Sil

[half]

It is too easy to get 100% reliable skill check with any skill.

This is either a gross over-exaggeration, or you are missing some details about how the skill system works. You say that it is 'too easy' to get '100% reliable' in 'any skill'. Not only is not too easy, but it has almost never been achieved, and as far as I can tell, only in one skill, and not by you. In order to clarify for others who might be interested...

How high do you think your evasion needs to be to be 100% reliable? 30? 40? The answer is 54 (or 108 against monsters you can't see). You need to be 19 above their melee, and the highest melee is 35. I'd be surprised if anyone has ever achieved this in Sil (the highest appears to be 38, letting you get hit almost half the time by Thuringwethil). Moreover, it doesn't protect you against earthquakes or fire breaths etc., and is halved when you can't see them etc. etc.

Melee/Archery need to be 49 (or 98 if you can't see the opponent).

Perception checks get arbitrarily difficult.

Will needs to be 30.

Song checks get arbitrarily difficult, but even at range one, you need 37.

Smithing doesn't have checks.

The only one that has ever been achieved is Stealth, where you need 32 or so, and this has only been done a couple of times! (Note that it is hard to say precisely, and there are times when an even higher number is needed, such as while singing).

The real problem with Stealth is just in the high end, not once you have achieved 100% chance, and it is more to do with synergistic Abilities like the Song of Lorien, the Song of Silence, and perhaps most of all: Vanish. We have already increased the perception of some late game creatures, but are fairly happy with Stealth at the moment. We want Stealth to be powerful as it is canonically the only way Morgoth's throne room has been reached. Moreover, Sil has a much richer stealth gameplay than most other roguelikes and we want to make the most of this.

There are several other balance issues with Sil that we are addressing, such as various things concerning each of Smithing, Charge, and Two Weapon Fighting. We pride ourselves on having a well balanced game, and welcome constructive criticism (or praise) of the systems, but please let's keep it constructive.

[Scatha]

Yes, stealth is even worse. Evasion was just an example. It is too easy to get 100% reliable skill check with any skill.

Also, low values of skills are completely useless.

I agree Stealth is a little bit of an offender here. We're spreading out the monster Perception scores a bit for the next version. This should both make a moderate investment in stealth more useful, and total reliability less attainable.

Since the player doesn't have enough resources to get 100% reliability with every skill, it isn't necessarily problematic if they can get total reliability with some skill, as long as this doesn't make the game trivial for them (or it could be that the challenge is in getting to that point of total reliability). Evasion was a possible offender here, but the combination of things which ignore or halve evasion, and monsters which have such high melee you can't realistically dodge them reliably is I think enough. Stealth goes in this direction, but fast enemies or corridors can provide problems with getting trapped/bumped into. Some of the Songs can be used to control fights, but it's not easy to get them high enough to be reliable. For the other skills, it's hard to see how a 100% success rate could by itself be a problem.

There are more uses for low skill investment, too. In fact for just about every skill there are reasons you might pursue this (even apart from as prerequisites for abilities).

Here are examples for the ones Patashu didn't mention:
- A few points in Melee may be a good investment for an archer using Versatility.
- Enemy evasion is halved against Archery, and just a few points here can keep this a viable ranged attack throughout the game.
- Even high protection characters, who are happy to get hit all the time, may still invest a bit in Evasion to reduce the severity of critical hits taken.
- A small investment in Stealth will lower the radius at which you wake up monsters. Even if you're never attacking unwary monsters, this can mean fighting small groups rather than half the dungeon.

[LostTemplar]

Some skills are ok, but not all.

[Patashu]

Yes, stealth is even worse. Evasion was just an example. It is too easy to get 100% reliable skill check with any skill.

Also, low values of skills are completely useless.

I am playing around with stealth atm and I agree, it's really really good, unlike melee/evasion it seems to scale faster than enemies do.

I can think of some uses for low skills, though:
-You only need a bit of perception or will to make the relevant checks go from 'will never succeed' to 'will succeed often enough to satisfy you'. Plus, if you're only seeking a low level skill in either you can make the dip.
-Getting enough forge for weaponsmithing for a bow.
-Getting enough song for song of slaying.

[LostTemplar]

I would argue that high stealth is more "broken" than high evasion

Yes, stealth is even worse. Evasion was just an example. It is too easy to get 100% reliable skill check with any skill.

Also, low values of skills are completely useless.

[Patashu]

How shadow casting creatures work:

Anything that emits light or shadow emits it with this distance function: longer distance + half shorter distance (rounding down)

...33..
..2223.
.211123
.21@123456789
..11123456789
...223456789
...334456789
...44556789
...55667789
...6677889
...778899
...8899
...99

0) Cast ambient light
If a tile is lit up naturally, it starts at 1, else it starts at 0.

1) Cast the player light
You cast a light level of (your radius + 1 + (1 if Inner Light is learned)) on your own tile, and for each shell outward subtract 1 until you reach the end of your radius. For instance, if you have a brass lantern and helm of brilliance you have a radius of 3, so you cast light levels in your radius like so:

4321
3321
221
11

And if you have inner light, that becomes:

5432
4432
332
22

2) Cast monster lights and object-on-ground lights
Monster lights are cast the same way your light is, except if their light is negative they subtract it instead of adding it.
Items also cast light if they're of brilliance, slaying (with their slay present) or if they're feanorian lamps/lesser jewels/silmarils.
Shadow cloaks cast darkness. (Do lanterns of shadows cast darkness? I think the answer is no)

Shadow radiuses for monsters are as follows:
-1: Shadow worm mass, Spider of Gorgoroth
-2: Shadow mold, Lurking horror, Shadow bat, Shadow spider, Creeping horror, Amethyst serpent, Lungorthin, the Balrog of White Fire, Shelob, Spider of Darkness, Gothmog, High Captain of Balrogs
-3: Ancient amethyst serpent, Gwathrauko
-4: Ungoliant, the Gloomweaver

So a shadow worm mass casts the following shadow, subtracting these numbers from the light levels of every tile:

21
11

And a shadow mold casts this shadow:

321
221
11

And so on.

3) Can you see the tile now?
If it is 0 or less, you CANNOT see it. If it is 1 or higher, you CAN see it.

For example the shadow mold in the tutorial can barely be seen standing next to it with song of the trees: lit room (1) -3 (shadow mold is -2 radius, so -3 on itself) +3 (you're +3 radius no inner light, so +3 next to you) = 1 (you can see!)
A true expert of the tutorial will ignore that he has song of the trees, grind just enough exp to get inner light and use his orc slaying sword with an orc archer to get 1(radius)+1(inner light)+1(orc slaying light) = 3 while standing next to the shadow mold and thus light it up

4) Formula for seeing lone shadow creatures
Call the shadow radius of the monster S
If you have inner light and are in a lit room, S light radius is enough
If you are in a lit room OR have inner light in an unlit room, S+1 light radius is enough
If you are in an unlit room without inner light, S+2 radius is enough

Don't forget about Song of the Trees (adds 1 light radius per 5 song score), Staves of Light, Bows of Brilliance or crafting <foo> of brilliance if you need more light!

Secret techs for combatting shadow monsters (all of these work):
-Wielding slay foo weapons for the extra light when their target is around
-Dropping items of brilliance/slay foo/permanent light sources around you or under you to add to the light level
-Luring light bearing monsters to where the shadow casting monster is

[Patashu]

Sil bug while doing the tutorial:

You hit the Orc soldier!!
Your Shortsword strikes truly.
You recognize it as a Shortsword of Orc Slaying (+0,1d7) [+1].
You hit the Orc soldier...
Your Dagger of Accompaniment strikes truly.
You recognize it as a Dagger of Accompaniment (+0,1d5).

[Patashu]

How to always get free hits (on normal speed monsters) with Sprinting:
Note: This technique should only be used when you are really, really, really afraid of getting hit by what's behind you but need to kill it.

There's a hidden state variable in Sil, which is 'how many turns you've spent in sprint modulo 3'. Every third turn no normal speed monsters move and the counter resets. The counter is kept across different sprints.

Demonstration of this mechanic:

0 Step - Now says fast 0
0 Sprint 1
1 Sprint 2
2 Sprint (nothing moves) 0
0 Breaksprint - No longer says fast 1

1 Step - Now says fast 1
1 Sprint 2
2 Sprint (nothing moves) 0
0 Sprint 1
1 Breaksprint - No longer says fast 2

2 Step - Now says fast 2
2 Sprint (nothing moves) 0
0 Sprint 1
1 Sprint 2
2 Breaksprint - No longer says fast (nothing moves) 0

So:
-If your sprint ended on the free move, take 2 moves before striking next sprint
-If your sprint ended one after the free move, take 1 move before striking next sprint
-If your sprint ended two after the free move, strike immediately next sprint

And since it takes 4 moves for 'Fast' to appear:
-If your sprint ended on the free move, run for 6 moves then strike
-If your sprint ended one after the free move, run for 5 moves then strike
-If your sprint ended two after the free move, run for 4 moves then strike

[debo]

The analysis on the value of extra points of melee looks roughly accurate at the low end, but it's a bit better than that at the high end, since it increases your chances of getting criticals.



It is correct that against a given enemy, more points in Evasion give increasing returns. Many games balance by having getting points in different stats equally hard, but diminishing returns from each skill. As well as increasing returns, Sil has increasing costs to increment any particular number.

Things stood like that for quite a while during development, but we realised that very high evasion was still too strong. We've now taken another step, of further differentiating opponents. Although there are increasing returns from evasion against each foe, there are some monsters against whom it never does that much (there are various situations in which your evasion is halved, and some ways to die which ignore evasion altogether), so there's some pressure to diversify.

I think our ideal is that an extreme evasion strategy should be just about workable, but not the strongest (and definitely not far-and-away the strongest). Based on characters on the ladder we seem to be approximating that reasonably well, but if you can demonstrate that evasion is still too good I'd love to know and we'd be happy to consider further changes.

Focusing on evasion is a losing strategy IMO because there are a bazillion monsters, and only one of you. Unless you have crowd fighting and critical resistance and evasion coming out of your ears, there are so many things that are going to kill you anyways if they get a lucky critical in... Ururaukar and giants being prime offenders in this. Also, some monsters like Cat Assassins are so damned fast and so damned accurate that the rolls are never going to go entirely in your favor.

I've had up to 30 evasion before, and I was still getting hit often enough to make things scary.

I would argue that high stealth is more "broken" than high evasion, because you can basically skip entire threat groups (holes in your resistances, effects that you didn't know monsters had, etc). I played mostly stealth characters in the beginning of my Sil time, and once you've got the pathing patterns of monsters down pat, you're pretty set. That's one skill for the player to develop.

If you're facing monsters head on, though, you need individual strategies to handle all the weirdness that will ensue. Plus enough experience to know how to behave when they interact... cool stuff, I think

[Patashu]

Sil bug: If you are holding a shield and no weapon, your inventory is full and you choose to wield a two handed weapon, the game notifies you that this will drop your shield from your inventory (even though it won't)

[Scatha]

The analysis on the value of extra points of melee looks roughly accurate at the low end, but it's a bit better than that at the high end, since it increases your chances of getting criticals.

Evasion (and many other skill checks) is "broken" in SIL, it is known and nobody cares, so feel free to abuse. Basically the more you have in a skill the more you gain from one point (this behaviour is normally considered bad and broken in other games, but here it is OK).

It is correct that against a given enemy, more points in Evasion give increasing returns. Many games balance by having getting points in different stats equally hard, but diminishing returns from each skill. As well as increasing returns, Sil has increasing costs to increment any particular number.

Things stood like that for quite a while during development, but we realised that very high evasion was still too strong. We've now taken another step, of further differentiating opponents. Although there are increasing returns from evasion against each foe, there are some monsters against whom it never does that much (there are various situations in which your evasion is halved, and some ways to die which ignore evasion altogether), so there's some pressure to diversify.

I think our ideal is that an extreme evasion strategy should be just about workable, but not the strongest (and definitely not far-and-away the strongest). Based on characters on the ladder we seem to be approximating that reasonably well, but if you can demonstrate that evasion is still too good I'd love to know and we'd be happy to consider further changes.

[Patashu]

Evasion (and many other skill checks) is "broken" in SIL, it is known and nobody cares, so feel free to abuse. Basically the more you have in a skill the more you gain from one point (this behaviour is normally considered bad and broken in other games, but here it is OK).

Have you beaten Sil yet?

[LostTemplar]

Patashu's treatise on hit rates (aka numbers for the numbers god)

Evasion (and many other skill checks) is "broken" in SIL, it is known and nobody cares, so feel free to abuse. Basically the more you have in a skill the more you gain from one point (this behaviour is normally considered bad and broken in other games, but here it is OK).

[Lantar]

I just came across this variant. Congrats for the inovative game and the masterpiece of game mechanics behind it (reminds me of old Table Top Rolemaster adventures)

Now diving into it, wish me luck and may the light of the Silmarils lead my way:!

[Patashu]

Patashu's treatise on hit rates (aka numbers for the numbers god)

def chancetohit(acc, eva):
x = 0
y = 0
for i in range(20):
for j in range(20):
if (acc + (i+1)) > (eva + (j+1)):
x = x + 1
else:
y = y + 1
return float(x) / float(x+y)

If you and your opponent have the same accuracy and evasion, you have a 47.5% chance of hitting.
If you are ahead by 5, you have a 70% chance of hitting.
If you are behind by 5, you have a 26.25% chance of hitting.
Full table:

1.0 <- 20+ higher
0.9975
0.9925
0.985
0.975
0.9625 <- 15 higher
0.9475
0.93
0.91
0.8875
0.8625 <- 10 higher
0.835
0.805
0.7725
0.7375
0.7 <- 5 higher
0.66
0.6175
0.5725
0.525
0.475 <- Same accuracy as opponent evasion
0.4275
0.3825
0.34
0.3
0.2625 <- 5 lower
0.2275
0.195
0.165
0.1375
0.1125 <- 10 lower
0.09
0.07
0.0525
0.0375
0.025 <- 15 lower
0.015
0.0075
0.0025
0.0 <- 20+ lower

So, what is the marginal improvement of each point of accuracy? About 11% if you started off equal, 6% if you're a lot higher, 15% if you're a lot lower:

...
6% improvement from +4 to +5
6.9% improvement from +3 to +4
7.9% improvement from +2 to +3
9% improvement from +1 to +2
10.5% improvement from +0 to +1
11.1% improvement from -1 to 0
11.8% improvement from -2 to -1
12.5% improvemnent from -3 to -2
13.3% improvement from -4 to -3
14.3% improvement from -5 to -4
...