Playing Angband for the first time

[chris]

I find the math in Pyrel's approach to be much simpler to deal with; if you find a ring of +5 accuracy then that's a 5% increase in your chance to hit, no matter what (unless you cap out, of course). And if a monster has 25% evasion, then that makes it 25% harder to hit, no matter what (barring cap-out, but I can't imagine a monster hitting the evasion cap because it'd be a pain in the ass to fight). It's very simple, with easy math.

Well, I can't say how Pyrel's system will work since there isn't a playable game at the moment. My response was more to the early gestalt of this thread that Angband's combat system is too complicated, and I reduced everything down to an extremely simple formula that refutes that notion: p = 1 - 2A/3S (The 2/3 factor could be removed, of course giving just p = 1 - A/S).

But calculation of S is too complicated, you say? I agree with Nick that the this should be rephrased as "Angband's system is rich": Race, Class, Experience, Stats (Str and Dex), Equipment and even magical enhancements are possible to the player, and all of these should be managed in a way that gives demonstrable in game differences in the combat system. Variants add dimensions for personality and combat techniques (e.g. dual wielding, two-handed wielding etc) that further enrich the system. But at its core, it is just Combat skill vs monster AC.

This is not to say that the meaning of S should not be communicated to the player clearly. It should. In Poschengband, I give breakdowns of melee effectiveness on the character dump that anyone should be able to understand.

But I don't like your +5% accuracy example. Maybe I am missing something, but a ring of +5% accuracy seems silly to me as it effects everything equally. 5% more likely to hit an unclad, filthy street urchin as well as 5% more likely to hit a steel clad mercenary, not to mention an ancient dragon of legends? This is simpler math, I grant. But it feels like the wrong approach to me. Whereas something that gives +5% melee skill makes much more sense to me as it makes me 5% better than I was before. Whether or not that renders me capable of handling that steel clad merchant or even that dragon of legends depends on what my initial skill was to begin with.

But I wasn't picking on Pyrel with my comment. Indeed, I confess ignorance on that score. Rather, I was trying to defend Angband. Geez, someone has to

[Derakon]

Hey, fair enough to pick on Pyrel. I keep talking about what it's going to do, but until people are actually playing it and verifying that the system actually works, it's all just theorycrafting, really. No offense taken.

That said, the "5% better at hitting what" thing can be dealt with. If you want filthy street urchins to be easier to hit than battle-scarred veterans than orcs than dragons, then you can just scale their respective evasion scores. I don't plan on doing it myself since I feel it's more important to have a few noticeably-evasive monsters than it is to have monster evasiveness scale with level, but you could do things that way.

My attitude here is that if you scale evasion and player accuracy with level (/gear/class/etc.), effectively, then you end up with basically equivalent odds to hit things as if you do no scaling at all, except when there's a level mismatch. If the player is higher-leveled than their opponent, then it doesn't really matter if they can hit them 100% of the time or just 75%; they're going to win easily anyway. Likewise, if the monster is higher-leveled than the player, then it doesn't really matter if the player can hit them easily or not because the monster will be hard to kill for a myriad other reasons.

For what it's worth, once Pyrel actually gets out there, I want to start playing with the different affixes we can generate and start tweaking the combat rules. Affixes that make you highly accurate against specific monster types would be easily doable and would go nicely alongside the slay affixes.

Finally, don't forget about monster absorption. This reduces the damage taken per-blow by a flat amount, and it should scale with the monster's level since the player's damage output also scales. A Cave Orc with an absorption of 2 is noticeably more difficult than one with an absorption of 0, but that same absorption on a Mature Red Dragon is effectively meaningless.

[Nick]

Regarding melee accuracy, the current system uses 1dS >= 2A/3 to determine a hit, where S is the player's melee skill and A is the monster's armor class. Provided S >= 2A/3, this yields a probabilty of 1 - 2A/3S for a hit. Notice 2 things about this system: [1] probability approaches 1 asymptotically as player skill increases (for a given A) and [2] by forming the ratio of A to S, the amount of skill required to overcome a given armor class A is proportional to that A.

Thanks chris, that's a really useful way to put it. Given that, the question just becomes one of how best to communicate that to the player.

Also, is there any reason why 2/3 AC is used, rather than just having all monster ACs scaled down by 2/3?

[buzzkill]

But I think that the thing that is interesting to a player who is tunnelling is actually how many turns it takes and not how likely they are to succeed in any given attack - which is not the as combat. I prefer it like this because turns taken is the measure of how boring trying to tunnel is.

It's misleading. 15 is a hard number as it's stated on the screen. The actual number of turns may vary greatly. I can see that in most situations this wouldn't matter much, but in certain situations, like fleeing, thinking that tunneling will take a stated number of turns could result in death when it takes longer.

Back to the flip side. Do we calculate the number of turns necessary to kill the kobold or do we let the player weigh the variables. We certainly could calculate it given that we know the chance to hit, average damage, and opponents HP's. "You will kill the kobold in 3 turns (+0 dagger) or 2 turns (+0, +0 arrows)." All the precise damage numbers currently given are largely irrelevant once the player knows this.

I'm not advocating for the second option. Incidentally, it is similar to what Brogue uses (as well as letting you know how many turns you might survive the battle). It just might make a nice option for players using Morgoth's Manual (full monster knowledge).

[Derakon]

It's misleading. 15 is a hard number as it's stated on the screen. The actual number of turns may vary greatly. I can see that in most situations this wouldn't matter much, but in certain situations, like fleeing, thinking that tunneling will take a stated number of turns could result in death when it takes longer.

Does the digging information say "on average"? That's all that's really needed IMO; let the player know that their actual results may vary.

I have no opinion on doing this kind of thing for combat, aside from noting that the reported damage values when 'I'nspecting weapons assume that all blows/shots will hit, which is a blatant lie...

[half]

Thanks chris, that's a really useful way to put it. Given that, the question just becomes one of how best to communicate that to the player.

Also, is there any reason why 2/3 AC is used, rather than just having all monster ACs scaled down by 2/3?

I agree that chris's explanation is very helpful in getting an intuition for this and that you should definitely scale down the ACs by 2/3 if that won't introduce any big problems. Perhaps the issue is player AC values?

I was wondering whether you could also just get rid of the 5% guaranteed hit and miss chances. (or was it a 7% hit chance ???). Does it ever really get to a situation where you reach 5% or 95%?

It also struck me that a similar and arguably cleaner system is just to have the odds of hitting being S:A. So it is 50-50 if you have equal skill to armour, 75-25 if you have three times as much etc. This asymptotes to 0% if your skill is dwarfed by their armour and 100% if the reverse. If Angband players like typical characters to hit about 75% of the time, the numbers could be balanced to have things typically end up in that region. You could have a minimum modified skill or amour of 1 point to avoid infinities. I find it hard to imagine a two variable hit or miss system that is cleaner than that.

[taptap]

(Average number of turns to kill.)

All the precise damage numbers currently given are largely irrelevant once the player knows this.

Given how the combat system in Angband seems to work, this may indeed be all you need to know. However, if a system is sufficiently transparent, you not only don't need this kind of statistical information, some players would feel patronized, if it were offered unsolicited. (I certainly would do in Sil.)

I agree with half on the guaranteed hits and misses, but deleted a comment I had already written because I have no real clue about Angband. Imo it is table top inheritance and arbitrarily fixes the system at one point which may interfere with scaling or other things you try to achieve. (E.g. in Sil automatic hits would make low armour / high evasion chars - which isn't that relevant to an AC as evasion system such as Angband - almost impossible.)

[Derakon]

It also struck me that a similar and arguably cleaner system is just to have the odds of hitting being S:A. So it is 50-50 if you have equal skill to armour, 75-25 if you have three times as much etc. This asymptotes to 0% if your skill is dwarfed by their armour and 100% if the reverse. If Angband players like typical characters to hit about 75% of the time, the numbers could be balanced to have things typically end up in that region. You could have a minimum modified skill or amour of 1 point to avoid infinities. I find it hard to imagine a two variable hit or miss system that is cleaner than that.

That sounds reasonable to me, or at least it sounds nicely straightforward. I do kind of feel like it ends up boiling down to Pyrel's system except with asymptotes (i.e. diminishing returns) and more room for big numbers. Certainly it's an improvement over what we have now!

[LostTemplar]

It also struck me that a similar and arguably cleaner system is just to have the odds of hitting being S:A.

I use similar approach in my variant, and it seems ok. I have made most of the combat calculation directly proportional to stats, e.g. damage is proportional to strength, cahnce to hit and miss to dexterity, etc. Some things are proportional to (100+x) and are effectively percentile bonuses. While combat equations may be extremely complicated, many factors may be involved, relative effect of changing one factor is obvious.

[Therem Harth]

Say LostTemplar, I didn't know you had a variant. Where's the repo? I want to check out the modified combat system (and maybe steal it for Neoband if you're okay with that).

[LostTemplar]

Say LostTemplar, I didn't know you had a variant. Where's the repo? I want to check out the modified combat system (and maybe steal it for Neoband if you're okay with that).

https://github.com/LostTemplar/Yggband A variant, used mostly to test combat stuff in a working game environment, currently no longer developed. Combat system is not modified, but rewritten from scratch. I somewhat got bored fighting with Angband codebase, so currently writing from scratch here https://bitbucket.org/LostTemplar/draugr_rl
Basic reason to rewrite a combat system was to make player-monster symmetry, and reasonable monster vs monster fights.
When I will finish with map generation I will merge one thing into another.