Improving Z+Angband

[Thomas Larsen]

I really like Z+Angband. Zangband was always my favourite variant, and Z+Angband has just made it even better. So thanks, "mangojuice", for developing it!

Here are a few suggestions:

• Have smaller, richer dungeons. Maybe even make them permanent to reduce scummability?
• Have more continuity between quests. At the moment, they don't really seem connected in any way. I suppose to do this well would require a plot, and I don't really know how (if?) that could work.
• Implement autoexplore (DCSS style).
• Create more incentives to go through the early levels of the game. At the moment it's far more tempting for me to make my first-level character buy a wand of lightning bolts from a store and seek out a troll or giant camp or somewhere dodgy in the wilderness---that way I can get my character up to level 20 in ten minutes or so.

Cheers,
Tom.

[Camcolit]

I definitely agree with the auto-explore suggestion. I played DCSS for a long time and the 'Philosophy' statement for Dungeon Crawl Stone Soup is something I think all games could learn a lot from, regardless of genre. I will post what I believe are the most important sections -

Crusade against no-brainers
---------------------------
A very important point in Crawl is steering away from no-brainers.
Speaking about games in general, wherever there's a no-brainer, that
means the development team put a lot of effort into providing a "choice"
that's really not an interesting choice at all. And that's a horrible
lost opportunity for fun.

Anti-grinding
-------------
Another basic design principle is avoidance of grinding (also known as
scumming). These are activities that have low risk, take a lot of time,
and bring some reward. This is bad for a game's design because it
encourages players to bore themselves. Even worse, it may be optimal to
do so.

Interface
---------
The interface is radically designed to make gameplay easy - this sounds
trivial, but we mean it. All tedious, but necessary, chores should be
automated. Examples are long-distance travel, exploration and taking
notes.


They apply these principles, which made DCSS the fantastic game it is, though I prefer traditional class systems rather than the skill system it has.

[Thomas Larsen]

I definitely agree with the auto-explore suggestion. I played DCSS for a long time and the 'Philosophy' statement for Dungeon Crawl Stone Soup is something I think all games could learn a lot from, regardless of genre.

Agreed.

Tom.

[Antoine]

I definitely agree with the auto-explore suggestion. I played DCSS for a long time and the 'Philosophy' statement for Dungeon Crawl Stone Soup is something I think all games could learn a lot from, regardless of genre.

I think *band developers should consider the DCSS Philosophy statement, but not necessarily adapt it. Angband is not Crawl and attempts to use a Crawl approach can lead you astray. Some would argue that this had been the case in some recent V development initiatives.

A.

[zaimoni]

I definitely agree with the auto-explore suggestion. I played DCSS for a long time and the 'Philosophy' statement for Dungeon Crawl Stone Soup is something I think all games could learn a lot from, regardless of genre. I will post what I believe are the most important sections -

....

Interface
---------
The interface is radically designed to make gameplay easy - this sounds
trivial, but we mean it. All tedious, but necessary, chores should be
automated. Examples are long-distance travel, exploration and taking
notes.

...

Auto-explore is the one major misfeature that Crawl has; it positively doesn't belong in Crawl. On any level where @ isn't overwhelmingly overpowered, autoexplore has a non-zero plausibility of instant-killing @ in one keystroke. [This is why auto-explore does belong in NetHack: NetHack actually allows the @ to always be overwhelmingly overpowered in the mid-game and later.]

For a game that is intentionally designed to be a threat to any @ at any point in the late game, auto-explore is merely a death-in-one keystroke button. The only thing it seems to be good for is real-time speed runs.

[Magnate]

I think *band developers should consider the DCSS Philosophy statement, but not necessarily adapt it. Angband is not Crawl and attempts to use a Crawl approach can lead you astray. Some would argue that this had been the case in some recent V development initiatives.

? I can't think what the last sentence is referring to, but I personally agree with all three limbs of the Crawl philosophy wholeheartedly. Having said that, I also agree with zaimoni that auto-explore is a colossal misfeature (and would be so in Angband).

[Mangojuice]

1: Smaller richer dungeons - have you tried playing with "ironman_small_levels" on? It's definitely a bit risky but things do feel tighter that way.
2: Oh man how I would love to. The only way I can really think to do this, though, would be to have canned plots instead of canned individual quests. I'd love to make the game world feel more real, but I'm not sure this will.
3: Meh. I'm not particularly interested in having Z+ be the first variant to import that feature; it seems a bit odd for a 'band.
4: It's been my experience also that rapid advancement in the early levels is possible, and though your character risks death, you risk very little because you've invested almost nothing in them. I think that's actually a good thing! But there's a nice way to create a greater incentive to focus on the early game - play in competition mode. You may be able to find a way to get your (one and only) character to advance quickly but if not you'll just have to play the low levels slower. And of course some race/class combos are harder to get a fast start with than others...

But what I really wanted to talk about was the idea of having non-random levels. So I like it, from the point of view of the world being totally consistent. But it's daunting. It's bad enough keeping track of monsters on the current level - how do we keep track of monsters on all other levels that have ever been visited (and walls, traps, objects, doors, etc., too) without a complete explosion in memory? Also, how do we deal with the stairs problem? Right now, stairs allow you to scum for low-hanging fruit. With fixed levels, stairs would give you an absolutely safe escape. Should monsters be able to chase you up or down stairs? I'm worried about messing with the game balance, but it's interesting to think about.

I really like Z+Angband. Zangband was always my favourite variant, and Z+Angband has just made it even better. So thanks, "mangojuice", for developing it!

Here are a few suggestions:

• Have smaller, richer dungeons. Maybe even make them permanent to reduce scummability?
• Have more continuity between quests. At the moment, they don't really seem connected in any way. I suppose to do this well would require a plot, and I don't really know how (if?) that could work.
• Implement autoexplore (DCSS style).
• Create more incentives to go through the early levels of the game. At the moment it's far more tempting for me to make my first-level character buy a wand of lightning bolts from a store and seek out a troll or giant camp or somewhere dodgy in the wilderness---that way I can get my character up to level 20 in ten minutes or so.

Cheers,
Tom.

[zaimoni]

But what I really wanted to talk about was the idea of having non-random levels. So I like it, from the point of view of the world being totally consistent. But it's daunting. It's bad enough keeping track of monsters on the current level - how do we keep track of monsters on all other levels that have ever been visited (and walls, traps, objects, doors, etc., too) without a complete explosion in memory? Also, how do we deal with the stairs problem? Right now, stairs allow you to scum for low-hanging fruit. With fixed levels, stairs would give you an absolutely safe escape. Should monsters be able to chase you up or down stairs? I'm worried about messing with the game balance, but it's interesting to think about.

I've done some prototyping on this for Zaiband. The savefile code needs drastic restructuring to enable this; the player and dungeon save data are critically entangled in V3.0.6, and I don't see any relevant changes that would make things easier in recent V.

As for the stairs problem: it's a tradeoff. The safe escape also severely impairs stair-scumming (you actually have to make it to a different staircase). Hengband and its descendants have weak level persistence; you need to either Word of Recall or use a different staircase up/down than the the up/down staircase you arrived at to lose the level persistence.

[Antoine]

I think *band developers should consider the DCSS Philosophy statement, but not necessarily adapt it. Angband is not Crawl and attempts to use a Crawl approach can lead you astray.

For instance, I am very skeptical of any attempt to remove "no-brainer" decisions from Angband.

No-brainer decisions are what Angband is all about. Specifically, what separates a competent player from a poor player is the ability to consistently make no-brainer decisions correctly.

The poor player's weakness is that, when he's been playing for 4 hours and it's 2 a.m., he thinks 'I'm bound to kill that mob in one more round of attacks - it's only got two stars yet right?'. Even though the no-brainer decision is to take a no-fail escape option.

A.

[LostTemplar]

Angband no-brainer combat decisions actually may become interesting with increasing of game difficulty or getting bad luck e.g. if in ironman game you have very limited supply of fail safe escapes or good healing, it may be worth taking some risk instead of wasting them at first sign of danger.

[Antoine]

For instance, I am very skeptical of any attempt to remove "no-brainer" decisions from Angband.

And then, why would you try to remove opportunities to grind from Angband?

Many (most?) Angband players like a bit of grinding now and then.

The ones that don't, find ways not to do it.

A.

[Bostock]

Auto-explore is the one major misfeature that Crawl has; it positively doesn't belong in Crawl. On any level where @ isn't overwhelmingly overpowered, autoexplore has a non-zero plausibility of instant-killing @ in one keystroke.

Example, please?

I've played a *lot* of Crawl and this has never, never, happened, be I overpowered or (vastly more typically) evenly- or underpowered.

[LostTemplar]

Running in angband can kill too, more likely then crawl autoexplore, but many players use running. This is a common choice between easy play and safe play, same as diving vs farming.

[zaimoni]

Example, please?

Mountain Dwarf Fighter on a level that can generate Ogres. Auto-explore, get interrupted when pathfinder avoidably moves next to an unexplored square. Ogre in square. Ogre gets free attack. Die. (This does seem to be much rarer in 0.7.x, which has my proposed partial pathfinder algorithm fix specifically for preventing this.)

Alternately, foolishly explore in such a way that three orc priests rather than one are sighted at once: smite, smite, smite, splat.

Ignoring one-move causes of death that are truly unforeseeable like a unique with a wand of disintegration who zaps the wand on first possible move (which is less than 0.1% of my games), autoexplore fatalities are invariably bad movement doctrine fatalities. Either directly (like the above), or indirectly (by cutting off your own escape routes for you).

I did propose a partial algorithm fix for instant death by moving adjacent to an unviewable square on RGRD, and from what I can tell that has made it into SS Crawl 0.7.x series. It took several tries to explain because the calculations involved in correcting the pathfinder, were trivial.

[zaimoni]

Running in angband can kill too,

Since the removal of the cut-corners option, running in Angband does not kill @ in ways that are avoidable with manual play. The instant there is a meaningful choice, it stops.

Crawl's autoexplore will make meaningful choices of what move to make, even going so far as grabbing items for you, until either the whole level is explored or a monster is sighted. The problem is that it likes to move @ in ways that either kill @ before he can react, or cut off escape options so that the mistake that kills @ ~200 moves later has already been committed.