or 11 it would transform to a +1, so instead of 12d or 13d it would be 10d+1 and 11d+1.

Variance stops increasing once the static modifiers are introduced. Variance for 10d is 1.58 and 11d is 1.66.

I wonder if people have an opinion, intuition, or even some evidence or other reasoning, on what the ‘ideal’ damage and variance would be when solving for ‘fun’. Perhaps average damage after average damage reduction should be closer to 3 instead of 2 for example.

As a reminder the 10 levels in spellbound are meant to map to the 20 levels in 3.x, so being 2 levels apart would be more equivalent to 4 levels. Another open question is if progressions should be geometric and smooth, or if there should be any discontinuous functions. One example might be to cluster levels 1-3 a little closer together before having a faster ramp starting at 4, or have a tight cluster 1-3, a faster ramp 4-7, then to slow it down again at 8-10.

Ultimately whatever progression I settle on will be more guidelines than say a hard increase as damage output and reduction will be determined by a few factors and not juist be a simple, unchanging lookup table
~~

Besides the damage rework there were a few other more minor takeaways; but I think my current goal is to fix the damage system and otherwise rework/edit the material I have to be more presentable and then to get it into something that will make collaboration, updates, and actual usage all easier. Perhaps a wiki or just github. Suggestions are welcome, and as always if anyone has any interest in contributing please let me know – there’s plenty of material left to write.

[deaddmwalking]

When you're rolling 20d6 and counting 'hits' to determine whether you did the expected 10 hits, or did badly with 8 or really good with 12, it's going to get ridiculous.

You'd probably be better off with a base damage of 2 (scaling +1 with level)+ 1d3. The 1d3 creates just about the same EXPECTED variability while ensuring you don't end up with a really out of expectation roll. It'd certainly be easier.

[Omegonthesane]

Pools of 20 dice are kind of a nuisance in Roll20; they're entirely intolerable with actual physical dice.

[merxa]

Yes, 20d6 in person is a hassle, that's why i was suggesting I would cap it at 10/11d and add static +1 bonuses instead. Is 10d, 11d also prohibitive in person? I could just cap it at 5d/6d and add static bonuses there, but not sure if there's enough variance (var 1.12, 1.22 respectively). One upside of rolling those dice is that getting all successes / all failures on 5d, 6d is something you would expect to happen once or twice an evening depending on how combat heavy the session is, and an 0-5, 0-6 RNG seems ok. Comparatively, 10d, 11d, you're going to see mostly 2-8 range, and entire sessions, even shorter campaigns may never see a 10d roll that's 0 or 10. But even the outside chance of doing 8, 9, 10 damage might be a bit overwhelming to a 10 wound box system.

capping a 6d and adding static bonuses is pretty comparable to the static base plus a 1d3 damage die. (i will freely admit, i have a bit of a kneejerk aversion to static damage so its rare for me to consider it without external suggestions).

[deaddmwalking]

In a game like Magic:The Gathering, nobody has a problem with a War Elephant always doing 3 damage and only needing to take 3 damage to kill. We understand that it is a simplification. But when you imagine a War Elephant trampling a dozen people, it seems likely that some will be seriously wounded (maybe even dying) and some will be unhurt. Assuming that they're all equally tough, randomized damage seems to track with our expected reality (even though this is entirely a fictional thought experiment).

If you want variable damage (perhaps even HIGHLY variable damage), then you need to make it possible. On 20d6 you're going to get 10+/-2 (ie 8-12) almost 75% of the time. You're going to get 10+/-3 (ie 7-13) almost 90% of the time. So what's the 5% of the time that you get fewer than 7 hits or more than 14 actually worth in the system? If this represents your 'nova ability' and you score 3 hits (1 time in 1000) are your players going to feel cheated? Is that offset by the chance that they get a few extra hits? My gut says from a psychological standpoint the answer is no. Rolling badly when you think you shouldn't usually sucks, and while rolling a little better than average feels good if there isn't any special riders it's not going to really matter - especially if it takes the same number of attacks to 'drop' an opponent! What's the benefit of getting 17 hits on the first attack and 10 on the second attack compared to getting 10 and 10 if it takes 20 hits to kill an opponent?

If you do want 'spectacularly lucky actions' to potentially drop an opponent in a single hit, you are going to need some type of variability. I'm not sure that you're receptive to it - if you're worried that doing 9 wounds to a 10 box wound system is too powerful, that's a sign to me that you're not okay with an opponent potentially being dropped in a single blow. It's also worth thinking about how Opponents are treated versus the PCs. In MANY systems you're better off having 400 mook archers hoping to get an exploding crit than having 1 level 400 guy or 40 level 10 guys - even a 5% chance or a 1/400 chance (20 followed by a natural 20) is going to show up sooner or later, and it's one of those things where a 10th level PC getting dropped by a goblin is going to feel MUCH WORSE than how good it feels when a 10th level PC one-shots a dragon. If you allow one, you're by no means obligated to allow the other - but you're going to have to include that in your design.

So, equal opposition (ie, mirror fight), how many attacks should it take to drop? Do you want it to be the same at low levels as at high levels? How consistent should it be? If A and B are identical, and they both always hit and they fight each other are you okay with victory being determined by who rolled the best initiative? What's the minimum number of attacks and the maximum number of attacks you'd be comfortable with?

You say damage was either too much or too little. Eliminating variance isn't hard (well, it may be hard if you dislike static damage, but then you've got incompatible design goals and you need to address that *wink*). Maybe a dice pool that pushes things toward the mean would be better, but it can still have 'too much or too little', just not quite as often. If doing too much or too little AT ALL is the problem, your proposed solution doesn't help. If doing too much or too little TOO OFTEN is the problem, it may help - but I think you really need to define the problem more explicitly. Why were too few hits a problem? Why were too many hits a problem? Was it a problem for everyone equally? Was there a difference between 'bad rolls' and 'good rolls' in terms of how the players felt (a psychological dimension) that you want to address? Do you want to put a hard cap where threats of a certain difference can't hurt you (ie, an opponent 4 levels higher than you is immune to your attacks)?

[merxa]

Solving damage output under system assumptions

Action economy
3 AP and 1 RP
HP 10

Combat should be 2 or 3 rounds, sometimes 1 or 4.
rules lite idea: if combat hits round 5, it forces a settled negotiation, examples might be truce, conditional surrender, flee, unconditional surrender, ally, re-engage(until round 10).

PC mirror match (simplest to solve for)

Assume 3 AP translates to 2.5 attack opportunities.

Attackers, on average, hit 66% or 60% of the time (⅔ or ⅗).

~ 1.5, 1.6 expected hits per round per character.

PC Defenses are expected to negate 1 attack or reduce damage 1/round, and perhaps 1/opponent on higher tiers

That would turn into ~.5 hits a round

Damage should be roughly 0-6+2 or an average of 5

Ultimately this is a big probability tree, but the fattest path suggests 2.5 damage a round, ending a mirror match in 4 rounds.
~

Now before i solve for pc vs say, 4 mooks, i’m really solving for 3-4 PCs vs team monster which is arbitrary, and it’s arbitrary in many ways. If the party dies every session, there’s not much of a game. So what do people consider thresholds for encounter difficulties? Let’s say we define a ‘deadly’ encounter as a coin flip, a true gamble, you are as likely to win as to lose. I suspect i’m already far afield of expectations, in truth PCs win, they generally always win the combat mini game. That is for many reasons, but the central conceit remains that it’s an adventure about the PCs.

So we might say, PCs win a level appropriate encounters 95% of the time, ie, once in a while, every 4 or 5 sessions, the PCs realize they are on the other side of the RNG and begin to scramble, be forced into a negotiation, be taken captive, fail an objective, have a less than optimal outcome, suffer any sort of set back – no matter how minor, stub their toe after their glorious victory, however it is the PCs achieve the ‘do not win’ condition.

The juice then comes from level inappropriate encounters, because PCs ultimately want to take extra risk for extra rewards, and maybe their plans and builds are so good it even the odds anyway. If i return to the +1 level +1 damage assumptions, that tips it from 2.5 to 3.5 to 4.5 to 5.5 per round before we adjust to hit numbers.

~

Returning to damage die pools, if the damage die pool caps at 6d and provides a static +1 per every additional 2d, it creates a dead zone 1d on odd damage dice. The game could turn that into a mechanic, ignore it, or throw in the additional 6d/7d complexity rule/resolution.

[deaddmwalking]

There's some word salad going on. I assume AP is 'Action Point' and characters are spending AP for each action (replacing move/standard, etc). I assume RP is 'reaction point' which is some type of 'immediate action' (ie, played on someone else's turn).

If you're trying to simplify the action economy by using AP, but then you have to give people an RP, you're not really simplifying things. Now you have two types of actions and you really have to think about how much of a problem is it if a character can keep attacking (spending all 3 AP on offense) versus if they have to use an AP for some other action. Since I don't understand your system from this description (or from the other threads I recently re-read), I'm speculating, but a character that uses 1 attack per round (which is usually negated) is going to be doing far less damage than you propose compared to a character that is making 3 attacks each round (of which one is negated).

Setting all that aside, you propose that ideally a mirror-combat will take 2-3 rounds. Then you calculate that it will most often take 4 rounds. You have identified that your most likely scenario doesn't yield the result you expect and then you blithely ignore that. If combat should usually take 2-3 rounds and it usually takes 4, that's a problem. As for combat entering a 5th round automatically ending in a settlement, that's hard to justify from a simulationist perspective - if the PCs are unhurt and the opponent is badly hurt (but not dead), why wouldn't the PCs continue 'grinding'?

Please keep in mind that you, as the Designer, can modify the assumptions to help ensure that things rarely go to a Round 5. A mirror match may not be the most indicative of expected opposition, but it's easier to simulate than 4PCs versus 1 monster. Eventually you will want to consider alternate scenarios, but your question must always be 'is this working the way I think it should'? It's possible that a defensive character facing another defensive character will have long grindy combat where neither one hurts each other and that might be okay if that's what you want. In that case, putting the offensive character against the defensive character might make sense - if they are supposed to have equal odds of winning you're probably going to want to give the defensive character some damage triggered by successfully defending against the offensive character - otherwise the offensive character always wins, it's just a question of how long and tedious the combat is...

In our system, at low level, characters have a single attack on their turn. However, characters also have two Attacks of Opportunity that they can make each round (the second at a significant penalty). We have a number of abilities that use AoO to power - these are specific abilities that a character chooses, such as the ability to make an AoO when someone enters your threatened area [hold the line], or to attack someone who just hit you [retribute], or to hit someone who just missed you [riposte], or to use two-weapon-fighting and getting to make two attacks instead of the normal 1 [two-weapon fighting]. From our perspective, this tends to make combats feel more dynamic and players remain more engaged (their character may take actions during their opponent's turn). It also means that average damage increases, making combats shorter (at least, if you're able to use these special abilities - which you're incentivized to find ways to do so). A low defense PC might choose [retribute] knowing they're going to be hit, but giving them a chance to make attacking them a regrettable decision; a high defense PC might choose [riposte], knowing that if they avoid the attack they're going to have a good 'attrition ratio'. We want characters to be able to use these abilities, but we also want to make sure there are limits to how often they can, and potentially force choices on the player. Using AP can potentially serve that function, but we felt that giving each character a collection of action types, each with a short menu of available choices, optimized the turn. There are some abilities that take a move action, and you don't sacrifice your attack (or an extra attack) by using them; there are some actions that are minor and you don't lose them if you move and attack. We would lose a lot if people had to choose between a minor action and an attack; and our system would too strongly reward all-out-aggression if you could transform all of your actions into attacks every round. I would suggest scrutinizing your action choice to determine if it encourages the activity you want. If I'm understanding and 3 AP can be 3 attacks (not moving or doing anything else) your PCs are going to want to let the monsters move up to them (wasting an AP on movement) so they can turn around and attack 3 times on their turn. Eventually, players will determine what optimal actions are and will gravitate toward them - if that encourages the style of play you want, that's good. If that encourages a style of play you don't want, that's bad.

Big picture, as a designer, you're trying to create a situation where PCs do the things that they're supposed to do (rogues sneak, wizards cast spells, etc) so you want to reward choices that align with those actions and discourage choices that don't. As an example of this type of thinking we wanted roguish characters to use light weapons like short-swords and not weapons like great-swords. Great swords do more damage, and you can potentially get sneak attack with them, so all else being equal, rogues have an incentive to use big-heavy-weapons. To discourage that we reward using light weapons; in all the situations you could get sneak attack you can also add Dexterity to damage with light weapons. A rogue might do 2d8+3 (plus sneak attack) with a great-sword, but they might do 1d8+7 (plus sneak attack) with a short-sword. To further incentivize the light weapon, we give rogues proficiency with light weapons; if they want to use a great-sword they must spend additional character resources. For us, rogues CAN use great-swords, but they usually don't - we feel like we've balanced the rewards for making the choice we prefer with allowing players to choose options they really want. If we had missed our mark we could have played with those incentives to balance them for what we felt was most appropriate to the assumed setting. But each change has knock-down-effects. Making light weapons more damaging means that Berserkers have an incentive to potentially wield short-swords instead of great-swords. Since Berserkers usually have a high Strength and low Dexterity relative to Rogues, they get more benefit from two-handed weapons (1.5x STR), and commensurately less benefit from light weapons. Our berserker would get 2d8+7 (plus rage damage) with the greatsword and 1d8+7 (plus rage damage) with the short-sword. If your incentives completely invalidate a character type that you want to support, you need to adjust them; likewise if your incentives make a single choice automatic for everyone, you probably need to reduce them. As the designer, you have that power.

For you, Merxa, you need to answer the question of 'what do I want' and then build solutions for that. But it doesn't make any sense to say 'I built something that doesn't do what I want, but oh well, changing it is hard so I'll let it be' - everything you've done up to this point is hard - you don't have to build your own system - so if you're going to do it, build a system that actually gives you what you want!!!

[merxa]

Appreciate the feedback deaddmwalking.

I don’t think I was blithely ignoring myself, just to repeat the prior assumptions combats should be between 1-4 rounds, and usually 2 or 3 rounds. I’m ok with combat rounds sometimes only being 1 round, but i don’t really want combats to hit round 5 or more (some exceptions come to mind, but I want to avoid complex tactical combats going this long, and if this were a rules lite game, i would consider some rule like the one i noted above which was with the intent of an aside for a rules lite idea). If any match is going to go longer, the PC mirror match should be it, at least in my opinion and desire – I want PCs to be tough by default and if the mirror match is by definition coin toss, I’d like it to be a slower one with as many reasonable choices provided as can be sustained by the combat engine. So yes, 4 rounds, the general limit, is ok for me. And if sometimes this complex combats hit rounds 5 or 6, that’s life i suppose, I’m trying to aim for 3 round average, and if the system ends up being more like 3.5 rounds on average, that seems fine, probably better then 2.5 rounds.

The current expectation is that a PC is going to get about 2.5 attack opportunities per round.


1 PC vs 2 mooks

Lets say hit chance jumps up to 80%,, so 2 attacks land a round, and if we keep the same damage, then the PC should be dropping 1 mook a round on average.

The mooks themselves need to have probably a poor chance to hit and reduced damage to not be overwhelming. If they get 5 attacks between them, i think we want the same ~1.5 expected hit chance, so 50% chance is 2.5 and pcs negate 1, and can damage stay the same? We don’t really ever want the PC losing this fight, especially on full resources, i suggested a 5% threshold, but thats probably still too high (any opinions on this value?). But returning to combat not going beyond round 4, we want their average damage to defeat the PC on round 5 in that case, so between the two of them they need to deal a total of 10 damage over 5 rounds. This suggests an average of 2 damage.

A wrinkle worth considering is that their damage output is expected to be highest in the beginning then drop in half as one of them gets defeated at some point. The expected outcome, if the mooks went first, is to deal 2 damage round 1, then 1 damage round 2, being defeated at that point. In scenarios where the mooks win, they really need to not drop the first few rounds to have much of a chance. We could split the difference, and suggest 2.5 rounds must be 2x and 2.5 rounds is x for a total of 3x = 10 over 5 rounds, so damage closer to 3.3 but i don’t believe the current system can easily hit 3.3, (it’ll either be 3 or 3.5), and we want pcs to win, so dropping it to 3 seems plausible. That would be

Round / Round number / damage/ total damage
Round 1 3 3
Round 2 3 6
Round 3 2.25 8.25
Round 4 1.5 9.75
Round 5 1.5 11.25

Lets calculate something silly, like 1 vs 5, right away we must acknowledge that the pc needs to bring down more than 1 a round to finish before round 5, 1.25 if 4 rounds, but we’d like to hit our ideal 3 rounds, so 5/3 (1.67). Returning to PCs doing 2.5 attacks a round, we’ll keep the 80% hit chance, against team monsters with a total of 50 hp, or 6 attacks must deal that damage output (8.3, but let’s round up to 8.5). This requires PCs to have some ability to spread damage, and for that spread to be more total damage than their single damage (at least 8.5 is higher than their mirror match of 5). This ability could have some resource cost, but PCs must have something like this for a 1v5 scenario to play out well. Some sort of cleave mechanic could also do this, where excess damage is applied to the next mob etc. But 8.5 spread across say 3 mobs is 2.83, so if pcs are doing 5 damage on average, some ability that does 3 across 3 mobs would have use cases, or alternatively 4.25 across 2 mobs. If we go back and say pcs would very rarely miss this group (getting an expected 7 attacks, 93% hit chance), that turns the damage spread to 7.14 or 7.5 instead, which might work a little better.

As for how much damage the mob does vs the PC, with 5 mobs the per round assumption is 5x+4x+3x+2x+x=10 or 15x=10 or .6 average damage (again, we’ll round down to .5, and this also suggests the floor as rounding down to 0 is nonsense, so situations calling for 6v1 or 7v1 won’t be well supported unless ‘swarm’ mechanics or some similar abstraction gets used instead) That average damage of .5 could be calculated in various to hit / damage ratios, but we want low variance, so low damage outputs is probably the way to do this, but if they get 2.5 x5 or 12.5 attacks between them round 1, and 10 round 2 etc, that would be 2.5 damage spread across 12.5 possible attacks, we could turn that into .5 damage over 5 hits or 1 damage over 2.5 hits, taking the second would be 20% hit chance, first would be 40%. But I Forgot the 1/attack /round reduction the pc is expected to have, that would need to increase hit chances to 48% and 28%, but with so many attackers, especially front loaded as it is, we could ignore this to again benefit the pc and ensure that, more than likely, they win this encounter.

I think i have enough sketched out to build more formal stat blocks and run some testing. But at the moment is looks like the low end of monster damage output is going to be something like 2d+1 and maybe that is where fists will start, at 2d.

[deaddmwalking]

To me it looks like you've done more analysis here than you've done in a while, but I'm surprised you forgot about negating hits. It makes a big difference if the PCs can negate 1 hit total each round versus 1 hit per opponent. If it is 1 hit total (as I surmise) than Team Monster, like Team PC, has a real incentive to coordinate/gang-up on a single member of the party.

I don't think it makes sense to set mook damage based on 5 mooks to 1 PC. Normally there will be multiple PCs, and having 20 mooks for 4 PCs may be overwhelming to track in most fights. If you have something like 2 mooks per PC + 1 boss (equal level to the PCs) as a standard encounter.

If you're insisting that the PC has 10 hit points and your preferred outcome is that the PC should be able to finish the mooks in 1-2 rounds and the PCs can focus on the boss monster in rounds 3-4, you're right to figure on expected damage decreasing as mooks are eliminated, but you'll want to consider the possibility of 'boss monster' damage as well.

While PCs should win, they shouldn't feel that it's a certain thing, or that they came through unscathed. Uncertainty is why people watch live sports - we know who SHOULD win, and the odds-makers tell you by how much they OUGHT to win, but the outcome is actually uncertain. Sometimes upsets happen. Sometimes they don't, but they look like they could.

There are games like Candyland where the outcome is pre-determined based on how the deck is sorted. There is no skill, there is no action that can change the outcome. The pre-determined result will be revealed through play, so while it's UNKNOWN it's not particularly EXCITING. At least, not for older kids.

Are mooks a threat? At least in sufficient numbers? What is their purpose in an encounter? Why have them at all?

I would suggest that mooks should be a threat otherwise the strategic optimum would be for the PCs to ignore them and focus exclusively on the boss monster. To be a threat, they should be able to hit hard relative to their toughness (that is, it is better to swat the mosquito than to let it bite you). For mechanics, how are you determining 'accuracy'? Previously I thought you were counting hits (4,5,6) on xd6. Giving more dice potentially yields more hits, but you're not considering any factors for who they're attacking? Ie, they are just as likely to hit a dragon and deal damage as they are to hit a mosquito? Other than the dragon having the ability to reduce damage more, both creatures have 10 hit points. So the mosquito needs 10 damage total (no reduction) while the dragon might need 80-160 hits (1/16 to 1/8 damage after reduction)?