Dead character feedback

[Derakon]

I've been playing a half-troll paladin for awhile. He's dead now. So, time to wrap things up!

The dump's here so you can see his stats. In no particular order...

* Giant Rocs are overpowered. I compiled a list that sorts monsters from levels 36-60 by their average melee damage, multiplied by their speed (so e.g. a 1.5x multiplier for mature dragons). Giant Rocs are 9th from the top, and the only non-unique monster that beats them is the Death Mold. For reference, they're native to dlvl 40. If you divide the damage by the monster's native depth, and exclude unique monsters, then you get this list, where the Giant Roc is 5th from the top.

Now, this list ignores resistances; depending on the type of attack a monster has, it may deal more or less damage. But Rocs still have absurdly burly melee damage for their depth. Given that I remember them being nigh-absolute pushovers, I think someone overcompensated at some point.

* Stat gain is very slow. I don't know what the desired rate here is, but note that I only had two stats above 18 (by internal numbers). I'm pretty sure I saw less than 10 stat potions this entire game, counting !Charisma.

* Note the heavy crossbow in my home. I couldn't hit diddly with that thing, and when I did typically the bolts did pathetic damage. Orb of Draining, especially with my unaccountably high WIS, was much more reliable.

* Also note the total lack of artifacts. This isn't necessarily a bad thing, but it occurs to me that most of the artifacts in the game are going to be serious disappointments by the time they do get generated. If you're going to get 5 artifacts all game, all of them showing up after you've passed 3000', are you really going to want one of them to be Paurhach? I don't know what the right solution here is, though.

* We may want to revisit prowess/finesse values for the different classes. Paladins at least are overbiased towards prowess weapons, I think. When I died at clvl33 my innate finesse was 88 and prowess was 252. Granted this wasn't helped much by my pathetic DEX score of 12 -- but if it'd been 18/50 instead I'd only be up 54 points. The thing to realize is that players are going to tend to favor weapons that favor their stats, so even a slight bias in their stats gets emphasized by their weapons. For example, even if my finesse went from, say, 50 to 500, my blows would only go from 1.05 to 1.5 if I'm using a Great Hammer.

* Potions of Restore Mana are too common. I also found a lot of potions of Healing for a bit, then suddenly they stopped showing up. Might want to look at a plot of distribution of !Healing as a percentage of all items, with respect to depth.

* Cavernous levels need some work. Sub-items:
** It's very hard to run away from fights in caverns. Caverns have very simple layouts, and the straight-line distance between two points is only rarely much shorter than the distance monsters must travel. Monsters will rapidly track you down if you try to flee, even via teleport.
** The monster density is very high.
** You can't really use the terrain to control fights. Normal combat in Angband relies heavily on one-tile-wide corridors with right angles in them. There's the occasional chokepoint in cavernous levels, but it still typically leaves you in full view of a large number of monsters.
** Given the monster density and bad terrain, group monsters are particularly vicious, especially since they often end up being able to play keepaway (not charging) while still having you in LOS to deploy breath weapons / spells.
** There's very few items on the floor.
All said, I think caverns would work much better as a room type instead of a type for the entire dungeon.

* At some point ID-by-use of weapons and armor becomes unfeasible (because swapping gear around becomes dangerous, if nothing else). Currently that means you start examining the item for desirable affixes, and IDing the ones you think might be worthwhile. This breaks down when you consider weapons that can have lots of hidden affixes that boost dice, damage, and finesse, which could make them far more valuable than the weapons that have noticeable affixes but no significant dice or pluses.

My suggestion would be to make dice and pluses reveal on pseudo at some character level; at least by clvl 30. This would go with auto-ID-on-pickup at clvl 45 or whatever, which I know's been suggested before.

[Magnate]

* Giant Rocs are overpowered. I compiled a list that sorts monsters from levels 36-60 by their average melee damage, multiplied by their speed (so e.g. a 1.5x multiplier for mature dragons). Giant Rocs are 9th from the top, and the only non-unique monster that beats them is the Death Mold. For reference, they're native to dlvl 40. If you divide the damage by the monster's native depth, and exclude unique monsters, then you get this list, where the Giant Roc is 5th from the top.

Now, this list ignores resistances; depending on the type of attack a monster has, it may deal more or less damage. But Rocs still have absurdly burly melee damage for their depth. Given that I remember them being nigh-absolute pushovers, I think someone overcompensated at some point.

It also ignores their resists, hitpoints, and so on. What you've done is recreated half of the monster power algorithm (the bit that works out total melee damage). You need to add power for monster spells and defences - see src/monster/mon-power.c - it hasn't been updated for a while, so it doesn't take into account evasion or absorption properly yet. In fact to be clear, it doesn't take evasion into account at all (which means it needs updating), and it probably undervalues absorption, as it's what used to be simply AC.

* Stat gain is very slow. I don't know what the desired rate here is, but note that I only had two stats above 18 (by internal numbers). I'm pretty sure I saw less than 10 stat potions this entire game, counting !Charisma.

That's a very recent change, announced by takkaria in the V forum. IIUC it's intended to offset the fact that stat gain is much quicker now you can get from 18/90 to 18/100 in a single potion.

* Note the heavy crossbow in my home. I couldn't hit diddly with that thing, and when I did typically the bolts did pathetic damage. Orb of Draining, especially with my unaccountably high WIS, was much more reliable.

Is this after your changes to ammo dice? More food for thought re archery.

* Also note the total lack of artifacts. This isn't necessarily a bad thing, but it occurs to me that most of the artifacts in the game are going to be serious disappointments by the time they do get generated. If you're going to get 5 artifacts all game, all of them showing up after you've passed 3000', are you really going to want one of them to be Paurhach? I don't know what the right solution here is, though.

Well, the stats show that this time you really are just unlucky (compared with your first v4 AAR, when you were discovering quite how overly rare artifacts were in that version). Artifacts are still rarer than in V (though takk's recent changes have probably narrowed that gap by making artifacts less common in V), but you should certainly find a few before 3000'.

* We may want to revisit prowess/finesse values for the different classes. Paladins at least are overbiased towards prowess weapons, I think. When I died at clvl33 my innate finesse was 88 and prowess was 252. Granted this wasn't helped much by my pathetic DEX score of 12 -- but if it'd been 18/50 instead I'd only be up 54 points. The thing to realize is that players are going to tend to favor weapons that favor their stats, so even a slight bias in their stats gets emphasized by their weapons. For example, even if my finesse went from, say, 50 to 500, my blows would only go from 1.05 to 1.5 if I'm using a Great Hammer.

But isn't this precisely what the system is intended to encourage? IMO paladins *ought* to be the most prowess-biased class, so that makes sense. I'm not sure what's really wrong here - were you doing too much damage too early?

* Potions of Restore Mana are too common. I also found a lot of potions of Healing for a bit, then suddenly they stopped showing up. Might want to look at a plot of distribution of !Healing as a percentage of all items, with respect to depth.

I think we probably need a look at consumables in general. They haven't really been touched since the fork, and were too common even in 3.3.x

* At some point ID-by-use of weapons and armor becomes unfeasible (because swapping gear around becomes dangerous, if nothing else). Currently that means you start examining the item for desirable affixes, and IDing the ones you think might be worthwhile. This breaks down when you consider weapons that can have lots of hidden affixes that boost dice, damage, and finesse, which could make them far more valuable than the weapons that have noticeable affixes but no significant dice or pluses.

My suggestion would be to make dice and pluses reveal on pseudo at some character level; at least by clvl 30. This would go with auto-ID-on-pickup at clvl 45 or whatever, which I know's been suggested before.

Well, pseudo is going, at least its current implementation. I'd be happy for dice and plusses to be revealed in the same way (after a time which depends on class, with warriors quickest then rogues/paladins then rangers/priests then magi), with the speed increasing with clev (so it's instantaneous for warriors by cl20, and for magi by cl40, say).

[Derakon]

It also ignores their resists, hitpoints, and so on. What you've done is recreated half of the monster power algorithm (the bit that works out total melee damage). You need to add power for monster spells and defences - see src/monster/mon-power.c - it hasn't been updated for a while, so it doesn't take into account evasion or absorption properly yet. In fact to be clear, it doesn't take evasion into account at all (which means it needs updating), and it probably undervalues absorption, as it's what used to be simply AC.

Okay, fair enough, but I suspect that the monster power algorithm will show giant rocs as being overpowered too.

That's a very recent change, announced by takkaria in the V forum. IIUC it's intended to offset the fact that stat gain is much quicker now you can get from 18/90 to 18/100 in a single potion.

Mm, I started this game before Takk's announcement, so I don't know when exactly it went into effect. Do we know about when we want the player to be, say, 50% through with stat gain?

Is this after your changes to ammo dice? More food for thought re archery.

After, but my changes to ammo dice were pretty conservative (I only doubled their size, basically), since I didn't really know what an appropriate value would be. Ammo doesn't really scale over the course of the game the way weapons do, after all.

Well, the stats show that this time you really are just unlucky (compared with your first v4 AAR, when you were discovering quite how overly rare artifacts were in that version). Artifacts are still rarer than in V (though takk's recent changes have probably narrowed that gap by making artifacts less common in V), but you should certainly find a few before 3000'.

Can we run stats with randarts turned on? Just to make certain they have similar drop frequencies as standarts.

But isn't this precisely what the system is intended to encourage? IMO paladins *ought* to be the most prowess-biased class, so that makes sense. I'm not sure what's really wrong here - were you doing too much damage too early?

It's more that there was no reason I would ever consider using even an unbiased weapon, let alone a finesse weapon, because my finesse score was so abysmal. I don't think it's unreasonable for a paladin to want to be able to use a longsword without feeling like he's sacrificing too much damage.

Well, pseudo is going, at least its current implementation. I'd be happy for dice and plusses to be revealed in the same way (after a time which depends on class, with warriors quickest then rogues/paladins then rangers/priests then magi), with the speed increasing with clev (so it's instantaneous for warriors by cl20, and for magi by cl40, say).

Maybe I'll take a look at this when I get some time. No guarantees though; I've not looked at the pseudo/ID code before.

[Magnate]

Okay, fair enough, but I suspect that the monster power algorithm will show giant rocs as being overpowered too.

I think the +5 speed must make a huge difference, along with whatever ev/abs it has. Before those changes, but after the elec hits, it was 173rd in absolute power and 115th in scaled power. So it was notably tough for its native depth, but not ridiculously so - its peers were master mystics, ancient golds and Dwar, Dog Lord of Waw. I think it would probably be in the top 100 now.

Mm, I started this game before Takk's announcement, so I don't know when exactly it went into effect. Do we know about when we want the player to be, say, 50% through with stat gain?

The trouble with questions like that is that they depend on diving speed - hence all the debates about setting one's own difficulty. I'd say we want stat gain to be 50% complete by around clev 40, and 100% complete after clev 50, but I think pro-divers will disagree (and I am in other ways very pro-diver).

Can we run stats with randarts turned on? Just to make certain they have similar drop frequencies as standarts.

That is possible, but nobody has the spare computing power to do it - it's an order of magnitude slower because of the randart generator. I've been postponing it until the generator rewrite, because it'll be a lot faster.

It's more that there was no reason I would ever consider using even an unbiased weapon, let alone a finesse weapon, because my finesse score was so abysmal. I don't think it's unreasonable for a paladin to want to be able to use a longsword without feeling like he's sacrificing too much damage.

I see. I guess this raises an issue with balance- and heft-changing affixes: sometimes they'll actually have a net negative impact on damage output, e.g. when they're applied to a weapon of the opposite preference.

So what you're saying is that the current values in the system are too extreme, and we ought to have slightly less of a gulf between finesse chars and prowess chars? So that sometimes a char will use an awesome weapon of the opposite preference? I wonder if this will undermine the success of the system (which I think depends on the extremes playing very differently), but it's worth thinking about.

[Derakon]

So what you're saying is that the current values in the system are too extreme, and we ought to have slightly less of a gulf between finesse chars and prowess chars? So that sometimes a char will use an awesome weapon of the opposite preference? I wonder if this will undermine the success of the system (which I think depends on the extremes playing very differently), but it's worth thinking about.

I wouldn't go that far. Certainly I have no problem with paladins never wanting to use daggers or rapiers. But longswords are unbiased -- 50% finesse, 50% prowess. Currently the only characters that I suspect would be willing to use such unbiased weapons are warriors (who have largely unbiased stats) and mages (whose class-based stats are so terrible that equipment and stat bonuses make up the bulk of their combat capability). That's wrong.

In other words, I want your average character to look at the three "classes" of weapons and think:

* This weapon is for the opposed combat modus. I don't want to use it.
* This weapon is for my combat modus. I'd like to use it.
* This weapon is unbiased. It's not as good as one for my modus but if it's good enough in other ways then I'd be happy with it.

[fizzix]

AFAIK the *only* change to Rocs is the +5 speed, which probably is the main reason for their difficulty. They were pretty burly in the past too, I've died to them before. That's not to say that they shouldn't be deepened or weakened.

That's a very recent change, announced by takkaria in the V forum. IIUC it's intended to offset the fact that stat gain is much quicker now you can get from 18/90 to 18/100 in a single potion.

Stat gain needs to either require *no* grinding or it needs to be unnecessary. Derakon has 660k game turns, assuming half of these were in stat-gain range, he should have higher stats

I proposed a while back a "minimum stat" approach where @'s effective stat (pre-equipment/RB-CB bonuses) is MAX(physical_stat, mstat(c_level)). So for example, you could imagine minimum stats to look something like:

CL0: 12
CL5: 13
CL10: 14
CL15: 15
CL20: 17
CL25: 18
CL30: 18/20
CL35: 18/40
CL40: 18/60
CL50: 18/80

I'm just making up numbers here, but the idea is that you can get character growth by either gaining levels or finding stat potions. However you can't get to your full potential with just one, you need both for that. This would also somewhat mitigate the problem of nexus stat-swap being game over.

[Derakon]

That's an interesting idea, that your minimum internal stat bonus would depend directly on your clvl. Might well be worth trying. Though you should start from 10, not 12, since that's where you start with point buy. Starting from 12 would basically give about 6 free points to the player from the start.

Oh, one other thing I meant to bring up: the Bless and Berserker spells are available to the wrong classes. Bless increases finesse and is available only to prowess fighters; Berserker increases prowess and is available only to finesse fighters. Of course, Bless also chews up three full spell slots (Bless/Chant/Prayer), so it's not like you can just swap the two.

Heroism, for reference, boosts both finesse and prowess equally.

[fizzix]

Though you should start from 10, not 12, since that's where you start with point buy. Starting from 12 would basically give about 6 free points to the player from the start.

Yes, you're right. For some reason I thought you started at 12 not 10.

[Magnate]

Yes, you're right. For some reason I thought you started at 12 not 10.

I really like that idea. We can't stop people grinding/scumming for items, whether stat potions or anything else, but mechanisms like this that make it less tempting are excellent.

[Derakon]

One possible concern with the "minimum stat by clvl" idea is that you can use stat-swap potions to drain a stat below your clvl minimum, and then when you gain a level that loss would itself be "lost", leaving you with a net gain.

Given that stat-swap potions are currently the only way to permanently drain stats (unless you count nexus, which is irrelevant for this discussion), and given that the stat-swap potions have themselves been a bit controversial, I wonder if perhaps they should be done away with altogether.

Alternately, we may simply not care about that particular abuse case. I'd feel better about that if stat-swap potions showed up at the same time as stat-gain potions.

[CunningGabe]

I maintain that the stat-swap potions should be replaced with potions that temporarily boost a stat (by say, +5), and that 5% of the time they leave a permanent boost when they wear off. Then we don't have these weird stat-swap issues, and id-by-use is easier.

[will_asher]

I maintain that the stat-swap potions should be replaced with potions that temporarily boost a stat (by say, +5), and that 5% of the time they leave a permanent boost when they wear off. Then we don't have these weird stat-swap issues, and id-by-use is easier.

I agree, except I don't see a need for the chance of a permanent boost for these.

[Scatha]

It's more that there was no reason I would ever consider using even an unbiased weapon, let alone a finesse weapon, because my finesse score was so abysmal. I don't think it's unreasonable for a paladin to want to be able to use a longsword without feeling like he's sacrificing too much damage.

I think we ran into a similar issue when balancing Sil. Our first guess for balance ran a straight line for the tradeoffs between two values as found on different items. This was a problem because character preferences generally translate to a derivative, and this would push them all the way to one end or the other.

Our solution was to replace the straight line with a curve. You might do that here by removing the "sum to 100%" rule, and letting balanced weapons get more than that (perhaps longswords 60% prowess, 60% finesse?), and perhaps extreme weapons less (90% / 5% at the ends?), while keeping 80%/20% as a 'typical' split.

Edit: I suppose you could also keep the balance numbers the same but increase the base damage for balanced weapons?

[half]

Our first guess for balance ran a straight line for the tradeoffs between two values as found on different items. This was a problem because character preferences generally translate to a derivative, and this would push them all the way to one end or the other.

Edit: I suppose you could also keep the balance numbers the same but increase the base damage for balanced weapons?

I was thinking exactly the same things:
1) this needs to not be a straight line,
2) the most obvious way is to make the finesse and prowess not always sum to 100,
3) the better way is to increase the damage for the balanced weapons.

PS Scatha is being very generous to me with the use of 'we'. Actually he worked this out and I actually took a little bit of convincing. It is definitely the right approach though.

[Derakon]

Good advice, thanks! The "balanced weapons get better dice" approach seems like it'd be easier for the users to deal with. I guess the question is, how much better...