Depth vs. Complexity

[half]

Extra Credits has extra discussions about how to lead a player into a complex game, including on building tutorials (pithy summary: keep interesting things happening even if they don't require much skill on the player's part; don't frontload everything). Since it's generally-agreed that Angband ought to have a tutorial, this would be a good starting point for someone thinking about working on one.

A very good starting point is to just play the Sil one, and then copy it (or whatever aspects of it you like). The technical side of it is that it is a save game file that is stored in the folder with the high scores. It requires a new item type (a note) that displays its description when someone walks onto it.

It also requires a few changes to debug mode. Firstly, it requires a new command to reset the experience; to forget the map, item flavours, and and monster memory; and to take it out of debug mode. It also requires a modified look command accessible in debug mode that lets you modify the dungeon. I have it such that if you are looking at a square and type a letter, it creates the first monster/terrain represented by that letter. If you hit space, it iterates to the next one of the same type, if you hit tab, it generates a new random one of the same type. There is also a key that toggles permanent light on the square. Then I started a new debug character and built the level! It is pretty easy to find all this in the Sil source code.

(There are a few other things such as setting a tutorial flag such that you can't save during the tutorial).

There's also some discussion of "first-order optimal strategies", which are basically the easily-learned, reasonably-effective techniques you use early in a game (e.g. in Angband, Potion of Speed + whack'n'back).

This is a pretty subtle point in game design and I completely agree with it. Some players have enough knowledge to be able to play games well and to improve the balance of them, but tend to over-balance the early game. People often say that we need the choices to be non-obvious. There is an important truth here, but it is not that *every* choice should be non-obvious. We need lots of choices with different shades of obviousness. Then people get a natural learning curve as they learn how to optimise more and more of these choice situations.

[Patashu]

Dodge + Flanking also makes it more desirable to fight in a room than in a corridor (something that's been discussed recently), which I enjoy

[Mikko Lehtinen]

Dodge + Flanking sounds awesome. Now what would be the best way to copy the mechanic to Mist... (Edit: I'm developing a possible mechanic here.)

I still haven't played Sil properly. I will.

[half]

How about a Potion of Whirl, or a mage spell of the same name. Whenever a monster attacks you in melee, you try avoid the attack by moving in a random direction. If there's no space in that direction, bump, the attack hits you.

I'm not sure being randomly moved around the room sounds fun. I'm also not sure it is worth it to avoid one attack. If it is worth it, perhaps in a positional tactical game it shouldn't be.

The way this works in Sil is that there are abilities called Dodge and Flanking. Dodge gives you [+3] to evasion (a fairly large bonus) if you spent the turn moving. This is not super useful on its own, but helps with escaping a little. Flanking gives you a free attack on a creature if you move between two squares both of which are adjacent to it. If there is more than one valid creature, it chooses the targeted creature by preference, and a random one if none are targeted. In Sil, attacking a creature moves the target to it.

Together these abilities lead to the kind of dynamic combat you are imagining, but without any random movement. Others also work particularly well with these, such as Whirlwind Attack, Sprinting, Charge.

[Mikko Lehtinen]

With "attack two monsters" mechanic I wanted to push the player to utilize tables, magic circles, forests and warding runes in rooms. Mission accomplished.

Berserk rage allowing attacks against all adjacent enemies is a good idea.

How about a Potion of Whirl, or a mage spell of the same name. Whenever a monster attacks you in melee, you try avoid the attack by moving in a random direction. If there's no space in that direction, bump, the attack hits you.

[half]

A whirlwind attack that allows you to attack all adjacent monsters may make group fighting advantageous when it's combined with other temporary buffs (HM's approach to killing morgoth in Sil). This could also be implemented in V, although 1 v many situations are often too deadly to even consider using this.

I think having the beserk rage effect make you attack all adjacent enemies when you bump into one would be a really cool and pretty balanced improvement to V. It would also be really quite easy to implement: just run a loop through all adjacent squares and run the attack function on them if there is a monster present (in Sil I went for bonus points and make sure to start with the square they bumped and then go clockwise or anticlockwise at random).

	if ((p_ptr->rage) && (adj_mon_count(p_ptr->py, p_ptr->px) > 1) && !p_ptr->afraid)
	{
		int i;
		bool clockwise = one_in_(2);
		
		// message only for rage (too annoying otherwise)
		if (p_ptr->rage)
		{
			msg_print("You strike out at everything around you!");
		}
		
		dir = dir_from_delta(y - p_ptr->py, x - p_ptr->px);
		
		/* Extract cycle index */
		dir0 = chome[dir];
		
		// attack the adjacent squares in sequence
		for (i = 0; i < 8; i++)
		{
			if (clockwise)  dir = cycle[dir0+i];
			else            dir = cycle[dir0-i];
							
			yy = p_ptr->py + ddy[dir];
			xx = p_ptr->px + ddx[dir];
			
			if (cave_m_idx[yy][xx] > 0)
			{
				if (p_ptr->rage)		py_attack_aux(yy, xx, ATT_RAGE);
			}
		}
	}
	else
	{
		py_attack_aux(y, x, attack_type);
	}

[half]

I think I've succeeded in Mist, but it took some extreme measures. I gave the player the ability to fight with two monsters in melee at the same time.

I think that is a pretty interesting approach. It doesn't feel too unrealistic either. It is not that the player is stronger against groups, just that they can't use their ability to attack in multiple directions in corridors.

Many "realistic" role playing games have opposite rules, giving all sorts of penalties to the one-against-many fighter. I decided tactical fun is more important than realism.

Sil takes the 'realistic' approach here. When a monster attacks in melee, you get [-2] to evasion for each monster on the three sides of you that are furthest away from the attacker and [-1] for any other adjacent monster. This means that if you are surrounded, then as well as taking 8 attacks for each one of yours, you are at [-10] to evasion to all of them! This makes being surrounded very bad, and the monsters in Sil know how to take advantage of this and they try to surround you. However, in practice you choose to move to a wall for 5v1 or a corner for 3v1, or ideally a corridor for 1v1.

The main effects are to keep weaker monsters interesting (the horde of orcs can still take down Boromir...) and to make tactical positioning more interesting (you'd rather have the two Trolls adjacent to each other than one on each side of you, so you manoeuvre to achieve this).

Sil compensates for this anti-room mechanic in three ways. Firstly there are ways to be able to attack all adjacent monsters: herbs of rage and the Whirlwind Attack ability. The latter only works if all adjacent squares are passable so it really entices people to the centre of rooms. Of course 8 of your attacks to 8 of theirs is still not brilliant. It is most useful in situations where you can't avoid a many-to-one fight and need to do it as well as possible. It is also useful when you don't want to waste turns in combat, when you are getting a bonus for killing creatures quickly due to Song of Slaying, and a few other cases.

The second thing that pushes people out of corridors in Sil is the lack of escapes: corridors are really dangerous if 5 orcs come up to you from either end. In contrast, in a room you can see monsters coming from further away and can flee even if a few of them remain.

The third (and largest) effect is the use of an improved version of the 'pack AI' so that intelligent enemies won't go 1 on 1 with you in a corridor unless it is in their favour (e.g. you are wounded or confused). You often end up fighting 3v1 at doorways or 2v1 at corridor corners until you are wounded enough that they will pursue you to 1v1.

Overall, it is still pretty corridor based, but there is quite a bit more of interest. In general, there is a surprising amount of geographical tactical complexity possible with just walls and floor.

[fizzix]

Yeah, making terrain features matter in combat is so hard that it may be not worth trying in Vanilla. Most fights happen in corridors.

I think I've succeeded in Mist, but it took some extreme measures. I gave the player the ability to fight with two monsters in melee at the same time. Many "realistic" role playing games have opposite rules, giving all sorts of penalties to the one-against-many fighter. I decided tactical fun is more important than realism.

Some features, like teleportation gates, would work well. Moving a portion of escape possibilities from the player's inventory/spellbook to the map might be fun.

There are ways to make 1 v many situations appealing for the 1. V already has one, which is ball or LoS spells which allow you to hit many monsters at the same time, maximizing damage/mana. There's also the rare situation where you have immunity to an element and you can damage monsters by having other enemies breathe on you. (I've used this once with summoned hellhounds and draugluin.) A simple approach which is applicable to V is to make something like bedlam actually work well (~30% of the monsters will be confused, and when confused, have a chance to attack an adjacent monster or breathe at a random spot.) You could sidle up to a giant pit, cast bedlam, and then scurry away while the monsters pound on each other. Or you could use escorts to take down tough uniques, i.e. gelugons and lungorthin.

A whirlwind attack that allows you to attack all adjacent monsters may make group fighting advantageous when it's combined with other temporary buffs (HM's approach to killing morgoth in Sil). This could also be implemented in V, although 1 v many situations are often too deadly to even consider using this.

There are other options but they may violate the depth vs complexity mechanic. For example, you can have player skills that give you extra combat expertise when more than one enemy is adjacent. You could have spells that you cast on an enemy and it then damages the monster based on how many there are around. You could have a leech life spell which sucks some life from all adjacent enemies and heals you in the process. You could have shielding spells which allow you to absorb some amount of damage, giving you temporary invulnerability and allowing you to put yourself in high damage situations.

When it comes down to it, most of the ways to make 1 v many situations appealing is to turn the enemies into (unwilling) allies.

[Mikko Lehtinen]

Yeah, making terrain features matter in combat is so hard that it may be not worth trying in Vanilla. Most fights happen in corridors.

I think I've succeeded in Mist, but it took some extreme measures. I gave the player the ability to fight with two monsters in melee at the same time. Many "realistic" role playing games have opposite rules, giving all sorts of penalties to the one-against-many fighter. I decided tactical fun is more important than realism.

Some features, like teleportation gates, would work well. Moving a portion of escape possibilities from the player's inventory/spellbook to the map might be fun.

[Magnate]

Ah yes, that will be a big improvement. Although, I'm thinking of also as more of a flavor thing with the condition that it can't make gameplay worse. I've really only focused on terrain so far, it's hard. I have a feeling that once we start thinking about monster placement, a lot of that work is going to be a collective decision.

d_m had some really fantastic ideas on this - I really hope he wrote them up somewhere!

[fizzix]

I wasn't really talking about terrain btw; I was assuming that the monster placement aspect of your dungeon generator would be a bit improvement on V (e.g. themed areas, lairs, former lairs now inhabited by other things, etc.).

Ah yes, that will be a big improvement. Although, I'm thinking of also as more of a flavor thing with the condition that it can't make gameplay worse. I've really only focused on terrain so far, it's hard. I have a feeling that once we start thinking about monster placement, a lot of that work is going to be a collective decision.

[Magnate]

I'm skeptical that terrain, no matter how varied, will add more strategic gameplay options. The main strategic difference is open room vs corridor, with the obvious result that in a 1 v many the 1 will always choose a corridor. Different terrains, like forest, ice, tables, etc. might increase strategy, but the effect is not as large as might be thought. My approach to dungeon generation assumes that varying terrain in V is mainly for flavor purposes, and the goal is to keep the dungeons interesting to explore (and maaaaybe internally consistent.) rather than level after level of the same room types.

That being said, I'm hoping to get back to pyrel development this week. I took a little break after January 1 to focus on other things. Now that mystery hunt is over, I should have more free time the next few months. Unfortunately, with mystery hunt 2013 over it means that mystery hunt 2014 just begun, which matters because I'll be on the writing team next year. I'm not expecting that to start eating free time until March-April or so and then eat all of my free time around October-December. So there's no time like the present for pyrel work!

Good to hear it. I lost my momentum too, so am waiting for it to come back.

I wasn't really talking about terrain btw; I was assuming that the monster placement aspect of your dungeon generator would be a bit improvement on V (e.g. themed areas, lairs, former lairs now inhabited by other things, etc.).

[fizzix]

I was disconnected from the world for pretty much the whole weekend, and I'm only now catching up.

regarding saving rolls: Yeah, this is way back in the thread, I know. I think opposing saving rolls are fine as long as the consequences for failing aren't dire. If the consequences for failing are death, and you expect your game to be long and permadeath, then it's not fair. You can't take 1% chances and expect to live. Sil is both shorter and has less dire situations (i.e. taking damage awakens you from entrancement.) I like the idea of being immune to certain effects because of a high saving throw.

Vanilla currently is not an overly strategic game. It's charm lies in defeating tough enemies, exploring unknown areas, and finding cool loot. The main strategic decisions are based on inventory and when to use consumables. The main difficulty is knowing what monsters do and what you should fight and what you shouldn't. Once you know that, most fights are scripted. This is not a poor game design, but it is very different from most modern roguelikes which place a high priority on interesting combat situations. They are also almost always shorter.

Yes, I agree with this. More heterogeneous monster placement would be a quick win. Fizzix's dungeon generator for pyrel should lead to quite a lot of really interesting board situations.

I'm skeptical that terrain, no matter how varied, will add more strategic gameplay options. The main strategic difference is open room vs corridor, with the obvious result that in a 1 v many the 1 will always choose a corridor. Different terrains, like forest, ice, tables, etc. might increase strategy, but the effect is not as large as might be thought. My approach to dungeon generation assumes that varying terrain in V is mainly for flavor purposes, and the goal is to keep the dungeons interesting to explore (and maaaaybe internally consistent.) rather than level after level of the same room types.

That being said, I'm hoping to get back to pyrel development this week. I took a little break after January 1 to focus on other things. Now that mystery hunt is over, I should have more free time the next few months. Unfortunately, with mystery hunt 2013 over it means that mystery hunt 2014 just begun, which matters because I'll be on the writing team next year. I'm not expecting that to start eating free time until March-April or so and then eat all of my free time around October-December. So there's no time like the present for pyrel work!

[Magnate]

The biggest cause for too high resource management complexity is simply that there are too many items and spells you could use in a given situation. Going through all of the options every time you need to make a decision is boring. There are so many options that players don't have the brainpower to keep them all in their mind. They just do the same old things that worked before. Only when death is looming near, the player gains the energy to look through all the inventory items and spells and make the time-consuming tactical analysis.

However, adding a strategic level to resource management (like half described earlier) does not necessarily increase that kind of complexity...? And if it does, we can counter by reducing the number of inventory items and spells. In my opinion, too, it would make the game more enjoyable!

I think I agree that too much choice takes fun away, but I don't think Angband has too much. Having said that, a lot of the spells in both realms are pretty 'meh', so they could be pruned without losing a lot.

I was pleased that half and Scatha actively strove to make every weapon different in Sil - Derakon and I did the same in v4. I think *meaningful* choices are more important than more or fewer choices per se.

Also, we could increase Board Complexity a lot without taxing the average player's brain too much. All the positioning rules in Mist are pure fun for me, and I've never come even close to analysis paralysis because of them.

Yes, I agree with this. More heterogeneous monster placement would be a quick win. Fizzix's dungeon generator for pyrel should lead to quite a lot of really interesting board situations.

[Mikko Lehtinen]

I think MaRo meant with Board Complexity cards that are already on the battlefield. In Angband terms: the dungeon map and everything on it.

Strategic Complexity in MaRo's terms is how to best use cards in your hand. In Angband terms: spells and items in your inventory. Let's call this Resource Management Complexity instead, shall we? In Angband it's mostly tactical.

My recent problem with Ironband Mana clearly falls in this category. Resource management can already be quite complex in Angband, and increasing the complexity further might make the game less enjoyable. Of course players have wildly different tastes in this.

The biggest cause for too high resource management complexity is simply that there are too many items and spells you could use in a given situation. Going through all of the options every time you need to make a decision is boring. There are so many options that players don't have the brainpower to keep them all in their mind. They just do the same old things that worked before. Only when death is looming near, the player gains the energy to look through all the inventory items and spells and make the time-consuming tactical analysis.

However, adding a strategic level to resource management (like half described earlier) does not necessarily increase that kind of complexity...? And if it does, we can counter by reducing the number of inventory items and spells. In my opinion, too, it would make the game more enjoyable!

Also, we could increase Board Complexity a lot without taxing the average player's brain too much. All the positioning rules in Mist are pure fun for me, and I've never come even close to analysis paralysis because of them.