Page 1 of 2

DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 00:16
by duvessa
It's common in indie game projects (and some commercial ones) for the gameplay to evolve as development goes on, and most commonly this takes the form of giving the player(s) more options. This necessarily gives them more power, unless the options are literally useless. To combat this in single player games, the designer often doesn't want to remove player options, so instead they make the opponent(s) stronger - platforming levels with longer jumps, monsters in an RPG with more health and damage, you get the idea.
In practice, this is not sustainable. In fact, I have a great example of it failing spectacularly in another roguelike. Tales of Maj'Eyal (which the crawl.develz.org sidebar links to) has been steadily increasing player power for the last few years, and soon people started complaining that the hardest difficulty level (Insane) was too easy. In response, the developer didn't nerf anything - instead, they added another, higher difficulty level (Madness) that was designed to be unwinnable: monsters deal enough damage to kill players from full health in one hit, monsters can sense the player from far away, and some monsters will heal to maximum health every turn. Well, it turns out that difficulty level is completely winnable, because one of the game's classes had become so powerful that it was functionally invincible if played a certain way. Scaling enemies to match player options was a complete failure.

I mention this because it appears to be what Crawl is trying to do right now. Player damage has been increasing version after version, and after 0.8, so have player defenses. Portal vault frequency increases, giving extra items. Rods, gods, and misc items get made into better options, without any compensating nerfs to consumables or wands. AI is made more exploitable without any other changes to player options. Constriction gets added and tacked onto monsters and players without any compensating nerfs to those monsters or players. Meanwhile the game keeps introducing new monsters that are stronger than anything previously in their branch (caustic shrike, iron giant, shock serpent, spriggans in swamp, fire crabs in lair) trying to combat increasingly strong player characters. In spite of that, players and developers continue to complain that the game is getting easier. Whether right or wrong, the fact that the claim can even seem plausible is a sign that something is wrong with this method of balancing.

There have been some very good nerfs to player options, like divinations removal, summoning nerf, and square los. But the long-term trend is, if you have two player characters from different post-0.5 versions fight a yak, the one from the newer version probably does more damage, has more max hp, takes less damage, and has more items and god abilities that they could use to escape or do extra damage. You could give the yak more hp, higher damage, and spells to shut off some of those escape options - but then you'll just have to do it again later when the player gets even more powerful, and eventually you end up with a ToME situation where the player has so much control that they just don't die, and you can't fix it by scaling monsters.

It's much better to decide on an equilibrium point for player power, and keep it there - don't move it up or down unless you have a good reason to deliberately change that power level (like the game being too complex to learn, or too simple to have any depth). When you add options to players, take away the same amount. When you change the difficulty or the complexity of the game, it should be a conscious decision, not a side effect.

I do not expect to persuade anyone to change the design direction of DCSS, but I hope this can at least serve as a sort of warning.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 01:59
by tabstorm
I don't think player characters are that much better than in 0.12, when I started playing (started online in .13). Here are the things I can think of that have made the game easier:

1.No item destruction. You could already avoid most of it if you were willing to engage in extremely annoying behavior but most people did not (This is a general trend in Crawl where most threats can be largely dissolved but doing it is so boring that you'd rather not)
edit: forgot - 1a. No item weight, so spriggans, deep elves, etc. can carry more consumables. Most low-str races outside of Sp/Fe aren't really world-beaters.
2. Removal of -cTele makes some end vaults much easier.
3. Some armor penalties (like EV growth and UC penalty in heavy armor) were reduced or smoothed. GDA was slightly buffed.
4. Total experience possibly increased with the addition of depths? Depths has harder monsters now than old D, though. But if you reach XL20, how can you die in a normal game? I can count about 5 deaths out of ~1500 games in which I died on a character above XL20 that wasn't in Hell at 14k turns or something like that.
5. rCorr on potions of resistance and temporary corrosion in general (Acid was much rarer outside of slime, you could always just use backup gear at a high XL to kill TRJ. I mean people kill him in melee at XL16 all the time nowadays by just using decent tactics)
6. Changes to enchant scrolls, e.g. no enchant failure.
7. Elemental evokers added
8. Statue form was buffed a little.
9. New summon spells, but I think the nerfs to summoning were probably a bigger negative to this school than the new spells were a positive.
10. Some uniques' depth shifted down (no more Mara in Swamp etc.)

I don't think any of these make the game a great deal easier except no item destruction. Meanwhile there are lots of harder monsters in earlier areas and some backgrounds have been nerfed like En, Ne, and many of the more powerful gods have had their piety gain slowed, and there have been many small difficulty increases.

Maybe this thread should've been made back in like 0.8, I don't really know what changes have been made from 0.8 to 0.12 that decreased difficulty. I honestly think the game has been this easy for awhile, players probably just didn't exploit the more degenerate mechanics in crawl like huge floors and upstairs letting you reset almost any potentially unsafe fight. I mean, summons were uncapped in 0.10, you could basically float on a cloud of imps all the way to Vaults. Players are just getting better.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 02:03
by duvessa
I don't think Crawl has become easier either, but that is because of monster/dungeon power increasing to match the expanding number of player options (and I do think this number has expanded). I am arguing against the sustainability of that. I am definitely not arguing that the game is getting easier, since, well, I don't believe that.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 02:06
by tabstorm
I don't think player options have really increased that much... What do you really have nowadays that wasn't in early versions other than elemental evokers and more consumables due to no itemdest?

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 02:14
by bcadren
Yea I certainly think the current state of trunk is harder than slightly older versions. MOSTLY because lair branches were made harder and easy XP branches were truncated. IE: Less XP before the game gets harder means though there may be more options available, fewer ones will actually be online.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 02:31
by chequers
Why do you think average winrate hasn't changed much if the game is getting easier? Or you think the change that has occurred is significant and/or will accelerate?

EDIT: here's a graph showing winrate per released version: https://shalott.org/graphs/e7b389a9c1dd ... abeea.html

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 02:59
by Hurkyl
chequers wrote:Why do you think average winrate hasn't changed much if the game is getting easier? Or you think the change that has occurred is significant and/or will accelerate?

Duvessa already pointed out how power creep gets compensated for, and why that leads to Bad Things.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 04:46
by archaeo
duvessa wrote:I do not expect to persuade anyone to change the design direction of DCSS, but I hope this can at least serve as a sort of warning.

Let's pretend you could expect that. What kind of concrete changes would you make to combat power creep? Is it really just as simple as putting your foot down at some point and requiring that any addition or change requires an equal and opposite removal?

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 05:49
by duvessa
Here are the easier rules I follow to prevent excessive power creep in my games. Sometimes you or someone else will break them, and you'll end up going back and fixing it later, but that's not terrible; the point is to make power creep in either direction easy to recognize and fix.
1. Before anything else, pick targets; have a point of reference to balance towards. In the context of DCSS there would be numeric targets for numbers like damage/turn and AC/EV/SH/HP in relation to where in the game you are, but it would also involve answers to questions like "how badly should the player be allowed to screw up and still recover by using items", and "how much control should players have over monster behaviour" (many AI changes have increased this control, I suspect unintentionally).
You don't actually have to make up specific numbers for things like damage/turn or AC; you can just say "however much AC players had in 0.17", and from 0.17 onwards, enforce it by not allowing large changes to the AC players get. I do this by keeping a list of all the changes that affect AC and periodically looking at it to make sure they come out close to 0.
Examples of significant changes to these numbers: infusion, spectral weapon, str damage buff, weapon base type buffs (all weapon classes except short blades and axes have better base stats than in 0.5), missile buffs, battlesphere, floor enchantment nerf, randart buff, divorcing aux attacks from unarmed skill.

2. If you are trying to maintain an existing balance, when adding a new X or changing an existing X, do not introduce a new best or worst X. Suppose triple swords are (close to) the best weapon in the game. In that case, it is very bad to add a new weapon, or buff an existing weapon, that would be on equal footing with triple swords. Same for adding a monster that's more or less dangerous than any existing monster in its branch, or a vault with more threat or reward than any existing vault in its depth. If you really, truly think the best X is not strong enough or the worst X is not weak enough, you chose the wrong target for X and need to change that target - as well as all of the existing X.
Examples of violations in DCSS that have thrown off the game's difficulty level in the player's favour: deep dwarf, fedhas, sack of spiders, lantern of shadows for several versions, throwing buff. The buffs to cloak of the thief and plutonium sword are also good examples, but they are so extremely rare that they don't have a noticeable effect on difficulty.
Examples of violations that were probably supposed to increase difficulty, but in doing so, also cement the power creep: caustic shrikes, juggernauts, iron giants, deep elf elementalists, monster digging in general ... you get the idea, I already listed a bunch of these.
Making useless X doesn't really throw off the difficulty level, but it means there was no point in adding the X in the first place (or in changing the X instead of outright removing it). Examples: force lance for a long time, glaciate, a few gods.

3. If you are trying to maintain the existing number and quality of player options, then yes, you have to do equal and opposite removals when you add more options. That's just the way it goes. You can't add 1 to 10 and still end up with 10, so if you want to do both, you'll have to subtract 1 as well.
There is a common perception that increasing player options is inherently a good thing, but that's completely untrue. First, decisions aren't made interesting by the number of possible choices; they're made interesting by the difficulty of choosing between the choices. If an option is not closely competing with another option for being the correct option, it isn't adding anything.
Examples of increasing player options: new consumables and evocables, evocations buffs in general. Pack AI. Retreating AI. Ranged AI, sometimes. Noise nerf. A lot of the stuff I listed above.
Good examples of getting rid of powerful but uninteresting player options: divinations removal, summoning nerfs, square LOS, cTele removal, tomb of doroklohe removal.



p.s. I am not saying that randart buff, str buff, etc. were bad, so please don't pretend that I am.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 08:58
by johlstei
It's much better to decide on an equilibrium point for player power, and keep it there - don't move it up or down unless you have a good reason to deliberately change that power level (like the game being too complex to learn, or too simple to have any depth). When you add options to players, take away the same amount. When you change the difficulty or the complexity of the game, it should be a conscious decision, not a side effect.

I'm not sure I'm with you here. Perhaps there is an optimal point for player power vs monster power, but I don't see any reason to think we've attained it in any given crawl version. There's no reason to believe that a certain player power level is ideal just because it is a level that it was at before. Even if we do decide we've approached such a point, overweighting the goal of "fixing player power relative to monster power" sounds like it will detract from other, perhaps more-important goals in crawl development.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 10:24
by sanka
For me a good and interesting balance and complexity is one of the most important goal for a game (at least for games like DCSS) - if it's too easy I got bored but if it's too high I also got bored. I can't imagine what can be a more important goal in game development.

I think that what duvessa said is so important that I would put it into the Design Philosophy somewhere.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 11:19
by kuniqs
Yeah duvessa is awsum and I love her.

If somebody can put a table comparison of yak's statistics from the past 10 versions or so, I'd be very grateful.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 11:36
by duvessa
kuniqs wrote:Yeah duvessa is awsum and I love her.

If somebody can put a table comparison of yak's statistics from the past 10 versions or so, I'd be very grateful.
Actually, I chose a yak as the monster in that example because yak stats have never changed, except for the global changes to HP/MR calculation.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 11:38
by Sprucery
I would assume (being a total non-developer), that
1) It is more fun to add something than remove something. If the player has been given more options to deal with stuff and you therefore feel that an increase in difficulty is in order, it is more fun to design and implement a new dangerous monster than to remove some of the earlier given options.
2) In a group development project, it may be hard to suggest that something that someone else has designed and implemented should be removed. (I have no idea if this is an actual problem in Crawl development.)

One thing that worries me is that if the game is developed with the strongest characters and so-called optimal play in mind, it may at some point become too difficult for unrecommended combos played unoptimally. It you cut the branches too much, my Trollish spellcasters may suffer too much etc. Of course, it should be challenge to play a weak combo, but it should not be impossible or require outrageous luck. To be clear, I don't think there's currently any problem in this regard.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 12:49
by kuniqs
Bring back Mountain Dwarves

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 13:06
by dynast
Sprucery wrote:One thing that worries me is that if the game is developed with the strongest characters and so-called optimal play in mind, it may at some point become too difficult for unrecommended combos played unoptimally.

I dont think that will ever be the case for DCSS. Instead of adjusting the difficult towards optimal play, just get rid of optimal playing aspects, like stairdancing or pillardancing.
Sprucery wrote:It you cut the branches too much, my Trollish spellcasters may suffer too much etc. Of course, it should be challenge to play a weak combo, but it should not be impossible or require outrageous luck. To be clear, I don't think there's currently any problem in this regard.

I dont know why players should be allowed to swim against the current, it is not fun and highlights how easy the game has become. Imagine if no monsters could ever hit you for massive damage because people decided to play spriggans and not train fighting.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 13:21
by Sprucery
dynast wrote: Imagine if no monsters could ever hit you for massive damage because people decided to play spriggans and not train fighting.

But I'm fairly sure the game can currently be won with a Spriggan without training fighting :)

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 14:23
by dynast
Thats why i said that was not the case for DCSS. Also, you can win, but you will be punished for your poor decisions and thats just fair.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 14:33
by Quazifuji
johlstei wrote:
It's much better to decide on an equilibrium point for player power, and keep it there - don't move it up or down unless you have a good reason to deliberately change that power level (like the game being too complex to learn, or too simple to have any depth). When you add options to players, take away the same amount. When you change the difficulty or the complexity of the game, it should be a conscious decision, not a side effect.

I'm not sure I'm with you here. Perhaps there is an optimal point for player power vs monster power, but I don't see any reason to think we've attained it in any given crawl version. There's no reason to believe that a certain player power level is ideal just because it is a level that it was at before. Even if we do decide we've approached such a point, overweighting the goal of "fixing player power relative to monster power" sounds like it will detract from other, perhaps more-important goals in crawl development.


I think you've misinterpreted duvessa's statement. The target isn't an optimal point for player power relative to monster power. It's an optimal point for player power in absolute terms. The point is that if you compensate for every increase to player power by adding an equivalent amount of monster power, rather than removing an equivalent amount of player power from somewhere else, you get power creep where players and monsters both become increasingly more powerful over time, and this has the potential to cause problems in the long run.

I also don't think duvessa necessarily meant to imply that the ideal target balance equilibrium we should choose is one that's necessarily existed in past versions. Just that we should pick a target, rather than letting average player and monster power increase patch after patch and assuming everything's fine as long as they're increased roughly equally.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 14:42
by advil
Here are the yak stats. You can see that proportion of deaths to yaks have approximately doubled since 0.7. However, I think this is a bit misleading for reasons discussed below.

  Code:
08:50 advil: !lg * !won 0.7|0.8|0.9|0.10|0.11|0.12|0.13|0.14|0.15|0.16|0.17|0.18-a / ckiller=yak  s=cv
08:50 OCTOTROG: 8010/2205229 games for * (!won ((0.7 || 0.8 || 0.9 || 0.10 || 0.11 || 0.12 || 0.13 || 0.14 || 0.15 || 0.16 || 0.17 || 0.18-a))): 1173/277396x 0.16 [0.42%], 965/220223x 0.15 [0.44%], 953/194791x 0.17 [0.49%], 867/242281x 0.10 [0.36%], 734/151889x 0.18-a [0.48%], 606/214743x 0.11 [0.28%], 575/189548x 0.9 [0.30%], 567/139576x 0.14 [0.41%], 505/140160x 0.13 [0.36%], 374/178976x 0.7 [0.21%], 352/111572x 0.12 [0.32%], 339/144074x 0.8 [0.24%]


Another prediction that as far as I can see this theory makes is that we should find more deaths at higher XLs in modern crawl. This is sort of true but is more complex than it first looks. The proportion of deaths at xl 27 has roughly doubled:

  Code:
09:26 advil: !lg * !won 0.7|0.17 s=cv / xl=25
09:26 OCTOTROG: 223/373794 games for * (!won ((0.7 || 0.17))): 171/194818x 0.17 [0.09%], 52/178976x 0.7 [0.03%]
09:26 advil: !lg * !won 0.7|0.17 s=cv / xl=27
09:26 OCTOTROG: 989/373794 games for * (!won ((0.7 || 0.17))): 727/194818x 0.17 [0.37%], 262/178976x 0.7 [0.15%]


However, as far as I can tell without importing the data into R and actually running the models this is entirely driven by a huge drop in deaths at XL 1. I couldn't figure out how to get a proportion version of this graph in sequell unfortunately but here is a graph that tells the story (generated from `!lg * !won ((0.7 || 0.17)) s=xl,cv -graph`). This would be best viewed proportionally because there are about 20k more games in 0.17 (and 20k more deaths), but even without that it looks as if the main difficulty change has really just been a huge difficulty drop at XL1, with the deaths then spread across the rest of the game. I guess you might have expected them to clump at low XLs, which doesn't seem to have happened, but I still don't really see how this is compatible with the "power creep" theory (which perhaps I don't fully understand, though).

This also suggests that the yak stats above can't be easily interpreted to mean much, since it is likely just reflecting the generally doubled proportion of deaths at all XLs above xl8 or so (the proportion curve is a bit closer up to that point than what you see with absolute numbers).

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 15:25
by Sprucery
dynast wrote:Thats why i said that was not the case for DCSS. Also, you can win, but you will be punished for your poor decisions and thats just fair.

Yes, I agree that it is not the case currently. I hope it stays that way. And punishing for poor decisions is absolutely ok.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 16:40
by Rast
advil wrote:it looks as if the main difficulty change has really just been a huge difficulty drop at XL1, with the deaths then spread across the rest of the game.


I'd be OK with this.

Are you sure it isn't just less players quitting/leaving dungeon?

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 17:03
by dynast
D:1 used to be brutal in older versions, mainly because of throwing darts/spears/etc. Oh and traps, of course.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 17:11
by PleasingFungus
The general dev philosophy behind crawl balance has been to not worry about fine details. If a change is a small buff or a small nerf, that's okay: crawl is a big game & can handle small balance changes in the short run, and in the long run, other changes will cause compensating buffs or nerfs. This generally works okay, but it does have the weakness that you can end up with having considerably more buffs than nerfs.

I think we should be clearer about what we see as the potential problems caused by power creep. My concern is that you end up obsoleting older content [items, gods, enemies, w/e], and thereby making the game less fun, by creating more chaff enemies, ignorable items, bad skill choices, etc. I can't think of much of that on the player side at present - gods that I'd never pick because some other god was added or buffed to be better, items that I'd never use because some other item was added or buffed to be better - but it seems plausible that older enemies have become less threatening by virtue of various player buffs. I don't know how you'd go about confirming that, though (are yaks less dangerous, or is the average player better at dealing with them?), and I'm not sure what you would do about it if you decided that older enemies had become less effective across the board.

I think it would be useful to discuss those two questions - I'd prefer for this not to become another thread where Sequell queries overwhelm actual discussion.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 17:11
by PleasingFungus
dynast wrote:D:1 used to be brutal in older versions, mainly because of throwing darts/spears/etc. Oh and traps, of course.

Also, significantly lower starting HP.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 18:06
by dynast
PleasingFungus wrote:Also, significantly lower starting HP.

I had no idea.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 18:08
by carbonbasedlifeform
This is very unpractical approach ( adding certain options by taking out equal amount of others). If you look at anything in particular there will be always slight unbalance and too many factors to look at( amount, chance of occuring, hierarchy among other options, place and time etc). Equilibrium exists theoretically but I think we still won't have to worry about power creep side effects at least not in close future.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 18:39
by Lasty
I agree that trying to leave player power and monster power unchanged over time except in situations where the existing balance is already identified as an issue should be a goal of design and development. Crawl development doesn't always succeed at doing that, but I think by and large crawl developers do attempt to do just that. However, I think each crawl developer has their own feelings about which aspects of player or monster power are already out of balance. Also, each developer has their own list of existing features that they feel have problems that need resolving, and resolving those problems sometimes impacts player/monster balance in ways that are hard to foresee.

For example, I'm responsible for the randart buff. I didn't intend to make it a buff, but it's possible that that is what it has been in net effect. What I saw was a chaotic system that was hard to tweak and hard to keep modern and hard to accurately anticipate the outcomes of. I cleaned it up so that it's much easier to work with and much easier to tweak any aspect that needs adjustment. Since I did that, there have been very few tweaks to the current balance; if overall the randart buff has been a buff, it would be good to identify what aspects have gotten stronger and twist those knobs a bit. It's definitely something I've been trying to keep an eye on, and I'd be interested to hear feedback on how it could be improved.

I'm also responsible for several new monsters, including juggernauts and caustic shrikes. I saw those as addressing a long-standing issue in crawl, which is power curve. Character power ramps up much faster than monster power around Vaults/Depths, making Depths largely a free XP zone; this was made worse by the fact that so many of the scary Depths monsters worked exactly the same: have very good defenses focused on AC/HP, move speed 10, have a few good spells. The area needed things to change the types of challenges presented as well as the dimension of the challenge presented. (Side note: I think entropy weavers work okay right now, but I think maybe stormcallers should get removed again; juggernauts need to be reworked.)

As far as I'm concerned, the main current balance-ish gameplay issues that need resolution are:
1) Spectral weapon is an unreasonably large boost to offense for relatively small investment.
2) Ozo's armour is an unreasonably large boost to defense for relatively small investment.
3) There are too many different evocables and most of them are very strong.
4) Throwing in particular and ranged combat in general is much too strong.
5) Lair gives too much XP relative to the danger of its monsters; after a couple levels of Lair, you almost always have enough XP to clear the remainder of Lair safely. I'd love to see it trimmed by 4 levels, and add scarier monsters on the OoD clock -- maybe dragons, which are more or less on-theme.
6) Energy randomization is a clunky solution to speed 10 chases.

There are probably a bunch of other things that also need some kind of attention, but those seem like some of the more glaring ones to me, at least off the cuff.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 18:45
by njvack
duvessa wrote:Here are the easier rules I follow to prevent excessive power creep in my games.

I don't want to derail this thread, but I would very much like to play games you've had a hand in -- if you'd link 'em in CYC or something, I'd be much obliged. If you'd rather not, that's totally cool.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 20:01
by duvessa
Lasty wrote:For example, I'm responsible for the randart buff. I didn't intend to make it a buff, but it's possible that that is what it has been in net effect. What I saw was a chaotic system that was hard to tweak and hard to keep modern and hard to accurately anticipate the outcomes of. I cleaned it up so that it's much easier to work with and much easier to tweak any aspect that needs adjustment. Since I did that, there have been very few tweaks to the current balance; if overall the randart buff has been a buff, it would be good to identify what aspects have gotten stronger and twist those knobs a bit. It's definitely something I've been trying to keep an eye on, and I'd be interested to hear feedback on how it could be improved.
That is not the randart buff I'm talking about. It wasn't a significant buff. I am talking about the buff to randart enchantment values that happened prior to that. In older versions, randart weapons and armour had lower enchantment on average. I do think this buff was necessary (randarts were rarely useful because of garbage enchantment) but it was a power increase nonetheless.

Re: DCSS has a power creep problem

PostPosted: Monday, 25th January 2016, 20:51
by Siegurt
duvessa wrote:
Lasty wrote:For example, I'm responsible for the randart buff. I didn't intend to make it a buff, but it's possible that that is what it has been in net effect. What I saw was a chaotic system that was hard to tweak and hard to keep modern and hard to accurately anticipate the outcomes of. I cleaned it up so that it's much easier to work with and much easier to tweak any aspect that needs adjustment. Since I did that, there have been very few tweaks to the current balance; if overall the randart buff has been a buff, it would be good to identify what aspects have gotten stronger and twist those knobs a bit. It's definitely something I've been trying to keep an eye on, and I'd be interested to hear feedback on how it could be improved.
That is not the randart buff I'm talking about. It wasn't a significant buff. I am talking about the buff to randart enchantment values that happened prior to that. In older versions, randart weapons and armour had lower enchantment on average. I do think this buff was necessary (randarts were rarely useful because of garbage enchantment) but it was a power increase nonetheless.

Are you talking about this:
http://s-z.org/neil/git/?p=crawl.git;a= ... 34d66467e0
  Code:
Don't allow body armour or weapon randarts to have negative enchantment

These were never wearable, since the value of these slots is so heavily
based on the enchantment.

?

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 00:07
by Hurkyl
Lasty wrote:5) Lair gives too much XP relative to the danger of its monsters; after a couple levels of Lair, you almost always have enough XP to clear the remainder of Lair safely. I'd love to see it trimmed by 4 levels, and add scarier monsters on the OoD clock -- maybe dragons, which are more or less on-theme.

Note this would increase the availability of dragon armors; I'm not sure if that's a good thing or a bad thing. It would also interact badly with the fact that autoexplore sometimes takes an eternity to explore lair levels.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 00:28
by milski
I do think that Crawl is a lot less likely to suffer from the sort of exponential power creep implosion that ToME had, since ToME relied very heavily on multiplicative buffs and stacking reductions and a lot of other absurdly bad ideas, while Crawl tends to be more logarithmic in nature (Besides Battlesphere and Spellforged Servitor, those are just straight up multiplicative buffs to spellcasters).

With that being said, power creep definitely seems to be occurring in Crawl, and even in games where it doesn't result in a total implosion, it can result in some balancing instability; DotA 2 had a very obnoxious competitive meta in 6.83 and 6.84 before 6.85 and (possibly) 6.86 reigned the game back somewhat. I agree that it's probably worth evaluating pure numbers nerfs to make up for power granted by options or removal of outright-bad outcomes.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 01:02
by dpeg
Many thanks for the original posting, duvessa! It's a discussion definitely worth having.

Myself, I am pretty sure that we'll avoid the worst excesses of power spiralling, simply because we're aware of the fact. In a way, Crawl has something of a nerf/removal culture. It's true that more content gets added than removed, but there are several good reasons for this: (1) some additions are side-way (see below) and (2) scrutinising existing content is work-like whereas adding your own ideas is unadultered fun. Also remember that the game is being developed by very many people over a long time, so there are different visions.

I'll give you some examples how we deliberately avoided the power spiral: instead of adding new branches (Depths was a split-off), the Lair subbranches are rolled. That way additional content does not lead to more experience and more loot. (duvessa has a point that portal vaults noticeably increased experience and loot. This can and should be addressed: the total CHANCE for all portal vaults in a depth range should be fixed and possibly reduced.)
Overpowered features have been nerfed, ever since DCSS 0.1. (Sometimes overnerfed, as with 0.6 Armour.) This is an attempt to keep the power spiral in check.
I don't think it is feasible for Crawl to try and fix a power level. There are too many things in flux at any given moment, and I reckon we're better off with a more loose assessment.

Rubbing the worst cases (only ever mention the worst cases, much better to tackle these than to talk about all bad cases day and night) into our faces time and again is appreciated!

Here's what I consider the worst offenders (some of these are not related to any recent changes, but they'd be straight-up nerfs, so they'd bring the power curve in the harder direction, which I like -- this is another discussion worth having, by the way):
1) the new evokables -- very nifty ideas, but a lot more power for cheap (semi-limited use, just one skill)
2) luring (very strong mini-tactic, also extremely tedious)
3) buff spells (power gain for little investment, no-brainer to use, interface-wise troublesome)
4) portal vaults (tone down frequency, smaller maps for early ones)


About that sideway new content: because you only have one species in a single game, and only one god at any time, more species and gods do not affect the power very much. This does not mean any new god/species is a good god/species, but you don't turn the power wheel much with these changes.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 01:38
by PleasingFungus
Lasty wrote:As far as I'm concerned, the main current balance-ish gameplay issues that need resolution are:
1) Spectral weapon is an unreasonably large boost to offense for relatively small investment.
2) Ozo's armour is an unreasonably large boost to defense for relatively small investment.
3) There are too many different evocables and most of them are very strong.
4) Throwing in particular and ranged combat in general is much too strong.
5) Lair gives too much XP relative to the danger of its monsters; after a couple levels of Lair, you almost always have enough XP to clear the remainder of Lair safely. I'd love to see it trimmed by 4 levels, and add scarier monsters on the OoD clock -- maybe dragons, which are more or less on-theme.
6) Energy randomization is a clunky solution to speed 10 chases.

I'd note that Spectral Weapon was just nerfed a few weeks ago, and that energy randomization is probably not a balance issue. That's quibbling, of course! I generally agree, particularly with 2, 3, 4, and 6.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 04:48
by pubby
I mostly just hate how clear cut everything and singular everything is now.

You win a fight solely because evokers.
You escaped a fight solely because teleport.
You lost a fight solely because caustic shrike.

It feels like there is very little interaction between the game's little effects because a few big ones completely dominate.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 06:27
by kuniqs
I almost never use evokers because I'm afraid of wasting them. In recent games I don't even pick them up, since they clutter my inventory and do nothing.
Wands are another thing. It's good to know Evocations6 double the base power of wands, Evocations12 triple the base power and so on.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 07:47
by duvessa
dpeg wrote:About that sideway new content: because you only have one species in a single game, and only one god at any time, more species and gods do not affect the power very much. This does not mean any new god/species is a good god/species, but you don't turn the power wheel much with these changes.
Except when the new species is stronger than all the existing species (Deep Dwarf) or the new god is stronger than all the existing gods (Fedhas).

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 08:24
by nago
I only know few versions ago the "depths" most threating common enemies were foo giants, and now recently was considered necessary to add things like caustic bees, iron octopode and why not also juggernaut.
And that 98% of times it's yet totally possible to threat them as tab fodder with minimal or totally absent strategic effort, because most char by depths - if not doing crazy stupid things like using +10 flail at xl25 because randart and shiny shield - have ridiculous high AC and offense anyway. And for the remaining 2% there are so many different consumables that one could wipe anything RNG can throw anyway easily.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 08:32
by Tiktacy
duvessa wrote:Except when the new species is stronger than all the existing species (Deep Dwarf)


I would agree with you, but I don't think they have added a species stronger than humans yet. Deep Dwarves might be really good, but nothing matches the versatility of a human. Trolls are very good as well, perhaps better depending on who you ask.

As for Fedhas... I can see where you are coming from, but I think you could make the argument there are gods that trog closely matches Fedhas in power. On the other hand, just because one older god matches the power of a newer one doesn't mean the creep isn't there, Ashenari and Gozag are pretty high up there in the power department as well and they are both "new" gods.

Either way, I think the addition of some of the new gods has been very healthy for the game even if its detracted from its difficulty. Ash and Ru are both extremely fun, well thought out, and creative additions to the game that I would be very sad to see go.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 08:36
by Tiktacy
nago wrote:I only know few versions ago the "depths" most threating common enemies were foo giants, and now recently was considered necessary to add things like caustic bees, iron octopode and why not also juggernaut.
And that 98% of times it's yet totally possible to threat them as tab fodder with minimal or totally absent strategic effort, because most char by depths - if not doing crazy stupid things like using +10 flail at xl25 because randart and shiny shield - have ridiculous high AC and offense anyway. And for the remaining 2% there are so many different consumables that one could wipe anything RNG can throw anyway easily.


I think the biggest issue with this is the fact that consumables can be horded so easily. Having specific consumables(blink/teleport scrolls for example) be removed from all areas aside from the main dungeon might help with this issue. Although it would probably do some major damage to the games already out of whack learning curve(but thats a different thread topic entirely).

Even so, I do think the amount of power the player has by the time go start tackling depths is a little bit out there. This issue might be easier to solve than some others though, you can always try having monsters scale with player experience if there is no other solution.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 08:43
by nago
Tiktacy wrote:This issue might be easier to solve than some others though, you can always try having monsters scale with player experience if there is no other solution.


Which is exactly the problem Duvessa has pointed out in the opening post. And also already explained why it's bad and totally not working.

EDIT: btw, Gozag was power-wise utterly shit for long time (flavour\design wise was probably even worse). Now he's kinda okay but still under-powered in comparison of many other gods.
Ash has a very long power curve, she's totally useless for long time\low piety, and then she's one of the strongest god around. Therefore overall, I wouldn't say she's a strong god either.
If you really wanna to point out a strong god added lately, he's Ru, but he's still balanced in comparison of strongest gods like Fedhas or Trog.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 09:09
by Tiktacy
nago wrote:
Tiktacy wrote:This issue might be easier to solve than some others though, you can always try having monsters scale with player experience if there is no other solution.


Which is exactly the problem Duvessa has pointed out in the opening post. And also already explained why it's bad and totally not working.

EDIT: btw, Gozag was power-wise utterly shit for long time (flavour\design wise was probably even worse). Now he's kinda okay but still under-powered in comparison of many other gods.
Ash has a very long power curve, she's totally useless for long time\low piety, and then she's one of the strongest god around. Therefore overall, I wouldn't say she's a strong god either.
If you really wanna to point out a strong god added lately, he's Ru, but he's still balanced in comparison of strongest gods like Fedhas or Trog.


I thought gozag was considered good because he helps get you out of bad situations in the early game that can't be easily avoided? I don't know a lot about him, I've just heard from some players that he is extremely good if used correctly. I'll have to look into trying him out to see for myself.

I am pretty sure ash is quite high up there compared to other gods. Compared to trog and fedhas maybe not, but compared to the average god in the pantheon she is amazing. Maybe getting another opinion on this would be good though. I've just always been under the impression Ash was a top tier god, but maybe I'm mistaken.

I really REALLY wanted to mention Ru, he is my favorite of all the new gods. I ended up deciding against it though, I sort of got the impression that Ru's abilities were more of panic buttons to get out of bad situations from making the wrong decision as opposed to just an overall powerful god like Ashenzari. However, unlike gozag, his panic buttons don't help as much in the early game(although this is just speculation, I'm planning on trying out gozag tomorrow to see what he is all about).

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 12:54
by dynast
You cannot worship gozag if you dont have enough food to get through the dungeon until you can fund the first shop, as when playing optimally you have to assume the possibility you wont find any food during that.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 14:11
by tabstorm
So I tried playing 0.10 yesterday. My character felt a little worse than an 0.17 character, but I'm not really sure why. I don't think the combat formulae have been changed that much. It might've been because I went into orc at XL10 with 10/12 and forgot I wasn't a Minotaur because I didn't want to deal with Nausea.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 17:43
by PleasingFungus
I don't think any parts of crawl are being designed with an eye specifically toward challenging dd or fedhasites. Certainly none of the monster changes mentioned here were made in response to them. It seems more reasonable to discuss the changes that affect a more significant portion of games, like evocables, weapon stats, etc.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 20:00
by dynast
I dont get it, why does DD or Fedhas have to be the way they are? My only complaint would be that Fedhas doesnt have any drawback, so whenever you play a game optimally and fedhas is one of the first altars you come by, not worshipping it is always a bad decision. If fedhas abilities were flipped out(you get evolution as your last ability, rain becoming the first) would it be like removing something from the game?

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 20:42
by asdu
tabstorm wrote:So I tried playing 0.10 yesterday. My character felt a little worse than an 0.17 character, but I'm not really sure why. I don't think the combat formulae have been changed that much. It might've been because I went into orc at XL10 with 10/12 and forgot I wasn't a Minotaur because I didn't want to deal with Nausea.


I tried playing 0.10 some time ago (trying to do speedruns with pre-nerf summoning) and I too noticed that my characters seemed to kinda suck, then I noticed my morningstar had 10 base damage (for the record, it's now 13). I knew that more or less all weapon damage had been buffed since then, but I didn't remember it to be this big.
A lot of other little things were like that: things that are good in "modern" Crawl were not as good or outright bad. Also lots of useless items (all the harmful consumables that have been pruned since, redundant weapon types, negatively enchanted artefacts, most fixedarts, frankly, etc) making it less likely to get the good ones.
On the other hand, older versions definitely had more exploitable brokenness (summons fighting out of LoS being the most obvious case).

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 21:04
by Lasty
Regarding negatively-enchanted artefacts, I don't think there's been meaningful power inflation -- hardly anyone ever used negatively-enchanted body armour or weapon artefacts, and people are only barely more likely to use +0 ones. The change was written so that body armour / weapon artefacts that would have been negatively enchanted ones are now +0, but there's no other change. Since the +0 artefacts are only very marginally more likely to get used, it's a pretty negligible power creep.

Re: DCSS has a power creep problem

PostPosted: Tuesday, 26th January 2016, 21:12
by Siegurt
Well weapon base damage had a fairly major overhaul at the time, some things got better, some got worse (exec axes used to be 20 base damage, rip ankus's)

The analogous weapon to the current Morningstar is the .10 spiked flail (dam 12, sped 16, -2acc)