Thursday, 8 May 2014

More necromancers

So a while ago I imagined a Necromancers game and proposed a broad skills-based system. Dan disagreed and has put forward a class-based game with specific abilities. After much distraction doing other stuff, I have an initial idea for this that I’m going to play around with.

Although I originally mentioned a BRP-like system, and I still think that would work, I've been playing around with ideas for theme-based organic magic recently. Which is to say, ways of combining small numbers of generic keywords to create a variety of interesting effects. I feel like this could avoid the need for massive lists of specific spells, and allow player creativity,


The basic idea is that magic – which should be the majority of the cool stuff going on – will be modelled as a combination system. There are Talents (what you do) and Spheres (what it relates to) which can be combined to model a wide range of necromantic magic, without creating specific spells. The Talents are Commune, Evoke, Manipulate and Assume. The Spheres are Darkness, Essence, Bone and Beast.

Spells are cast by rolling up to 5 d6s, at the caster’s choice. Each roll of 4+ is a success. On a double, the spell misfires in some way – low numbers will tend to sputter out or go awry so they have little effect, while high numbers go out of control, possibly doing what you intended but also wreaking havoc. As a result, rolling more dice will produce a more powerful spell, but will also increase the risk.

By itself, the misfire rule would be annoying. However, the caster’s aptitude makes a great difference. This allows you to ignore doubles, based on your Mastery, making it advisable to stick to your favoured spells.

Ignore numbers

You rank the Talents and Spheres from 0-3, and when casting a spell, add these numbers together to find your, um... Mastery? Sure, why not. You can ignore any doubles less than or equal to your Mastery. This means that your best combination can always be safely cast (3+3=6), your worst combination will be pretty risky except at low power, and most spells can be cast with 2-3 dice for only moderate risk.

Of course, this means there’s always a risk for anything but your very best combination. Or do I want to have improved skill cancel out specific numbers of doubles, making casting below your skill level always completely safe? Let’s look at some numbers.

For 2d6, the probability of a double is 1/6

For 3d6, the probability is... ouch. Okay, I can’t actually find any guidance on this question from people who are willing to talk the right level of maths. Which is to say: I can find people who answer specific questions about this with specific numbers, and I can find people who refer obliquely to complicated maths, but not anyone who will lay out clearly how I would calculate this stuff for myself. I’m pretty sure it’s a factorials thing, but I’m happier with my ability to tediously lay out charts in Excel than with my ability to guess at statistics. And doing factorials with fractions is a nightmare.

So it looks like probabilities go something like: 0.1666..., 0.444..., 0.722... and I can’t be bothered with the ghastly cut-and-pasting required to do 5 dice, but it should be around the 0.9 mark. Assuming I didn’t mess up. The figures match the specific figures I got elsewhere, so great.

This means that if you’re casting a four-dice spell, you have a 72% chance of something going wrong. Unless you’re casting your very best spell, there’s at least a 12% chance of a miscast, and this will generally be more like 36% as the average spell rating will be 3. Okay, you’re not likely to cast your rubbish spells that often, but basically you have a substantial chance of problems on any but your most favoured spells, although in many cases this will be a case of overpowering them rather than wasting your action, because low doubles get lost first.


What about cancelling dice instead? That is, on favoured spells you can ignore specific dice for the purposes of doubles, rather than doubles with certain numbers? So if you rolled 4, 4, 4, 5, 6 you’d be able to ignore the triple if you had two Mastery. I’d need to adjust the ranks, probably using 0012 rather than 0123.

That’s... at least equally difficult to calculate, if not more so, because I have to care about number of pairs rather than their existence. Thankfully, someone has done it because Yahtzee exists, although again they only vaguely mutter about binomial expansion without explaining it.

With only 3 dice, the most likely common minimum (rolling 2 dice for 4+ doesn’t seem like a gamble many people will take often), there’s a 3% chance of needing 2 mastery and a 42% chance of needing 1.

For four dice (I did this one first, which is why it’s longer): Four identical = 6/1296, Three identical = 120/1296, Two pairs = 90/1296, One pair = 720/1296

So there’s a 6/1296 chance (0.00463) of needing 3 Mastery, a 16% chance of needing 2 Mastery (a triple or two pairs can both be cancelled by negating two dice), and a 55% chance of needing one Mastery.

With 5 dice, there’s a vanishingly small chance of needing 4 Mastery, so small that I don’t think it’s worth worrying about ever. There’s a 5% chance of needing 3, a 38% chance of needing 2, and a 46% chance of needing 1.

On the whole, then 2 Mastery will be enough the overwhelming majority of the time. I suppose I could introduce a rule that when cancelling pairs, you have to cancel both dice of the first pair before you can cancel the second pair, but that seems a bit clunky.

Highs and Lows

A simpler alternative would be the ever-popular 1s rule. Rolling a 1 would make the spell fizzle, but Mastery allows you to ignore 1s. Similarly, 6s might make the spell a bit more powerful than you intended, but Mastery absorbs that (whether you want it or not) although they still count as a success. If you roll a 1 and a 6, you could decide which to cancel first. You’d of course get more 1s and 6s with more dice, but with Mastery you can ignore many of them. This is probably a lot simpler, I wish I’d thought of it before. But I probably want to use d10s rather than d6s here, and go for a 5+ success.


Here's a table of the Spheres and Talents, with some example spells that might be possible.

Darkness Essence Bone Beast
Commune See in darkness
Detect shadow beings
Detect souls
See ghosts
Speak with dead
Find bodies
Sense skeletons
Forensic necromancy
Sense sinister animals
Speak with beasts
Evoke Conjure darkness
Cause blindness
Cause fear
Animate dead
Death bolt
Conjure skeletons or skeletal constructions Summon bats, wolves, rats, cats
Manipulate Alter shadows
Create shadow servant
Influence creature
Compel Ghost
Warp enemy
Shatter bones
Turn into cloud of bats
Assume Travel through shadows
Shadow aura
Heal pain or cuts
Drain life
Ghostly “rider” aids you
Grow skeletal claws or wings
Mend bones
Take on bestial traits

Draft system

So, under this draft I'm using the high-and-low system.

During character generation, you assign numbers 0,0,1,2 to each of the Spheres and again to each of the Talents. When casting a spell, you determine which Sphere and Talent combination it falls under and add these scores to find your Mastery.

To cast a spell, you roll between 0 and 5 d10s. A score of 5+ grants you one success, making the spell more effective. To achieve significant effects, you need larger pools. The choice of how many dice to roll is entirely up to you; you decide how much power you want to try and draw on.

Rolling extra dice is risky because you're more likely to lose control of the spell. Any roll of a 1 is a Fizzle, as the spell dissipates prematurely or goes awry in some mundane fashion; there may be minor effects from the magic, but effectively it's a dud. Any roll of 10 is a Miscast, as the spell's energies overwhelm you and you lose control; the spell goes off and will typically achieve more or less what you wanted (10 is still a success, after all), but it's also going to have unexpected consequences.

Mastery helps you to control your spells without untoward happenings. You can ignore a number of Fizzles and Miscasts equal to your Mastery - this doesn't affect the number of successes you get (the dice aren't discarded), so rolling 10,10,10,2,4 with Mastery 3 is still a successful cast.

Additional successes don't have to change the result of the spell. If you only want to achieve a minor result, and roll four successes, it doesn't force you expand the spell's effect - it's not much of a success otherwise.


  1. I'd be a bit concerned about a system under which rolling more dice increases your chance of failure. I haven't run the numbers on this but since every die adds a one-in-five chance of something going wrong, I can see that getting very unpredictable very quickly. Think how irritating we found (okay, primarily I found) the ones-subtract-successes thing in Demon.

    1. I'm well aware of that - that is specifically what the Mastery is intended for. Unless I've messed up horribly somewhere, you are *entirely* safe provided you don't try to push your limits, and have pretty good odds of succeeding even at stuff you're mediocre at. The idea is to encourage you to specialise in certain types of necromancy to help define the characters, without actually banning you from doing other stuff.

      Ah, yes, I remember. The system I shifted to does mean you can't cast even your best spells with five dice without a *very* slight chance of something going wrong, as you can only ignore 4 dice. Unless you roll five 1s, though, you'll probably pick a miscast over a fizzle, so it's success with attendant problems.

    2. Fair point, and if I'm reading it correctly, it's not like you have to pay points to *get* extra dice, so in that sense it's less annoying - rolling extra dice is specifically supposed to be a gamble.

    3. That's right, it's entirely your choice - how much dark power do you want to call on? How much can you handle? How badly do you want this? I kind of felt that being good at stuff should make you better at stuff, rather than more powerful, if you see what I mean. Plus, it fits with a lot of canon - amateurs messing with necromancy is never a good idea.

      I may actually swap the target number to a 3 if I'm sticking to D6s, because casting spells is kind of the whole point. That might help encourage people not to always grab for five dice just to make sure they cast at all.

    4. Again, I might suggest that using an add-up system rather than a success system might help here. The problem with 3+ is that it still has a largeish failure chance for small numbers of dice, but also makes it very likely that you'll get large numbers of successes.

      If instead you add up the dice, and have 4/8/12/16/20 corresponding to 1/2/3/4/5 successes, you drastically reduce the chances of a flat failure. Your odds of getting one success on two dice a 75% at 4+ and 89% at 3+, while your chances of rolling a 4+ on two dice is 92%.

  2. Thinking about it in more detail:

    I think I'd be inclined to stick with D6es, and to ditch "fizzles" completely. You already have a chance of the spell failing simply because you didn't roll enough successes, adding an auto-fail because you rolled a 1 just doesn't seem to add anything interesting compared to an auto-chaos on a 6.

    Incidentally you could switch from a success count to a sum (roll as many dice as you like, add up the result) which might allow for a little more leeway.

    Overall a very solid-looking system though. You might want to have a look at Ars Magica (which uses a very similar Noun-Verb setup).

    Something else that struck me is that if you want a modern game where you're all Necromancers, White Wolf's /Geist: the Sin-Eaters/ is, in fact, exactly that.

    1. True enough, seemed like an idea at the time but on reflection it's not really interesting. I can't remember if I was going to do something else with it. Die size I'm not really bothered about either way; D6s are easier, I went with d10s both because I was using fizzles, and because it allows that little bit of generosity on successes.

      Ah, it would allow for more leeway, but it would also need me to rebuild the game. At the moment it's very coarse-grained and relies on interpretation; I can sort of gauge what a 2-success spell might look like compared to a 4-success one, and the raw numbers can be applied to the equally coarse-grained attributes. But even two dice allows an 11-point spread and five allows a 25-point spread, as well as introducing a strong bell curve with very unlikely but extreme outcomes. Mechanically handling differences between results would mean making other parts of the system finer-grained to compensate, I think. I suppose you could just add up and every N points means one success, but that's adding maths, which you don't like :) Especially if you're aiming for 4s to keep the probabilities the same.

      Glad you like it. I do mean to look at Ars Magica sometime, but I have a lot of RPGs still to read that I already own...

      Hmm. Geist might be interesting, I did enjoy Demon. Although I'm guessing you probably aren't campy fun necromancers, and possibly not really very necromancery as a PC, if Demon's anything to go by?

    2. As I mentioned above, I think you can get effectively the same effect by having treating 4/8/12/16/20 as equivalent to larger numbers of successes. It isn't any finer-grained, because you don't have to sweat the difference between a 4 and a 5 or a 16 and an 18, and it has the advantage that you don't need at least X dice to get X successes.

      As for Geist, I haven't played it, but I think it might feel more necromancery than Demon felt demony because a necromancer *is* a human being with special powers, so the fact that you wind up playing, well, a human being with special powers, fits pretty well.

    3. I'm a bit puzzled here, because pretty recently you were arguing very strongly against mechanics that call for maths, and now you're in favour of one that's relatively fiddly (gauging fours is generally harder than fives or tens). I'm not trying to have a go, but I'm curious about the shift of opinion there, if you feel like discussing it.

      I'm not sure that I actually consider it an advantage not to need X dice to get X successes. Not sure either way, mind you. In some cases it might make sense (lucky timing on a spellcast, hit a vulnerable spot, etc) but in others it seems like it works against the (very rough) magic system, especially with any kind of enduring effect, because the power of your spell is no longer limited to the power you put into it.

    4. Ah, I just remembered why I kept Fizzles - because otherwise it's quite tempting to cast high-powered spells outside your chosen spheres and talents, since there's a good chance that the spell will achieve your aim even if it miscasts - that's written into the miscast rule, in fact. With fizzles, adding more dice is more of a gamble because a Fizzle means you wasted your time.

      I could probably handle that by coming up with some guidance for what miscasts should actually do, not sure yet.

  3. I'm a bit puzzled here, because pretty recently you were arguing very strongly against mechanics that call for maths, and now you're in favour of one that's relatively fiddly (gauging fours is generally harder than fives or tens). I'm not trying to have a go, but I'm curious about the shift of opinion there, if you feel like discussing it.

    It's a balancing act. I agree that counting successes is faster than adding dice (and I suspect that counting 1/2/3/4/5 is easier than saying "is it above this, this or this threshold") but I think there's an advantage here because of the way the statistics play out, that is, it's easier to get a 90+% success chance for simple tasks, with relatively small numbers of dice, while having a comparatively low chance of a miscast. It also just feels more thematic to me, because each extra die *necessarily* makes the spell more powerful, rather than having a 50% chance of making the spell more powerful and a 50% chance of doing nothing.

    The probabilities here are nontrivial, but as it currently stands your chances of getting X successes on X dice are 50%, 25%, 13%, 6% and 3%, which makes your chances of actually getting large numbers of successes relatively low, regardless of how high your Mastery is, or how many dice you roll. It partly depends on how often you expect PCs to *want* to achieve 5 successes.

    1. To be honest, I’m not sure how much thought I want to put into this one rather than Monitors. I mostly threw it out there for the spell-making system and I’m not that invested in thrashing out a full set of mechanics at the moment. As you might have noticed, I left everything pretty blunt and vague elsewhere.

      Off the top of my head, I’d argue that if a summing system seems preferable then it would actually be best to revert to the d10s idea – and restore 1s as a failure condition – because you can then work on 5s rather than 4s.

      In terms of successes, again, I’d have to put quite a lot more thought into what spells actually do and how those numbers interact with other numbers. Presumably PCs will *want* to achieve as many successes as possible all the time, so it’s more about how often they *need* to do that, or how much of a drag getting small numbers of successes is. As I said though, probably not going to dig into this too deeply right now.

    2. Sorry, not intending to keep dragging you back to this, it's just that I actually really like the core mechanic so I'm keen to see what can be done with it.

      With an adding system, I might suggest that it could be worth keeping the D6es, but counting by 5s to (a) simplify maths and (b) increase the chance of miscasts. I've been thinking about it and I think that with a five die limit, Mastery is just a bit too good. It's currently never worth rolling less than your Mastery, and it's pretty safe to roll up to 6x your Mastery (3x with Fizzles on D6es, 5x with Fizzles on D10s, 10x with no fizzles on D10s). A character with Mastery 2 will only Miscast if they roll three sixes on five dice, after all, and that should only happen (I think) 1.9% of the time.

      I might be inclined to up the dice limit to significantly above the Mastery limit, and either have an adding system, or else simply require that characters not score more total successes than their Mastery (this would make it safeish to roll up to 2x Mastery, but you could push it if you wanted to).

      Still, really cool system, and there's a lot to play around with in it.

    3. Oh, no problem, I don't mean to shut down the discussion - just that one full-blown RPG at a time is probably my limit, so (for example) actually working out how many successes you'd need, which involves working out what successes really mean in terms of other bits of mechanics like social and combat interactions, which involves giving some serious thought to the point and feel of the game, is a bit more intense than I want to get just now.

      Glad you like it though, and it's good to get your thoughts (and your maths!). I'm liking the idea of larger pools and adding fives - technically there's actually no reason you need a cap at all in this system, because past a certain point either you win everything forever or you mess up badly.

      In a capless model though, I'd probably want each bit of additional power to make your life worse if you miscast. In that case perhaps have very small numbers of extra dice cause slight glitches, smallish numbers cause fizzles, and larger numbers cause increasingly serious problems? I dunno, it's late.

    4. I think *some* kind of cap would be helpful, if only so you could play with things that lift that cap - e.g. you can roll up to 6 dice with your personal power, but up to 12 if you Do A Big Ritual are Perform An Appropriate Sacrifice.

      For what it's worth, many (many many) years ago I wrote a game called /Affairs of Wizards/ with a similar setup, and one of the key premises in that was that it *only* had a magic system, that is, the game wasn't interested in anything your characters didn't do with magic. (It was based partly on an attempt to refute the old saw that Gandalf does hardly any magic in Lord of the Rings, my attempt was to argue that Gandalf is actually doing magic *all the time*, it's just that he mostly does it by talking to people).

      I think for a system like that it's important to be consistent about what a "success" means. Since it's about *raw power* it seems clear to me that a success means a more powerful magical effect, and *not* an effect that is closer to the character's intent. In which case a miscast with more dice would always be worse, because it would (almost) always have more successes.

    5. I actually disagree, to me it’s unintuitive for “successes” to translate into more raw power regardless of the situation and intention. I see a clear distinction between how much magic you attempt to control (dice rolled) and how useful an effect you can derive from it (successes). Often that would translate into a more powerful effect, like impressing a crowd or learning more from a divination. But I find it unhelpful to enforce that, because otherwise deliberately producing minor effects will be *as difficult* as producing major ones, since you’d have to carefully guess the right number of dice to avoid overdoing it – and having high Mastery in the chosen spells would not make you much more likely to achieve your aim. There will also be situations where “doing it better” is most intuitively interpreted as doing something more elegantly, more subtly and so on rather than with more raw power.

      That disagreement isn’t necessarily a problem, because I do think a miscast should be more serious when you were drawing more power. It probably calls for a chart.

    6. Ah, you see to me since "successes" come directly *from* the amount of dice you roll, it seems unintuitive for them not to mostly translate to an amount of raw power. Since the only way to roll multiple successes on a spell is to roll lots of dice, it follows that the only way to get a lot of successes is to channel a lot of raw power.

      It would seem intuitively wrong to me if the way to exert really fine control over a task was to throw a lot of raw power at it.

    7. Fair enough, but the issue I see there is that otherwise there doesn't particularly seem to be much player control over achieving their intent, only indirect influence over how powerful a spell they cast. Admittedly, not sure how much of a problem that's likely to be without fleshing out more of the game.

    8. I think to an extent that's true in any RPG. I mean in - say - Cthulhu you don't have much control over whether you achieve your intent other than success/fail.

      FWIW I wasn't necessarily suggesting that a non-miscast should automatically overshoot a players intent, just that it doesn't make sense to me to require lots of successes to be subtle or discreet, since the way you get lots of dice in the first place is to draw on a lot of power.

      You *could* have a system where your aim was to score exactly n successes, and Mastery could allow you to negate excess successes as well as miscasts.

    9. Based on that example, we may be arguing semantics here. I mean, in Cthulhu you try to do X with your skill and roll to see if you succeed; the way I was interpreting your comments, it seemed like you'd be deciding to cast a spell with the aim of achieving X, and rolling to see what the spell actually did.

  4. Oh also: doubles probabilities. I think it's fairly straightforward, in that it's 1 - the probability of them all being different. So for a D6 it's:

    1: 1 - 1 (one die is never the same as itself)
    2: 1 - 1*5/6 (the second die has a 5/6 chance of being different from the first die)
    3: 1 - 1*5/6*4/6 (the third die has a 4/6 chance of being different from each of the first two dice)
    4: 1 - 1*5/6*4/6*3/6 (should follow fairly clearly from here)
    5: 1 - 1*5/6*4/6*3/6*2/6
    6: 1 - 1*5/6*4/6*3/6*2/6*1/6
    7: 1 - 0 (the seventh die must always equal one of the previous dice)

    For n D6es the probability is:

    1 - 6!/((6-n)!*6^n)

    For n DXes it should be:

    1 - X!/((X-n)!*X^n).

    1. Ah, thanks, there we go. I had the broad strokes, but translating them into factorial-speak is something I haven't done for ages.

      For future reference, to do this in Excel you use the formula:

      Where row 1 contains die sizes starting in column B, and column A contains dicepool sizes starting in row 2.