What should Vanilla take from Sil?

[fizzix]

Sleep for breeders seems like too much of an edge case to be worth it. I'd like sleep to be much more powerful, so it has a high chance of success, and can be reliably used as an escape. It would then be superior to teleport, and can also allow for "sneaky" play in the early game, something that's difficult to do even for super stealth characters like kobold rogue.

[Raajaton]

Also sometimes when running away from something bad when there is a weakling that can be slept between you and it. It has saved my char few times in early game.

Yeah, I have used it to great effect while escaping through a narrow corridor. It's pretty hilarious when you manage to sleep the enemy right behind you, creating a road block for everything else.

[Timo Pietilä]

I've yet to come up with a use for Sleep though; all I can think of is using it to run away from monsters, and there are lots of other, more effective ways to do that, even if you disregard the failure rate.

Sleep is useful against out-of-control breeders. Staff-version of it. Also sometimes when running away from something bad when there is a weakling that can be slept between you and it. It has saved my char few times in early game.

[LostTemplar]

I hope it will not take restriction of one object per square.

[Derakon]

Slow and Confuse can be worthwhile in the early game, where they have a chance of working even against dangerous uniques (e.g. Wormtongue or Nar), and make such monsters vastly less threatening. It usually takes several attempts, but you can generally get it to stick eventually. I've yet to come up with a use for Sleep though; all I can think of is using it to run away from monsters, and there are lots of other, more effective ways to do that, even if you disregard the failure rate.

[wobbly]

Some equivalence to how Sil handles monster will would be great. Currently anything in Angband worth slowing/confusing/sleeping is completely immune. It's like a part of the game completely fails to work as is.

[wobbly]

2) The light-dark effect. (the importance of light and darkness producing monsters) At lower levels this can be pretty much imported directly. In some ways it already is, we just don't have monsters that can produce dark. At higher levels ESP is an issue, (and it also makes an issue with stealth.) So we would need the same thing for ESP, you would have "mind shielding" which would make it harder for monsters to detect you, and monsters would have the same.

Balrogs in Sil have that particularly nice effect where they light up there own square with fire & are surrounded by darkness. I've had a character with high light (around 5 radius) fooled by Durin before by 2 shadow spiders hiding with-in his darkness radius. So part of getting it to work properly is getting the AI to use darkness as well.

[debo]

1) unaware monsters.

+1 billion. A very simple step in this direction could be to get rid of the idea of certain monster types that _always_ spawn awake *cough* HOUNDS *cough*.

(Actually, I can't even remember if hounds still always spawn awake in Vanilla.)

But yeah, it's always kinda bothered me that awake = tracking you. The unwary state is helpful. I'm pretty sure Crawl also has sleeping, unwary, aware states on monsters.

[fizzix]

Ok, here are two things from Sil that I think would be reasonable to port to angband near immediately.

1) wandering monsters, sleeping monsters, unaware monsters. Right now Angband has two states, asleep and aware. An aware monster will track you always if its in range. It will never go to sleep unless my magical means. This limits what you can do with stealth, and stealthy characters are fun to play.

Wandering monsters in Sil, while still somewhat predictable, are far better than in any other roguelike I've played. Allowing them to leave the level on their own volition is a really nice touch.

2) The light-dark effect. (the importance of light and darkness producing monsters) At lower levels this can be pretty much imported directly. In some ways it already is, we just don't have monsters that can produce dark. At higher levels ESP is an issue, (and it also makes an issue with stealth.) So we would need the same thing for ESP, you would have "mind shielding" which would make it harder for monsters to detect you, and monsters would have the same.

One of the difficulties is that when Sil if you blunder into a monster you can't handle, you're usually not dead. The monster may not have seen you, or you can beat a retreat to the stairs. In angband, if you blunder into a monster, you may often be dead. So detection is a key to survival in angband. I'm not sure if this is good or bad, but it's something to think about.

[debo]

Huh? The worst case is 8.

Maybe you mean that it is not too difficult to guesstimate the whole shape of the distribution function. I agree. Even if the distribution isn't uniform, it still holds true that E[X]=(max-min)/2, since both sides of the equality are linear. In addition, the more dice you have, the more the distribution is peaked around its mean, and the number of dices equals the first of the two numbers in the range.

What I meant was that, given a specific enemy, I can pretty easily calculate the probability of it rolling 20 on its melee and me rolling 1 on my evade. From there I can compute how many crits that would be, and estimate the probability of max damage because the distribution of xdy is not hard to scribble anyways. It's a bit harder to do the final worst-case probability calculation for protection, though. This makes the question "what is the max damage I could take here, and what is the probability of that happening?" harder to answer.

kryft did make a simulator for Sil combat a while ago, which does these sorts of things for you anyways. You get a feel for it after a while regardless.

[taptap]

I wrote about my 1st encounter in another thread (http://angband.oook.cz/forum/showthread.php?t=6488).

The Angband combat system is based on the highly abstract system of D&D. While I can't see how Angband can depart from that system and remain Angband, there should be options for simplification within that framework. Odd multipliers could be integrated in the numbers, guaranteed hits* (the 5/7%) could be removed. And this should precede any addition of new complexity.

* Most randomness based game mechanisms in Sil only work because there are no "guaranteed success / failure results" in rolls. E.g. stealth or high evasion/low armour play would be impossible.

A huge improvement would be realistic weapon weights and removal of retarded weapons like "lead-filled maces" - even if it makes 0 impact on actual gameplay.

[fph]

Sil displays your protection as a range of values like 8-21, even though the distribution isn't uniform because it's made up of a bunch of different dice. It's good enough to guesstimate the worst case.

Huh? The worst case is 8.

Maybe you mean that it is not too difficult to guesstimate the whole shape of the distribution function. I agree. Even if the distribution isn't uniform, it still holds true that E[X]=(max-min)/2, since both sides of the equality are linear. In addition, the more dice you have, the more the distribution is peaked around its mean, and the number of dices equals the first of the two numbers in the range.

[debo]

Even just a feasible range of values for wand/device damage would be helpful, even if you have no idea of the distribution. Sil displays your protection as a range of values like 8-21, even though the distribution isn't uniform because it's made up of a bunch of different dice. It's good enough to guesstimate the worst case.

[Nick]

This is fantastic, don't get me wrong, it's amazing that people will be able to modify the game that much more easily. You can get a ton of mileage out of game mods that just take existing behaviors and tweak the numbers. But it's not that much more clear than the current behavior, when you're in the game. I guess in my ideal world there'd be a little screen that looked something like this:

Small kobold attacks player: miss
(100 - ((monster skill 1) - (player AC 8))) < 1d100
Player uses Wand of Stinking Cloud: 2 damage
(8) * (100 + (player device skill 17)) / 100 * (.25 poison resisted) = 2.34

But coding up all of the logic to sensibly display these numbers and their labels would be a lot of work.

This is great - I expressed myself badly, and you took it to mean something else, which sounds like a really good idea

What I meant to say was that once the arcane calculations are exposed in the code, in a lot of cases it will be possible to make them simpler and more intuitive and uniform with only tiny effects on gameplay.

But having an option to display calculations like that is a wonderful idea.

*** SCOPE CREEP WARNING! ***

[fizzix]

Note that the original plan had been to get full experience for killing a monster and half for dealing with it in some other way. We had ideas for when to allocate that half (e.g. if the monster is now far away or on a different level), but it turned out to be easiest just to give you the half on first encountering them -- a little bit different in terms of flavour, but basically the same in terms of gameplay.

You could consider giving experience to a player when a monster flees the level as if it had been killed. Maybe full experience for a fleeing unique, and that unique is then no longer generated (you lose the drop, but he's decided angband is no longer a safe place)

I'll respond to the main post after giving a bit of a think.