Proposal: allow for a customizable Species

[XuaXua]

Inspired by this complaint, which I disagree with, I wonder if a special species factory could be created in the same way as a the Wanderer class.

Given that there is really no "true" difference between backgrounds (beyond starting equipment), this proposal allows for things to get more complicated by allowing a generated (or user-designed) species.

When a Species is designed, taking "humans" as the base template, attempts are made to "balance" a new species against that template.
You can have flight? You have -1 on three spell skills.
You have Axes and Maces at +2? Reduce Evocations and Invocations 1 each.
You can develop random game-changing mutations at level 7? -1 on all skills.
Your HP is +2? -1 on MP, Dodging and Traps.
You get +3 Traps? Take -1 on Short Blades or Poison Magic.
You're a small or large species? You get +8 Skill adjustments to distribute as compensation.
Your inherent mutations force you to Wear Barding? Gain +2 Ranged Weapon Skills points.

Obviously my build-craft is completely "off" here, but that's where the next part comes in. Somewhere, an arbitrary balance was decided on to design the species (having eyeballed these, there is obviously not a set tit-for-tat character design structure), and that can be adjusted to work to the advantage. Granted, some species are easier to use by default and break the mold (Lava Orcs), even considering the penalties they've been given (slow initial speed, poor spellcasting skill, etc.) don't outweigh the benefits, and vice-versa (insert a bad species here; I'm sure minmay will suggest one).

Using humans as the baseline (because what else do we have?), the first step in this process would be to take the existing species and deconstruct them, determining what
"costs" were used where to compensate for abilities. Skills are obviously not of equal value. Some skills (Fighting, Armour, Fire Magic, Charms (but only because of Haste/Swiftness?)) are less situational / more valuable than others (Evocations, Traps), and their skill cost might be less. In fact, this might be the time to set the Evocations baseline back to 0 (everyone received +1 Evo last version to enhance Evo; instead set +1 to 0 and compensate for the upscale within the game mechanics)

Certain species may have builds which come out to more or less points than expected. As a Human might be considered "STANDARD", these species would be considered EASY or CHALLENGE. Once a points baseline is determined, EASY would have +X additional creation points, STANDARD would have +0 creation points, and CHALLENGE would have -X fewer creation points.

Now you ask, "Well, won't so-and-so break the game this way"? To which I answer, "WHO CARES? REALLY?" The Custom Species (CS, I'd guess) would be something to play to screw around with and have fun.

Let's see if this is a viable idea with potential before I get too invested in it.

[Marbit]

In order to do what you are suggesting you would need a function that describes how good would a species be depending on the skill aptitudes, attibute distribution and the starting mutation set that it has. It seems to me awfully complicated although I have often toyed with the idea of doing that.

I might try to do that in a couple weeks; I'm taking a multivariate statistic class but I don't promiss anything. Even getting the data in crawl is a challenge as morgue files are difficult to parse with statistical software and you dont't want to collect the information manually from hundreds (if not thousands) of morgue files.

Then again you can just tinker manually in wiz mode or make a patch and experiment yourself to a certan degree what would happen if...

[xzanthius]

Haha... while you were musing about customizable species I was about customizable backgrounds.

[pratamawirya]

Haha... while you were musing about customizable species I was about customizable backgrounds.

and you both have names that start with "x", what the heck...

[XuaXua]

Customizable backgrounds is easier since, over time, all backgrounds are the same background;

the only differences (offhand0 are

- starting skill point distribution
- starting equipment (and quality of)
- starting diety (or not) or pre-allocated piety

Without revealing my hand, I've done this sort of character decompilation before (and been noted for it) for multiple pen-and-paer RPG and board game systems that don't have them. I was hoping there would be some sort of data or general guideline to start with, though. And considering things in trunk change semi-regularly (No more Stabbing skill and all species get +1 Evo!), fully locking down a build system would be tough.

Then-again, locking one down might make it much easier to build outward from one and see how changes to said system compensate.

[Bim]

I principle, I love the idea. It'd fit with random gods and allow for interesting combinations. If nothing else it'd mean a lot less 'I want this species' discussions and arguments whilst allowing players to prototype their ideas before asking for them to be added to the 'preset' list.

However, a lot of this requires differentiation between 'good' and 'bad' skills/mutations/abilities and crawl tends to deal in set trade-offs to balance things out (slow moving to balance higher defensive capabilities etc.). I can't see how this could be built in easily, although I'm sure there are ways to do some broad balancing to make it a useful tool, if not an actual part of a release.

Really would be awesome to use though.

[tasonir]

This could in theory be amazing but it would be a lot harder to get right than custom backgrounds. The potential for "dump stats" is just huge. Planing on going axes? Well you can take any amount of negatives to polearms, staves, short and long blades, unarmed, and maces. Planning on going for magic? You probably would love to have lots of negative skills with ranged weapons, which you'll never touch. Going melee? Well that fire skill is pretty useless to you.

Also if nagas are anything to go by, slow movement and half AC from body armor is worth about 9 good mutations: +9 ac scales, 5 MR per level, +20% hp, rpois, see invis, poison spit, large size (without losing slots), bardings (more rare but better than boots), constriction. Granted they also have -2 in defenses, but that isn't that big of a deal.

[Azrael]

I love this idea. I really don't understand this obsession with "balance" so many people have; Crawl is not a multiplayer game, so who cares if some CS are much stronger than others? If it's really a problem, ban them from tournaments. Same thing for customizable backgrounds.

[Lasty]

I love this idea. I really don't understand this obsession with "balance" so many people have; Crawl is not a multiplayer game, so who cares if some CS are much stronger than others? If it's really a problem, ban them from tournaments. Same thing for customizable backgrounds.

There's already Wizard Mode for people who want an arbitrarily-good start which isn't scored. The reason this is unsatisfying to most people who want high-powered stuff added to the game is the same reason balance matters: the game is only designed to be fun when its challenging. Not every race/background combo is equally challenging, but all of them are (intended to be) within a spectrum of challenges that are interesting to play out. The goal of balance isn't to make all playthroughs equally challenging, but rather to keep most playthroughs within the spectrum of challenge that is interesting.

[mumra]

I don't think there's a problem with balance; some races/combos are easier than others, this is welcomed rather than intentionally avoided.

I think the concept is interesting (as with having a customisable background) but the implementation is much harder for species. And making things like racial mutations modular is really tricky because of the way many things are hard-coded in the source, without a massive overhaul of lots of things. This would be an overall improvement of the codebase of course, but it's certainly far from trivial.

As I said in the customisable background thread, the interface for this would actually be pretty handy for testing things in wizmode too.