Power spiral

[duvessa]

"Power spiral" is the tendency for strong characters to improve at a faster rate than weaker characters. This results in an increasingly large gap between characters' power levels. This is bad because it makes balancing the game increasingly difficult. It's hard to come up with something that challenges post-Lair characters when the power gaps between post-Lair characters are so wide. This issue has been brought up before in relation to portal vaults, but I want to go over some other sources of it too, and how to fix or at least mitigate them.

Things in DCSS that do NOT cause power spiral
First I want to go over things that are often accused of creating power spiral, but actually don't create permanent differences in character power. They can create temporary differences, but said differences disappear completely after a short time.
- Vaults with OOD monsters, or just hard-but-avoidable monsters in general. A strong character can beat them without consumables and become stronger from the XP... but a weak character can simply do other levels until they're strong enough to beat the hard monster without consumables, and then backtrack and kill the monster and get the same XP.
- Piety decay and piety costs with non-gifting gods. This is because piety is capped. Weaker characters reach 200 piety slower than strong characters but when they do reach 200 piety they completely catch up.
- Variance in item/monster spawns. Some games will just find better items on the ground than others, or get more monster XP than others. This is fine, and in fact good, and it has absolutely nothing to do with power spiral, because those item/monster spawns are not dependent on how good the character is.

Things in DCSS that DO cause power spiral
- Consumables. Stronger characters need to use consumables less often, so they accrue increasingly large piles of consumables (making them even stronger) compared to the weaker characters that need to use consumables more often.
- Piety decay, piety costs, and XL-based piety with equipment gifting gods.
- Gozag. Unlike piety with all other gods, you can accrue unlimited gold, and thus increase your supply of god abilities without bound - it's the consumable issue all over again. Call Merchant also has the equipment gifting problem.
- Portal vaults. Because you cannot backtrack to portal vaults, a character strong enough to clear one gets a permanent advantage over one that does not.

Fixing power spiral with consumables
Power spiral is inherent to the concept of consumables, and the only way to fix it is to add some fairly ridiculous rules. You can solve it by capping consumables, which means introducing rules like "no more than 2 potions of heal wounds may exist at any given time, if there are already 2 in existence then newly generated ones are destroyed". That's about it.
So, not going to happen, probably shouldn't happen barring some total overhaul of consumables.

Fixing power spiral with equipment gifting gods
This can be done without shaking up balance much, but not in a very palatable way: give gifts at all piety levels, make the chance of receiving a gift constant regardless of your current piety (currently you get slightly more gifts the higher your piety is), and make the piety gain for these gods independent of XL or use a separate formula for gift timeout that doesn't use XL.
This removes the effect of piety decay, piety costs, and XL-based piety on gifts. Same kills/exploration, same gifts. You can have the gift timeout start at 100 or whatever so that you still can't get a gift immediately after starting worship.
This whole thing is not pretty though.

The other common proposal for equipment gifts is to make them temporary. This has been Discussed To Death(tm) and I don't have anything new to say about it.

Fixing Gozag
Basically, you need to get rid of Call Merchant and cap the player's gold. Capping gold has a lot of obvious issues, to the point that it's not worth it, so I give up on Gozag.

Fixing power spiral with portal vaults
You can't, not without defeating the entire point of portal vaults by allowing backtracking.
But if it can't be fixed, it can at least be mitigated.
The first thing to change is the timers. Obviously, it's easier for stronger characters to beat the timer. Previously, most portal vaults were unannounced and had a "long timer" that took thousands of turns, and only started their "short timer" when they came into view. The idea was to let players autoexplore normally, while still not letting them go get a bunch of xp/items from other branches before coming back to do the portal.
This was thrown out for very good reasons that you can probably guess, in favour of announcing all portal vaults instead...but with timers way too short to reach them with autoexplore.
Manual exploration is, in fact, not interesting - that's why we use autoexplore - and using the messages to trilaterate the portal's location is a chore. So I support revealing the portal's entrance vault upon announcement, as though it were magic mapped. And then lengthen the timers and/or actually display how much time is left.

Now that you have all characters reaching the portal, there's still the matter of the portal destination vault. A character that clears the destination without using consumables has a permanent advantage over one that doesn't, in both items and experience points (and possibly god gifts).
To prevent power spiral here, you'd have to have a destination that can be completed without using consumables with any reasonably built character. Some sewer and ossuary destinations are already at this point. Wizlab and Desolation destinations are also at this point unless you entered Elf early for some reason.
But that kind of sucks! If a reasonably built MuMo can clear the vault without consumables then that means the vault is trivial, and there was no point in having the portal in the first place. So what you really want to do is strike a balance where reasonable characters can get most of the vault reward but there's still some challenge, and the power spiral is an acceptable loss.
If you don't think it's an acceptable loss then portal vaults are pointless so you just remove them. So instead let's say you do consider it an acceptable loss. Here is where I would put the "dangerous" portal vaults right now:

Bailey: These use a lot of orc knights, which many characters can't kill, but on the other hand this is the one place in the game where orc knights have a chance to be interesting, so maybe it's worth the power spiral. I will claim that bailey_axe_minmay_hex_keep and bailey_polearm_3 go way too far, however.
Hex keep has a warlord which is just stupid, it's a less killable version of an orc knight. It also has ridiculous loot, which exacerbates its power spiral issue further, though if you are not a naga you can at least jack the loot easily and "just" lose the warlord's XP and inventory.
bailey_polearm_3 was already the most dangerous bailey in its original version, and then the deep elf monsters got buffed twice, and now it's just impossible without like LRD or something.
I'll give a special mention to bailey_polearm_5, because if you do it as intended you fight a bunch of berserk resistant agile monsters at once, but if you use a launcher to incrementally wake up each wave without causing shouts, it's pretty easy.
Hex keep and bailey_polearm_5 also have ridiculous loot, which exacerbates their power spiral issue further.
Of course this all goes out the window if the bailey generates after you've already completed Lair because for some reason they can do that: their range is D:7-14 and Orc:1. Why can they do that?

Desolation: This is easily clearable by any conceivable character at its depth, no power spiral issue here. Just the 20 gigawatts of electric power wasted every time you press 5 in Desolation.

Gauntlet: Easily the best portal vault, which is extra nice since it replaced what was easily the worst feature in the entire game. I thought that I was going to complain that the monster list goes too far and creates power spiral as a result...but it turns out that I just got unlucky the one time I found a gauntlet, and death scarab is actually the only crazy monster on the list.

Ice cave: icecave.des has a comment all but stating that power spiral is actually desired here, which is pretty funny because these days, characters are definitely strong enough to beat ice dragons/frost giants at ice cave depth. I think the ice fiends and blizzard demons are restricted to the "hard" versions of ice caves; if I'm right about that, then even those are usually beatable. It's hard to tell because reading ice cave vaults is really hard.
Aside: it's funny how in ice caves a blizzard demon is like 500x more dangerous than an ice fiend.

Ossuary: Not causing significant power spiral as-is, some destinations are just trivial though (slow monsters that you can kite forever). I will pick on the ones with wraiths because a character that can kill a wraith gets a huge reward over one that can't...but almost no characters can kill a wraith at ossuary depth anyway.

Sewer: Good as-is except for the ones that try to get you to fight a crocodile. Unlike a wraith there are a lot of characters that can kill a crocodile at sewer depth (curare, meph, summons, conjure flame...), and a lot that can't, and the former get a huge reward for it. The crocodile gets sped up by the water so you can't even kite it around while picking up the loot like you can with an ossuary wraith.

Volcano: Easily the biggest offender. Seeing a stone giant in a volcano was what spurred me to make this complaint in the first place. The old volcano monster set was just weird, the new volcano monster set is just extreme overkill. It's like a monster set you would choose for a regular non-portal vault that you can come back and do later. Expecting mid-Lair characters to consistently be able to fight stone giants, hell knights, red draconians, hell hogs, and molten gargoyles is crazy.
And then there are volcano_caves and volcano_overflow, which destroy the loot if you don't get it fast enough, while also having the aforementioned nutty monster set added to them. volcano_overflow even has the entrance on the opposite side from the exit so if your character can't fight those monsters you can't even enter a volcano portal safely.

Wizlab: Not causing significant power spiral, although they're encountered at a point where power has already spiralled out of control.

Ziggurat: This is in a weird position where it would cause a ton of power spiral if not for the fact that you should never actually enter one if you're trying to win, since it can generate the biggest monsters in the game even on level 1.



Thoughts? Are there sources of power spiral that I missed?

[sanka]

Fixing power spiral with portal vaults
You can't, not without defeating the entire point of portal vaults by allowing backtracking.

Well, maybe if you do not enter the portal vault in time the gate may close for let say, 5 levels (character levels) or something instead of forever. So between now-or-never (which has the power spiral issue) or backtrack whenever want (which defeats the point, true).

[Nekoatl]

I don't think portal vaults even need timers, really; there's still the fact that they close when you leave in order to pressure players into deciding how much of it they're willing to risk trying to clear. Also, the strategic decision about when a character should risk entering one is more interesting than simply deciding if they should risk entering one the moment they step into (or get shafted into) a level containing one. Delaying entry into the portal vault means a higher chance of a successful clear, but also means a delay in when the benefits can be leveraged against the rest of the game's challenges. Regardless, I do think some of those portal vaults are excessively dangerous for how early they can generate, so thanks for calling them out in such detail.

I don't see any point in worrying about god gifts. With the exception of ammunition (which hopefully shouldn't matter if you're spending your god choice just to make sure you don't run out) and Sif's books (which eventually cap out once all spells are learned), god gifts compete with each other for equipment slots. Once you get a good weapon, it doesn't matter how many other weapons are generated that aren't as good. Sure, the more gifts that are generated, the better your chances of getting a good weapon, but it's still a matter of chance with a soft cap on long-term power. If it's still a concern, eliminate piety decay and generate alters more consistently in the early levels of the dungeon so there's less disparity in how many gifts are awarded at any given point of progress. There's not even a point to keeping piety decay in the game (excepting Uskayaw). It made sense back when monsters would keep getting generated if you just sat around on a dungeon level waiting for them, but now that they're all generated right at the start, decay just punishes characters that need more downtime between fights (and players that prefer to autoexplore) for the sake of solving a problem that no longer exists.

Speaking of chance, when massively powerful artefacts generated early in the dungeon can trivialize the majority of the game, I don't see a whole lot of point in worrying about power spiral as a game balance issue if we're not willing to address that one. I agree that randomness in Crawl is important, but sometimes it goes too far.

[duvessa]

Removing timers from portal vaults is another thing that's been Discussed To Death and won't happen. I'll ask that this thread doesn't turn into yet another rehash of it.

[TheMeInTeam]

Given the disparity between starting species/backgrounds we're stuck with some spiral unless you want to make all timed portals happen later in the game. The main reason Wizlabs and Desolation of Salt are consistently doable is that they're late enough in the game where you've had enough chances at resources/xp to do them. Stick a Wizlab in Orc 1 and it's a similar problem to Volcanoes.

XP loss from portal vaults isn't a big issue. Even stuff like Dg and Mu wind up at XL 27 in 3 rune runs if they do Elf + something other than abyss for 3rd rune, and catch up quickly as harder monsters show up and give more XP. The main difference is loot, but I'm not convinced this is significant enough to worry about it. Tactics > loot for most of the game, and you have significant pure RNG loot drop effects plus the RNG of whether the portal vaults even generate to wash out the loot advantage/disadvantage to a degree...plus multiple species are so equipment restricted that even consistently completing hard vaults confers less of a reward (felids, octopodes, even ogres/tengu/trolls to a degree). These are actually designed to be at a disadvantage WRT power spiral from dungeon resources, yet some of them are high tier species anyway.

[Shard1697]

Huge agree on portal entrance vaults upon announcement. I think this will shrink the gap between weak and strong characters in finding the portals hugely, and I'm very over trying to find these based on the messages.

[Rast]

"Power spiral" is the tendency for strong characters to improve at a faster rate than weaker characters. This results in an increasingly large gap between characters' power levels. This is bad because it makes balancing the game increasingly difficult.

I was with you until this part. It's a single player game. Balance shouldn't be a goal. It is OK for an average or weak player to sometimes benefit from power spiral or other forms of luck.

[chequers]

I think duvessa's point is that the systems in crawl mostly reward strong characters and not weak ones. It's quite easy for a MiBe to finish almost any bailey but nearly impossible for many other characters. Further it might be ok that some species are better than other, but I think there are persistent imbalances between types of character that should be addressed.

I do think it's worth distinguishing between "power spiral" and "poor play punishment" though. It's ok for poor (or overly conservative) play to cause the player to fall behind. For example, duvessa said "A character that clears the destination without using consumables has a permanent advantage over one that doesn't, in both items and experience points (and possibly god gifts).". I think this situation is fine, you used a consumable to reduce the risk to yourself in the short term, the tradeoff is you can't use the consumable later. What should be addressed is situations where perfect play is not enough, or where the (medium-term) margin for error differs wildly between characters.

[VeryAngryFelid]

I think we should clarify one very important thing when talking about different characters. As all of you know there are no explicit difficulty levels in crawl so players choose different characters for that. For instance, KoBe has roughly the same playstyle as MiBe except MiBe is much easier due to higher HP and better aptitudes in relevant skills. As such I believe it is acceptable and probably even desirable that power gap between them increases while game progresses. What is not acceptible is when power gap is increased between characters with different playstyles. For instance, if some baileys can be cleared only by pure melee characters and should be avoided by most book backgrounds, then we have a clear problem. I mean there is gap caused by species and gap caused by starting background/playstyle. Increasing first gap is fine, increasing second one is not.

[rigrig]

you used a consumable to reduce the risk to yourself in the short term, the tradeoff is you can't use the consumable later.

The point is that the same situation is less risky for stronger characters, so "optimal" play for a weak character would be to use a consumable, while a strong character could conserve it.
So future situations become comparatively even less risky for the strong character, because they have more consumables to fall back on should things go pear-shaped.

[Siegurt]

I think we should clarify one very important thing when talking about different characters. As all of you know there are no explicit difficulty levels in crawl so players choose different characters for that. For instance, KoBe has roughly the same playstyle as MiBe except MiBe is much easier due to higher HP and better aptitudes in relevant skills. As such I believe it is acceptable and probably even desirable that power gap between them increases while game progresses. What is not acceptible is when power gap is increased between characters with different playstyles. For instance, if some baileys can be cleared only by pure melee characters and should be avoided by most book backgrounds, then we have a clear problem. I mean there is gap caused by species and gap caused by starting background/playstyle. Increasing first gap is fine, increasing second one is not.

Well, I "playstyle" within limits, if my "playstyle" is "rush headlong at every creature I see and attack it with an untrained club, attacking it until one of us dies" maybe having that get harder as the game goes on compared to less stupid playstyles isn't a bad thing.

Also some baileys are hard for a player with little to no range attack and/or reaching to clear, but easy for a player with such attacks, if baileys *on average* (And I'm not saying they are, but *if* they are) are balanced across play styles, so that if you encounter an unknown bailey you have pretty even odds of clearing it no matter what your play style is (and that may change radically once the type of bailey is known, depending on your play style) is that a negative sort of power spiral?

Is it a problem if a given run with a given character type will accelerate power-spiral wise, but another run will not do so as strongly, or only if you can "game" the system by selecting a certain type that's going to consistently out-power another one long term?

I *think* duvesssa's expressed concern is when the lower end of the power curve is very very far from the high end of the power curve, when a series of small advantages gets turned into an aggregate larger advantage later, not because it matters when comparing characters to each-other, but rather because it makes creating game content later on difficult to balance, because the distance between the low and high ends gets much harder (You don't want it to be trivial for high end, nor impossible for low end characters)

Where we're talking about "where you could optimally or near-optimally get to for a given start and RNG happenings", rather than supporting tactically unsound decisions (if you make a bad choice and are punished for it by having to use a consumable, and are therefore a bit behind in the power curve long term and perhaps paint yourself in a corner ultimately because of it, I think that's fine) I personally don't think the distance between a well-played low-power character and a well-played high-power character is far enough right now that the balance of the later game is a serious problem, but it is not *far* from being a problem, narrowing the gap would certainly not hurt, and making the gap any wider would certainly make things worse than they are.

[TheMeInTeam]

You can't get away from "weak characters need more consumables" in practice though. That's a substantial part of what makes them weak options in the first place. There are significant power differentials in species and backgrounds set before the player makes a single move. Part of the idea is that these continue to vary the experience throughout the game.

I doubt whether making strong vs weak initial choices converge into similar degrees of difficulty/approaches by mid-late game just to make mid-late game more "balanced" is a desirable goal. I'd rather let the player set how much of a power spiral they are likely to receive or lack in advance, by picking a strong or weak character (also influenced by god choice). This is especially true if we're already in a position where nearly every species/background is winnable, which is something Duvessa has asserted (and seems to be accurate).

Given that character choice controls difficulty to a degree and how much power spiral is even possible, making post-lair actually balanced could be argued as a bad thing for the game rather than desirable because it would reduce variety between species/backgrounds.

but rather because it makes creating game content later on difficult to balance, because the distance between the low and high ends gets much harder (You don't want it to be trivial for high end, nor impossible for low end characters)

The xp curve and up-scaling of enemy gear make this improbable. Early game lucker dogs are going to get increasingly less benefit from said XP/gear relative to someone who lacks those things, as any future gear becomes less likely to be useful compared to what is already equipped, and the XP gap is continually shortened until everyone is XL 27 and has the option to grind in the abyss or do extended if they actually care for more XP.

[VeryAngryFelid]

Well, I "playstyle" within limits, if my "playstyle" is "rush headlong at every creature I see and attack it with an untrained club, attacking it until one of us dies" maybe having that get harder as the game goes on compared to less stupid playstyles isn't a bad thing.

I am not sure how it is related to what I wrote but I disagree with your claim. When playing other games, easy difficulty still remains easy and it is normal, player feels free to restart the game at different difficulty if they learned enough and feel boring. While DCSS for some weird reason has TrBe who presses o, tab and 5 until first hydra and then suddenly struggles in Zot and especially extended.

Also some baileys are hard for a player with little to no range attack and/or reaching to clear, but easy for a player with such attacks, if baileys *on average* (And I'm not saying they are, but *if* they are) are balanced across play styles, so that if you encounter an unknown bailey you have pretty even odds of clearing it no matter what your play style is (and that may change radically once the type of bailey is known, depending on your play style) is that a negative sort of power spiral?

This is bad game design. Players may spend consumables to reach timed portal before it expires and then suddenly realize that they should immediately leave. There should not be "melee only Baileys" or "ranged only Baileys". We don't have "melee only Zot" or "ranged only Zot", do we?

Is it a problem if a given run with a given character type will accelerate power-spiral wise, but another run will not do so as strongly, or only if you can "game" the system by selecting a certain type that's going to consistently out-power another one long term?

There is important difference between getting good loot and getting useless (or even harmful) timed portal. The former does not lead to disappointment. Mimics were removed for a reason.

I *think* duvesssa's expressed concern is when the lower end of the power curve is very very far from the high end of the power curve, when a series of small advantages gets turned into an aggregate larger advantage later, not because it matters when comparing characters to each-other, but rather because it makes creating game content later on difficult to balance, because the distance between the low and high ends gets much harder (You don't want it to be trivial for high end, nor impossible for low end characters)

Let me disagree again. I don't see why it is ok to have trivial and almost impossible early game but is not ok to have them late game. Player chooses difficulty for whole game, not just for first few floors.

Where we're talking about "where you could optimally or near-optimally get to for a given start and RNG happenings", rather than supporting tactically unsound decisions (if you make a bad choice and are punished for it by having to use a consumable, and are therefore a bit behind in the power curve long term and perhaps paint yourself in a corner ultimately because of it, I think that's fine) I personally don't think the distance between a well-played low-power character and a well-played high-power character is far enough right now that the balance of the later game is a serious problem, but it is not *far* from being a problem, narrowing the gap would certainly not hurt, and making the gap any wider would certainly make things worse than they are.

I am not sure why people keep talking about late game balance. Early game is balanced indeed but late game throws at you monsters which kill you during explore from full HP (ancient lich, orb of fire, draconians with blink close and dimension anchor, cursed skull with those mushrooms to name a few) and extended is even less balanced as it has spells which don't care about anything.
So to summarize, making easy early game harder late game and making hard early game easier late game is NOT balancing, it is violation of players preferences to have easy/hard game which they selected during game start.

[TheMeInTeam]

Even with all that late game throws at the player, roguelikes do provide resources and convergence over time regardless. I would argue that *right now* late game is already easier than early game for most species on the weak side of the power curve. The resources might not be there just yet in d1-d4, but by Zot you have a wide range of available spells, tons of gear drops all over the place, options ranging from throwing nets to silence vs liches, full piety on a god of your choice, and enough XP to specialize into what you need to win if you are familiar.

In that sense, late game is already more "balanced" than early game, in that you get at least some guaranteed convergence, to a lesser extent for gear-limited but still there.

I'd be okay with altering timed portals so we don't have to move around and read text for approximate distances though. While doable that gets old fast.

[VeryAngryFelid]

Even with all that late game throws at the player, roguelikes do provide resources and convergence over time regardless. I would argue that *right now* late game is already easier than early game for most species on the weak side of the power curve. The resources might not be there just yet in d1-d4, but by Zot you have a wide range of available spells, tons of gear drops all over the place, options ranging from throwing nets to silence vs liches, full piety on a god of your choice, and enough XP to specialize into what you need to win if you are familiar.

In that sense, late game is already more "balanced" than early game, in that you get at least some guaranteed convergence, to a lesser extent for gear-limited but still there.

I'd be okay with altering timed portals so we don't have to move around and read text for approximate distances though. While doable that gets old fast.

Is my English that bad? I was saying that balance is not when late game becomes easy for hard combos or when it becomes hard for easy combos and you replied with "but late game becomes easier for hard combos". Late game balance is when if early game is hard, late game remains hard too and if early game is easy, late game remains easy too.

Edit. I have impression that some players believe that balance is when game becomes equally hard for all characters. That might be a good idea only if DCSS had explicit difficulty levels, otherwise good players will quit of boredom and bad players will quit of frustration.

[Siegurt]

Even with all that late game throws at the player, roguelikes do provide resources and convergence over time regardless. I would argue that *right now* late game is already easier than early game for most species on the weak side of the power curve. The resources might not be there just yet in d1-d4, but by Zot you have a wide range of available spells, tons of gear drops all over the place, options ranging from throwing nets to silence vs liches, full piety on a god of your choice, and enough XP to specialize into what you need to win if you are familiar.

In that sense, late game is already more "balanced" than early game, in that you get at least some guaranteed convergence, to a lesser extent for gear-limited but still there.

I'd be okay with altering timed portals so we don't have to move around and read text for approximate distances though. While doable that gets old fast.

Is my English that bad? I was saying that balance is not when late game becomes easy for hard combos or when it becomes hard for easy combos and you replied with "but late game becomes easier for hard combos". Late game balance is when if early game is hard, late game remains hard too and if early game is easy, late game remains easy too.

Edit. I have impression that some players believe that balance is when game becomes equally hard for all characters. That might be a good idea only if DCSS had explicit difficulty levels, otherwise good players will quit of boredom and bad players will quit of frustration.

Well, it just sounds like you're talking about a different problem. What you're talking about is a "balance" which is not the problem mentioned in the OP, the original post specifically deals with "power spiral" problems, where a weaker character is further disadvantaged or a stronger character is further advantaged as you get further into the game.

To phrase it in difficulty terms, power spiral is when selecting 'easy' makes the early game somewhat easier than 'hard' but the difference between 'easy' and 'hard' gets more and more pronounced as the game progresses, because the challenges remain of constant difficulty, but the ability of 'easy' mode characters to collect incremental power increases is much larger than the ability of someone who selects 'hard' (because the 'hard' mode selection forces you to abandon fights or possible loot and/or use consumables that the 'easy' mode character would not.) The complaint is not that it's unfair (or unbalanced), but rather that as the game progresses it becomes increasingly difficult to create things that are challenging enough for 'easy' mode characters, but not so challenging to become impossible for 'hard' mode (Since DCSS doesn't have explicit difficulty modes, the challenges presented are the same no matter how far along or what starting selection you made)

There is important difference between getting good loot and getting useless (or even harmful) timed portal. The former does not lead to disappointment. Mimics were removed for a reason.

Mimcs still exist (maybe you meant the 'monsters' were removed?), but timed portals (and altars and loot) can still be mimics, which can certainly lead me to disappointment.

[TheMeInTeam]

Late game balance is when if early game is hard, late game remains hard too and if early game is easy, late game remains easy too.

Okay, I understand now. I don't think you can realistically avoid said convergence in a roguelike such as crawl though. The natural tendency will be for loot/XP to normalize and make hard starts easier rather than leaving them hard...while also making some of the easy start advantages less meaningful. It doesn't completely go away...late game Felid or Kobold is still harder than Minotaur or Centaur in a 3 rune game. But they're closer.

What's our basis for wanting to alter this significantly? I don't see that crawl is in particularly bad place in this regard, enough to merit moving it in one direction or the other.

[ion_frigate]

Okay, I understand now. I don't think you can realistically avoid said convergence in a roguelike such as crawl though. The natural tendency will be for loot/XP to normalize and make hard starts easier rather than leaving them hard...while also making some of the easy start advantages less meaningful. It doesn't completely go away...late game Felid or Kobold is still harder than Minotaur or Centaur in a 3 rune game. But they're closer.

Oh it can be avoided, it just requires something that's considered absolutely against DCSS "philosophy". Ever played IVAN? That's the game that starts throwing goblins with +7 short swords at you if your gear starts getting too snazzy, or if you wear a permanent invisibility item. At the extreme end, it can generate these horrible invisible uniques (Sherarax and ZQ-29) who have sufficient Dex to rapidly dismember and behead you. This mechanic definitely clips power spirals and keeps the game difficult throughout, but character-dependent monster generation is considered verboten in DCSS.

[VeryAngryFelid]

Even with all that late game throws at the player, roguelikes do provide resources and convergence over time regardless. I would argue that *right now* late game is already easier than early game for most species on the weak side of the power curve. The resources might not be there just yet in d1-d4, but by Zot you have a wide range of available spells, tons of gear drops all over the place, options ranging from throwing nets to silence vs liches, full piety on a god of your choice, and enough XP to specialize into what you need to win if you are familiar.

In that sense, late game is already more "balanced" than early game, in that you get at least some guaranteed convergence, to a lesser extent for gear-limited but still there.

I'd be okay with altering timed portals so we don't have to move around and read text for approximate distances though. While doable that gets old fast.

Is my English that bad? I was saying that balance is not when late game becomes easy for hard combos or when it becomes hard for easy combos and you replied with "but late game becomes easier for hard combos". Late game balance is when if early game is hard, late game remains hard too and if early game is easy, late game remains easy too.

Edit. I have impression that some players believe that balance is when game becomes equally hard for all characters. That might be a good idea only if DCSS had explicit difficulty levels, otherwise good players will quit of boredom and bad players will quit of frustration.

Well, it just sounds like you're talking about a different problem. What you're talking about is a "balance" which is not the problem mentioned in the OP, the original post specifically deals with "power spiral" problems, where a weaker character is further disadvantaged or a stronger character is further advantaged as you get further into the game.

To phrase it in difficulty terms, power spiral is when selecting 'easy' makes the early game somewhat easier than 'hard' but the difference between 'easy' and 'hard' gets more and more pronounced as the game progresses, because the challenges remain of constant difficulty, but the ability of 'easy' mode characters to collect incremental power increases is much larger than the ability of someone who selects 'hard' (because the 'hard' mode selection forces you to abandon fights or possible loot and/or use consumables that the 'easy' mode character would not.) The complaint is not that it's unfair (or unbalanced), but rather that as the game progresses it becomes increasingly difficult to create things that are challenging enough for 'easy' mode characters, but not so challenging to become impossible for 'hard' mode (Since DCSS doesn't have explicit difficulty modes, the challenges presented are the same no matter how far along or what starting selection you made)

There is important difference between getting good loot and getting useless (or even harmful) timed portal. The former does not lead to disappointment. Mimics were removed for a reason.

Mimcs still exist (maybe you meant the 'monsters' were removed?), but timed portals (and altars and loot) can still be mimics, which can certainly lead me to disappointment.

It is related. The power spiral can partially fix currently broken situation so it is a good thing to have in dcss.

[VeryAngryFelid]

Okay, I understand now. I don't think you can realistically avoid said convergence in a roguelike such as crawl though. The natural tendency will be for loot/XP to normalize and make hard starts easier rather than leaving them hard...while also making some of the easy start advantages less meaningful. It doesn't completely go away...late game Felid or Kobold is still harder than Minotaur or Centaur in a 3 rune game. But they're closer.

I am not so sure. I had several Felid games where hard early game was followed by hard middle game and I didn't survive to see late game. I was spending consumables all game and eventually ran out of them. This is good balance and should not be "fixed" by making game easier for hard combos (or harder for easy combos).

[TheMeInTeam]

When I'm talking "tendencies", I mean expectations in games *on average*, not that a particular game can't be stingy and get you killed before you even see late game if you're sub-perfect.

The more levels, consumables, spell books, and gear you drop, the greater the likelihood of late-game convergence...a sort of in-built anti-spiral as each time you add something strong to your character's setup you are less likely to proc something stronger than it later.

You don't usually get *total* convergence, things like -40% HP, faster speed, increased LoS, and inability to equip gear still matter. But XL 27 characters are, on average, going to be closer to each other than XL 1-10 characters by a significant margin. That's the nature of getting stronger on an XP curve + procedurally generated loot.

[bel]

There's one idea I have for consumables which might reduce the associated power spiral.

Currently, consumables are generated based on exploration. I have discussed elsewhere why exploration-based piety is bad, and similar comments apply here.

What if we generate consumables only on monster kills? The rate would be dependent on something like (monster difficulty) / (player power). There could be special tweaks based on where the monster generated; for instance, Ossuary, Vaults:$ and Elf:$ could have bonuses.

To see how this would work, suppose a player A enters a portal vault, and thus gets more XP and more loot as compared to player B who doesn't enter a portal vault (the portal didn't generate, or they were too slow to get to it).

At comparable points in time, Player A's "player power" would be higher than player B's "player power". So A will get fewer consumables in the future (similar to how you get lower returns on XP if you have more XP to begin with). This will help player B to catch up to player A.

Of course, this proposal will break objstat. But we could balance item drops based on player morgues.

[cliffracer]

This seems like the opposite, as strong characters are much better at killing everything in a level while exploration is something of an equalizer. A minotaur will never need to ninja a rune, but the fact that you don't need to kill everything on the level is always on the table for felids. Stealth almost always correlates to being small and squishy.

Apart from the very specific example of portal levels, I don't think consumables are "power spiral". Unlike permanent upgrades like experience or equipment, having a larger stack of haste potions doesn't reduce your consumption or allow you to find more. Overall I would say Crawl doesn't have an issue with power spiral due to soft and hard caps on player power--XP requirements increase exponentially and cap at 27 so having more at any point in the game is meaningless in the long term, getting stronger equipment means later equipment is less likely to be stronger. This is only broken in the case of manuals, and they're intentionally rare enough to not sway the entire game and don't circumvent the level cap or entirely remove XP scaling.

I'd agree that it'd be a good idea to have piety that reduces the gift timeout or results in a gift still be gained. I don't think piety loss to inactivity is a huge killer, but I do think players hoarding piety instead of knowing to use abilities frequently is a common newbie trab. It also wouldn't be impossible to move Gozag to a similar system. They'd just need to open new shops (or add to a GozagMart catalog) based on total gold collected/spent (Gozag Platinum Rewards Club!) instead of a dedicated pay-to-open ability.

[petercordia]

I know this is off topic, but your suggestion for Gozag sounds really interesting. I don't know what I'd do if I had to bribe a branch or buy something expensive to make Gozag open a new shop, but I'd love to give it a try (if someone's willing to implement it.)