Branch ends and their rewards

[tazoz]

There is a certain missconception that giving out large amounts of loot is bad, there has been a gradual reduction of loot over time in the branch ends and even in Pan one of the pan lord vaults has been stripped of loot.

Why is this bad? giving large rewards for defeating a branch gives the player the motivation to risk going into more dangerous areas. For example because snake and swamp branch ends have no substantial loot, players sometimes leave those areas until after they take out vaults, elf and slime as they have no real reason to take them out earlier, if those branch ends provided the player with some sort of reward then players would risk taking on those branches much earlier leading to a more challenging and fun play through. The reason why elf for example is considered so dangerous is that often players take it on before they are prepared to do so because of the high rewards at the end. In reality the branch is quite easy when you are properly enough prepared.

There is an even bigger problem when you consider pan, abyss and the hells, the loot in those branches is completely inproportional to the risk involved in exploring those areas, this has lead to a seperation between 3/4/5 runers and 15 runers, there is very little in between, a player knows that if he isn't prepared for the extended end game there is very little chance that he will get the gear neccesary in the extended end game to brave it through. Notice that the only hell branch that is taken on by weaker characters is the iron city and that is because the loot is decent enough to take the risk.

The only extended end game branch that has a good spread of loot is the tomb but as the risk is so high, a character can't use it as a stepping stone to be able to brave the other end game areas.

To clarify things a bit, eventhough abyss can be an easy rune to get, no one takes on the abyss before vaults because exploring the abyss makes vaults harder because of drain on resources and mutations and you get no rewards for your efforts. (except the rune which has the same value as the vaults rune...)

Ironically enough I've found myself risking Pan just for the possibility of finding a Zig just because Zigs actually provide you with loot that will make a character viable in the extended end game. (of course this has often lead to dead characters but I don't see much point in ascending with 5 runes.)

[KoboldLord]

Orc 4 got shops added recently, and it seems to have helped. Orc 4 now nearly always has an interesting tactical situation to work with, and with four shops to hope for there's reasonable incentive to clear it before its strongest inhabitants are downgraded to popcorn status. Similarly, Swamp 5 and Snake 5 could probably use Shoals 5-level loot so that there's actually reason to go early.

I'm not sure there's any real problem with the loot in the post-endgame, though. You can get through those areas on much fewer resources than maybe you'd prefer, if you find yourself needing to. It would be nice to max out fire resistance for Gehenna and cold resistance for Cocytus, for instance, but two pips will generally do in a pinch. Tomb is pretty dangerous when unprepared, but if you've got Abjuration and a silent ranged attack you can safely clear it for your fourth Rune. It is true that there's no game left to use the loot on once you've finished the post-endgame, but I don't think adding more loot will help with that.

[dpeg]

Giving out too much loot is bad because it reduces choices. Also, in my opinion the amount of loot has been drastically increased since DCSS 0.1 and I am not happy about this. The reason are all the great, hand-made new vaults. Many, many of them come with items. Next are the portal vaults which increase loot some more. Also think of the recent Orc:4 loot burst. Because of this loot increase did we hold back from giving any compensation for the level cuts (in Lair etc.).

The idea for having no loot in Snake and Swamp is very simple: if you play the easiest game, a 3-runer, you only get the minimal amount of loot. If you want more loot, you have to go somewhere else. It is true that players forfeit Snake, Swamp and other branch ends until much later, and we're not happy about it. If I am not mistaken, our solution will not be to give loot, however. Rather, there will be a lock on D:14 and you can only progress if you show a rune. (That's a pretty mild version for a start.)

Pan and the Abyss are problematic for several reasons, and will be addressed. I don't think the hells or Tomb or Slime are problematic. You can do one of those branches if circumstances (skills, or gear, or god) are right. And at least skills and god are in your hand.

I am not an all-runer, so I may be missing something, but I don't think that loot should be amplified. In my opinion, it is rather the opposite. The late and extended game should be poorer, at least in xp and perhaps also in loot. But that's a personal opinion, would be interesting to hear others on the topic.

[tazoz]

Dpeg, you are right loot has been increased everywhere except where it's important for there to be loot, take troves for example, is there any challenge getting the loot contained within one? It's mostly a decision, do I roll the dice and sacrifice one object for another and sometimes it pays off and you get all the armour you ever needed. (I literarly had a 15 runer that found all his items in a single trove).

It's important to add rewards proportional to the amount of challenge a player faces. I'm just pointing out that there is a problem with the spread of rewards and the focus should not be in reducing the amount of rewards a player gets for dealing with the more challenging areas in the game but taking out the large cluster of items contained every where else.

Take for example the unique Pan lords, players just run past them and skip them, thus skipping the challenge as they give very little reward for killing them. (except maybe the staff of dispator.

[dpeg]

Troves are meant to provide interesting strategical decisions. If they don't work, we'll remove them.

I still don't see why Snake:5 should come with loot. Players get a lot of xp and all the monsters' items. The problem is that the rune is only needed much later, so we will change that. You prefer the carrot to encourage players to do Snake:5 early, but it seems we prefer the stick.

[tazoz]

Dpeg, I'm just pointing out a few things, do with it what you like. Thanks for all the hard work you and the other developers have been doing, the game is great and I've had loads of fun playing it over the years.

[dpeg]

Hey, it's fine. You are spot on that players not doing Snake:5 early is a problem. We were aware of that as well (see https://crawl.develz.org/wiki/doku.php? ... on:hopping), we just came up with a different solution. Thanks for the input!

[galehar]

Hey, it's fine. You are spot on that players not doing Snake:5 early is a problem. We were aware of that as well (see https://crawl.develz.org/wiki/doku.php? ... on:hopping), we just came up with a different solution. Thanks for the input!

This reminded me to update the page to add my idea there. Here it is:

• D:14 is a special level with a big thematic vault. Size is between half and full level.
• Difficulty is about the same as Snake:5 or Swamp:5.
• At the end of the vault, there's a portal to D:15 which ask for a rune.

This is a nice way to get new content for the game with less work than full blown branch. Any theme is acceptable for the vault, except one thing: no castle. If we get several good ones, then we can have a random rotation.

[dpeg]

Sounds awesome, galehar!

[jpeg]

It's important to add rewards proportional to the amount of challenge a player faces. I'm just pointing out that there is a problem with the spread of rewards and the focus should not be in reducing the amount of rewards a player gets for dealing with the more challenging areas in the game but taking out the large cluster of items contained every where else.

Take for example the unique Pan lords, players just run past them and skip them, thus skipping the challenge as they give very little reward for killing them. (except maybe the staff of dispator.

Something to keep in mind is the problem of the power spiral: Someone who's already powerful can brave the dangerous areas and get more powerful. Ziggurats are a good example of this.

[smock]

Yes, there's now too much loot. I'd vote for less loot per place with loot, not fewer places with loot. 4 shops in Orc might be too much. I'd be motivated to take it on it it had only two shops, one of which is guaranteed to be a good shop. Certainly, all the portals added to the game are a boon - I still want more, not fewer! Embedding portals as vaults is another option.

I love how you must kill the Royal Jelly to get the Slime loot. This can be implemented elsewhere. As in "You pick up Asmodeus' key. This should open the door to some great treasure!" or placing a unique who is "holding" the rune and some loot at a branch end, stationed where the rune currently is. Aizul in Snake or Frederick in Swamp might work. Why not have a spider's nest vault on D14, with three staircases and a one-rune-lock? Not too creative, but there is a good reason why many other games operate like this.

Having uniques guard troves (by holding the key to the trove) would more interesting than finding a +5 ice dragon armour. Thematically it would fit a bit better with troves, too. Also, taking a key from a unique in a trove or bailey might be a fun thing to log. Visiting and deserting an ossuary is less interesting than slaying Menkaure in an ossuary and taking the loot. OOD monsters could be used, too.

Ninjaing in selected places could still be allowed, of course.

[Marten]

Speaking of branches, I am much more wondering whether the variability of Lair sub-branches could be improved any further on the account that the current setting just makes me want to play Merfolk, so I can clean the guaranteed Swamp/Shoals with flying colours.

What I think would be awesome, is for the Lair to have 2 sub-branch slots - one for the usual (Sw/Sh/Pi) and the second for a sub-branch determined by the highest attribute when the game started - Str- Hills (Giants, Titans, whatever), Int- Grove (Spiriggans, yay!), Dex- Thieves' Den (Maurice styled human thieves) (all natural environments, so they could compliment Lair well), where the player could definitely find some loot to compliment his playstyle.

This would solve the reward issue for players, who don't have edaquate equipment by the time they wish to move onto Vaults, wouldn't it?

[XuaXua]

Yes, there's now too much loot. I'd vote for less loot per place with loot, not fewer places with loot.

Well, if you loot for loot then you'd loot loot, and as an aside, Crawl loot doesn't contain a lute.

[XuaXua]

Hey, it's fine. You are spot on that players not doing Snake:5 early is a problem. We were aware of that as well (see https://crawl.develz.org/wiki/doku.php? ... on:hopping), we just came up with a different solution. Thanks for the input!

This reminded me to update the page to add my idea there. Here it is:

• D:14 is a special level with a big thematic vault. Size is between half and full level.
• Difficulty is about the same as Snake:5 or Swamp:5.
• At the end of the vault, there's a portal to D:15 which ask for a rune.

This is a nice way to get new content for the game with less work than full blown branch. Any theme is acceptable for the vault, except one thing: no castle. If we get several good ones, then we can have a random rotation.

If there was something like this every several levels or so, one could conceivably remove portal timers, and this would seriously impact speed runs.

[dpeg]

Xuaxua: speed run is defined by the current game, there will always be players who pull astonishing stunts.

[TwilightPhoenix]

What if snake/swamp had some minor loot to go with the rune, but all just stuff you could find laying around on the floor anyway? So, along with the rune, you might find a few consumables which may or may not be ones you need, maybe a little bit of food, a couple of pieces of weapons/armor/jewelry which will most likely be stuff you don't need and most likely not be a randart. Basically, an extra pile of floor trash with maybe around 6 items total. You might get very lucky and find something awesome, but it's not like that luck couldn't have happened anywhere else in the dungeon either. Odds are you'll probably walk out with a couple of extra scrolls and potions, assuming you don't use them during your escape.

[XuaXua]

I think more monsters should be armed with the artifacts, including rings and amulets.

I would feel more of a sense of accomplishment pulling the Singing Sword off a dead orc warrior than just finding it where someone got annoyed enough to discard it.

[betamin]

Galehar's idea rocks, I love themed difficult levels.

[KoboldLord]

Giving out too much loot is bad because it reduces choices. Also, in my opinion the amount of loot has been drastically increased since DCSS 0.1 and I am not happy about this. The reason are all the great, hand-made new vaults. Many, many of them come with items. Next are the portal vaults which increase loot some more. Also think of the recent Orc:4 loot burst. Because of this loot increase did we hold back from giving any compensation for the level cuts (in Lair etc.).

You seem to be disregarding a vast middle ground between having a small incentive to press on and the fire hose of loot that actually causes problems. Orc 4 can and does give out too much loot, and Swamp 5 too little, and these situations can co-exist quite comfortably even within the same game. Perhaps Orc 4 needs to have one guaranteed-good shop and three consumable shops, and Swamp 5 needs a little good loot next to the Rune.

And speaking of reducing choices, what choice does Swamp 5 present you on a typical game? The typical Swamp 5 is fairly unchallenging and completely unrewarding, disregarding special cases such as Mennas, Boris, and Mara having one of their little tea parties just outside the Rune vault. Your relevant choice is to slog forward now, spending lots of real-time with no useful loot to show for it, or procrastinate and do it later. If you choose the later, you get nothing to show for the excursion in substantially less real-time.

Carrots are good and sticks are good, at least in moderation, but just because they are often good does not make them the solution for every problem.

The idea for having no loot in Snake and Swamp is very simple: if you play the easiest game, a 3-runer, you only get the minimal amount of loot. If you want more loot, you have to go somewhere else. It is true that players forfeit Snake, Swamp and other branch ends until much later, and we're not happy about it. If I am not mistaken, our solution will not be to give loot, however. Rather, there will be a lock on D:14 and you can only progress if you show a rune. (That's a pretty mild version for a start.)

I'm not really opposed to the D14 Rune Lock plan because it's not really all that different than my typical play anyway (Shoals/Snake/Swamp 5 aren't actually all that hard, certainly not worse than mid-Vaults). I'm really leery of the reasoning cited, though. Players choose to avoid part of the game that they perceive as a waste of their time, and the solution is to railroad them there anyway? Shouldn't there be some consideration to fixing the cost/benefit analysis that is clearly askew?

I am not an all-runer, so I may be missing something, but I don't think that loot should be amplified. In my opinion, it is rather the opposite. The late and extended game should be poorer, at least in xp and perhaps also in loot. But that's a personal opinion, would be interesting to hear others on the topic.

If you feel that loot needs to be removed, how about my nomination of all those D6 jewelry and book shops? I'm pretty sure those are more damaging to the game than looting from the Swamp 5 Rune vault a random book, a mediocre randart ring, and a piece of armor slightly better than what you have.

I love how you must kill the Royal Jelly to get the Slime loot. This can be implemented elsewhere. As in "You pick up Asmodeus' key. This should open the door to some great treasure!" or placing a unique who is "holding" the rune and some loot at a branch end, stationed where the rune currently is. Aizul in Snake or Frederick in Swamp might work. Why not have a spider's nest vault on D14, with three staircases and a one-rune-lock? Not too creative, but there is a good reason why many other games operate like this.

Having uniques guard troves (by holding the key to the trove) would more interesting than finding a +5 ice dragon armour. Thematically it would fit a bit better with troves, too. Also, taking a key from a unique in a trove or bailey might be a fun thing to log. Visiting and deserting an ossuary is less interesting than slaying Menkaure in an ossuary and taking the loot. OOD monsters could be used, too.

Ninjaing in selected places could still be allowed, of course.

The novelty would wear thin pretty quickly. Slash'EM puts a level boss in every single side level, and you stop noticing or caring after a while. Cerebov wouldn't keep his street cred very long if you had to be prepared to kill him to pass through his domain, because you would then make sure you always were prepared, and he'd end up dying every time you saw him.

[theropod]

The D:14 lock significantly increases the impact of the RNG witholding poison resistance. I've done Snake:5 without rPois with an endgame-level character with teleport control, but unresisted poison arrow on a midgame character is a bit much.

[smock]

The novelty would wear thin pretty quickly. Slash'EM puts a level boss in every single side level, and you stop noticing or caring after a while. Cerebov wouldn't keep his street cred very long if you had to be prepared to kill him to pass through his domain, because you would then make sure you always were prepared, and he'd end up dying every time you saw him.

I think a unique guard would be a good choice for troves nonetheless.

Not sure I get your logic re Cerebov's street cred. He'd still be the toughest.

[galehar]

The D:14 lock significantly increases the impact of the RNG witholding poison resistance. I've done Snake:5 without rPois with an endgame-level character with teleport control, but unresisted poison arrow on a midgame character is a bit much.

That's true. But maybe the solution would be to tone down Shoals:5 difficulty so it's a viable mid-game alternative. Since it forces you to pick a Lair rune early, then I think difficulty between the Lair branches will need to be adjusted.
Increase the chance to find a swamp dragon hide in Swamp can help too. But when spider turns into a branch, if you get spider/snake and no rP... ouch. Maybe make the Lair roulette choose between Snake/Spider and between Swamp/Shoals.

[dpeg]

galehar: I mentioned on ##crawl-dev already that with the rune lock we need to adjust Shoal's difficulty. Regarding rPois: I think it is best if the players can somehow work around it, like with the Swamp dragon armour. More dragon in Swamp would be good. For Spider, I've suggested that some corpse produced there could give temporary rPois.

[mumra]

galehar: I mentioned on ##crawl-dev already that with the rune lock we need to adjust Shoal's difficulty. Regarding rPois: I think it is best if the players can somehow work around it, like with the Swamp dragon armour. More dragon in Swamp would be good. For Spider, I've suggested that some corpse produced there could give temporary rPois.

That's a really nice idea, it would additionally introduce an element of inversely managing hunger so you're hungry enough to eat chunks as needed (without gourmand), making the food clock more relevant.

[Galefury]

I'm pretty sure that was referring to permanent stuff. I think using corpses for this would be a bit weird anyway, it might lead to people leaving certain monsters alive to get their resistance later. And the hunger management mumra described sounds more annoying than interesting to me.

Just giving most types of spiders a chance to drop spider eggs that decay like chunks but are edible when satiated (only giving sultana-like nutrition) and provide temporary poison resistance would work better. It would dodge most of the annoying side effects while still encouraging players without permanent poison resistance to play at a faster pace.

[Moose]

Interesting thought ...since 27 seems to be such a popular number in crawl, you could subdivide the main dungeon into three realms, each comprising nine levels.

Levels 9, 18, and 27 could be special in that they are themed, and require a rune to pass through. This would remove some of the seemingly arbitrary requirement of three runes to enter Zot, and make it a more progressive accomplishment.

If the Lair (and probably the Orcish mines) were guaranteed to be generated prior to level 10, then this could work, but would require a lot of retuning branches.

[dpeg]

Intrinsics from corpses? Wasn't there something in the design section of the manual about how that was a bad idea?

Sure, and I wrote it.

But having understood why Nethack's intrinsic system falls flat does not mean that we can't come up with ideas on our own. If whatever consumables used in Spider provide rP for a duration of around 500 turns with announced expiration, then that's something you can work with. And it does not fall into one of the old traps. There may be drawbacks to this idea that I don't see -- in this case I very much would like to hear them. Comparing the idea with something broken is not good, because we already sidestepped that peril. It's in the design section of the manual.

[absolutego]

intrinsics from corpses doesn't sound right. (edit: i posted this without reading dpeg's last post, and i can't really elaborate. it just doesn't feel right.)
and i've done snake without rPois, but i rarely go into swamp without it (or clarity). the only way around drakes is healing + teleport (or semi-controlled blink), seeing how they spam it more than transmuter ghosts and what with the reduced mobility. and i don't see how spawning more swamp dragons is the solution to that problem.

i don't dislike the idea of spicing up the mid-game (quite the opposite), but it should be handled with care. is spider makes it, there will be three lair branches for which rPois is nearly necessary (of which snake may be the easier to do without, at least for a caster/ranged fighter). that may not be the best step towards a rune lock.

[dpeg]

minmay: Since I wrote that section, I feel entitled to explain it. In Nethack, you can gain resistances (they're all binary) from eating appropriate corpses: fire giants for rF, bees for rP etc. The intrinsics gained in this way are permanent. This means there is an early mini-game of acquiring the relevant resistances from corpses, after that you don't have to bother about them anymore at all. That's the broken aspect. It should be said that a few resistances cannot be gained in this way, and all the Nethack forks I am aware of attack the problem (in different ways). So this I what I meant and what cannot be regarded as good design, in my opinion. You can see that the original creator of Crawl already learned this lesson from Nethack: some resistances are not binary anymore (rF, rC); there are various ways of getting a resistance (item, mutation, form, nowadays potion of resistance) but each of them comes with a price (using the item slot, can lose the mutation etc.); resistances do not work fully reliably.

So when I propose a consumable (here a food item that rots away) which gives temporary poison resistance, I am aware of all these problems. My reply is: the sources are finite (it could be some monster which only spawns in the particular branch); you cannot keep it forever like a potion (you could try to keep some monsters alive for later, but I don't think that makes much sense for rP); if you use the consumable, you're priorities change, as you'll try to gain ground quickly rather than slowly, this usually means using up other resources.

absolutego: I understand the principal objection better than the thematic one: if the resistance is granted by eating Foo, you'd expect that it's gone after the digestive system has done its duties, no?

[absolutego]

it's more along the lines that this mechanic doesn't exist in crawl, and while it could, there's no need to.
if the proposal was about an item similar to royal jellies that cures poisoning instead there would be a debate, but it'd be a different one. this is a can of worms, which i guess is part of why it made the no-no list in the first place.

if you're trying to work around the need for rPois in the lair branches, well, why not remove that instead?

[dpeg]

I don't think I buy the argument that something should be thought about only because the mechanics does not exist in Crawl. Thinking beyond the borders is absolutely crucial, otherwise you'd get new monsters, branches, spells and gods that look like more of the same.

The point is this: different branches require (i.e. become much easier with) certain things. You can do Snake without rP, but it is much easier with it. Since we are thinking about the rune lock, forcing players to actually finish a branch earlyish, we might want to throw them some bones regarding these requisites. Having rP from a ring or an artefact will be much better than the less reliable replacements (swamp dragon, consumable). But you can try to go with these -- this is another option in some case. Think of it like the Lugonu option in the Abyss: you're best off if you don't need it all, but it's still there, and if you come by an altar, you can contemplate whether to use it. These things make the game easier for the player because they provide a valid option, but they also make it more complex, because you have to assess the chance that it works, what it costs etc.

I'll stop explaining here. I have a long history of trying to explain design and concepts to skeptical players. This led me to drop my SA account and to quit using the ##crawl channel. I don't want the same to happen here.

[absolutego]

c'mon, it's a civil debate. and you can just ignore me, you know.

i like meaningful choices (the lugonu example works, and i say that from experience), but this feels like it'd be in practice a patch to work around the need for rPois in too many lair branches on the rune lock scenario. if it's the seed to larger things, well, i'd sure like to hear about that.

i was mostly replying to galehar's proposal of a snake/spider & shoals/swamp lottery, in that i'd rather do snake than swamp without rPois. if the lottery was snake/swamp & shoals/something else it could work: if you don't want to brace the former without rPois you can do the latter, which may be even harder (as shoals currently is). but that's high-level redesign, i know.

[KoboldLord]

You don't really need rPois for Snake. It obviously helps a lot, but MOAR DAKKA does the job just fine in an especially pro-active manner, and as long as you don't aggro the whole level at once you can usually retreat and rest off active poison. Naga you can outrun, and big snakes you can shoot before they bite you. As for the cases that are exceptions to the "usually", well, that's what healing potions are for. You'll burn quite a lot of permafood if you don't have poison resistance, of course, but it isn't like a typical game will run into a shortage.

[dpeg]

absolutego: the thing is that I spend many words trying to explain something, only to hear "nah, won't work". I had this so often, you wouldn't believe it. (The redesigns of Elyvilon and Trog, the concept of Ashenzari, the very idea of portal vaults, a species without healing, self-changing labyrinths, everytime we removed something substantial and whatnot. Radical concepts can go wrong -- but if we don't even try, then design stays timid and lackluster.)

[Thasero]

I would be skeptical of the temporary-resistance-from-food, not because the idea of gaining a temporary bonus from food is necessarily wrong, but because it seems like a pretty odd way to solve the problem. It reminds me of the old saying about "yak shaving" in programming.

Let's back up and remind ourselves of how we got to "maybe there should be a special monster that drops special meat that gives temporary poison resistance."

1. Players frequently skip the last levels of Snake and Swamp because the risk/reward ratio is low compared to other available parts of the dungeon. This is viewed as undesirable.
2. Increasing the reward for these levels is undesirable because of the risk of loot inflation; the player already tends to end up very rich by the endgame.
3. One possible solution is to prevent the player from entering other parts of the dungeon until they have cleared at least one of the oft-unvisited branch ends.
4. However, the way the branch ends are generated gives a high probability the player will be forced to clear a branch that is very difficult without resist poison.
5. One way to mitigate this is to provide a source of poison resistance which is closely associated with the branches you must clear.
6. One form of such poison resistance would be temporary poison resistance granted by eating chunks from a monster located in one of the poison-heavy branches.

I think there are a lot of ways to fix the problem which seem more 'natural' then a new category of meat chunks - there are plenty of places to attack the problem. In reverse order:

6. Create a different form of poison resistance, such as a 'potion of poison resistance' that is like a potion of resistance but more limited, or add a creepy-crawl themed god who grants poison resistance as a worship benefit and will have an altar located near the Lair sub-branch entrances.
5. Rather than increasing the chance that the player will have poison resistance, decrease the chance that the player will need it. Nerf the poisonous monsters, making them weaker or less numerous in the branch, and possibly replace them with other non-poisonous monsters that make the branches more varied. Remember that one of the reasons players avoid these branches is because the difficulty of the branch end is higher than the difficulty of other available floors.
4. Rather than altering the need/demand for rP in these branches, alter which branches appear. Mix up the Lair with other branches that don't require rP. Ex: "The Eyrie" contains lots of birds and you will want strong ranged attacks and/or levitation to reach them. When you think about restricting the player to playing Lair sub-branches, you notice that there's a good chance they'll be fighting against lots of poison. Maybe the Lair should be more thematically diverse.
3. Rather than changing the Lair, loosen the level of the restrictions, or allow non-Lair branches to also substitute.
2. Rather than restricting the player, change the cost/benefit ratio of the branch ends. If it would be bad to increase the amount of loot found at the end of these branches, then rather than adding loot wholesale, subtract loot from commonly-visited branches and add them to the under loved branches. Alternately, accept that the reason players don't visit the branches is that it's a bad deal, and just make the branches easier so that it hurts less. Note that part of the analysis is that players visit other levels first because it is easier to win the game if you choose other levels first. Conversely, that may mean the Snake Pit/Swamp are too hard for their relative location in the dungeon.
1. Maybe we don't have a super-great plan for getting players to go to these branch ends right now, so just let it slide until something can be done as part of a bigger reshuffle. Worry about the levels the players DO see - if players miss something, that may be aesthetically displeasing to the designers, but it doesn't ruin anyone's experience.

If I look at the possible solutions, I don't really find myself thinking, "the common thread here is that we need a new category of inventory object for players to eat".

I do think there are some downsides to adding meat-chunks-of-temporary-rP:

First, the general rule for meat is that it either has no effects, or has increasingly bad effects, depending on the monster. This would add a new kind of meat that breaks this rule. Inventory-based consumables that confer a temporary effect are usually potions, not meat. Consistency makes the game easier to understand for new players.

Second, adding a temporary consumable bonus effect to the food system has side-effects. This makes the Amulet of the Gourmand and carnivore mutations more useful. Being able to eat chunks at any times is already a pretty good bonus - does it need to be better?

Third, it would be difficult to balance. In order to work, the timer of the effect would have to be pretty long. If it's too long, the player won't really need to care about gathering chunks-of-rP. If it's too short, the purpose is defeated.

Fourth, it requires that a special kind of monster be created that helps you with the poison resistance problem... and which un-spoiled players will not know about. Remember that you typically don't get the chance to chop up monsters unless you have actually won the fight!

If an un-spoiled player pokes their head into Snake Pit, gets hammered by a bunch of poison spits or snake bites, and runs for the stairs, what will they think? "I should avoid this until I have the appropriate defenses" is the current, correct answer. The chunk-of-rP solutions means that instead we are hoping players will think, "I should ignore the apparently-bad tactical situation I am diving into and just try to kill every monster, because Crawl is a game where killing every monster you find is encouraged." Then eventually they will kill the monster that gives them temporary poison resistance. The player is reassured that the solution to dangerous monsters is to fight harder, and the next time they are banished to the Abyss, they attempt to kill every enemy they meet in hopes of finding the one that will turn into chunks of meat that grant a temporary bonus to Resist Demons.

[dpeg]

Thasero: Your analysis is very good, and I'll reply because you put so much thought into it.

First, there are very often a great number of solutions. I mentioned one possible idea, nothing more. No idea why everyone is going ballistic, but as I indicated, I've seen this before and I am more inclined to ignore it.

Second, it isn't your fault, but I think you underestimate the work to set up a branch. It's massive. A lot more than making a portal vault, a god or a species. Snake is working very well, I'd contemplate many things before changing its monster set.

Regarding your list of objections, I read it but I looks very familiar to me. You spent the effort, so here are some replies: For quite some time, we want more functional food. (Ambrosia is one example.) We already have food that mutates, so an effect is not exactly a revolutionary idea. There is an advanced plan for a food reform, which among other things makes Gourmand less trivial. I don't believe the bit on balance. Temporary/consumable effects are easy to balance; permanent effects and spells are problematic. We have introduced the potion of resistance at some point, and duration is an issue, but just a number possibly to tweak at some point (this happened for resistance, too, I believe). Regarding spoilers: the same applies to the dragon armour trick; we can communicate one to the player, hence also any other method.

As I said, it is just one particular idea, which may never materialise. I don't know why prospective changes sometimes draw an insane amount of attention, but I believe that this idea could be decently fleshed out. As could be a great number of other ideas, but someone else has to develop them.

[Thasero]

Oh no, I realize that more branches would be drastically more work. It's just that if the problem is "we don't like certain branches" then "maybe the contents or order of the branches should be different" seems like the first solution. If the solution causes certain problems - like resist poison becoming more important - my first instinct would be to say "let's go back to the drawing board on the branch changes" rather than "let's add a new mechanic for getting resistances and tie it to the branch changes".

I think the reason this change meets resistance is that it sounds very much like a special case, or a temporary fix. I'm imagining this conversation in the future:

Stone Soup maintainer: "Why do we have this one dungeon branch with special monsters that drop special meat? It doesn't fit in."

Player A: "Well, it doesn't make sense now, in the year 2037 when cars fly through the air and Stone Soup is beamed directly to our cyborg-brains, but it used to be the meat was the only way to make sure players had resist poison."

Stone Soup maintainer: "But why treat poison resistance differently? Meat doesn't give fire resistance or cold resistance or any other consistently useful equipment-like bonus - except for being food, of course."

Player B: "Back in 0.10 the branches were changed so you might have no choice but to go through the Lair or the Swamp in order to win. Lacking poison resistance could make or break your character."

Stone Soup maintainer: "OK, but that's not true anymore. After all, dungeon branches are a big part of the game, and every version makes large or small tweaks to how they work and what's inside them. Dungeon branches are so central to the game that it would be very hard to write an update that DIDN'T affect them in some way. Having an extra way to get rP might have been really important in 0.10, but it didn't stay as important for 0.11, 0.12, 0.13, and all of the other versions that have come afterwards. I'm going to remove this feature, because pruning features that used to make sense but have lost relevance is an important development task."

Player A: "WHAT!?!? You're going to make a change to the game? No!"
Player B: "I hate change!"
Player C: "This proposed change to Stone Soup is the worst thing to happen since an alien battleship destroyed humanity's first colony on Mars, igniting the Interplanetary War of 2031."
Player D: "Let's all go on the forum and complain about this change until the maintainer promises to change it back!"

etc., etc., the maintainer you save may be yourself, etc.

[jpeg]

First, the general rule for meat is that it either has no effects, or has increasingly bad effects, depending on the monster. This would add a new kind of meat that breaks this rule. Inventory-based consumables that confer a temporary effect are usually potions, not meat. Consistency makes the game easier to understand for new players.

Second, adding a temporary consumable bonus effect to the food system has side-effects. This makes the Amulet of the Gourmand and carnivore mutations more useful. Being able to eat chunks at any times is already a pretty good bonus - does it need to be better?

Third, it would be difficult to balance. In order to work, the timer of the effect would have to be pretty long. If it's too long, the player won't really need to care about gathering chunks-of-rP. If it's too short, the purpose is defeated.

Fourth, it requires that a special kind of monster be created that helps you with the poison resistance problem... and which un-spoiled players will not know about. Remember that you typically don't get the chance to chop up monsters unless you have actually won the fight!

Hmmm... actually, there's one way it could work (maybe that's what is meant by the proposal, but I've never seen the details): it would be possible to add a new type of monster native in Snake (and possibly spider) that is neither a snake nor naga and whose presence in the branch can be explained by having attained a poison resistance that is so strong that you can even temporarily become resistant yourself by dining on their corpses (which would be mentioned in the description). If the new monster poses a different kind of challenge, this could even make the branch more interesting, but it does require a lot of thought.

In my opinion, said monster absolutely may not be poisonous itself nor be closely related to poisonous monsters (which would also rule out e.g. spider eggs) because this seems likely to cause confusion about why some poisonous monsters leave corpses that are poisonous themselves and others that grant poison resistance.

For other resistances, players could be tempted to keep some of those monsters around for later, but Lair is such an early branch and poison is soon replaced by other threats, that this seems a bit unlikely. The point about Gourmand Thasero makes above is a much more important one: do we want players having to dance off their satiation so as to be able to consume a poison resistance conferring chunk when they need it? Weighing the two options, it appears to me that allowing the player to eat the item at any time they need it (permafood, potion) ranks a lot higher than having the item perishable (chunk). If it doesn't decay with time, the number would either have to be rather low or the duration would have to be rather short, so you don't end up amassing large stacks of them. Ideally, there would be just enough to last you through the entire branch if you occasionally opt not to use the item to save it for the later levels.

At the same time, we already do have one type of potion that gets stale and disappears with time, so another potion sharing that capacity would not be out of the question. It could lose its power after some time and then become useless or harmful.

It's hard to think of ways around the poison heavy branches that don't involve poison resistance. The usual means of ranged combat, speed, evasion, avoiding combat, etc. apply, but of course this only covers part of the character builds, and braving Snake with a fighter that can take a lot of hits but constantly has to run off to regenerate doesn't sound like it would be a lot of fun. It's entirely possible to clear a poisonous branch this way, but it's tedious and especially dangerous on the deeper levels. Keep in mind that the point of the entire plan is for the players to gather a rune, so we're not talking black mambas here, but naga mages.

[galehar]

The real problem with meat is that you have to be hungry to eat it which leads to tedious hunger management to be able to gain the rP from it. That's why I like Galefury's idea of giving spiders a chance to drop eggs when you butcher them. Eggs rot like chunks but can be eaten when satiated.
And we're not doing some rhetorical long term design here. We're planning features for 0.10 and Spiders is almost finished.
There's also the idea of making rP give only 90% resistance instead of immunity. This would reduce the difference between having rP and not having it.

[Jeremiah]

This is possibly a stupid idea, but how about giving the player a choice of which branches to visit, instead of it being determined by the RNG - ie entrances to Swamp, Snake, Shoals and Spider Nest are all generated, but the player can only enter two. (The other portals disappear when the player has entered two.)

The player could then make a choice between risk and reward - ie enter an easy branch for the best chance to get the rune, or enter a harder branch for more loot as well as the rune.

At present, with being able to delay the branch ends as long as you like, it would probably always be a no-brainer to go for more loot, because you could wait until you were powerful enough to clear the branch anyway, but with something like galehar's proposed lock, then the choice would become more relevant.

It would also perhaps make the rPois issue less important, as, eg if you have no rPois, but have a way to deal with lots of merfolk and water, you could go to Shoals instead of any of the other branches.

Maybe this wouldn't work yet... but might at some hypothetical future time when there is a choice of more branches to enter with more different requirements and challenges.

[Galefury]

I love the branch roulette. It is awesome and there should be a lot more of it. Not just in lair, but everywhere. Alternatives to elf and crypt would be sweet. The lair roulette provides randomness on a strategic level, and it sucks that this property of crawl is lost after lair. Something that could be super interesting is a two out of three mechanic for lair, vault and a new branch (one shallow, one deep), including all their sub-branches and scaling those branches to depth. Or simply switching around lair and vaults sometimes with proper depth scaling (Tomb and Slime could stay as they are, two versions of other sub-branches). Of course this is not going to happen because it would be an incredible amount of work, requiring two versions of all midgame branches and numerous new enemies. But I think it could provide some very interesting gameplay.

Regarding special poison resistance monsters and spider eggs: a specific monster with high likelihood of providing poison resistance invites delaying the kill if your resistance is still up, which is annoying. If most monsters can provide resistance, but with low probability or fairly short duration (IMO better because it allows tactical use) this problem is avoided. But because most monsters are poisonous it also means you have to kill poisonous monsters to get poison resistance. I don't see a way out of the dilemma. Whatever mechanic is used (if any), I think it should be introduced before the branch it is used in, for example in lair as a rare monster, or in the entry vault of the branch. This reduces need for spoilers.

In any case I think chunks should not be used due to hunger management and gourmand issues. Some sort of rotting permafood or potion is much better. Antivenom glands, spider eggs, simply an unexplained potion of poison resistance, anything but chunks. This also helps distinguish it from Nethack intrinsics a little.

[dpeg]

Jeremiah: This, of course, won't happen. It is the prototypical demand for variance reduction (other, as prototypical demands are: "I want to choose my draconian's colour!", "Why are the demonspawn mutations random? This sucks!"). The point is that at the start of the game, you don't know yet what branches will await you. This requires some flexibility, something we like. If the player could choose, then (a) we have to spend tears and blood in order to make the branches exactly equal in danger and reward and (b) nonetheless we'd soon see that despite offering four branches, almost all players only visit two of those.

galehar: I like the eggs. (Off-topic: It occurred to me that Spider should have a new trap type, the web. I can come up with details.)

galefury: Hey, "it sucks that there's no branch roulette after Lair" We have just started with this idea in DCSS 0.5 (or so). As you know, it takes time to set up alternative branches. Once there are ideas, there can be alternative branches to anything (bar Zot, I guess).

[absolutego]

spider eggs as some sort of vaccine: gross even by crawl standards.
they could grant shot-duration, partial resistance, increasing the more you eat (between predefined levels if you find high granularity annoying). that way you could manage rP beyond the binary regimen, whether it's toned down to 90% resistance or not. it wouldn't dilute the threat and it'd be harder to paint yourself into a corner, which is the danger here (easy with rP, too hard without). it could also be annoying to micromanage, of course. just thinking out loud.

[dpeg]

I recall the fact that Crawl's royal jelly food restores all drained stats.

[mumra]

galehar: I like the eggs. (Off-topic: It occurred to me that Spider should have a new trap type, the web. I can come up with details.)

There's already a discussion on the wiki (I had some of my own ideas about webs as an environmental obstacle).

Perhaps the proposed chunks could simply cure poison rather than grant immunity, this would be more in line with consumables such as health pots, rj, etc., and provide a new way of curing poison since the spell was removed. I find health potions quite boring as a generic cure-all for near enough anything, maybe spider should feature some poison types that can't be simply cured with health pots (or avoided with rPois) and therefore require use of the chunks / eggs.

[galehar]

provide a new way of curing poison since the spell was removed.

It's resist poison which have been removed. Cure poison is still there, it's even in the VM book.

[Yet Another Stupid Noob]

So if one aspect of the problem is that poison resistance is very useful (arguably nearly criticical at certain times/branches) and then less so as the game progresses, why not add potions of cure poison.

Potion: Removes all existing poison and makes the drinker immune to poison for a short while. I think that 20-30 turns (actions) is about right. Enough time to take care of the immediate threat and not much more.

This potion has similar utility as healing potions (and should take some of those drop "slots") and remains useful for about as long. Once past orc/lair, the amount of poison being slung around (both offensively and defensively) drops noticably. This potion acts to make managable those few encounters where rpois changes the battle from "deadly" to "trivial" (and I'm looking at you transmuter ghosts). You use these potions like a wand; to act as a tempory bridge over a difficult encounter.

You still get utility from a rpois item, since you won't chug a potion for every giant ant or kobold, and eating poisonous chunks can be huge.

[jpeg]

So if one aspect of the problem is that poison resistance is very useful (arguably nearly criticical at certain times/branches) and then less so as the game progresses, why not add potions of cure poison.

Potion: Removes all existing poison and makes the drinker immune to poison for a short while. I think that 20-30 turns (actions) is about right. Enough time to take care of the immediate threat and not much more.

This potion has similar utility as healing potions (and should take some of those drop "slots") and remains useful for about as long. Once past orc/lair, the amount of poison being slung around (both offensively and defensively) drops noticably. This potion acts to make managable those few encounters where rpois changes the battle from "deadly" to "trivial" (and I'm looking at you transmuter ghosts). You use these potions like a wand; to act as a tempory bridge over a difficult encounter.

I don't think such a potion would be an interesting addition to the game. Currently, the player has to face the choice between spending a potion of healing to cure poison or to try to sit it out and save the potion for hp restoration at some later point. Of course, the potion of healing only cures a comparatively small amount of hit points but at the time where poison poses a large problem this amount is still highly relevant. It actually gets worse if the proposed potion both cures poison and confers temporary poison resistance because that reduces the usual trade-off to the normal "use now or save until much later" decision that applies to all potions. If an item only cures the effect (as healing does) it's in the player's interest to delay using it as much as possible, at the very least until the threat is eliminated, while at the same time doing so will make you rapidly lose hit points; conversely, an item that only provides in-advance resistance (such as the potion of resistance) needs to be used before the effect actually comes up. In both cases, the player has to weigh the chances and make a decision based on the situation.

I could see the addition of an early-game, more common version of the potion of resistance that only applies to poison. It would lose its usefulness at some point (but so do a number of wands), so there'd be some incentive for the player to use them early on. However, as long as they're not too common, using them wouldn't be a no-brainer either. For example, players might want to save them for the Lair and its poisonous sub branches, or the Hive (especially once the planned overhaul goes through).