RFE Inspection Messages for +Stat Items

[Scatha]

I think your analysis of health looks interesting, fizzix, but there is a general point that you (and the monsters) have lots of different stats, and general power level is some increasing function of these, which is usually at least linear in each one (and can be substantially superlinear with things like an evasion/protection system or combat rolls as in Sil). So if the stats rise linearly with depth you'll naturally get something which rises at least like a significant polynomial. Sil's system behaves pretty similarly to an exponential (it may even be super-exponential -- I think a lot of characters could stand next to an enemy from 10 levels earlier and just pass, and would starve to death before dying from the attacks). There are some parts of Sil's system which rely on it being a short game, but if anyone is interested in adapting it to a longer game I know how to change them to make it essentially scale-free (this is something we considered a while back in development), and would be happy to talk about that.

To Magnate and Nick:
Obviously those are pretty flattering reasons for not trying Sil! I'm a little sad, not least for the rather selfish reason that you both seem to have good instincts for game design and it means we're deprived of criticism and ideas from you.

More generally, it's obviously better to have a spread of games trying out different ideas than clones of the same game. But Sil has made a lot of changes. I think it's quite possible to divide these into a spectrum from those which are fairly uncontroversially good, through ones which should maybe be adapted by more games in the genre but probably shouldn't be universal, and to things which are quite specialised to the game and may well not want to be copied.

Let me attempt to draft such a spectrum. Obviously the location of individual items is pretty subjective, and I'm sure some people will disagree with some of my placings, but hopefully I'll have these largely correct.

Relatively uncontroversial:
- Tutorial and manual (but these are quite a lot of work; in the case of Sil, almost all due to half).
- Simplifying and streamlining the commands required (to make it easier to learn).
- Choosing clean mechanics and making these transparent to the player (there probably is a space for games with obscured mechanics, but at least V and v4 seem to be going in this direction).
- Philosophy of encouraging interesting choices for the player.
- Realistic weapon weights.

Widely beneficial:
- Making the start of the game challenging (or having an accessible mode where this is so, or presenting dangerous opportunities early which you can choose to avoid, etc.). If I can't find something difficult or where it feels like there's a potential for it to go wrong in the first 15-30 minutes of a game, I'll often not continue playing. This is where a lot of the interest of the game is, so present it up-front.
- Consistent flavour.
- Generally keeping numbers visible to the player (health, damage, bonuses etc.) as small as possible while giving enough resolution for the mechanics to work (+1 to just about anything is a nontrivial amount in Sil).
- Decaying experience with kills of the same type of monster (doesn't need to decay as fast as Sil); more generally incentives to venture out of your comfort zone.
- Almost never having more than one attack per character/monster per round (to avoid message spam).

Would be good to see more of:
- Lack of instant escapes. (I doubt V wants this, as it seems that the escapes and frequent situations where you can die in one round that they entail are part of the identity of game, but the lack of instant escapes does help the tactical depth.)
- The stealth system, or something close.
- Experience for non-combat things like encountering monsters.
- The essence of the combat system (v4 is going in a similar direction).
- The forced descent, or other mechanic to prevent grinding.
- No town (actually there are quite a lot of games with this already).
- Game length (again by no means unique to Sil, but it is a major departure)

Might be good for some other games:
- The light/darkness system (could be good if another game wants magical darkness, but perhaps that's a bit niche)
- Classless system with lots of options to build your character by spending experience.
- Player health not really changing through the game.

Fairly specific to Sil:
- Low magic, and what magic there is subtle (I suppose Beleriand might go in this direction for the same flavour-driven reasons as Sil, but I think magic is part of the appeal of a lot of fantasy).
- The story/setting (trying to steal a Silmaril)
- Having the races at such disparate power levels (again, flavour driven)

Well, that's quite a long list, and I'm sure I've missed some important changes (feel free to suggest where I've gone wrong), but hopefully it's a useful starting point for people thinking about what would be good or bad to borrow from Sil.

[Derakon]

Originally Posted by Scatha

Relatively uncontroversial:
- Tutorial and manual (but these are quite a lot of work; in the case of Sil, almost all due to half).

Documentation always ends up being more work than you think it'll be (he says, while writing documentation for a program at work). And once you write it it has to be kept up to date with changes in the game. The real trick is having useful documentation that people will actually read. Some kind of tutorial mode seems to be the go-to method these days.

- Simplifying and streamlining the commands required (to make it easier to learn).

At this point, I think the only convincing reasons we still have multiple "use" keys are to a) reduce the odds of accidentally using the wrong item, and b) enable more keymaps (since you can do both @m1 to bind your spellbook, and @z1 to bind a rod, for example). If we had a better keymap system that let you say e.g. "Cast a spell from Magic for Beginners, whatever slot it's in", then we'd be a lot closer to being able to do away with all of the varied use keys.

Otherwise, what other commands do you think we could remove from Angband?

Widely beneficial:
- Making the start of the game challenging (or having an accessible mode where this is so, or presenting dangerous opportunities early which you can choose to avoid, etc.). If I can't find something difficult or where it feels like there's a potential for it to go wrong in the first 15-30 minutes of a game, I'll often not continue playing. This is where a lot of the interest of the game is, so present it up-front.

Vanilla needs to have some room for players to experiment with the game mechanics, I think. But there's no need for this to be more than one or two levels' worth of the dungeon. Right now we have a fairly smooth difficulty ramp from level 1 through level 20; if we cliff-faced that at level 3, say, then the game would get "interesting" much more quickly.

- Generally keeping numbers visible to the player (health, damage, bonuses etc.) as small as possible while giving enough resolution for the mechanics to work (+1 to just about anything is a nontrivial amount in Sil).

Yeah, yeah, the +100 weapons in v4 were a mistake.

- Almost never having more than one attack per character/monster per round (to avoid message spam).

v4's combat system makes half the classes, more or less, rely on attack spam. But we could probably compress this to "You hit the Foo 4 out of 6 times" or something. Monster attacks usually could be handled similarly, though there are a few monsters who have varied effects on their attacks (e.g. hit to hurt, touch to drain charges, and bite to drain experience).

Would be good to see more of:
- Lack of instant escapes. (I doubt V wants this, as it seems that the escapes and frequent situations where you can die in one round that they entail are part of the identity of game, but the lack of instant escapes does help the tactical depth.)

I'm inclined to agree here; what I think would probably work better for Vanilla would be to make it harder to stockpile these escape items, mostly by making them harder or impossible to buy. I know that there are lots of people who would disagree with me, so let me get sidetracked for a moment:

One big way that Angband and similar games get interesting is by getting you in over your head. You know that whole "ignorance breeds fear" thing? Well, when you don't know if you can handle a challenge, you become afraid -- and being frightened, in a game, is interesting! But when we have all these escape items, any time you're worried that you might get in over your head, you just hit the reset button and can carry on playing inside your comfort zone. We should be trying to push you outside your comfort zone as much as possible, because that makes the game more interesting.

That means that easy access to easy escapes needs to be removed. Easy escapes (broadly, ones that work quickly and also almost certainly provide safety) should be valuable items that you hoard and have to consider carefully before using, in case you get into even greater trouble down the road.

Now, we of course have to counterbalance that by making certain that we don't just present you with impossible challenges that you now cannot escape from. That's only rarely fun.

- The forced descent, or other mechanic to prevent grinding.

Generally I don't see why we should actively prevent players from grinding if that's what they want to do. Rather we should ensure that the player never has to grind -- grinding then effectively becomes a way for players to vary their "difficulty level" over the course of a game. There are three roadblocks currently in Vanilla that can force grinding, though: the need for Free Action / basic four resistances, and the need for "enough" damage and consumables to kill Sauron and Morgoth. The latter is somewhat flexible -- there was a credible attempt to kill Morgoth with a hand-enchanted shovel, but sadly Ditchdigger ran out of healing -- thus, endgame damage and consumables trade off between each other. Generally, the existence of multi-hundred-HP damage sources in Angband is a problem.

- No town (actually there are quite a lot of games with this already).

Despite prophecies of doom from various forum members, the Vanilla town is not going away anytime soon.

Fairly specific to Sil:
- Having the races at such disparate power levels (again, flavour driven)

The race you choose is a form of difficulty selection in all Angband variants, as you basically decide how heavily you want to min/max your character. A gnome warrior will have a harder start than a half-troll warrior will, and the high-elf and dunadan races are effectively an "easy mode" of sorts. We could readily exacerbate this and increase the impact that race has on gameplay; currently the effects are usually somewhat subtle.

[ekolis]

Originally Posted by Derakon

If we had a better keymap system that let you say e.g. "Cast a spell from Magic for Beginners, whatever slot it's in", then we'd be a lot closer to being able to do away with all of the varied use keys.

ToME2 had something like that, IIRC, though its spellbooks functioned differently - they were more like Nethack-style spellbooks in that each book contains only one spell, and you don't need to carry the book around to cast spells that you've already memorized. But the memorized spells themselves did become menu items in various submenus according to what type of spell you were casting, so the casting mechanic was the same from the perspective of keypresses.

So in ToME2 you could cast "manathrust" (the equivalent of magic missile) by typing something along the lines of "maa" just like in Angband, but the first "a" didn't represent "book 1"; instead it meant "attack spells" or something like that.

As for the "search" functionality, you could type an @ sign at the "m" menu to search through your spell list, so you could cast manathrust by typing "m@Manat" and pressing enter (assuming you didn't know any other spells that began with "Manat"). And then of course you could bind keys to these searches

[AnonymousHero]

Originally Posted by ekolis

ToME2 had something like that, IIRC, though its spellbooks functioned differently - they were more like Nethack-style spellbooks in that each book contains only one spell, and you don't need to carry the book around to cast spells that you've already memorized.

You're misremembering. There's no memorization, nor explicit spell learning of individual spells. You just have to have a high enough level in the appropriate skill and have a single-spell book, a tome (collection of spells) or an item containing the spell you want to cast.

Originally Posted by ekolis

But the memorized spells themselves did become menu items in various submenus according to what type of spell you were casting, so the casting mechanic was the same from the perspective of keypresses.

So in ToME2 you could cast "manathrust" (the equivalent of magic missile) by typing something along the lines of "maa" just like in Angband, but the first "a" didn't represent "book 1"; instead it meant "attack spells" or something like that.

Your memory is also slightly off here. "m" means invoke the "magic" menu -- typically the menu contains at least two choices "a) cast a spell" and "b) copy a spell [to an item]". (The letters may sometimes be swapped which is... unfortunate.).

So "maa" usually means "cast a spell from the first book". You would then select the spell with a-z.

You could inscribe books just as in angband such that "ma2" would specify a specific book inscribed with "@m2".

Originally Posted by ekolis

As for the "search" functionality, you could type an @ sign at the "m" menu to search through your spell list, so you could cast manathrust by typing "m@Manat" and pressing enter (assuming you didn't know any other spells that began with "Manat"). And then of course you could bind keys to these searches

Not quite correct. You would macro "m@Cast a spell\r@Manathrust". Since spells were available from various sources (single-spell book, item or tome) this is by far preferable to any other method of macroing spellcasting.

[Derakon]

The proper way to handle that, of course, is to have a keymap recording system that assumes that any selection from a list of choices means to always make that selection, not to always select that slot. So if you turned on recording and then did "maa", instead of recording "cast the first spell from the first spellbook" it would be "cast Magic Missile from the spellbook Magic for Beginners." And if you didn't have Magic for Beginners, or if somehow Magic Missile weren't in it, then the keymap would fail.

The only case this breaks down in, really, is when you're shooting ammo at your closest target, since you can run out of your tier-0 ammo and want to gracefully fallback to your tier-1 ammo. Fortunately Vanilla already has a "target nearest enemy and shoot first ammo" keymap in the 'h' command.

[Nick]

Originally Posted by Scatha

Well, that's quite a long list, and I'm sure I've missed some important changes (feel free to suggest where I've gone wrong), but hopefully it's a useful starting point for people thinking about what would be good or bad to borrow from Sil.

That's a great summary, thanks. I'm planning some similar things and some very different things for Beleriand - I'll probably post on the state of my thinking some time soon.

I have played Sil a little; I guess my main feeling is that I don't have the time/thinking space to do justice to learning Sil properly at this point.

[fizzix]

Thanks Scatha! That's a useful post.

As Derakon says, I don't think Angband will ever force player descent. However, what we will do is ensure that a forced descent ironman game is always viable for the people who want that challenge. Some people like a slow game that let's them remain at a depth for as long as they want, and I'd be loath to turn that player off from Angband.

As far as decreased XP per type of monster. V does have decreasing XP benefit as your level goes up, which sort of has the same effect. Sil has a very different approach to levels and XP so I'm not sure how much of a comparison is useful here. There are other ways to bring players out of their comfort zones. The most common is to limit the monsters with permanent or semi-permanent levels. Another approach touches on the monster genocide idea, namely a player should be able to kill off all monsters of a certain type. The approach here is the converse of that. Hell Wyrms are lucrative so you limit their number.

I think the beginning of Angband has a decent ramp to difficulty. Perhaps the reason players here don't think it's difficult is because they are not beginners. I remember dying on dlevels 1 and 2 when I started. I don't actually think early dungeon ease is a huge problem (3.4 is harder than 3.3, I'm not sure where v4 falls.)

Lastly on escapes. Right now Vanilla needs escapes because the game presents too many situations that are downright unhandleable. Arrive in a dungeon in a room of timehounds? Maeglin summons Ungoliant and the Tarrasque? Arrive near a graveyard? All of these situations require escapes. I'm personally of the opinion that we need to reduce/remove these situations. Make pits/graveyards smaller. Make time/gravity hounds less dangerous and have smaller pack size. Allow the player to weaken summons by killing monsters. I'm not a fan of spawning monsters either. I think you should be able to clear out a dungeon level without having monsters constantly arriving on it. But these are my own opinions and they engender much disagreement. In summary, I would like us to be in a position where we can start limiting escapes, but we need to change the essential flow of the late game dungeon to implement that.

[Scatha]

(I edited my previous post to mention game length, which is a significant departure of Sil from V I'd managed to forget.)

Originally Posted by Derakon

At this point, I think the only convincing reasons we still have multiple "use" keys are to a) reduce the odds of accidentally using the wrong item, and b) enable more keymaps (since you can do both @m1 to bind your spellbook, and @z1 to bind a rod, for example). If we had a better keymap system that let you say e.g. "Cast a spell from Magic for Beginners, whatever slot it's in", then we'd be a lot closer to being able to do away with all of the varied use keys.

Otherwise, what other commands do you think we could remove from Angband?

This was less a specific proposal than a suggestion that it's worth keeping in mind; the complexity of the command set is often cited as a reason people don't get into roguelikes.

Sil is certainly not the only recent roguelike to make some effort to make the commands more accessible, but since its changes were from an Angband-like system, they might be relevant here:
- Added a unified (u)se command and promoted this to new players. The old, specific use commands almost all still exist.
- Similarly added a unified method for terrain interaction.

Vanilla needs to have some room for players to experiment with the game mechanics, I think. But there's no need for this to be more than one or two levels' worth of the dungeon. Right now we have a fairly smooth difficulty ramp from level 1 through level 20; if we cliff-faced that at level 3, say, then the game would get "interesting" much more quickly.

Fair point about giving people room (Sil essentially takes the approach that it's okay for people to die a lot near the start while they're learning, but in some ways that's unnecessarily cruel), and interesting idea. The problem I see with cliff-facing like that is that it's not transparent to the player; why should they expect level 3 to be so much tougher? And if it is, why shouldn't level 4 be tougher too? This is avoidable if you can somehow label it as a different part of the dungeon.

Edit: Perhaps it's okay to do this to players anyway, even if a little unexpected. It's early enough in the game that it's easily learned, and it is natural in some ways.

v4's combat system makes half the classes, more or less, rely on attack spam. But we could probably compress this to "You hit the Foo 4 out of 6 times" or something.

Sounds to me like that would be an improvement. (Sil manages something similar to v4 combat but has the light weapon users achieve parity through criticals rather than extra attacks.)

Monster attacks usually could be handled similarly, though there are a few monsters who have varied effects on their attacks (e.g. hit to hurt, touch to drain charges, and bite to drain experience).

We had a similar issue in Sil (but no monster had more than two attacks). Our solution was to make them attack just once each round, but randomise which attack they used. I don't know whether that would be of use here.

One big way that Angband and similar games get interesting is by getting you in over your head. You know that whole "ignorance breeds fear" thing? Well, when you don't know if you can handle a challenge, you become afraid -- and being frightened, in a game, is interesting!

I agree with your whole digression on danger and escapes, but wanted to highlight this section as I think it's particularly insightful. Of course frequent and easy escapes can only go if the deadly situations that require them are also much rarer. (I don't know the history here. I'd assumed that escapes were easy and then more deadly situations were added to keep the game interesting, because it was harder to imagine it flowing in the other direction, but perhaps they've both always been around.)

Generally I don't see why we should actively prevent players from grinding if that's what they want to do.

I certainly agree that there is a space for games which allow grinding (and I think Angband should probably be one of them), but let me explain why you might not want to:

Allowing grinding in some sense makes the game a continuous "choose your own difficulty level" as you go. This can be great for people pottering around having fun with their character. But it makes the game less of a challenge to be beaten. At least personally, something I enjoy is trying to optimise my play. This runs into problems if "optimal" play is to grind a lot, which I don't enjoy.

Perhaps that's something I should have mentioned in my list:
- Ensure optimal play is interesting.

You can have a challenge like "win the game in as few turns as possible", but this has a couple of problems:
- Without external reference (e.g. the ladder) or playing a lot of times you have no yardstick to measure yourself against.
- It's terrible as a goal when you're learning the game and have no idea what the late game is like.

The forced descent is a way of saying: "You don't have to choose your difficulty level, we've done that for you." This may annoy some players, but it's great for those enjoy meeting and overcoming external challenges. It also helps us from a game balance point of view to try to construct situations which stay dangerous for a majority of characters through the game while not being unfair on others.

Interestingly, Angband has an option which prevents grinding: Ironman. Is playing Ironman a sensible way to learn the game? I'm not sure. There's a lot of weight in default settings; if you thought that no grinding was the best way to play then making Ironman the default with an option not to would significantly change the perspective of new players to the game, while doing very little to old players who know how they prefer to approach the game.

Originally Posted by fizzix

As far as decreased XP per type of monster. V does have decreasing XP benefit as your level goes up, which sort of has the same effect. Sil has a very different approach to levels and XP so I'm not sure how much of a comparison is useful here.

What kind of decay function does V use? I spent a while thinking about these approaches some time ago. You can use decay functions to get the effect that when you've reached a certain level the current level gives you nothing. This is a similar, but not quite the same as Sil, which actively rewards seeking out new foes, and might induce you to try to fight one of a certain type of nasty enemy for the bonus for the first kill rather than run away every time you see them.

Of course you're right that Sil's experience system as a whole is quite different, but I think this part translates over quite easily, freely of the other aspects.

[Magnate]

Originally Posted by Scatha

What kind of decay function does V use? I spent a while thinking about these approaches some time ago. You can use decay functions to get the effect that when you've reached a certain level the current level gives you nothing. This is a similar, but not quite the same as Sil, which actively rewards seeking out new foes, and might induce you to try to fight one of a certain type of nasty enemy for the bonus for the first kill rather than run away every time you see them.

V uses a very simple system: the XP for each monster is divided by clev. So when you hit cl2 everything is worth half as much. When you go from cl10 to 11 everything is worth about 10% less. Later on in the game it's not really noticeable.

I agree that there is considerable scope for improvement in V here. The problem is, every time I think about reforming xp I end up re-creating Sangband.

[Scatha]

Originally Posted by Magnate

V uses a very simple system: the XP for each monster is divided by clev. So when you hit cl2 everything is worth half as much. When you go from cl10 to 11 everything is worth about 10% less. Later on in the game it's not really noticeable.

I agree that there is considerable scope for improvement in V here. The problem is, every time I think about reforming xp I end up re-creating Sangband.

Ah, I'm afraid to have to point out that this system doesn't really achieve anything. Or rather, it's equivalent to just multiplying the experience gap between level n and level n+1 by n (modulo some boundary conditions depending on exactly what happens when a single kill gives you a lot of experience and takes you over a boundary). At least to me it would then feel more elegant and transparent to just have experience from a monster unchanged with level and rescale the level boundaries (I think this can work fine). Or perhaps to push all of the scaling to this decay so the level scale can be linear (e.g. 0-999 = level 1, 1000 - 1,999 = level 2, etc.).

Apart from the Sil mechanic, one example of a system which has a genuinely different kind of effect would be:

If at character level n you kill a monster of level m < n, then the experience gained is divided by some increasing function of (n-m) (e.g. (10/9)^(n-m)). If you kill a monster of equal or greater level you get the usual amount of experience.

The effect of this would be that you are fairly rewarded for killing monsters of equal or greater level to you (and as deeper monsters will tend to be worth more experience, particularly rewarded for defeating these, as per the current system), but when you get to a certain point the experience for shallow monsters really dries up.