Give a paranoid confirmation prompt when the player is levitating or flying over a liquid and attempts to remove an item that would result in them losing that levitation or flying and falling in.

A game mode that allows you to keep playing after death. There are two ways this could go:

• Continue playing as the monster who killed you. This game mode would likely be non-scoring, and your actual character’s death would be written out in the normal manner.
• Continue playing as your own corpse. In this mode, you are effectively a blind, deaf, and sessile monster who can take no actions other than resting. The game finally ends when your corpse rots away. This is mainly interesting combined with an implementation where allies can zap your corpse with undead turning to resurrect you; if they do so before you rot away, then you can keep playing. It’s not clear what would happen if you are petrified instead; a statue can survive indefinitely, and the odds of a pet being able to stone-to-flesh your statue are low but not zero, so possibly the criteria should instead be “you die when there are no more pets on the level”. This still has a problem in that some pets could survive indefinitely, though.

Remove the recharge counter from objects. Most tools can be infinitely recharged, so they don’t need one. For wands, replace the recharge count with a chance of a wand becoming eroded when the wand is recharged, or of exploding if recharged when already eroded.

Not specified what happens for the few items that don’t follow this pattern, such as magic markers and the wand of wishing. For the wand, it will likely work to make the erosion guaranteed the first time, and the explosion guaranteed when it’s already eroded. Magic markers could possibly just be un-rechargeable.

New conduct, permaconflict, which must be enabled by setting an option. If you play the game with this, you constantly generate conflict, which likely makes some aspects of the game much easier and some much more difficult.

Not specified whether the permanent conflict would cause extra hunger like the ring does, or not cause any like artifacts do.

Zapping a wand of polymorph at Master skill gives you control over the form the target monsters polymorph into, whether that is yourself or others. Other monsters still get their chance to resist the polymorph entirely.

It would not give you control over what an object would polymorph into, for obvious reasons.

However, if there are monsters that can use the wand of polymorph at master skill, this has the potential to become very annoying very fast, since they would be able to turn the player into a sessile monster, make them break their armor, or turn themselves into something very threatening and out of depth.

When you unpolymorph, you temporarily keep any intrinsics you had from that form and which you no longer have. How long it is before these intrinsics time out isn’t specified, but probably contains some random component so you can’t plan around when the intrinsic will lapse.

This was suggested for a hypothetical druid role, but noted that it doesn’t have to be a role-specific mechanic and could apply to other roles as well.

Have a small visual element on the screen like a border or a 2x2 area in the corner. When NetHack completes a screen redraw or is otherwise ready for input, update this element with a symbol crawling around the border or bouncing around the corner to provide a visual indication to the player. This is intended for laggy connections, so you can tell whether your input was processed or not.

Aligned priests can use turn undead against undead monsters they want to attack. “Want to attack” is not currently well-defined for priests vs undead in NetHack, but it would probably follow these rules:

• High priests don’t turn undead, for simplicity.
• Priests of Moloch turn the player if they are polymorphed into an undead, but don’t turn other undead monsters.
• Other priests turn undead that are near them, whether that is the player or not. This opens up an interesting strategy where you can flee from undead into a temple and the priest will help you out.

Allow some Gehennom filler levels to be non-persistent in a limited manner. They will generate new maps and terrain upon being re-entered, but will retain all the monsters and objects they had, which will be placed in random spots. Since this compensates for the mysterious force by delaying the ascension run and offering more time for harassment and monster spawns, remove the mysterious force as well.

Add some way for the player to spawn a sleeping, non-mindless monster on the magic portal in the Elemental Planes. This provides an alternate method of portal detection, and additionally allows the player to track the moving location of the portal on the Plane of Water.

Conflict should cause monsters to prioritize moving toward and attacking targets they would attack anyway (like the player) over targets they wouldn’t (like other monsters), instead of it being fully random.

The purpose of this is to maintain the offensive use of conflict (monsters deal damage to each other) while nerfing the defensive use of conflict (monsters are too busy attacking each other to worry about or even approach the player), and so make conflict overall not quite as powerful for trivializing tough areas of the game like the Sanctum and Astral.

Remove the ring of levitation. This is because it is too good at solving interesting problems from the player: for the low cost of a ring slot used only situationally, the player can bypass water, lava, most traps, and wield cockatrices with almost total immunity. It can then be taken off when not wanted and put back on whenever situationally needed again.

As an alternative to removing the ring, add some cost to using it, such as requiring them to be invoked to get levitation, subject to a cooldown to use them again, although that is strange and unprecedented for non-artifact items.

The boots of levitation have more or less the same problems as the ring, only slightly less so because they are heavier and swapping out boots takes a little longer. These could also be removed, and replaced with an artifact pair of boots that confers levitation, themed after the talaria winged boots Perseus borrowed from Hermes. (This also has some potential for strangeness, because Perseus’s statue appears in the game and sometimes contains levitation boots; would the boots always be this artifact if it doesn’t exist already?)

Orctown generating should force a temple to be generated somewhere else in the main dungeon, possibly around level 15. Without this, there is a possibility the player’s first opportunity to buy divine protection will be late in the game at the Valley of the Dead.

A ramification of this change is that the game would have to know it is generating Orctown at the beginning of the game, rather than deciding it when Minetown generates by picking among its variants randomly.

An uncursed scroll of enchant armor that targets a +0 (or below?) piece of armor will raise its enchantment the same amount as it would if it were blessed (1 to 3 points).

The purpose of this is to incentivize players using them earlier in the game, and therefore smoothing out the gain in AC over time, rather than hoarding them and only ever using them when blessed and on ascension kit items, which tends to mean a huge gain in AC all at once. This isn’t intended to make using them earlier the clearly best strategy, just to make it a more balanced choice.

If the player is about to be hit with a death ray and has not yet expended their saving grace, the saving grace protects them from the instadeath. Probably by converting the ray hit (and any reflections from the same ray?) into a miss; otherwise the game would have to explain why a non-resistant character got hit by a death ray and failed to die.

Mirrors carried openly in your inventory act as a weak source of reflection: if you have no other source of reflection, they will block an incoming ray (possibly only a death ray), at the expense of breaking and costing you 2 points of luck as breaking a mirror usually does, except if your luck is already negative it won’t be penalized further. Arguably it shouldn’t apply a luck penalty at all unless the player caused the ray; after all, you aren’t punished for a monster zapping a wand of striking and breaking a mirror on the ground.

This would allow better protection against random attack wands early in the game, when mirrors can be found but a permanent reflection source is rare.

The potion of wonder gives both a positive and negative intrinsic when quaffed, and both a positive and negative object property when something is dipped into it.

Reduce the ability of holy water to bless large objects and stacks, by telling the player “Your [item(s)] is too large to dip into one bottle”. From there, it could either prompt “Use X bottles instead?”, tell you you don’t currently have enough potions, or suggest you apply multiple potions instead of dipping.

Small and single objects would still be blessable by using one potion.

HP recovery is faster when your HP is at a lower percentage than it is at a higher percentage (i.e. you heal fast, but it gradually slows as you heal). This would mean it’s an advantage in terms of turns spent to rest only when injured.

There should be some way to enter the Valley of the Dead without requiring fighting through or looping around the Castle, in order to allow a player struggling to get up to the Quest experience level threshold to access the resources in the Valley.

Add a “randomizer mode”, a non-scoring mode enabled by some option you put in the config file that randomly mutates the stats of each monster – for instance, in one particular game, gnomes could have fewer hit dice and thus lower HP, while soldier ants gain an ice attack.

In the spirit of traditional randomizers, this could also randomize loot, for things like artifacts and invocation items being mixed up and findable in regular chests, monsters generating with and dropping gear they shouldn’t have, etc.

Amulet of full healing, which heals you completely (or as much as an uncursed potion of full healing) when your HP drops to 0, but does not stave off non-HP-based forms of death.

Dipping ammo in various potions can endow it with a permanent object property related to the potion.

Change the HP and/or Pw regeneration formulas (natural regeneration, not from external sources) so that it regenerates faster in the early game but more slowly in the late game.

This fixes existing balance problems – that the hero doesn’t regain them fast enough early on, and by later on they regain them so fast it trivializes problems – but would be very strange to justify flavor wise.

Magic markers are made more common of a tool, but they hold fewer charges to compensate.

A new take on a Bard role, generally themed around leading a party of other adventurers through the dungeon, with some other additional ideas:

• The bard and healer cannot both be focused around petmongering, so one of them would have to change. Most likely the healer, as the bard’s traditional abilities are only good on pets.
• Instead of the standard little dog or kitten, the bard starts with a “party” of 3-4 player monsters, who are randomly chosen from other roles.
  • This would make the primary strategy focused on having the party do the fighting for you, letting them level up, providing them with weapons and equipment, and not letting them die.
  • Healers and archeologists, and possibly tourists, don’t make very good additions to the party, and so should perhaps be ineligible.
  • Party roles could be randomly chosen from distinct sets of “strong melee fighter” (Bar or Cav or Kni or Val), “ranged combat” (Ran or Rog or Sam or possibly Tou), and “spellcaster/cleric” (possibly Hea or Mon or Pri or Wiz). These could also just be hardcoded.
  • The party will not start with the full starting inventory of their respective roles. They will generally have the same weapons and armor, but will probably not have any of the scrolls, rings, wands they normally start with, and starting spellbooks should be either eliminated or unrandomized. In other words, you shouldn’t be incentivized to slaughter your allies for their stuff.
• Healer strategy should shift away from petmongering - the easiest way to do this is change the protection formula in such a way that the protection racket doesn’t exist; so that the player isn’t incentivized to keep XL as low as possible until they can buy protection.
• Difficult for Healers to find a new niche. (Avoidance of combat is an option, but dtsund’s Class Overhaul Proposal calls for Archeologists to fill that niche). Perhaps something that focuses on pacifying monsters by healing them, but this has issues: you shouldn’t be able to ‘‘tame’’ by healing since that turns Healers back in the direction of acquiring lots of pets, and most monsters start at full HP and don’t need any healing.
• Bards (and probably all players) should be able to equip their pets by #looting them, or a similar mechanism. (This would also apply to mounts.) Intelligent monsters will prefer player-given items above all others and will never willingly replace them for other gear. Possibly make this behavior dependent on options, because some players may want their pets to be pragmatic in their gear choices.
• Player monster pets should perhaps not be able to detect curses.
• Songbooks, + class items which contain a song that the bard can learn.
  • Balancing the songbook generation rate against spellbooks might be tricky, because although bards can still cast spells and other roles can play songs, each is not very likely to have that much use for the other set of books. Hardcoding different probabilities based on whether the character is a bard or not doesn’t seem very clean.
  • Possibly the bard should be able to discover songs by trial and error, although simple brute forcing is definitely not the way to do it. Maybe all spellbooks start identified.
  • The bard could need to practice the song by using the songbook in order to restore “song memory”. Song memory does not work quite like spell memory, it works essentially like direct success rate. There might be some mandatory delay between practices, meaning you can’t take an unlearned song and practice it up to 100% immediately. Or there could be a possible failure of practice, maybe a dexterity penalty (“Dumb move! You strain your fingers.”) to serve to prevent practice-spam. Playing a song would serve to
  • Songbooks do not eventually go blank, and can be used for practice indefinitely.
  • Or to keep it realistic with learning actual songs, you need to have the songbook on hand early on when you want to play it, but as you play it more and gain practice, you can eventually play the song without it anymore.
• The bard should possibly be able to sing some of their songs without requiring an instrument.

See also: Bard implementations.

Charm monster is generally considered too powerful and too low-level, and should probably be nerfed. Ideas:

• It only works if you’re not wearing any armor. (This is apparently how D&D does it.) Alternatively, its chance of successfully charming is reduced the more armor you’re wearing.
• Make it more reliable than it is now, but it always fails if you have other pets. (You can then only have one pet at a time without using scrolls of taming.)
• It only tames one monster at a time. (Make it a directional beam with a maximum range of 1.)
• It only pacifies monsters, and can’t be used to tame an always-hostile monster.
• Make it do what the temporary pet code in the Bard patch does: the monster remains tame for a fairly short amount of time and then reverts to whatever it was originally.
• Split into three spells: pacify monster (peaceful for charisma*2 turns), charm monster (tame for Charisma*2 turns), and dominate monster (tame permanently or for a long time).
• It only works on monsters, and doesn’t work on intelligent beings.
• It can’t directly tame a hostile monster; it can only pacify it. Peaceful monsters can then be tamed by subsequent casts, but attempting to tame a peaceful intelligent monster may anger it instead. The pacification step is dependent only on monster MR, but the taming depends on the player’s level relative to the target. The scroll of taming should remain comparatively powerful.
• Scale with skill: unskilled attempts to pacify one adjacent monster, basic to pacify all adjacent, skilled to tame one adjacent, and expert to tame all adjacent. All of these are subject to normal monster MR checks.
• Whatever nerfs do get applied, they should ensure that the scroll of taming is more powerful and reliable than the spell.

Add a Druid role: intended to be a more balanced form of SLASH’EM’s doppelganger race, druids are highly attuned to nature and possess innate shapeshifting abilities.

• You begin the game with polymorph control and get polymorphitis at XL 5 or so.
• Possibly you start with a wand of polymorph, or potions of it.
• Inventory could be severely restricted, so you have to figure out how to do most things through polymorph.
• You have the power to change into a tree, which will make most monsters ignore you. While in tree form, you are immobile but regenerate HP and Pw faster. Orcs might attack you with axes, a la Tolkien.
• Without extrinsic polymorph control, your polymorph control limits you to changing into forms with the M1_ANIMAL flag.
• Possibly have limited access to #polyself outside of wizard mode, so they can change intentionally and not randomly. Would work best as an ability with a timeout, or cost Pw outside of wizard mode.
• Possibly, the monsters you can polymorph into must have a certain base level or difficulty that’s tied to your XL somehow (maybe XL/2). If their base level is too high, you can’t polymorph into them.
• Your controlled polymorphs always succeed - you will never accidentally fail and “feel like a new woman” with the 20% chance that all other roles have.
• You are seriously bad at combat in your regular form, having very little weapon skill (nothing can be advanced to Expert, possibly not even to Skilled), and physically weak in your normal form. Combat in a polyform should be incentivized enough so that it isn’t really worth it to wear any armor. Maybe you should get to-hit and damage bonuses while in a wild form?
• Diminishing returns on polymorph time limit that prevent you from reusing the same form(s) over and over and over again. (This may actually be applicable as a general YANI for polymorph control).
• The critical balance needed is to make polyselfing powerful and awesome to play with and use for typical combat, but simultaneously polyselfing needs enough restrictions that it doesn’t just turn into Master Mind Flayer: The Game.
  • One proposed restriction: you can only turn into monsters you have already encountered, or if that is too lax, into peaceful or tame monsters adjacent to you at the time. The druid should be able to pacify most animalistic monsters to take advantage of this (temporarily?)
• The starting pet is a woodchuck (or wolf). Pantheon is probably drawn from Celtic mythology, though hopefully not overlapping with Knights.
• Some heavy polymorph / monster / player as monster tweaking is probably required, in the sense that there should, ideally, be valid reasons to exist in any given polymorphable form.
• You should keep the intrinsics from a polyform after you no longer have that form, but only temporarily, so they time out eventually.

Falling rock traps can dump multiple rocks onto you at once, up to (depth/4). Damage is taken for each rock. If a trap would drop 4 or more rocks on you, it may instead drop (rocks/4) boulders on your head instead. Helms should be less effective at reducing damage from falling rocks, and almost useless at reducing damage from falling houlders.

If multiple boulders are dropped, they might be dropped on nearby spaces as well, so that a pathway might end up blocked.

Magic traps’ blinding and summoning monsters effects should be split up, since both at once tends to make them much more fatal, especially to early players. Alternatively, make the summoned monster selection with a lower-difficulty pool of monsters.

It’s a problem that utility spells with a high failure rate, such as identify at 95% fail, is a mere inconvenience to players, since they can sit in a closet with a stack of food and wait for their Pw to recharge until they succeed. This is not really a problem with combat spells, since the penalty for failing to cast the spell correctly is a disadvantage in combat, where time and Pw matter. Some possible ways to address it:

• Large penalties for failing to cast a spell, like a 1% memory loss and 50 hunger.
• Failing to cast a spell locks it for a certain period of time. It would be hard to make this not turn into an even slower grind for utility spells, since the lockout period could just be waited out.
• Make spells never actually failable, but Pw cost is increased proportional to the failure rate, specifically: real Pw = base Pw / success rate. Under this formula, a 15 Pw spell at 95% fail becomes a 300 Pw spell at 0% fail. This means the Pw cost should be exposed in the spellcasting menu rather than failure rate, even though failure rate still needs to be calculated. Maybe use different colors in the spellcasting menu to denote spells you can cast now, spells you can cast by waiting to recover more Pw, spells you can’t cast even at your current maximum Pw, and forgotten spells.

Instead of books having spestudied charges which leads to a lot of issues with polypiling, you read them once and they disappear, but you get 3 times the spell memory. This is intended to remove tedious spellbook micromanagement.

• Spellbooks would need to be considerably more common than they are in vanilla. In FIQhack, where this was originally discussed (and later implemented), spellcasters generate with books, and the level of dropped spellbooks scales well as the game progresses since more dangerous monsters carry higher-level books.
• This addresses polypiling by allowing the player to polypile the book as much as they want, but they can only read it once. (Does not fix the problem of polypiling a book multiple times and identifying but not reading the results – perhaps an increased chance of going blank is in order.)
• Slightly makes things tougher for people who finish their games in under 60,000 turns, since they can’t use a spell for free confusion. Which may actually be a good thing.
• Also a problem for people who play very long games, because they will need to find or write replacement spellbooks; however, this was sort of already the case.
• May make polypiling unknown spellbooks too popular, since an unknown spellbook is likely not to be the one you want, or you have price-IDed it and know it’s something you will never use, and you can now poly without consequence. Perhaps address this by giving polymorphed spellbooks an elevated chance to become blank, and blank spellbooks do not poly into anything else.

The wield slot is fairly underused among object classes, and there are a number of interesting bonuses we can add to spells if the corresponding spellbook is wielded.

• Decreases failure rate of the wielded spell (utility spells might need to be made higher-level/harder to cast to maintain balance).
• Or decreases Pw cost of the wielded spell.
• Allows you to cast the wielded spell even if completely forgotten.
• Allows you to cast the spell without knowing it already - the Z menu gets an additional “-“ option to cast from the spellbook in hand if you don’t know what that book is.
  • Casting from an unknown spellbook will immediately check for reading success as if you had read the book: if you succeed, it then moves on to other checks like whether you have enough power, etc and you may cast the spell. If you fail, you get a random failure effect appropriate to that spellbook.

The current resistance system is unsatisfying, because most resistances separate the effects of something into a binary outcome: worse effects with no resistance, totally ignorable effects with it. Most resistances also are get-once-and-be-protected-forever, which is uninteresting in the long term.

Proposed solutions to this generally assume that there will be some distinction between resistance and immunity, which offers better (or total) protection but is not normally obtainable as a permanent intrinsic effect. This also enables partial resistances (as found in SporkHack) to be used, though some ideas still treat them as binary - either you have it or you don’t.

• Extrinsic immunity isn’t as good as inherent immunity.
• Resistance halves HP damage and cuts attribute loss down to 1 point.
• Role benefits should probably only be resistance and not immunity.
• For partial resistance, damage is easy enough to do (express your resistance as a percentage and cut damage by that percentage), but many effects, particularly those protected by MR, are binary, and will be tricky to adapt.
• Resistances could start at 100% when you get them, but decrease by 1% every time it protects you from something.

An Alchemist role, revolving around potions and alchemy, and likely requiring overhauls of several systems.

• The main form of combat is to create phials of potion (where 1 potion splits into many phials), which can be tossed at enemies to cause potion effects on them. The main potions used are things like acid, paralysis, (lit) oil, and confusion.
  • Phials come in all the same types as normal potions but have somewhat weaker effects.
  • You can dip a stack of empty phials into a potion to fill them with that potion (up to some maximum).
  • Thrown phials have the same exact splash damage effect on things as throwing the potion would produce, but drinking it is a small enough dose to only cause the splash damage effects to the drinker. Requires some balance so that splash damage is useful against monsters but not useful for the player to drink for beneficial phials.
  • Phials can be dipped into potions, but nothing can be dipped into phials.
• The quest artifact is the Philosopher’s Stone.
  • The quest leader is Nicolas Flamel and the nemesis is an Avatar of Death (like Death, but weaker).
  • Fun possibilities for transmuting materials here if the object materials patch is in effect.
  • Its base item type could be a ruby (but perhaps not because dnethack makes the Heart of Ahriman a ruby) or a garnet, based on its color in Harry Potter, or an opal.
• Can also create alchemic gizmos: smoke bombs from potions of blindness, firebombs from potions of oil, all lighter and more numerous and useful than the base potions.
• Start with all potions identified.
• Could start with an alchemy kit, which is a rare tool find for other roles. NeroOneTrueKing proposed a set of mechanics for an alchemy kit:
  • Has 3 compartments. The first accepts only potions/phials of polymorph (or perhaps it just has charges and you can recharge it by using potions of polymorph).
  • The other compartments must each only contain items of the same material.
  • A success chance is displayed based on the amount of polymorph potion available, the weight ratio of the two compartments, and the materials themselves (metal can transform into metal relatively easily).
  • A successful use swaps all of the materials of the two compartments.
• Needs mechanics such that in an Alchemist’s hands, no potion is useless.

The invoke effect of the Philosopher’s Stone is heavily debated.

• It creates potions of the Elixir of Life. This potion cures disease and restores lost attributes, but most importantly it grants temporary intrinsic lifesaving. If you die with intrinsic lifesaving, you lose the remaining time for the intrinsic.
• Creating potions is way too powerful since the player will be able to bank them, so nerf this: perhaps you can only get Elixir of Life by dipping the Stone into a potion of full healing, or something. Or Elixir goes bad after a while and reverts to water, making it usable for before a fight but not for stashing.
• The Elixir should not be more powerful than a potion of full healing, or the invoke effect of the Staff of Aesculapius - healers should be the best at actual healing. Probably, getting lifesaved from intrinsic lifesaving will only restore enough HP to stay alive a little longer.
• The Elixir gives a large temporary boost to HP regeneration.
• No Elixir of Life; it instead turns potions of sickness into (extra? full?) healing.
• The Stone can turn metal objects into gold or can turn rocks into gold pieces when rubbed on them.
• When dipped into gain level or gain energy, transforms the potion into polymorph.

A really great Alchemist implementation would probably involve a full alchemy overhaul which adds herbs and fungi, harvesting ingredients from corpses, cooking, and interesting ways to combine everything. The challenge with making this is how to make it useful for other roles and not just the alchemist, and at the same time not overcomplicating the game with the new additions.

Make the identification game less regimented. Currently, a rational spoiled player (who will not take the risks associated with direct use-ID) is restricted to indirect use-ID (and monster use-ID) of most objects until such time as they can find a general, book, or scroll shop, at which point they can price-ID the scroll of identify and other items. Once identify is known (and could be blessed), the good items (as determined by price ID) can be fully identified. This is problematic because it’s not at all a gradual process, and it feels like it should be.

Some ideas have been floated to remove price ID outright (presumably, a shopkeeper would pay the same amount for each item of an object class as they currently do with amulets). This has some advantages; price ID is tedious, encourages stashing to some extent (because stashing near a shop where you can ID new things is good), and roguelike players who don’t primarily play NetHack tend to hate its price ID system. In this case, either the scroll of identify should start out identified for all characters, so that identification doesn’t take forever to get off the ground, or some other mechanic should be added so identify is easy to pick out for spoiled players (such as retaining price-ID for only that scroll).

A designer may not want to get rid of the price identification system entirely. But there are some options for amending it. Overall, it should become less regimented and predictable, and less necessary of a strategy. NetHack’s entire idea of price tiers doesn’t appear to be based on much of anything, and having items organized into tiers makes it easy to disregard entire sets of items once their base prices are known. Ideas include:

• Change price ID so that each item is assigned to a “band” of possible prices. There are four price bands, and each item class has its base price randomized based on its band. The price bands show up on unidentified items, e.g. “an expensive scroll labeled KIRJE”. Some items might fall into more than one price band.
• Implement price tiers (e.g. all scrolls cost either 100, 150, 200, or 300), and items are placed into 1-3 possible tiers at the start of the game.
• Fuzz the prices of items so that it’s not possible to definitively say what an item is, only that it’s cheap or expensive.
  • One way to do this is to scatter the base price of each previously tiered item each game on an interval so that it overlaps with the intervals from other previous tiers, perhaps a random number from 1/2 the original price to 2x the original price.
  • Or fuzzing is done on a per-shopkeeper basis. Each shopkeeper sees each type of object as having a different base cost, based on hashing their monster ID. Asidonhopo sells a scroll of fire with a base cost of 86; Enniscorthy sells it with a base cost of 103.
  • Or, do away with tiers entirely and simply scatter the base costs permanently. A ring of free action might be 240 or 260 base, because it’s on the higher end of usefulness in the former 200 tier. Some fuzzing would probably be needed here to keep things from being unambiguously identified.

In an effort to make use-testing spellbooks less of a terrible idea, all characters should get a warning when reading a spellbook that’s way out of their comfort zone and they have very little chance of successfully studying it.

Make scrolls and spellbooks in particular partially identifiable outside of a shop without having to formally identify them or guess based on frequency (which are currently the ‘‘only’’ ways to identify them outside of a shop). Ideas:

• Higher level spellbooks could be heavier. D&D does this. However, unless the game shows the weights of items to the player, it will be tedious for the player to figure out the exact weight of a book by picking up and dropping items of known weights.
• Make spellbook appearances more complex based on level. A simple color indicates a 1-2 level book, an unusual color or appearance indicates a 3-4 level book, a very odd or ridiculous appearance or material indicates a 5-6 level book, and a completely over-the-top appearance indicates a level 7+ book. (Example: “red”, “steel”, “bone”, “jewel-encrusted”). Shuffling of the random appearances would need to be changed so that the books retain an appearance in their original bracket.
• Make scroll label length (or, more complicated, its number of syllables) roughly correlate to its cost. The correlation could be fuzzed a bit, so MAPIRO MAHAMA DIROMAT is probably a 300 zorkmid scroll, but is certainly no less than 200, and NR 9 is probably something really cheap, but might be 100 zorkmid.
• When you read a spellbook, you are given a menu with three options:
  • Give the book a cursory glance-over. This can fail with low Int/XL but is fairly unlikely and has very minor failure effects. If successful, it identifies the spell level of the book. Takes 1/10 the usual spell study time.
  • Briefly study the book but don’t try to learn its spell. This can fail with mediocre Int/XL but will be reliable at high ones. Moderate failure effects. If successful, it identifies the spell contained in the book. Takes 1/3 the regular spell study time.
  • Study the book normally with normal failure effects and normal spell study time. Learns the spell if successful.

With respect to other proposals about crafting (involving a craftsmen’s guild or even scattered craftsmen), the process of traveling back and forth to and from it might become busywork. So if players could craft things on their own, that might be better.

• Would probably need properly flavored terrain for workstations, like a furnace or smelter, for certain crafts, otherwise flavor is too weird. (But this creates the problem of having to travel back and forth to it again.)
• Fixed, non-simple recipes (e.g. existing scroll crafting is a simple recipe) are a heavy spoiler tax and should be avoided.
• This could be simple for certain types of items, as in requiring some silver or gold and a diamond to craft a diamond ring, but doesn’t explain where any magical effects come from, and also might weirdly tie crafting recipes to randomized appearances which have previously been meaningless.
• Probably requires a few more “base” crafting materials, some source of leather, some source of wood, etc.
  • Corpses could decay into bones, which disappear themselves with a longer timeout, but in the meantime can be used to craft bone stuff.
• One of the craftable items should be saddles, because they’re very hard to find for non-Knight characters who want to ride.
• Tallow candles should be relatively easy to make, because they can be made out of fat from slain animals.

Add a Craftsmen’s Guild to the game, as a method of advanced crafting (unlike the player’s simplified crafting such as alchemy crafting potions into other potions, or combining a blank scroll with a magic marker to make magic scrolls). The guild either has its own dedicated special level, or occupies part of a preexisting special level.

• The guild is populated with master craftsmen who work on the principle of “valuable junk items + very large sums of gold = desirable items”. They can also do some things available to the player, like increasing weapon enchantment, without needing the specific consumables needed for the player to do it.
• Would serve as a replacement or stand-in for the black market. Resolves some of the problems with balancing black market shopkeepers: if you kill the black market staff, you can have every item in the market, so black market staff must be insanely powerful. If you kill a master craftsman, however, you don’t get anything of value. Therefore, master craftsmen don’t have to be insanely powerful, or even good fighters.
• Not as good as wishes; there are some things they cannot make for you, like magic lamps, and many magical tools. You cannot specify enchantment or blessing either. For items for which enchantment doesn’t matter, the material cost will be higher.
• The interface works by #chatting to them. They will ask you what you want to create, and you enter it (using the wish parser logic to extract the proper object class), and then they will tell you the items and gold they will need for it. These could either be randomized or use certain fixed ingredients with some randomization or use completely fixed recipes; if randomized it will be deterministic for that item for that craftsman in that game. You are then prompted to select the items from your inventory.
• For gold and other balance reasons, this would probably work better the farther it is into the game; maybe it could even be found deep into Gehennom. Maybe you even have to liberate it from a demon lord.
• Craftsmen might include:
  • Metalworkers (see below)
  • Wandmakers (produce wands)
  • Scribes (produce scrolls and spellbooks)
  • Alchemists (produce potions)
  • Toolsmiths (produce mundane and some magical tools)
  • Jewelers (produce rings and amulets)
  • Sculptor (produce statues and figurines)
  • Leatherworkers (produce leather items and saddles)
  • Chefs (produce food)
  • Arcanologists (transfer charges/enchantments from one item into another, consumes only gold and not items)

Metalworker notes

Metalworkers are probably the most important sort of craftsman for many characters. There could be multiple types, each specializing in one type of metal, or just one type that does everything.

• Silvering items (or gilding, or copper-coating) should require you to use up junk silver items equal to some amount of weight proportional to the weight of the new item. The zorkmid cost of silvering should be very high.
• A forge room type (probably containing lava and maybe some junk items like iron chains) could be added which has one or more smiths in it (silversmiths, goldsmiths, blacksmiths).
  • Goldsmiths seem rather pointless right now, since there is only one gold item (the ring). They would be more useful if gold equipment existed and conferred some benefit.
  • Blacksmiths don’t have much to do as far as turning items into iron goes, but they could produce iron weapons of a type the player wants, or repair and improve iron weapons and armor (increasing its enchantment).

Cantrips are level 0 spells that cost d2 or d3 power to cast. In order not to let them be unbalancing, they don’t train skill and are mostly useless, except in certain circumstances or for low-Pw spellcasters who can’t do much of the bigger stuff yet. ais523 suggests that good candidates for cantrips might be things that have little combat use, and whose effects could be duplicated by backtracking or other tedious things, but would be useful to avoid boredom. Given their cheapness, they should probably not train skills, and may not even need spell schools.

Most ideas for cantrips seem to be a little too powerful and would do better off as normal spells (and may be listed as independent YANIs for normal spells); however, those that seem like they would fit are listed here.

• Increase the odds of monsters dropping a corpse in the next few turns.
• Create an empty unlocked chest. (This would need some restriction so you can’t farm and sell them; perhaps you can only create up to 10 or 20 chests per game with this.)
• Mark a map square with a symbol, perhaps a comma, that is no different from normal floor but shows up on #overview and the normal map view.
• Deal some small damage, such as d5 or d6, to a nearby creature. (Acid Bubble from D&D is a nice corresponding cantrip.)
• Dig out a single square.
• Create a cancelled hostile yellow light. It cannot explode at you since it is cancelled, so it becomes an autonomous light source of radius 1, which can’t follow you down levels like a pet can. Killing it should not grant you any experience; otherwise this would be easily farmable.
• Light a radius 2 or 3 area permanently, like a weak form of the spell of light.
• Mage hand: you indicate a direction and the nearest item on the floor in that direction, assuming it’s under 100 weight or so, is brought to you and is placed in your inventory. This can be used > to get things out of pits.
• Create an illusion of yourself at your spot. The illusion never moves. Monsters that see it assume it is you and will attack whichever one is nearer (the illusion is instantly destroyed if attacked).

Address startscumming by introducing a point buy system for starting characters: unrandomizing starting stats and inventory and letting the user spend points on stats and equipment however they like. (Note: dtsund’s class overhaul proposal also addresses unrandomization of starting inventory by simply removing all the random chances to get items, but it doesn’t involve point buy.)

Starting stats

• Any point buy for stats is likely to involve a tedious interface. Developers could get around this by making it options-only, perhaps in a way that allows the user to say “buy to get my Strength up to 10, and my Int up to 9, and randomly pick the rest”, but specifying it per-role could get annoying.
• It’s important to allow people who want to start and splat games rapid-fire to do so, and it would be preferable if new players didn’t have to learn the complexities of a new interface.
• The rc file must allow the user to specify starting stats per-role in this case, or specify that they want random stats.
• “Dump stats” like Cha, and Int and Wis for non-spellcasters, allow people to buff their important stats at the expense of unimportant stats, which may indicate some underlying balance issues with these stats or that the stats can have too wide of a starting range.

Starting inventory

• Even in a non-point-buy system, for starting inventory in cases like for Monk and Wizard where the player gets a random spellbook with equal weight among the possible choices, the player should be able to select which.
• Any point buy system for inventory should probably be options-only, since designing an interface to do so would be complicated.
• Naturally, items like magic markers and rings of polymorph control cost lots of points, whereas things like apples and oil lamps cost only a few.
• Some items should probably have maximums.
• This would come at the expense of players who actually want different nonrandom starting inventories and don’t want to have to edit their rc file every time.
• If the player’s options leave some or all of the available points free, the game will randomly buy items until there are no points left.
• Unspent points should probably convert into a certain amount of gold.
• Possibly the default items are expressed as a list which consumes most or all of the points and is sorted from highest priority to lowest, and ones the user specifies get moved to the beginning of the list, which will push the low-priority items into the range of not being able to be purchased.
• Certain items, like the knight’s lance and probably most armor, should not be purchasable, and the character always gets them.
• FIQ proposed a system in which you can specify one item, and the game will try to reroll your starting inventory many times until it gets that item. Since this is effectively built-in tool-assisted startscumming, it gets tracked as a conduct. However, making it a conduct could be self-defeating in that people who don’t want the game to show they broke the conduct will not use this system and continue to startscum.

A radically different system from how alchemy currently works, known as “1 + 1 = 2”: no matter how many potions are in the stacks being dipped, only one from each stack is consumed, and two alchemized potions are always produced (unless they explode).

• Intended as a balance fix for the player’s current ability to make stacks and stacks of full healing, gain ability, or gain level potions, and promote use of found potions instead of diluting them.
• Significantly, potions of holy water mostly work like they do now - the holy water does not mix with anything and is consumed, while the dipped item becomes blessed - but you can only dip a singular item into the holy water, including other potions. This loss of the ability to mass-produce holy water creates a significant balance change.
  • Or it could sometimes make two holy waters, but be biased against it, and allow both waters to become uncursed in the process.
  • Still unbalanced, since holy water can otherwise be mass produced through water prayers and blessed confused remove curse.
• Undiluted potions can be duplicated by dipping them in normal water, e.g. one potion of gain level + one water = 2 diluted potions of gain level.
  • Diluted potions may need to have weaker effects to compensate for this, or undiluted potions stronger effects.
  • Or potions in general should become rarer, and possibly generate diluted.

When a nymph steals an armor item, she remains frozen for as long as it takes you to remove it, thereby preventing her from stealing all your other things while you are removing it.

Generate a single out-of-difficulty monster on every level, as an experiment to see how the game balance changes (not necessarily as a permanent feature). Or instead of picking something out of difficulty, generate a monster with the maximum permissible difficulty.

Dipping an item into any fountain may erodeproof it with low probability. (Or else have this only apply to magic fountains with somewhat higher probability.)

Several elaborations on a system for more reliable (and less tedious) gifts from altars:

1. Your god will always give you a gift on the first sacrifice from a given coaligned altar (not if attempting to convert an altar). But after that, they will not give you any more gifts on that altar no matter how much you sacrifice. (Other sacrifice effects will still happen as if you failed the die roll for a gift.) This eliminates all the tedium associated with altar camping for gifts (though not all altar camping in general), but is a significant change since getting one corpse to an altar is much easier than getting multiple ones, and streamlining the game to this extent may be undesirable.
2. As above, but the gift will only be given for the first monster whose difficulty is above a certain threshold (possibly partially dependent on depth, and also with some random factor). Thus, if you sacrifice and get no artifact, you know you have to try a harder monster.
3. The total difficulty of monsters sacrificed is accumulated; the first sacrifice to put you over some threshold (say, 30 cumulative difficulty) is guaranteed to give you a gift. The totaling may or may not be per-altar.
4. Your god has a set list of rewards, each of which is tied to some monster difficulty level. Some of these rewards are artifact gifts; others may be things like luck. You only get a gift when you sacrifice a monster of the requisite difficulty, and thereafter monsters of equal or lesser difficulty to that are ignored by the god for reward purposes (they may still be used for mollifying and decreasing prayer timeout). If you make a sacrifice that earns you multiple rewards at once, you either get them all at once or you get the biggest reward and the smaller ones on subsequent sacrifices. This would enable game developers to fix a number of artifact gifts (say, 3), by encoding that many rewards for them. This system might have problems in games with no deep usable altars where it is harder to get randomly generated difficult monsters.
5. Variation of the above: at the beginning of the game, three random monster difficulties are chosen (with some limiting on ranges). Sacrificing a monster of a difficulty above any of these thresholds automatically gives you a sacrifice gift and removes the highest remaining threshold you met.

Any item can be dipped into a potion of levitation to permanently reduce its weight by 20%. This uses up the potion, and it will only affect the item’s base weight, not the weight of any contents (i.e. a bag of holding thus dipped will weigh 12 + contents instead of 15 + contents). If the object is canceled at a later point, it loses this property and reverts to its normal weight.

The effect cannot be stacked; attempting to dip an already-levitated object into another levitation potion will have no effect and will not use up the potion.

A possible variation is to also confer this effect for potions of enlightenment, since “enlightening” the item could be interpreted as making it physically lighter.

Dungeons levels generate random, small Sokoban “rooms” - actually themed rooms, never on a path between the two stairs, possibly using “diagonaloban” rules which permit boulders to be rolled in any direction. There is a minor prize in each room at the end of the puzzle.

Internally, rooms like this should have their own room type, which will help with putting in anti-cheating measures:

• You cannot teleport into a square whose room type is this room. (Teleporting out of the room should be fine.) Likewise, the room is not a valid destination for arriving on the level via levelport, falling from above, etc.
• The walls are nondiggable and non-phaseable.
• Air currents are in effect over pits/holes in this room type.
• The room should have a narrow entrance with a bend in it, so that boulders can’t be pushed in from elsewhere on the level.

It’s unclear what the methods to prevent the standard forms of cheating (breaking boulders, reading scrolls of earth, carrying boulders as a giant, etc) should be. Possibly they will just carry cheating penalties like normal.

Track the amount of time the player has spent on a given dungeon level. As it increases, random monster spawns gradually get less frequent (unless you have the demigod flag set, which overrules this) and more difficult, starting at one or two thousand turns spent on the level.

Faction system, which lays the groundwork for more complex monster interaction rules:

• Every monster stores an integer which represents its faction ID.
  • Faction ID 0 is used for the player and tame monsters; others are sequentially assigned as needed.
  • Other groups of shared faction IDs would be things such as “all gnomes in the mines”, “all dwarves in the mines”, “all orcs in Orcish Town”, etc.
  • If a group of monsters spawns, they all share a faction ID.
• There is a data structure which stores a list of two faction IDs and an associated rule (fight, peace, etc)
• Monsters of the same faction don’t attack each other, unless possibly they grudge one another.
• Monsters of different factions interact as normal, unless the data structure specifies that their behavior should be overridden.
• When a monster in a faction is attacked by something it didn’t previously want to attack, the data structure is added to with a (attacker faction, defender faction, “hostile”) piece of data. E.g. if a kobold throws a dart and hits an orc in a group, all the orcs now hate that kobold.
  • Possibly, if only some monsters see the act of aggression, they split off into a new faction (which is aggressive, while the old one remains non-aggressive), and the data structure is set so that when any monster in the old faction sees a monster in the new faction, it automatically changes to the new faction.
• Monsters that read cursed genocide or create monster could then have the summoned monsters’ faction be hostile to the summoner, for symmetry with players.

Make indirect identification more common and reliable through the game, but make the scroll of identify less likely to identify multiple items and impossible to identify your entire inventory. Thrones can still identify your whole inventory as a random effect.

Make more monsters immune to magic missiles.

Make magic cancellation not work with a fixed rate versus all types of attacks it works on throughout the entire game. Instead, scale a monster’s ability to penetrate your magic cancellation based on its level, base level, maximum HP, or hit dice.

Hit dice and base level are nice if it’s desirable for all monsters of the same species to have the same MC penetration, but hit dice and maximum HP are a little weird since they aren’t meant as offensive stats. Maximum HP and actual monster level are nice because they can be discovered with a stethoscope.

Make it so that dying in a polyform carries a severe, possibly permanent penalty, such as losing some Constitution, maximum HP, or forcing you to a low percentage of current HP regardless of what you had when you first polymorphed.

Since lycanthropy is an “abnormal” polymorph, dying while in your wereform kills you completely. (This closes off the silliness of turning into your wereform and throwing your weapon at yourself until you die and turn back to normal).

Possibly, weak lycanthropes such as wererats and werejackals should be removed so that getting infected by one isn’t essentially a death sentence.

Make lycanthropy less random in when it triggers, allowing the player to know clearly when an episode is approaching, but make the effects when transformed more dangerous to the player. While transformed, the player is overtaken and controlled by an animalistic AI, giving no control to the player until they untransform.

Player monsters on the Astral Plane can be named after people on the top ten list (note that they don’t all have to be; some can use the same names they have now). Possibly, pick a different name if a player monster would be generated with the player’s own name, in order to avoid problems on single-user installs where everyone on the top ten list could have the same name as the player. Then again, that could be cool.

Fix the effective level difficulty of all Sokoban levels to either the level past the one with the stairs, or the level with the stairs, or a constant 9.

It’s a common refrain that Sokoban was too easy in 3.4.3 with its easier difficulty as you progress through it, and that it was too hard in 3.6 with its harder difficulty. Fixing it at a constant difficulty is a popular middle ground.

Damage reduction from AC happens before half physical damage instead of afterward.

Every time you use a unicorn horn to cure a problem, the chances of any unicorn horn curing that problem in the future slightly and permanently decrease. Effectively building up an immunity to the effects of unicorn horns.

Challenge mode in which all non-sessile monsters are faster.

Whenever the Wizard of Yendor revives, he has immunity to whatever he was killed with the last time.

When you are turning to slime, higher luck increases the odds of a container trap or magic trap being a tower of flame, and increases rather than decreases the odds of any tower of flame trap (container, magic and fire) hitting you.

All spheres should explode in a 3x3 explosion like gas spores do. They should do this regardless of whether they died by attacking or some other method, so they should have an appropriate AT_BOOM attack.

Allow multiple randomized item descriptions to correspond to the same actual item identity (e.g. a yellow potion and a red potion might both be healing).

Some limited way to charge rings that isn’t the general-purpose magical charging that also charges wands and tools. Maybe add a “scroll of enchant ring” that, like armor and weapon scrolls, only works on rings.

See also L’s Scroll of Enchantment patch, which merges ring, weapon and armor enchanting into a single scroll.

Whenever you arrive on a level through a portal, either the portal moves a short distance or you appear a few spaces away from it. Or turn all 2-way portals into paired 1-way portals which can have different arrival coordinates defined.

Move Sokoban much later in the dungeon, and make time not pass there. This Sokoban would consist of larger levels, but the boulder puzzles don’t cover the whole level.

Rather than making weapons train 1-to-1 on all successful hits, make them train differently:

• Weapons train only when their natural damage die rolls a 1 / rolls a max (ignoring other weapon damage bonuses), so that weaker weapons train faster than stronger ones, and weapon skill for stronger weapons comes in later in the game rather than having all players reach Expert relatively early in the game. This has the oddity of incentivizing orcish weapons over regular, elven or dwarvish though.
• Make the enchantment of your weapon irrelevant to training or even make training slower with a positively enchanted weapon; if the weapon has magical homing assist technology, you’re not really building skill in how to use it that much.

The mysterious force triggers every certain amount of turns; this sends you either back to the downstairs of your current level or the upstairs of the level below. Changing levels resets the timer.

Monks get a small amount of free bonus movement points when they kill a monster bare-handed, and also gain a small amount of free bonus movement points while making movement actions and not wearing body armor.

Whenever a lich is created, a special phylactery amulet is created somewhere on the level. When the lich is killed, the phylactery begins a timer to resurrect into the lich again. The player can somehow destroy it to finish the lich off once and for all.

Weapon differentiation - possibly a replacement of the old D&D “damage vs. small / damage vs. large” system currently in use. Weapons would do the same damage regardless of what size the target is, but might get a damage bonus or penalty based on if the target is vulnerable or resistant to piercing, slashing, or whacking/blunt weapons. Being resistant to this weapon type means that the damage is halved; being vulnerable to it means that the damage is multiplied by 1.5. Some obvious examples:

• Most undead are resistant to piercing damage. Skeletons are additionally vulnerable to blunt damage.
• Thick-hided animals are resistant to slashing damage.
• “Blunt is best against hard things, slashing is best against soft things, piercing is best against fleshy things”

While wielding a polearm, moving orthogonally towards an enemy that is 3 spaces away (making it 2 spaces away, after the move) makes a free attack on that monster.

A bunch of scattered ideas for handling candles, given that the hero almost always wants to have only one lit at a time:

• Track all candles separately unless they have the exact same number of turns remaining.
• Merge all candles automatically, but still keep track of each one’s lifetime. (This would enable lighting multiple candles and then having one of the stack burn out.) When you light part but not all of a stack, the game internally picks the shortest candles.
• When candles merge, their lifetimes are averaged together.
• You can’t light more than one candle at once at all. (Not a great idea since it would destroy the mechanic of lighting a large amount of candles for more light).
• When candles merge, the candle with the shortest life automatically sacrifices as much of its lifetime as needed to “top up” the rest of the stack. That is, if the lifetime is 400, a candle with 50 turns remaining will merge with a candle with 375 turns remaining, making a stack of 2 candles with 25 turns remaining; when those 25 turns are up, it becomes a stack of 1 candle with 400 turns remaining.

The portal on the Plane of Fire moves around, one space in a random direction each turn. Never onto lava or another trap.

Special candle found generated in Gehennom. While one is lit on a level, it suppresses spawn rates and makes the difficulty about what it is now (without the candle lit monsters will generate at a much higher difficulty). They should be fairly nonrenewable, perhaps wishable but not polymorphable. This limits your duration of time in “easy Gehennom” before it gets harder. The candles work the same outside Gehennom, but are less useful (maybe they specifically suppress demon spawns).

Playing a wooden harp (and possibly a wooden flute) near intelligent peacefuls may cause them to give you some money or items from their inventory, but there’s also a chance that they become hostile (without an alignment penalty though).

Structure player speed on a four-tier system: normal, fast (available intrinsically), very fast (available extrinsically via speed boots), and extremely fast (temporary and available only via consumables).

Unconsecrated or minor altars. They are either to some unnamed lesser god, or to no one in particular. They can be converted to your god like normal, but don’t have any penalty for kicking or attempting to engrave or anything while still unconsecrated.

Losing alignment makes it permanently harder to regain alignment. Do this by instead of using a linear scale where current alignment equals alignment gained minus alignment lost, use a ratio where current alignment is alignment gained divided by alignment lost (probably with some constant factor in there somewhere).

Weight the probability of Orcish Town down lower than 1/7. Possibly as a starting point, reintroduce Frontier Town as minetn-8 so that the probability will become 1/8.

When you throw ammo without the proper launcher, the game makes it clear that you’re doing it wrong, e.g. “You clumsily throw the arrow”. This also abuses Wisdom.

Silence auras that prevent both you and monsters spellcasting. Maybe also a directed silence spell, which silences an individual monster. While silenced, monsters can’t be chatted to, and nothing they do will make noise (possibly only if in a silence aura).

While under the effect of silence, the player’s stealth is elevated.

Spellcasters who curse items also have a separate way to remove your knowledge of items’ beatitude, so you don’t know what might have just been cursed.

Instead of instakilling, death rays cause a delayed instadeath that require consumables or prayer to fix. Touch of death should also probably fall under this.

An unaligned alignment quest.

Rebalance multishot damage so that multishooting +7 projectiles doesn’t give a massive damage bonus over a single +7 melee weapon. Possibly only apply the damage bonus to the first projectile.

Enchantments on armor and weapons time out (rather slowly). Also, any erosion damage to a positively enchanted item will be absorbed by that positive enchantment, which then vanishes in proportion to the amount of erosion. Perhaps the enchant scrolls should be made more common or cost less ink to balance this.

A “reset button” for Sokoban, which restores all the original holes and resets all the original boulders, and places the player on the downstairs (or is reachable only from the downstairs). Probably not activatable after the level is solved. Flavored as some sort of mechanism.

Float stone, a gray stone which has negative weight. Non-stacking, and rare. If you get enough float stones to more than cancel out your own weight and the weight of any possessions you’re carrying, you levitate. Possibly doesn’t work when put into containers, or at least doesn’t get multiplied by the effect of a cursed bag of holding. May gain positive weight when cursed, or greater negative weight when blessed (or vice versa).

More harassment effects: the chance of respawning the Wizard is a fixed 1/7 and the rest of the effects are equal probability. Multiple non-duplicate effects may happen at the same time.

• Unchanged: curse items, summon nasties, vaguely nervous, aggravate monsters
• Darken much of the level, and extinguish, destroy, or cut the remaining lifetime of light sources. Maybe also blind the player.
• Drop boulders around the upstairs, or iron bars.
• Spawn pools centered on the player, with sea monsters in them.
• Spawn lava around the level, not under the player.
• Create a hole or pit under the player.
• Create a stinking cloud centered on the player.
• If you are carrying the Amulet, it teleports to a random square on your level or the level below.
• If you are carrying the Amulet, it (and possibly any fake Amulets) becomes heavier. This either lasts until the next harassment triggers, or is permanent but the total amount of weight added to the amulet has some sort of cap on it. This is interesting primarily because it targets carry cap, which is a valuable end-game resource.
• A variety of powerful player debuffs such as dealing half damage, taking double damage, blocking HP or Pw regeneration, super hunger, etc.

Make fireballs like old-school D&D: the explosion isn’t a certain radius; instead it has a defined volume, and will expand out from its origin point via random breadth-first search. This makes it dangerous to use in confined areas.

The confused scroll of identify, if non-cursed, identifies a random unknown object class and adds it to your discoveries list, even if you’ve never seen any of that object.

Make death messages more diverse and flavorful. Base it on the attack type of the monster and possibly the weapon used, so that “killed by a jackal” becomes “clawed/mauled/shredded by a jackal”, “killed by a gnome” becomes “bludgeoned to death by a gnome” if it was wielding an aklys, “stabbed to death by a gnome” if it was wielding a dagger, etc.

Note that this creates a headache for tournaments that track unique deaths.

Iron chains are given either whip or flail skill. Probably flail, since they’re heavy.

Artifact whose invoke power creates attack wands.

Escape as its own key should be used as an actual escape command, so you can do things like try to get out of pits and engulfers with it.

More variety in ranged unreflectable elemental attacks, as a sort-of-nerf to reflection (all ranged elemental attacks currently in the game are reflectable). For instance, have enemies shooting fire arrows at you.

Modify Sokoban maps so that a monster can never get stuck in an awkward place except by the player doing something foolish. Possibly make the corridors leading to the goal double-sized or have alcoves off to one side where a monster can move.

Saltwater or murky water, intended as a nerf for “just dilute most potions”:

• Potions of saltwater behave exactly like potions of water, except that they cannot be used to bless or curse things. Holy (and unholy) water must be pure.
• All in-game sources of water except for magic fountains (possibly regular fountains as well, but they might be too plentiful) are saltwater. Diluting potions in them will yield only potions of saltwater.
• Can also get pure water if a unicorn horn dip would already result in water.
• Not specified what cancelling potions should do. Possibly most potions just cancel to saltwater, juice or sickness. It’s weird because flavor-wise it shouldn’t produce anything but pure water, but it obviously won’t fix the current problem if you merely have to cancel potions instead of diluting them.
• If called murky water instead of saltwater, the “murky” randomized potion appearance should change.

Alternatively, don’t actually make saltwater a separate potion; just use saltwater as a flavor reason for why the player can’t intentionally dilute potions to water. (Can use a separate flavor argument, that potions are generally corked while not used, to explain away the lack of dilution for water damage.) Also, make most potions cancel into non-water base potions (such as confusion and hallucination to booze), with the notable exception of polymorph. It’s also harder or impossible to get water from random alchemy or polymorph.

A third option is naming blessing-capable potions “purified water” and leaving the mundane potion of water as it is. Things such as unicorn horns can neutralize toxins but can’t actually remove them, so the resulting water isn’t purified.

Entering the Valley and entering the second level of Gehennom (as well as possibly entering other branches) give the hero an automatic level up.

In shops, picking up an item spawns a peaceful monster that vanishes once you pay for the item properly. If you make any attempt to steal, it turns hostile. (The general principle: the more you want to steal, the harder it is to do so.)

Special room that is lined (not filled) with statues of you, and contains no monsters, items, traps, or anything else interesting. All doors in contain traps that will make them lock behind you. Intended only to make the game more creepy.

You can attack yourself by using the F command and selecting . for yourself.

Ammunition that is +X will break if and only if it scores a hit on a monster above a certain difficulty. The formula suggested was that it can only break when 5X+3 < difficulty, but this is just a wild guess at the numbers and isn’t balanced.

Change the color of the player’s @ when their HP gets low or decreases sharply.

Determining whether you escape or dodge trap effects should be based on stats like Dexterity and maybe role and XL, rather than Luck.

Wands may generate pre-recharged.

Elves get low-light vision, which doubles the radius of light sources, rather than infravision. It also gives them vision on dark squares that are orthogonally or diagonally adjacent to a lit square.

Find some justification for making the flavor of Sokoban a branch where the ceiling collapsed and opened up pits and dropped rocks everywhere.

Make death magic resistable by an amulet or ring. Possible names and corresponding exact effects vary:

• “life protection” (bad name, life saving is very similar)
• “unliving” (makes you not count as living and life saving won’t work)
• “undeath” (actively makes you count as undead, you are vulnerable to turn undead effects and lawful gods may be annoyed at you)
• “versus death magic”

Restore the Amulet Delivery Service to the game: demon lords in the Sanctum, but not the Wizard, will fight the current holder of the Amulet of Yendor for it. Only do this if lugging a demon lord down to the Sanctum is considerably harder than just fighting through the Sanctum is itself, and no demon lord is actually guaranteed to win against the High Priest.

Keys only work on locks in one dungeon branch.

Fort Ludios is a visit-once level: once you leave by the exit portal, the Dungeons of Doom portal disappears and the level becomes inaccessible forever. (Maybe if you leave any of the invocation items or the Amulet there, the portal refuses to let you out so the game doesn’t become unwinnable.)

Single-level branches whose portals, found in vaults, disappear after one use; they are subdivisions of the Magic Memory Vault, and modeled after ADOM “vaults”. These are basically challenge levels, smaller and less rewarding than Fort Ludios (the master branch), that contain some nice loot and harder-than-usual monsters. Portals to them can also be found in vaults in Gehennom.

Potions that you can drink from multiple times.

Applying or invoking the Candelabrum with insufficient candles prompts you to attach more candles to it rather than lighting it.

Luck ideas:

• Luckstones don’t remove luck timeout, they merely extend it.
• Luck is much less controllable, impossible to get any clues about except through enlightenment, and various random events (like what?) can give or remove luck. A skilled player would know how to bias the odds towards luck-increasing events. Controllable sources of gaining luck like sacrifice and throwing gems at unicorns are removed. It no longer times out.
• The luck timeout rate is less predictable, and a luckstone will help but it might still time out quickly if the RNG hates you.
• The base number that luck times out to is equal to the total number of luckstones in your inventory (so carrying 2 noncursed luckstones means luck times out to +2). Luckstones’ weight is increased.
• Remove luckstones in favor of an amulet of luck. Luck timeout can no longer be slotless.
• Luck times out at a rate of 1 point per (600*(number of luckitems + 1)) turns.
• Do away with luck timeout entirely in order to make it impossible to micromanage luck. Luck is now conferred entirely by luckitems. The problem with this approach is that NetHack has about two ways to gain timing-out luck and many, many ways to lose it.
• Luck can be gained by enchanting luckstones.
• Different types of luckitems which stack with other luckitems but not with each other, like a four-leaf clover you can get from leprechauns.

New unaligned artifact dagger Almagest: uses the powers of scrolls to augment itself, and bears a scroll label signifying which scroll ability it is imbued with. Opinions vary on whether this label should be randomly generated by the dagger (and the player may choose to read it, upon which it gives the uncursed scroll effect and disappears for 1000 turns, eventually returning with another label) or whether the label should be deliberately imbued by the player by rubbing it on a scroll. It gains +d4 to damage and +d4 to hit when it has an inscription.

There are additional special effects if the player has formally IDed the scroll whose label is on the dagger:

• Identify: 25% chance of probing
• Light: +d6 to undead, light source
• Enchant weapon: +d4 damage
• Enchant armor: gives 3 points of AC when wielded
• Remove curse: Protects fully from all curses when wielded
• Confuse monster: 25% chance of confusing
• Destroy armor: Ignores enemy armor when calculating to-hit
• Fire: +d8 fire damage
• Food detection: Enemies are much more likely to leave corpses
• Gold detection: Enemies may drop small amounts of gold when killed
• Magic mapping: 5% chance of activating clairvoyance on hit
• Scare monster: 15% chance of scaring on hit
• Teleportation: 10% chance of randomly teleporting on hit
• Amnesia: Extra mindflayer tentacle attack, no effect when read
• Create monster: 5% chance for create familiar effect on hit
• Earth: 5% chance of creating boulder over enemy
• Taming: 5% chance of charm monster effect on hit
• Charging: Does 2x Pw damage and adds it to your own, 5% chance of cancellation
• Genocide: Instakills as Tsurugi, +1 damage
• Stinking cloud: Permanently poisoned
• Punishment: Weighs 100, deals +d12 damage. Autocurses and triples alignment losses.

Amplification trinsic which is a counterpart to reflection: an attack that hits you will be amplified and continue past you (greater range and greater effects).

Statues block line of sight.

New slot for belts:

• Belts themselves could be armor or tools, but probably armor. They mainly serve to hold items for quick access. It’s not defined whether they are intended to hold an indefinite number of items like any container, or a limited amount. The only problem with making them armor is that they would by default be enchantable for more AC, but belts should probably not be enchantable. If it makes more sense that several magical belts would hold charges, tools would be the appropriate class.
• Items carried in a belt weigh nothing and are mostly or totally protected from destruction, and using them doesn’t take the action it usually takes to get an item out of a bag.
• Types of belts could be scroll-holding, potion-holding, wand-holding, ammo-holding (with a possible “endless quiver” item or artifact), gem-holding (to get partial resistances from certain gems), chastity (providing immunity to foocubi seduction) or no special powers but some moderate AC.
• One randomized belt description could also provide protection against leprechaun theft.
• Randomized descriptions could include: cloth, leather, worn, studded, buckled, braided
• Tourists and Convicts also start off with garish swim trunks/striped slacks, which do the same things as Hawaiian shirts/striped shirts. The swim trunks also allow swimming. Both of these protect from instadeath when you sit on a cockatrice corpse.
• Belts of giant strength, elven grace, and troll toughness. These increase your Str/Dex/Con either to 25 or by +enchantment (if belts can even be enchanted, which isn’t currently defined.)
• Swordbelts/sheathes. These allow you to ready an additional weapon or give you an additional weapon slot.

More belt ideas here.

Zombies always try to path in a straight line towards the player (assuming they can see or otherwise are aware of the player), and get stuck on terrain.

If you have all three invocation items and have not yet performed the invocation, the Oracle always gives you the message about going to the bottom of Gehennom and performing the invocation. If you are missing one or two of the items, she will give you a message focused on (one of) the missing items. If you have the Amulet, she will give you an otherwise unreceivable message about continuing upward through the Elemental Planes.

Rename the potion of gain level to “potion of level up”.

Dwarven characters always passively see all gold on a level, with the possible exception of gold carried by a monster.

Reduce the base price of the scroll of blank paper to something low and unique. It should probably cost less than the scroll of identify, but not be incredibly cheap.

Instead of having Sokoban as a branch, generate occasional Sokoban puzzles in the Dungeons proper. It must not be mandatory to complete them to pass through the level.

Falling rock traps are flavored as unstable ceiling (so that the flavor doesn’t get weird for situations like large open levels) and the message changed accordingly. Alternatively, create a separate unstable ceiling trap for that purpose.

You can break and destroy a loadstone in inventory by applying a pickaxe to yourself, or rubbing the pickaxe on the loadstone.

Give a YAFM when confused and trying to read a scroll of blank paper: “Being confused, you mispronounce the lack of words…”

Polymorph traps, when they polymorph the player, always choose a form considered to be “playable” - i.e. has hands, or is fast enough to run from threats, or is well-defended enough to sustain damage until transforming back, must have speed > 0 - but you do not re-form with your original HP when your polyform is killed, you merely die.

“Omnicidal” monster attribute: it attacks peaceful, hostile, tame, and possibly other omnicidal monsters. Suggested in relation to monsters reading cursed genocide or creating monsters.

Include a kitchen sink at the top of the Sokoban branch (so that anyone wearing a cursed levitation item on that level has some recourse to get back down).

Alignment record changes: chaotics get -1 penalties over time for failing to do any chaotic behavior, neutrals get +1 alignment bonuses more slowly over time by not doing anything, and lawfuls get neither of these and need to get their bonuses on their own. Alternatively, lawfuls accumulate bonuses by doing nothing (not breaking the law).

Farlook allows you to see a list of all items on a square as if each were the only item on that square. Items hidden under others are not obscured. Also works if there is a monster standing on the square; you can still see the items under it.

Cut all map sizes in special level files down to at most 20 lines tall, so that there is extra space for messages and status line in an 80x24 terminal. Few levels currently use all 21 lines.

Make permaconverting harder to do by accident, since confusing the actions of rededicating an altar and rededicating yourself seems unlikely. Maybe make it happen if you sacrifice or pray on a crossaligned altar while confused.

Decide once and for all whether alignment record should be god-specific or god-agnostic, because NetHack currently tries to do both and fails. Things like the massive penalty for killing a Quest leader make way more sense for lawfuls than chaotics.

In variants or proposals where you can kill your quest leader to get the Bell of Opening, if you have permanently converted alignment already, killing your Quest leader does not give you a massive alignment penalty.

Blessed potions of polymorph, when quaffed, grant you immediate polymorph control over just that one polymorph either 50% or 100% of the time. To compensate, cursed potions of polymorph will ignore changing into a dragon and will try several times to roll a monster that will make you break out of your armor. Or the cursed potion could have a wildly variable duration, so you have no idea when it will end and cannot rely on it.

Knights start mounted, or else are trained enough to mount their horse from turn 1.

Make candle stacking for more light actually useful and realistic: the number of candles required for another square of radius increases quadratically instead of exponentially, so 1 candle gives radius 2, 4 give radius 3, 9 give radius 4, 16 give radius 5, etc.

Later extended into a more general idea allowing light sources of any type to stack: candles each give 5 lumens (because a single candle’s light reaches up to sqrt(5) squares from the center), lamps give 10 lumens, potions of oil 2, and the full Candelabrum gives 20. To compute your total light radius, add up all the lumens and take the square root. (Note that this exact formula might scale too fast: 4 candles equals the Candelabrum).

Elaborations on the Fort Ludios minesweeper proposal by ais523:

• Should spaces with 0 neighboring mines get 0 gold? Good because it makes the game possibly more instantly recognizable as minesweeper, bad because it makes the overall amount of gold less predictable
• Need to guarantee somehow that the person stepping into the room doesn’t trigger a mine instantly, and set up the level accordingly.
  • The column nearest the entry door could be clear of mines and gold.
  • The whole border could be clear of mines. However, this means that the total amount of gold must end up being an even number - not ideal.
• What happens if a monster steps on a mine?
• Possibly this needs an Oracle consultation to explain it to unspoiled players. Or make an engraving when you enter - possibly “xyzzy” as a reference to the cheat code in Windows Minesweeper. Or start off with one mine uncovered and identified on sight.
• The number of adjacent mines could be expressed in the ones digit of the gold amount. Possibly with a large round number added, so every space is 3001 or 2002, which hopefully clues the player into the fact that it’s a significant number. But this might be too restrictive.
• How to reach the magic number 69105, and do so in a way that guarantees it can be generated? May need to bias the level generator so that this must happen.

A Barbarian rage mechanic (possibly an effect on Cleaver replacing its current damage bonus) where a damage bonus is assessed based on how long ago you last killed a monster. Originally a buff to Vorpal Blade where it beheads anything it hits if the last behead was N turns ago, but this is unbalanced.

Reduce the cost of a major Oracle consultation; the current price could be too high for unspoiled players to consider it worth it.

Potion equivalent of extra healing for magic power, or else some consumable way to recover 20-40 power in one go; the normal potion of gain energy is currently more useful for raising maximum power than it is for restoring it. This new thing need not be more effective than gain energy is at raising maximum energy. (Perhaps “potion of restore energy”, which does not raise maximum at all.)

Thrones will not disappear in a puff of logic unless at least one interesting effect from sitting on them has happened.

Remove the bell from the list of musical instruments that Elf characters can receive at the start of the game. This is because it’s atonal and its only half decent use is to attract pets (note that elves can’t get tin whistles, so clearly summoning pets is not important enough to qualify something for initial inventory).

Invisibility works completely when the invisible monster and the detecting monster are far away. When within 5 (or 4?) squares of each other, its location is known to within 1 space (fuzzing the real location by 1 space in all directions). When the invisible monster takes a direct offensive action, or with some probability when it is adjacent, its position becomes known. It still appears as I, but the I will move wherever the monster goes until it becomes 5 or more spaces away, at which point invisibility once again works fully. This could work differently if the invisible monster is stealthy.

Lycanthropy fixes:

• It triggers much more rarely, every couple thousand turns, but while you are in your beast form you lose all control.
• Or it triggers much more predictably, with messages making it clear that a change is coming, and the duration of the change is both more limited and more predictable.
• Or it triggers on a certain damage threshold or after taking a particularly strong hit, determined by your level, HP max, and phase of the moon.

Ideas around a system in which you can cook corpses:

• The general idea is to unbreak the hunger system by making regular corpses unsafe to eat, so the shortage in resources isn’t in food itself but is in safe methods of cooking it.
• Cooking sets a corpse’s rot timer to 0, lets them go longer without spoiling, or unspoils them.
• May also remove poison from some kinds of poisonous corpses.
• Cooked corpses are unsuitable for sacrifice and less likely to give intrinsics, or don’t at all (because the vitamins are destroyed).
• Can apply wands of fire to cook corpses.
• Or maybe eating raw non-vegetarian corpses is just plain dangerous, causing vomiting or sickness or other bad effects.
• Corpses can be cooked by killing monsters with fire or hitting monsters with fire attacks while they are carrying corpses.
• The problem is that nobody has yet proposed a mechanism for cooking corpses that would work well. Having firewood or some similar thing would massively affect the early game between characters that can and cannot manage to find it, and roles that start with or without it. Wands of fire have been proposed as a resource but those are valuable. Introducing a new item like firewood means that a whole lot of mechanics have to be reviewed for whether they should interact with or create that item.

Race and polyform should not be distinct; they should be merged into a single permonst or corpsenm-like variable. Role is independent of race, and perhaps every monster should be described with a monster/role pair, with the role being null for most monsters (i.e. gnomish wizards are just gnomes that are also wizards). This might eliminate the entire concept of player race, internally, but transferring some of the racial effects on players (e.g. elves’ sleep resistance at XL 4) might be tricky.

Have a record in the xlogfile of whether the game was likely a startscum, for analysis purposes. It’s easier for the game to track things like fountain quaffing and using obvious dangerous items. How exactly this would happen is not determined; the game probably shouldn’t try to make a decision itself, there will be some false negatives and positives, but storing fields for everything like “number of times quaffed from a fountain” is probably bad too. Also, if players are aware of a scum-detecting algorithm, some will deliberately work around it.

More monsters should be locked to the grid and can only move orthogonally, like grid bugs.

A Word of Recall spell, which after a sizeable delay (3d5 or 5d6 turns) levelports you to one of several possible locations: Minetown, dungeon level 1, the Oracle, the Castle, the Quest. You will not travel anywhere you have not already visited. Does not work if you have the Amulet.

Rays should reflect erratically or not at all from non-wall rock terrain, Walls are smooth, but rock is rough.

Silver weapons should be rarer than they currently are outside Gehennom and fairly common inside it.

Add a silvery potion, either as a random appearance or a dedicated potion like dNetHack’s potion of silver starlight, which will silver ammunition dipped into it, perhaps temporarily. It might also silver other weapons (spears, daggers) or other items. Random appearances have the possible problems of the potion potentially having a defined dip effect already (polymorph, for example), and it potentially not appearing in the game if there are extra random appearances. Dipping fragile silver items, like silver arrows, into a potion of acid could create a silvery potion.

Armor should make use of its corpsenm (monster index) for efficiency and to do interesting things that are not currently done. For example, this would allow dragon scales and scale mail to be merged into two items that use the various dragon monster indexes to figure out its special effects, and it would allow for fireproof low boots created from polypiling a crocodile corpse to be flavored differently. In variants where applying dragon properties to other pieces of armor is being considered or planned, this would provide an easy way to do so.

Also see dtsund’s dragon scale mail proposal.

You can jump over swimming monsters in water, and over tiny monsters.

Blessing a corpse prevents it from being tainted or rotten (the “Blecch! Rotten food!” effect).

More acid resistance sources in FIQHack, to balance yellow dragons’ exploding acid breath. Possibly a ring or amulet of acid resistance.