Sil on Roguelike Radio

[Darren Grey]

Hullo all. I'm not normally a *band fan, but Sil rather caught my fancy. Andrew Doull and I have just covered it on Roguelike Radio:

http://www.roguelikeradio.com/2013/01/episode-59-sil.html

I wasn't all positive about the game, as there are a number of things I didn't like about the way combat is handled, but overall I am hugely impressed with the design focus. I'd like to get into the game more and cover it in more depth in a later episode. I'd be keen to hear some counter-points from veteran players, and maybe details of important areas where we've missed out.

[jdh]

Here are a few thoughts in response to various points. The podcast was more than an hour and you covered a lot of ground, so this may be a little scattered.

The AI running away can be seen as an annoyance, or flavor (since not all enemies run away) and a challenge to be dealt with. You can use arrows to deal with it, as mentioned; but throwing weapons (daggers, spears, throwing axes) tend to be more reliably found in the early levels, and will serve well if it's really only one more hit. You can also pin enemies into corners, where they tend to turn and fight.

Or, you can use the second-floor forge to make Gloves of Treachery (Opportunist ability gives you a free hit on enemies moving away from you) or a weapon or shield of Fury/Wrath. Enemies which are enraged don't run away, if I remember correctly, and they don't stand outside corridors either. You can also take the Opportunist skill but it usually takes a little longer to get to.

Or, treat it as an opportunity and use Elbereth and weapons with slays (of Gondolin, etc.), Staff of Majesty or the Majesty ability to encourage fleeing. Weapons with a slay tend to make enemies which see them being used afraid. (High light level also helps with quite a few enemies! This is particularly noticeable at the end of the game.)

The particular annoyance of archers running away can be dealt with by pinning them in corners, if you don't have Opportunist or a good ranged attack. Try to find a corridor with a right angle in it, though a room arranged right will also work. If an archer is following you and you back up five spaces, you can walk back to melee range and they'll only get a hit or two in, and only run away when they get scared. To me, this is actually a bug in the AI because it is not so fun to do this dance, but it's more fun than being killed by arrows.

Stealth: the skill level + the d10 rolls (one for Stealth, one for enemy's perception) mean you can't count on being unseen without high stealth -- but do create something resembling a bell curve (I'm a bit weak on the actual statistics), making it increasingly unlikely you will be. The other side of this is that later on, there are several faster enemies who can easily run into you and notice you (this can also happen earlier when you're unlucky coming around a corner), meaning you need another way to deal with them. One way is vanishing (which the vanish skill helps): you can take advantage of enemies standing around corridor mouths to run away and have them forget about you. You could also use Song of Lorien or a Staff of Slumber if you have enough Song or Will.

About smithing: as noted in the recording, 1 point in a skill makes a big difference. Look at HM's artistry start video (perhaps someone else can provide the link) for how a modest investment in smithing can help you. +1 evasion doesn't seem like much in the early game when it only costs a few hundred experience, but by the end the points you're getting from equipment are worth thousands of experience. It can also help to invest in melee+evasion first, up to at least 10 in each, before trying to build smithing: it doesn't matter how high your smithing skill is if you don't live to use it.

Many of the criticisms seem to be about combat, but I'd say a large portion of the game is managing fights and particularly being able to avoid or escape them. It is not that hard to win as a stealth-based character if you make yourself run away first, before the situation goes bad, instead of trying to fight everything. Keen Senses + Listen work well to give you some advanced notice of enemies before you walk into them. You could take Armorsmith + Enchantment and make probably +2 Stealth boots, cloak and armor on the second floor forge, giving you a 6-point boost (quite significant, enough to avoid most early enemies).

You also overlooked, or at least I didn't hear you mention, the inventory management game. Identifying unknown items is a particularly enjoyable part of the game for me, though I know some players go straight for Lore-Master.

Last point: you can get toys quite early. You may get lucky with an artefact or special item (I've found Robe of Idril on the first or second floor before), but even if not you can make some interesting things at the first forge: Crown of Command, Robe of Majesty, Gloves of Treachery. Various songs can also be fun to play with (Elbereth, Slaying, Trees, Freedom) though they tend to be more fun later, when you have a good investment in Song.

[HallucinationMushroom]

One of the important themes that you won't really get to experience until mid-game, is that of light vs. darkness. I find the concept very interesting... monsters that exude darkness so much so that you cannot even see them. Combating this darkness is crucial for survival, which can be done in a few different ways, and lends the dungeon a good deal of atmosphere that seems lacking during the opening game. It ties in nicely with the objective of the game, which is to return a Silmaril to Middle Earth, which shines with great light. Obvious good vs. evil, light vs dark parallel, but it works and I like it.

[Scatha]

Hi Darren,

Thanks to you and Andrew for the episode. Some really good comments in there.

As jdh, I apologise for the scattershot replies.

-

I found some of the UI comments very interesting. I think this can be a tricky part of design, and half has done a lot of good work on it, but at least I'd never thought of putting more gameplay commands additionally in the window menu, which seems like it could have a lot of mileage.

I'll flag a couple of interface matters which relate to things commented on:
- You can get directly to the character screen with the @ key
- If you open the savefile of a dead character, you can just hit enter through to the game. It's not quite the quick-start option you were asking for, and you still need to allocate experience, but it's something.

Of course as you missed these there's a sign that we should consider doing more to tell the player about them.

-

Your comments about dice were interesting. Some replies:

- It's quite easy to be in situations where your Stealth is 10 higher than enemy perception on a given turn, so they have no chance to notice you (there is an entire ability, Vanish, which only functions in these circumstances). There is a middle ground of dangerous turns.

- In stealth and combat, having significant variance from the dice helps to keep it feeling exciting and to generate interesting situations (even sometimes in what 'should' be relatively easy fights). At least we really enjoy the risk-management game: is it worth taking a small chance of disaster to kill the wounded giant/steal the sword from the dragon's hoard? This is a big feature in another game half and I like a lot, Battle for Wesnoth. This kind of feature does need a short enough game to work, and Sil might be pushing the upper length of that.

- The dice for weapon damage being perhaps too complicated is a separate point, and one I don't have such a strong reply to, except that it is a natural way to bring in some of this variance.

- You talked about taking a long time to kill things. This is an issue with underinvestment in offensive capabilities (melee/Strength/weapon/abilities). It's a bit of a longstanding problem: how do we flag to players to avoid making these character choices?

-

As you guessed, more of the interest comes along a bit later in the game. I think the start of the game is one of the weaker parts, and you had some interesting ideas for things that could be done to address this. I'm not certain whether we'll pursue this, but it is something to keep in mind. Maybe just slightly more/more interesting floor drops in the first couple of levels would help a lot (kept low enough to discourage start-scumming).

-

We are considering reducing level size a bit (my instincts are closer to yours here than half's are), but I will say something in favour of empty rooms: as well as adding flavour and helping the stealth game, the space improves tactical choices. I often lead fights around a section of the dungeon.

-

There are a few more options for the start of the game than you imply (but perhaps these aren't obvious to new players). For example, try getting Song of Elbereth at the beginning. It won't work on the early (unintelligent) animals, but otherwise changes the fight dynamic considerably. The generalist approach is also quite possible, but perhaps not as easy as streamlined builds.

-

Finally, for anyone who cares about the history: I've been involved in the development for five or six years (around where the version numbers begin in the early changes document). I think half had started branching it off five years before that.

[fizzix]

As you guessed, more of the interest comes along a bit later in the game. I think the start of the game is one of the weaker parts.

I have some ideas here.

1) terrain gives you free variability. New terrain options can have little gameplay effects, but do wonders for keeping players interested. Consider introducing junk items related to nearby monsters. Or flavor terrain.

2) Sil's experience format prevent you from adding new monster types without throwing things off and requiring balancing, but you could pick some monster types that only show up in certain games. They would all be roughly equivalent in difficulty, but the fact that each game only includes a sub-section of them means that the beginning of the game will be somewhat different each go.

3) use light to your advantage. Have lit points in the early dungeon and some darkness producing effects. The light mechanic seems the best to try to extend into the early game.

I'll let you know if I think of more. The beginning is really the only part of Sil that I've personally actually experienced, having never really gotten further. I have watched debo and HM's videos.

[Scatha]

I'll let you know if I think of more. The beginning is really the only part of Sil that I've personally actually experienced, having never really gotten further.

Thanks for the ideas. The fairly difficult start gives another reason to put more interesting game elements early, as they may be missed by many players otherwise. Weighing against this you want to avoid complexity overload at the start of the game.

More generally the difficulty of the early game in roguelikes is that it is doing double duty:
- Providing an introduction to the game to new players (but they should play the tutorial first).
- Providing interest to experienced players on repeated playthroughs.

The fact that we seem to have quite a few players who have played several times but not got past the early game means that it also wants to serve:
- To encourage people to get better at the game.
- To have enough variety and interest to keep people's attention while they keep dying.

[Darren Grey]

Thanks for the ideas. The fairly difficult start gives another reason to put more interesting game elements early, as they may be missed by many players otherwise. Weighing against this you want to avoid complexity overload at the start of the game.

I think the game is lost there already with the character creation system. I got the feel very strongly that this is a game you have to learn before you play, not learn as you play. There is too much to be missed in character generation - if you don't read the manual you simply won't be successful.

The game should just play to its strengths and focus on its core experienced audience, in my opinion. There's no point trying to make an early game that satisfies both veterans and complete novices. Any experienced roguelike player like myself won't mind a steep learning curve. Anyone new to roguelikes will struggle with many many things in Sil already.

[debo]

I think the game is lost there already with the character creation system. I got the feel very strongly that this is a game you have to learn before you play, not learn as you play. There is too much to be missed in character generation - if you don't read the manual you simply won't be successful.

Or play the tutorial -- which is really well done. I played the ever-loving-shit out of Sil last year and it wasn't until about July that I read the manual thoroughly. The tutorial was enough to get me going.

I do haunt the forums a fair bit, though.

Still, the honest truth is that you are probably going to die umpteen times before arriving at an initial build/playstyle that works for you.

[Darren Grey]

The tutorial is great for learning the commands and getting a feel for the game, but it doesn't teach you how to build a character, which is something you get very quickly confronted with. You have to read the manual to get a good feel for that. Or play and die a lot I guess, but that's inefficient. This isn't a criticism, mind, the manual is great and I loved reading it. I wish more games were designed with this sort of knowledge-based play in place.

[getter77]

Taking a page from the somewhat verbose/overt nature of character building that Frozen Depths employs might be a good remedy. http://koti.mbnet.fi/frozend/

[Mikko Lehtinen]

The rulebook or in-game help could have several example character development paths that you could choose to follow. Perhaps with some explanation for why a given ability was chosen for the build.

[Darren Grey]

Thanks to you and Andrew for the episode. Some really good comments in there.

Some Sil fans disagree

I found some of the UI comments very interesting. I think this can be a tricky part of design, and half has done a lot of good work on it, but at least I'd never thought of putting more gameplay commands additionally in the window menu, which seems like it could have a lot of mileage.

Overall it has an excellent UI, especially for a roguelike. Just a few things that could be tweaked. Some other things I'd suggest:

- The main menu should be reordered New Game / Continue / Tutorial / Quit. Tutorial should be top the first time you install the game and then move down the list. Continue should be top if the last action in-game was save&quit. This makes it easier to just press return to get into the game.

- On character creation when selecting stats and skills to put points into it would be good to have some context info on each one displayed. So whilst Strength is highlighted it would have text below saying what Strength affects. Or even better there would be a side-window with some of the really detailed info from the manual, especially for the skills. Otherwise one is left consulting the manual quite a bit during char creation.

- Have char creation remember your gender preference, ie. always default to same as last char.

- The combat stats window is a bit of a mess I think. I don't find it practical in battle - too many numbers to keep track of turn by turn. It would be nice to have a touch of Brogue's "You have a 56% chance to kill this enemy in 3 rounds" style of information. This is much quicker to absorb and helps make an informed judgement.

- "Repeat same character" option on death screen. Every roguelike should have this

- Multiple lines on the message bar. Having to press for "more" all the time really slows the game down. Perhaps remove more prompts entirely if the player has the separate message window open.

- If you open the savefile of a dead character, you can just hit enter through to the game. It's not quite the quick-start option you were asking for, and you still need to allocate experience, but it's something.

Heh, this is very useful to know. A good way to teach it might be to include some of these files already with the game, as example builds for people to use. You can base them on the likes of Feanor (smith), Fingolfin (melee + song), Luthien (stealth + song), Idril (stealth + archery?), or maybe just make up characters. Tell new players to try out these recommended builds and you'll teach them about the character profile system whilst also giving them a quicker route into the game.

- It's quite easy to be in situations where your Stealth is 10 higher than enemy perception on a given turn, so they have no chance to notice you (there is an entire ability, Vanish, which only functions in these circumstances). There is a middle ground of dangerous turns.

I want to know when! Some system so you know when you're safe and when you're not. Technically you could go and calculate that already with the manual information etc, but that's a burden on the player. Have the "S" in stealth mode change colour dependent on how safe you are. White when no enemies around, green when impossible for enemies to see you, yellow when they have 80-99% chance to see you, orange for 60-79%, red for <60%. Would obviously base it on worst chance for the enemies around.

I think you need this sort of clarity in any Stealth system. Lots of Stealth focused games like Tenchu or MGS have this sort of safety meter front and centre. The best stealth roguelikes have had 100% clarity on when you can be seen or not.

- In stealth and combat, having significant variance from the dice helps to keep it feeling exciting and to generate interesting situations (even sometimes in what 'should' be relatively easy fights).

I disagree :P But this is a matter of design philosophy.

I guess a big part of my problem is that the numbers are so cut down and set elsewhere, but vary so hugely in combat. The combat system feels really out of place next to the stat system. The stat system is very tight and well-designed whilst the combat system feels loose and messy.

It's also odd to use the 1d20 system for attacks instead of the 1d10 system that's universal elsewhere in the game. I'm guessing you're worried about people building up too high evade?

At least we really enjoy the risk-management game: is it worth taking a small chance of disaster to kill the wounded giant/steal the sword from the dragon's hoard?

Nope, never! This is a permadeath game that takes several hours to complete. You can't take such risks. Get to a position of overwhelming power or minimised risk and then proceed. Which is why Risk the game is so dull - with the huge number of dice rolls it's all about who gets to that overwhelming position first.

Ever played Small World? Wonderful board game with very deterministic combat, all in very small numbers. Immensely more fun than Risk. I think Sil could take some inspiration from it if you were to consider reforming the combat.

- The dice for weapon damage being perhaps too complicated is a separate point, and one I don't have such a strong reply to, except that it is a natural way to bring in some of this variance.

A little variance I don't mind, but when you have the combat modifiers be tied to this variance as well it gets a bit silly I think. Why have Strength increase in the dice rolls instead of having it just add +1 damage for 2 lbs of weight of the weapon? Easier to describe, easier to keep track of during normal play. The system at present is complexity for complexity's sake - it adds no tactical depth to the game and only makes a barrier to tactical decisions.

Just my opinion etc etc

Maybe just slightly more/more interesting floor drops in the first couple of levels would help a lot (kept low enough to discourage start-scumming).

At the very least give the player some starting items based on the character they create. If they've invested in archery they should have a bow and arrows at the start of the game. Why have a sword in the first room instead of already equipped? Character creation takes long enough, don't force me to spend two dungeon levels of exploring before I feel like I'm even getting to start my build

And as has been said elsewhere having some of the light vs darkness dynamics come into play earlier would be nice.

Finally, for anyone who cares about the history: I've been involved in the development for five or six years (around where the version numbers begin in the early changes document). I think half had started branching it off five years before that.

Would be good to have you both on the show some time. It's quite peculiar to have a game in development for so long come out and make a big splash. Or to come out at all really. Really is light from the shadows! Would love to hear some of the dev history and what has inspired you guys

[fizzix]

This discussion actually highlights one of the main issues I have with skill-tree based games. Namely, that making a good build is often a daunting task for a new player and is also a poor way to learn. If you die, you won't know if it was because your build was bad or you made a poor choice or you just got some rotten luck. Games are most fun (for me, YMMV) when there is a basic working build that I can use to learn and explore the game mechanics, and then branch off from that build as I gain expertise. If the game came with some standard guides, like a smithing build, a protection build and whatnot, then it would help new players that don't necessarily want to trial-and-error figure out how to make something reasonable themselves. And the ones that do want to trial-and-error it are free to not look at these guides.

I'm sue you could convince HM, Psi, debo, etc. to make a couple short descriptions of how to make certain builds work. These could be shipped with the game and tagged appropriately with spoilers or something.

I had thought this was something mostly unique to me, but listening to some of the discussion on roguelike radio and here makes me think that other people might prefer similar information.

[HallucinationMushroom]

The game currently keeps track of when and what abilities are purchased, along with what items the player forges, and you can tell what the initial stat investment was in addition to the typical artifact/unique-monster notes. Greatly helpful for following, if you know about the ladder that is, which is listed on the Sil homepage. It would be nice if the notes took into account initial skill investment, and made a note of every time a skill went up. This way you could copy a build verbatim, if you wanted to. I will sometimes manually put notes for my challenge character on current skills, melee/protection for future reference. Many players add a lot of helpful notes too.

[debo]

I don't think I'd like the stealth game quite as much if there was such "perfect information" available. I'd be willing to try it, but I do sort of like the feeling of "ok, I think this is good enough... OH SHIT HE SAW ME"

On the other hand, if perception affected how well your character was able to do these calculations and expose the information to the player, well... that might be sort of cool

I like Sil because of the flavor of being stuck in the dungeon, I don't always like all the stat-guts poking out of a game. I don't think any elf in the silmarillion peeked his out out a cave door and said "oh shit that Orc has a 36% chance of seeing me -- I'd better tread carefully"

I realize I'm backwards though, and it's entirely possible that being exposed to these sorts of more modern-y approaches would make me wonder how I ever lived without them. *shrug*

Edit -- also, stealth depends on things like how close you are to a wall, etc, and on a monster's perception. So would you need to know the likelihood of every monster in a room seeing you (independently), for every step you take? Or just on inspection? I think the latter would be fairly useless except in specific high-danger cases (e.g. I'm about to try stealing Glamdring from under Glaurung's butt.)

OK now I'm rambling