Why is Angband fun?

[Philip]

So, I kind of suggested I have some (stolen) thoughts in another thread, on this particular topic. Near as I can tell, I got these concepts from an episode of Roguelike Radio, but there are a lot of episodes and there aren't any transcripts as far as I can tell.

The basic narrative of a game of Angband is pretty vague and unimpressive, compared to most RPGs. At some non-specific and impossible time, in a fortress called Angband containing almost all of the notable evil characters, and many of the notable good characters of the Tolkien mythology your character is supposed to kill Sauron and Morgoth and thereby win the game. The only interaction you can have with a character is fighting (shopping doesn't count because shopkeepers aren't characters, they are vending machines). The graphics are simplistic, and the game mechanics are extremely abstract and largely taken from tabletop RPGs, notably tiles and turns. There is little to no writing, because writing a narrative for the player to consume would be redundant.

The player writes their own narrative. I don't mean in a tabletop RPG way, though if someone plays Angband like that there's of course nothing wrong with that. However, in my experience, my history with the game itself is the main way the game creates memorable moments or emotional reactions. The first win is not fun because you have slain Morgoth, the Dark Lord, and Arda will be free. Arda doesn't exist within the bounds of the game. The first win is fun because you finally beat the goddam game after so many characters, promising characters even, died, leaving you to start from the beginning. Those characters who died, and those who won, are part of the meta-narrative of the game. The only way they are related to the current playthrough is by way of the player (because of permadeath), since no two characters are alike (procedural generation).

Now, if this theory is true, and the meta-narrative is in fact the most important part of the game, this has some important implications.
The game needs to be about making each playthrough not (just) an engaging experience, but most importantly, a good story. In particular, IMO, this is important with respect to "unfair" events. Getting killed by Kavlax, the hunger clock, Gravity Hounds, a Drolem, or an OOD monster may not be the most fun in the moment, but within the metanarrative, it's critically important. It creates intermediate goals for every subsequent playthrough, or motivates the player to defeat the ultimate bad guy (the game, of course).
The game should be hard, to build up this meta-narrative. If you win within three games of first playing Angband, that value is lost.
Continued maintenance is not only good because it makes the game better, but because the game is part of the meta-narrative, so any changes to it make the meta-narrative more interesting.
The purpose of procedural generation is to have stuff happen in a different way each game, it is not an end in itself.
Adding a narrative with clever writing would be a waste of everyone's time.

I would quite like to know if my ideas here line up with the ways other people have fun with Angband.

[Pete Mack]

Angband is a game of economics--the Tolkien theme is almost happenstance, and indeed is not preserved across all variants. Players loom to allocate resources efficiently, minimize risk, and maximize ROI in various ways. You end up with risk minimizing specialists on the one hand, and reward maximizing specialists--who accept the occasional death as a cost of risk--on the other.

[Bucephalus]

First of all, I agree with everything you've said, especially about memorable moments—untimely deaths and utterly deserved deaths alike, near misses, badly misjudged circumstances that somehow didn't kill your character, and everything in between. Deep down, this is what Angband is about.

I'd like to address something you didn't mention, which is item drops. Why risk ending your hero's glorious quest in the middle of his arc to stupidly march into that vault? Because Ringil might be in there. Why fight Kavlax when he probably won't drop anything your character even needs? Because maybe he will.

This is called gambling, and I think it's also part of what makes a roguelike. You keep doing the same thing on the off-chance that something good will happen, and it does often enough that you keep doing it, but not so often that you say, "This is stupid." and stop.

To wit, the latest versions of Tome fail the stink test here, since everything is dropping great items so often it becomes a roguelike vestigially attached to an inventory management game. Angband strikes a good balance, but YMMV.

[t4nk]

Originally Posted by Philip

Now, if this theory is true, and the meta-narrative is in fact the most important part of the game, this has some important implications...
The game should be hard, to build up this meta-narrative. If you win within three games of first playing Angband, that value is lost.

Mmm, Philip, but the easier versions of the game (such as V itself) are more popular than the harder ones (such as variants of OAngband). How do you explain that?
Anyway, I do agree that it's good and necessary to kill the player sometimes, and I think that Angband is fundamentally too long for that. See, if the average probability to die on a level is 1% and the game has 100 levels, than the expected winrate is 36.6%. The difference between 1% and 0% is (psychologically) tiny* (at least, for me), but the difference between 36% and 100% is yuge
Anyway, I, in fact, won Vanilla within my first three attemps (I don't count the games when I was just learning the commands and what scrolls of Phase Door do). I do think that V is too easy and boring and long...

* More precisely, the difference between 99% probability of survival and 100%.

edit: Some will say, "but you can make Angband shorter and more dangerous by diving". Indeed, and this is, IMO, the best thing about it. I think the game should gently nudge the player in that direction. I recall Derakon objecting to it, saying something like "too many people already find Angband geared towards speedrunning". I'd say, a lot more people find Angband grindy.

[Philip]

Originally Posted by t4nk

Mmm, Philip, but the easier versions of the game (such as V itself) are more popular than the harder ones (such as variants of OAngband). How do you explain that?

Oangband hasn't been developed in maybe a decade? The UI is garbage. It leans heavily on being irritating (item destruction through immunity, for example), which I think is a worse emotion to invoke than sorrow or anger. FAangband is also not in active development, and it had a somewhat niche concept in the first place. I would say the ComPosCheng branch is rather hard (for most race/class combinations) and frequently unfair, and is still rather popular.

Anyway, I do agree that it's good and necessary to kill the player sometimes, and I think that Angband is fundamentally too long for that. See, if the average probability to die on a level is 1% and the game has 100 levels, than the expected winrate is 36.6%. The difference between 1% and 0% is (psychologically) tiny* (at least, for me), but the difference between 36% and 100% is yuge

36.6% is a way too high winrate. If each character who wins is avenging on average two other characters, the stakes are too low.

The difference between no chance of dying and any chance of dying is infinite. The difference between 100% chance of winning and any other chance of winning is also infinite. The game only has a point so long as you have any chance of dying, and thus, any chance of winning. The length of the game is not excessive, so long as the aim of development is in fact to distribute a large chance of death over many levels. However, I don't think I could make a logically consistent argument to conclude the optimal length of the game from the metanarrative argument, since it's not clear to me that distributing deaths over many levels, or extending the game, is necessary to it, and in any case, that's not really the reason Angband has 100 levels anyway.

Anyway, I, in fact, won Vanilla within my first three attemps (I don't count the games when I was just learning the commands and what scrolls of Phase Door do). I do think that V is too easy and boring and long...

It wouldn't feel as long if it weren't boring, and it wouldn't be boring if it weren't easy.

edit: Some will say, "but you can make Angband shorter and more dangerous by diving". Indeed, and this is, IMO, the best thing about it. I think the game should gently nudge the player in that direction. I recall Derakon objecting to it, saying something like "too many people already find Angband geared towards speedrunning". I'd say, a lot more people find Angband grindy.

Indeed, the ability to create additional challenges is a good way to keep the game at an optimal level of difficulty and emotional engagement for experienced players. Current V is good at this.

[t4nk]

I would say the ComPosCheng branch is rather hard (for most race/class combinations) and frequently unfair, and is still rather popular.

Maybe. I only played the first several dungeons of it with warrior types and it was about as difficult as Vanilla - that is to say, not at all. Perhaps it does get more difficult later. I'll take your word for it.

36.6% is a way too high winrate. If each character who wins is avenging on average two other characters, the stakes are too low.

Are they? Well, currently in V the maximum possible winrate is very close to 100%. DCSS is the same. BTW, I used to play it online: http://crawl.akrasiac.org/scoring/players/zix.html. And I didn't even tried hard to win (that is, played poorly on purpose, to speed up the game a bit). Much higher winrate is possible if you're really determined and patient. Yet DCSS is probably the most popular roguelike currently. I played it quite a lot (and mostly offline).
Thus, my experience tells me that it's not true that a roguelike must be difficult to have good replayability.

The difference between no chance of dying and any chance of dying is infinite.The difference between 100% chance of winning and any other chance of winning is also infinite.

Wait... no? I think that's not true mathematically and also not true psychologically. Try this: what is the difference between 100% chance of winning and 99.9(insert any finite number of 9s)% chance? Clearly there is a x% chance of losing which is bigger that 0 yet just about everyone would say "it's basically the same to me". So the question what that x is.

[Voovus]

Originally Posted by Philip

The game needs to be about making each playthrough not (just) an engaging experience, but most importantly, a good story. In particular, IMO, this is important with respect to "unfair" events. Getting killed by Kavlax, the hunger clock, Gravity Hounds, a Drolem, or an OOD monster may not be the most fun in the moment, but within the metanarrative, it's critically important.

Objection, yer honour. While I agree that player deaths are important for the metanarrative, I don't think "unfair" events should be. For instance, dear ol' Rogue didn't have this issue. Yes, your first hobgoblin, first centaur, first troll and the first five griffins would kill you - but these weren't annoying instakills, and the characters that died these deaths felt like heroes. (The only "unfair" deaths I recall was from ice monsters, and these appeared very early in the game, so at least you wouldn't randomly lose a well developed character.) I'm not saying that I can propose a fix for Angband, but only pointing out that annoying deaths shouldn't have to be a certainty of Angband life.

[Derakon]

You're taking a narrative-based approach to the game, and that's totally valid, but it's not how everyone plays. I take a much more gameist approach; I like the game because of the nature of the decisions it asks me to make and the way it rewards those decisions. You know that bit in the Matrix where the guy says "I don't even see the code any more, just blonde, brunette, redhead"? It's the opposite for me -- I don't see black orcs, I see "group, weak to light, has ranged attack and moderate health". Dangerous monsters are fun not because my hero gets to face off against a legendary dragon or whatever, but because the stakes are raised on the decisions I'm expected to make. And similarly, the legendary artifacts are fun because they represent progression and an increase in power, not because I've found the same sword that Gandalf wielded (and anyway, Glamdring is pretty lousy...).

[the Invisible Stalker]

Originally Posted by Derakon

You're taking a narrative-based approach to the game, and that's totally valid, but it's not how everyone plays. I take a much more gameist approach; I like the game because of the nature of the decisions it asks me to make and the way it rewards those decisions. You know that bit in the Matrix where the guy says "I don't even see the code any more, just blonde, brunette, redhead"? It's the opposite for me -- I don't see black orcs, I see "group, weak to light, has ranged attack and moderate health". Dangerous monsters are fun not because my hero gets to face off against a legendary dragon or whatever, but because the stakes are raised on the decisions I'm expected to make. And similarly, the legendary artifacts are fun because they represent progression and an increase in power, not because I've found the same sword that Gandalf wielded (and anyway, Glamdring is pretty lousy...).

I agree with almost all of this, but it seems to be meant as a contrast to a point of view that I don't see anyone taking. There probably are people who really buy in to the role-playing aspect, but they don't seem to be represented on this thread.

Elaborating a bit more on your point about the fun lying in the decisions you have to make, I think that gives some guide to how hard or easy the game should be. The game shouldn't be so deadly that you are doomed to fail and your decisions therefore don't matter. On the other hand it shouldn't be so forgiving that you can consistently get those decisions wrong and still win. That criterion still leaves a fairly wide range of difficulties. It allows instadeath, for example, as long as it's not so common as to make winning nearly impossible, but it doesn't require instadeath.

[the Invisible Stalker]

Originally Posted by Philip

I would quite like to know if my ideas here line up with the ways other people have fun with Angband.

I agree with most of this. I do think there's a fundamental problem with Angband though, that its early game encourages bad habits. This is something Eddie used to complain about a lot and I would probably still have many of those bad habits if I hadn't found his rgra posts. That can make the game quite frustrating. One way around that is to tweak the game so that you can still win with bad habits, but a better way is to make it easier to understand why you just died. By that I don't mean proximate causes like 'killed by an orc' but deeper reasons like 'stayed too long on a level with a low stealth character'. I don't really know how to do that, which is one of many reasons I'm not a developer.