[Crawl-ref-discuss] 0.20 dev priorities

[Corin Buchanan-H. <co...@bu...>]

Hey folks,

I see a few big topics for crawl I'd like to work on during 0.20. These
are the topics I see as being the biggest problems for the game right now,
topics where a good solution could have a huge impact on the gameplay
experience. I'm laying this out to share this vision of crawl's direction
in order to open it up for discussion, and also as a request for
assistance.

Please let me know what you think, and in particular, let me know

== Topic 1: Clocks and Luring ==
Right now there are a variety of very strong tactics that center around
spending time to isolate monsters. Many of these tactics are problematic
because they eat up significant quantities of real time without being
interesting after the basic technique has been understood. Because these
tactics are strong and also uninteresting and time-consuming, they can be
understood to be scummy. The root cause of all these tactics is that time
isn't a meaningfully limited resource, and in fact spending time generally
earns the player resources (hp & mp recovery). There are three timers in
the game right now, hunger, piety decay, and out of depth spawns. Piety
only declines for a large subset of gods, and it declines slowly. Hunger
constrains only the very greediest of spellcasters and berserkers, and
barely penalizes luring. The out of depth monster clock can actively
benefit the player by giving them higher-value and better-equipped
monsters to kill without being forced to explore new, unknown terrain;
it's also fairly invisible and can be confused with out of depth vault
spawns, meaning that few players have a good sense of what it does. All
three clocks penalize recovering items from other floors and resting
significantly more than they penalize luring, and probably because of
that, it would probably create worse playfeel to speed up any of these
clocks significantly.

On that premise, I believe that what we need is an effective clock,
ideally one that doesn't penalize revisiting explored areas to recover
items but which does penalize luring. The most basic clock I can think of
would be increasing the OOD clock while causing OOD-spawned monsters to
give no XP and drop no items; however, a significant number of the
"normal" spawns on levels come from this clock, so this could be tricky to
calibrate.

dpeg has already put together an interesting-looking proposal to create a
more-dangerous condition (did that get sent out?) that monsters can enter
when time passes (either time aware of the player, or time on the same
floor, or time anywhere -- see his proposal for more info). The
more-dangerous monsters would not be in any way more rewarding, but their
basic stats -- hp, damage, speed -- would be increased. I think that this
plan is a great step in the right direction, and one that still feels
consonant with existing crawl to a large degree. I hope to try out a
draft.

I have a more radical proposal in mind that I intend to try out as well:
the "doom clock". In this proposal, hunger (and food) and the OOD clocks
would be removed and replaced with an interface-visible threat level -- or
doom level, if you will -- e.g. "threat level:
low/medium/high/extreme/probably fatal", or numerical values "threat
level: 1/2/3/4/5". The display would also indicate a brief summary of how
much average threat you had accumulated over the last 10 turns or so, in
order to give a sense of which activities are particularly risky.

Every turn, the threat level would be evaluated and give a chance for a
threat to occur, much like a hell effect, but generally much less common.
These events would be flavored as the dungeon and its inhabitants becoming
aware of you and taking action against you, and your own rising sense of
alarm leading to carelessness and mistakes. Threat events could include
durable summon OOD spawns, either near the player or not; trap placement
around the player and/or trap effects being applied directly to the
player; the sorts of things that currently are associated with miscasts,
hell effects, and/or god wrath. As such, the threat level system would
replace randomly placed traps and hell effects could become a special case
of the threat system. dpeg's plan above could also be an effect of the
threat level.

As a player's threat level increases, the frequency and severity of threat
events would increase. Time passing wouldn't necessarily increase the
threat level. Currently, I'm thinking the threat level would be adjusted
by the following:
* Whenever the player takes a portal to an explored level with a monster
tracking them.
* Whenever a monster moves towards a player, whether or not in view;
killing the monster could refund some of the threat perhaps.
* Excessively high-level magic, berserking, god powers, and other large
hunger-inducers could briefly increase the threat level by a fair amount
and then ebb back to normal.
* Contamination sources could (also|instead) increment the threat level.
* Exploring new tiles could increase the threat level modestly (to reflect
courting danger) or could decrement it (as a reward for exploring). I
favor the former, as it might inspire players to not full-clear levels.
* Taking a portal to a new level would reduce the current threat level by
a large percent (cut in half or so?), allowing a player to potentially get
out of extremely high threat levels by diving.
* Any other behavior we want to penalize could increment threat.

I hope to put together a draft during the 0.20 development cycle. I
believe the work needed is:
1) Remove hunger and all food.
2) Remove all random placement of traps and make all traps in vaults known.
3) Add threat level tracking to player.
4) Add threat level display to the UI.
5) Implement threat modification rules.
6) Add threat effects.
7) Move OOD spawning into the threat system.
8) Add documentation.

== Topic 2: Reform ranged combat ==
Right now, ranged combat is in a terrible place. It is almost identical to
melee combat, aside from the massive advantage that full-view range adds.
It also has a dedicated enhancer (gloves of archery) and gets no benefit
from the Might/Berserk statuses, and it is limited to some degree in the
early game by ammo. In terms of how it fits into the offensive tools in
the game, it has heavy overlaps with melee (same interactions with skill
and most effects, same availability for the majority of the length of a
given game) and with wands (long-range damaging attack with
strategically-limited ammo). It's mostly stronger than melee, but few
players embrace it; many (including me) specifically because using it
means switching weapons frequently to conserve ammo, feeling more
inventory pressure thanks to ammo, and the irritating tedium of picking up
spent ammo after each fight. So, it doesn't occupy a particularly unique
design space, it's too strong (or close to it), and it's largely ignored
thanks to the unpleasantness of using it.

I think there's only one solution that addresses all the problems:
1) Move the launcher to a dedicated slot to remove constant weapon switching.
2) Mulch all ammo to remove the annoyance of picking up ammo.
3) Store ammo in the player's inventory like gold, taking up no slots and
not individually manipulable.
4) Attempt to forge an identity for ranged combat by making it scale
heavily with skill while keeping ammo scarce all game; it becomes a
limited resource that scales from modest to very strong on players willing
to invest heavily.
5) Mulch all ammo that monsters drop to prevent farming monsters for ammo,
thereby allowing us to manage ammo drop scarcity.

This solution addresses every point of tedium and attempts to make ranged
weapons meaningfully distinct from other attacks by leaning heavily into
the concept of limited use. If this plan doesn't create a ranged combat
that is compelling, I am convinced that crawl cannot support a compelling
ranged combat and we should remove it. I am also convinced that removing
ranged combat would be an improvement over the status quo.

I have a branch in progress (ranged_reform) and have completed the following:
* Mulch all ammo.
* Mulch all ammo dropped by monsters.
* Remove the Blowgun launcher.
* Remove all needles except frenzy, curare, and poison.
* Remove all ammo brands and convert the remaining needles to distinct
ammo types rather than brands of a single type.
* "Goldify" ammo -- I have the base of this complete, but got hung up
fixing the Q interface to allow you to pick from goldified ammo.

Still to do:
* Make all launchers OBJ_LAUNCHER.
* Create a slot for OBJ_LAUNCHERs.
* Address the new launcher situation in a ton of code that references
wielded weapons.
* Update the (Q)uiver logic to allow you to select a launcher and then
automatically choose the correct ammo, or instead pick an ammo for
throwing.
* Prevent artefact launchers, or at least sharply limit the properties
they can receive.
* Rebalance launcher/ammo properties and skill effects.


== Topic 3: Wands and rods ==
I've already seen good discussion about gradually moving in a good
direction here. My preference would be to remove /tele, /heal wounds and
/hasting entirely. My view is that these are redundant to a great degree
and give excessive access to what should be limited resources. I'm
absolutely fine with this making mummies significantly harder -- I support
harsh conducts having few ways of ameliorating them.

I also really like the idea of merging disintegration and digging into a
single wand that is max-3-charge rare wand that deals heavy AOE damage and
carves out a big chunk of terrain, probably using a shotgun targeter. I
think tactical digging is interesting, but killholing is too abusable, and
therefore we should try to remove it, and the two wands are sufficiently
redundant that having one wand that fills this role would be best. I also
like the idea of having a severely limited, highest-end damage wand.

I have no strong position on rods versus wands; I like a lot of the rods
and don't particularly mind that you have to wield them, but I know this
is a big interface issue for some players. If I were going to remove a
wand, Shadows would be my first choice, as Evo already gives a lot of
access to strong summons, and this particular one feels a lot like Shadow
Creatures and ?Summoning, even though it's technically somewhat different.

== Topic 4: Level size and length ==
I've been enjoying the cuts to number of levels in crawl, and I think we
should continue the project of trimming it down gradually. I'd love to see
Lair lose 1-2 levels still, Lair rune branches lose 1 level each, and
Vaults and Depths each lose 1-2 levels during 0.20.

I think it would be a good idea to also reevaluate some of the map layouts
that result in extremely large levels and try to aim for a maximum level
size that's something like 60-70% of the current maximum.

== Topic 5: Charms ==
We've discussed this many times. I don't have anything new to add at this
moment, but I do think it's one of our significant remaining issues.

== Topic 6: Curses ==
Similarly, I've got nothing to offer here right now, but I think this
system could stand to be significantly improved. I tend to favor the idea
of implementing curses as a non-sticky negative feature on otherwise
decent items, encouraging players to accept the negative feature in
exchange for a power boost in other areas. On the other hand, that's kinda
how artefacts work. I have no well-thought-out proposal here at this time.

Thanks all!
Corin