Adapting mechanics from Sil to other (longer?) games

[Scatha]

There has been some discussion about how you might put some elements of Sil into a different game. During development I've thought at various points about what makes things work well, and I thought I could usefully share my thoughts on this. While I hope we've mostly made good decisions about Sil, they certainly aren't the only good decisions we could have made. I'm going to describe how you might go about doing some things differently, or incorporating certain mechanics into another game. One of the design goals with Sil was to keep closely to the Tolkienian world, but I guess adaptations are likely to drop that restriction.

The game mechanics are described in the manual , and half wrote a thread summarising the combat mechanics. I'm going to assume some familiarity with the mechanics for the subsequent discussion.

I'm mainly describing how you might start with Sil and vary it back in the direction of other games rather than how to insert the mechanics from Sil into other games. The other could of course also be done, but you might be better looking at the manual. For example I think a lot of people like the Stealth system in Sil, but I don't see any obvious variation that people might want/need to make, so I don't discuss it.

[Scatha]

Experience, Levels, Classes

Sil gives experience for:
(i) Descending
(ii) Encountering monsters (but less for monsters encountered before)
(iii) Killing monsters (but less for monsters killed before)
(iv) Identifying items

The presence of (i),(iv), and especially (ii) helps to make non-combat characters viable. The decay on (ii) and (iii) helps to limit the usefulness of grinding and give some control over the power level of characters at a given depth.

Sil eschews character classes and levels, but it would actually be very easy to put a class system on top of the current mechanics. A character class could just be a specification of how experience should be distributed between different skills. For example:

Warrior

Melee      35%
Archery    10%
Evasion    35%
Stealth     5%
Perception  5%
Will       10%
Smithing    0%
Song        0%

or

Ranger

Melee      15%
Archery    30%
Evasion    20%
Stealth    15%
Perception 10%
Will        5%
Smithing    0%
Song        5%

You could even allow the player to create a custom class by choosing their own allocation, if you wished. The increasing costs in experience to buy each skill point would mean that it's not obviously correct to just put it all in your most important sklls.

If you were going to use such a class system, levels might also be sensible. These could be very simple: at certain breakpoints of total experience, you 'level up'. At this point all of the experience you have accumulated since your last level is distributed between your skills according to class, and those skills passing a threshold will increase. The class system I've so far described doesn't make use of abilities, and you could use levels as a place to give the player points to spend on abilities. Perhaps 1 point per level, perhaps slightly more (depending on how large a thing each level is supposed to be). I think it is important to keep the increasing costs, so that the first ability in each tree costs 1, the next costs 2, and so on, since abilities in a given tree are often very synergistic with each other and/or with having a high score in the associated skill.

A common RPG trope is for health to increase with level, but it is an important part of the Sil combat system that health represents the degree of injury you can sustain, and should not increase too much through the game. I'd recommend either having no increase of health with level, or a very minimal one (a bonus one health per level, for example).

Sil uses (multiples of) triangular number costs in experience for buying skills and abilities. This works well, but it requires us to quite tightly control the amount of experience characters get. Characters getting to the bottom all have between 30,000 experience (pacifists diving very quickly) and 100,000 (extremely strong characters spending time to kill a lot), and 65,000-90,000 can probably be considered the range for normal characters. If you don't want to be so careful about this, using exponentially increasing costs for skills (with a base perhaps around 1.2? Perhaps slightly lower?) would make the system more resilient. In this case you might be better with a separate system for ability purchase. This would combine with the class system I describe above in a way which deals well with scaling levels in a longer game. If you did this you'd probably want to scale up the experience from monsters exponentially with depth rather than linearly.

[Scatha]

Magic

Sil is deliberately a low-magic game, but it would be quite easy to add more in. The most natural way would be to have a skill for each school of magic, and an associated set of abilities including the spells. Indeed that is pretty much what Smithing and Song are already; quite specialised sorts of magic.

If taking this route, I'd advise clearly defining the scope of each school and being careful about not giving it effects outside of this. You want to avoid having any one skill alone be capable of dealing with everything.

I think the spells would want to satisfy two properties:
- Each spell should depend in some way on the associated skill (could be opposed rolls, or increased effect; see the way we've done the different Songs for examples).
- It's generally best to avoid having spells which are upgrades of other spells. You might make exceptions when the obsoleted one is a prerequisite for the new one.

For example here's a possible ability tree for a school of magic.

Destruction
(X = score in Destruction skill)

Stabbing Pain
Prereq: Destruction 2
3 mana
Roll X against target's Will for 2d4 damage and slow target for 2d4 turns. Any target in sight.

Immolate
Prereq: Destruction 3
2 mana
Your next melee or archery attack will deal (X/2)d4 fire damage to target if it connects

Lightning Bolt
Prereq: Destruction 4
5 mana
Fires a beam for 6d4 fire damage. Targets get to roll Evasion vs X to avoid damage.

Ice Blast
Prereq: Destruction 6
4 mana
Deals 4d4 cold damage to all adjacent enemies. Roll X against their Will to knock them back a square.

Instrument of Destruction
Prereq: Destruction 8
Prereq: Immolate or Lightning Bolt
(Passive) Get a bonus of X/5 to your melee and archery

Dragon's Breath
Prereq: Destruction 10
Prereq: Immolate
8 mana
Breath a cone of fire for (X/4)d4 fire damage (decreasing with distance). No chance to dodge

Earth's Fury
Destruction 12
Prereq: Ice Blast or Dragon's Breath
10 mana
As activating a staff of earthquakes with X instead of Will

Death
Prereq: Destruction 14
Prereq: Stabbing Pain
8 mana
Touch an opponent (melee vs evasion) to force them to make a save of Will vs X or die

Note that of course that's not carefully balanced: I just made up some spells and wrote down some numbers that looked reasonable. The main point was to show how you could get lots of different effects which don't dominate one another even within the school of magic which is about killing things. Indeed I have probably gone too far; it would be more elegant to at least somewhat unify the mechanics of what saves are allowed against damage.

Sil is particularly low on the availability of magical escapes. This affects how combat plays. If you wanted to have magical escapes, such as teleportation, you might consider having them as a limited consumable rather than a reusable spell.

Voice (mana) is dependent on Grace. You might want to keep this dependence, or reduce the size of the effect. You might also give an increase to voice/mana with level, but similarly to health I'd be wary of putting a large effect here, since high-level characters will still be using early spells, for big effects since they have higher magic skills, but presumably the same mana cost.

Other schools might deal for example with illusions and mind-affecting spells, with healing and protection, or with knowledge and detection. But that might well be too many! If magic was an important component in the game and Grace became obviously the best stat, you might think about splitting it into two mental stats.

[Scatha]

Game length

Here are three measures of game length:
(A) how many levels there are
(B) how long the game takes to play
(C) how much the player powers up over the course of the game

(C) is the hardest to pin down precisely, but the most relevant for questions of how to balance the game mechanically. (A) is pretty irrelevant. (B) does matter a bit, since in a longer game it's more important to avoid instant or unavoidable deaths.

Sil has reduced (A) and (B) substantially. As a way of measuring (C) you might ask something like the following: how many times over do monsters become tougher such that a single monster is as much of a challenge as a horde from the earlier depth? I think this happens of the order of 5 or 6 times in Sil (Wolf --> Orc Warrior --> Warg --> Cat Warrior --> Vampire Lord --> Great Fire-Drake), which means every four levels of depth or so. I'm not sure how many times it happens in Angband.

(B) and (C) are somewhat separable from each other, but not entirely. If (B) is too high compared to (C) the game may stagnate. If (C) is too high compared to (B) it may feel silly. I think Sil keeps (C) relatively high in playing time terms, and that this is mostly a good thing. The higher (C) becomes in absolute terms, the more work there is in balancing the game. Note that Sil also keeps the rate of increase of player power fairly constant through the game, while it is common for this to slow down later in the game. I'm not sure how much this is a good thing, and it may depend on how much of the enjoyment of the game is from improving your character compared to dealing with combats while at a fixed power level.

The turn clock in Sil tends to provoke some strong feeling: some players love it, and others hate it. You wouldn't actually change the game all that much by removing it.

If you were making the game longer you'd of course need more monsters (you might also want to vary existing monsters). My main advice would be to extend the current progressions in monster stats as a starting point, and then tweak. I also have a couple of spreadsheets we used to evaluate how dangerous monster attacks were which I can share if anyone's actually doing this.

[Scatha]

Combat system

I think the majority of the combat system from Sil could easily extend to a different or longer game. There are some things you might want to do differently.

Mostly the combat system in Sil is scale-free: adding N to everyone's scores doesn't change anything. There are some exceptions, for example in various situations a score is halved, or some Abilities let you add two scores together. If you were going to make the game longer in the sense (C) you might consider some of these halvings with a fixed penalty (-10?). If you were going to make the game much longer you would almost have to do this. Likewise you would want to be very careful of Abilities which let you use a sum, as they could become extremely powerful (in some cases this might be fine).

We've quite carefully balanced the different equipment choices (light vs heavy armour and weapons) against each other, so you might well want to take these, at least as a starting point. I also have some spreadsheets we used to balance weapons which I can share if anyone wants.

Opposed rolls

Sil uses opposed rolls (skill+modifiers+d10 vs skill+modifiers+d10) for almost all of its mechanics (Combat uses d20s because it turned out the combat skills were just far too good with d10s). I think this works very well overall, but it has some peculiarities. If you can beat your opponent's skill+modifiers by 10 (or 20 resp.), you automatically succeed. In particular this makes it a little fragile for balancing. (But it does help give interesting decisions with increasing marginal returns for specialising in a skill.)

Possible variation:
Replace the dX in the opposed roll with a number drawn from a geometric distribution with parameter p.

Why? Makes the balancing more robust, since it becomes impossible to become untouchable on any axis. Makes the value of each skill point easier to understand (and more uniform, which could be good or bad). For example if p=0.1, if you are likely to win a roll, each extra skill point reduces the chance of failure by 10% (relative, not absolute).

Disadvantage: It's less easy to explain to the player.

Note that p can be used as a balancing parameter, but for elegance and comprehension reasons it's probably best not to use too many different values in the game.

[half]

You wouldn't actually change the game all that much by removing it.

I think this is a good point. Removing the time limit in some ways trivialises the game (e.g. makes borg-like play possible), but mostly people just choose to make the game harder for themselves anyway.

I'd consider some form of option for this. For example, I have no problems if people want to change the code to remove the time limit (which is easy to do). I might even make it easier to do (a compile time option), or add a play-time option. Though if it were the latter, I'd probably only do it if I was reinstating the cheat menu, since I would want it to be clear that it was the opposite of a challenge option, and that it can encourage boring gameplay.

[debo]

This so awesome, Scatha. Thank you so much!

I had a bit of a pen-and-paper Sil campaign going with a few people (hilarious, I know) -- ruleset was limited but that's what got me curious about how to scale the game to a longer haul. Your commentary on classes/archetypes and magic was very instructive as well, and was similar to the tree I was starting to bark up

[Scatha]

This so awesome, Scatha. Thank you so much!

I had a bit of a pen-and-paper Sil campaign going with a few people (hilarious, I know) -- ruleset was limited but that's what got me curious about how to scale the game to a longer haul. Your commentary on classes/archetypes and magic was very instructive as well, and was similar to the tree I was starting to bark up

You're welcome. Kind of fun setting thoughts down, and there's a chance they'll end up being useful to someone someday. Do let us know if you pursue that ruleset further!

[T-Mick]

That was a good read. I love the mechanics of Sil, and reading about them is always interesting. I especially like the part about how classes could be done. I may use that in the future, if you don't mind of course. Honestly, though, I didn't understand the part about opposed rolls and "p". Not my forte, but no matter.

[Scatha]

Hmm, I think I didn't write very clearly in the section on opposed rolls. But it's essentially technical geekery, and if you didn't follow it my guess is you wouldn't find it that interesting.

As far as I'm concerned you're welcome to take any of the ideas here and develop them.

[getter77]

I can't think of a good reason why the base Angband, nor most other Roguelikes really save for special cases, would not benefit from pretty well wholesale copying of the EXP gaining mechanics at the very least---probably perhaps only needing to tinker with the tapering off effects as the kills pile up for given types when there is 100 floors or whatever versus Sil. Hell, I already pretty much advised as much for LambdaRogue some months back.

It is indeed a Eureka aspect in terms of "Damn, why hasn't this been with us since a decade+ ago?!"

[Scatha]

I can't think of a good reason why the base Angband, nor most other Roguelikes really save for special cases, would not benefit from pretty well wholesale copying of the EXP gaining mechanics at the very least---probably perhaps only needing to tinker with the tapering off effects as the kills pile up for given types when there is 100 floors or whatever versus Sil.

I think the main obstacle to this working more widely is not the number of floors, but the number of enemies you can encounter at any depth. You need to expect to meet several of most types of foe for the decay to be meaningful.

[fizzix]

I can't think of a good reason why the base Angband, nor most other Roguelikes really save for special cases, would not benefit from pretty well wholesale copying of the EXP gaining mechanics at the very least---probably perhaps only needing to tinker with the tapering off effects as the kills pile up for given types when there is 100 floors or whatever versus Sil. Hell, I already pretty much advised as much for LambdaRogue some months back.

It is indeed a Eureka aspect in terms of "Damn, why hasn't this been with us since a decade+ ago?!"

I like this mechanic also.

Most games do have an EXP falloff, since you gain less EXP as you gain levels. So killing orcs for exp isn't really viable at clevel 25. You kill them for the stuff they drop. However, a strict falloff has the same difficulty of balance as a lot of the Sil mechanics when the player is allowed to hang out on levels forever.

Still, I wouldn't be opposed to someone drafting and implementing a new EXP system for v4. It could also give EXP for finding items and seeing monsters like Sil does. Might be tough to balance though.

[Nick]

I have seen the XP fall-off suggested before (and was planning to use it in Beleriand); the bit I think is genius is the experience for encountering monsters. That I am pinching.

[Magnate]

Still, I wouldn't be opposed to someone drafting and implementing a new EXP system for v4. It could also give EXP for finding items and seeing monsters like Sil does. Might be tough to balance though.

Yes, I'd like to see this kind of thing too - it could go hand-in-hand with a new scoring system.

The one aspect of Sil I don't like (and remember this is on paper/in theory, since I haven't actually played it yet) is the twin sets of rolls for combat. I've played a lot of combat games, and I prefer there to be one roll for attack and a constant value for defence (AC or evasion or whatever). Two opposed rolls makes the mental arithmetic of estimating your chances much too hard. Of course this may be the idea!