Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Clean Character System for a Roguelike

4 views
Skip to first unread message

LucFrench

unread,
Apr 20, 1998, 3:00:00 AM4/20/98
to

I've been thinking about building a Roguelike recently, and a point came to
mind; namely, I've played Daggerfall; I liked the concept, but the system was
buggy and holeish.

So, how would one go about designing a "clean" classless CRPG? Especially a
Roguelike, where YASDs are common, and backing up your saves is techinally
cheating.

Thanks
Luc French

Torben AEgidius Mogensen

unread,
Apr 20, 1998, 3:00:00 AM4/20/98
to

lucf...@aol.com (LucFrench) writes:

There is a version of Angband (Sangband) that uses skills instead of
classes, but it is (in my opinion) not done the right way. In
Sangband, you gather general experience which you can choose to
improve any skill. Furthermore, you earn experience in the usual D&D
manner by killing creatures (and a bit by picking locks and defusing
traps). There is even a degree of classes in Sangband, as you can
choose only one type of magic. Additionally, there are some silly
skills in Angband (in particular "fennling", which is both unnatural
and hugely unbalancing).

I prefer a system whereby you can only improve a skill by explicit
use, learning or training in that skill. You don't get better at
picking locks by killing monsters and vice versa. Furthermore, there
should not be any exclusion of skills including magic, except maybe
for some natural opposites. And these should ideally not be absolute
exclusions but rather interference that hampers the opposing skills or
magics.

In the most finely grained system you have separate skills for each
weapon and spell, each capable of being learnt independently of each
other (but possibly allowing "similar" skills to be obtained cheaply).
A slightly coarser system might group weapons and spells into a dozen
or so classes, but still allowing each to be learnt individually. A
suitable set of magic types might be:

- Healing magic.

- Protective magic.

- Informational magic.

- Damage dealing magic.

- Mental (invisibility, telepathy, confusion etc.)

- Transportation (teleport self, thing or creature).

- Time (speed, slow, skip time etc.)

- Enchantment (enchant weapon or armour, make light etc.)

- Summoning (summon or dispell monsters, spririts etc.)

You could use different dividing lines, e.g. based on elements.

The idea with these quite specialized magic areas is that you can be
primarily a weapons user but know a little healing or protective
magic. Or you can learn all (or most) types of magic and not learn to
use any kind of weapon.

By letting a skill only be improved by use, training or learning with
a teacher, you not only get a more realistic system but also make
people consider if the want to invest time in using/training a skill
at low levels to gain an advantage that may only be worthwhile when it
reaches high levels. If, for example, you are good with an axe, then
you may consider twice before starting training with swords as it will
take a while before you skill in that will be comparable to your axe
skill.

As for cheating by copying a save-file, there is really nothing you
can do on a system where the user can make a complete file-system
backup, so you might as well give up on trying to avoid that.

Torben Mogensen (tor...@diku.dk)

Gerry Quinn

unread,
Apr 20, 1998, 3:00:00 AM4/20/98
to

In article <199804201404...@ladder01.news.aol.com>, lucf...@aol.com (LucFrench) wrote:
>I've been thinking about building a Roguelike recently, and a point came to
>mind; namely, I've played Daggerfall; I liked the concept, but the system was
>buggy and holeish.
>
>So, how would one go about designing a "clean" classless CRPG? Especially a
>Roguelike, where YASDs are common, and backing up your saves is techinally
>cheating.
>

Rogue itself had no classes. You just got hitpoints and (I think)
bonuses to hit and damage as you gained levels.

I wonder if the classes only came in with multi-character parties?

- Gerry


===========================================================
http://indigo.ie/~gerryq/Brewster/brewster.htm
Brewster Kaleidoscopic Screensaver for Windows 95
The only saver that simulates a real kaleidoscope
===========================================================

Travis S. Casey

unread,
Apr 20, 1998, 3:00:00 AM4/20/98
to

Torben AEgidius Mogensen <tor...@diku.dk> wrote:

>lucf...@aol.com (LucFrench) writes:
>
>>I've been thinking about building a Roguelike recently, and a point came to
>>mind; namely, I've played Daggerfall; I liked the concept, but the system was
>>buggy and holeish.
>
>>So, how would one go about designing a "clean" classless CRPG? Especially a
>>Roguelike, where YASDs are common, and backing up your saves is techinally
>>cheating.
>
>I prefer a system whereby you can only improve a skill by explicit
>use, learning or training in that skill. You don't get better at
>picking locks by killing monsters and vice versa. Furthermore, there
>should not be any exclusion of skills including magic, except maybe
>for some natural opposites. And these should ideally not be absolute
>exclusions but rather interference that hampers the opposing skills or
>magics.
>
>In the most finely grained system you have separate skills for each
>weapon and spell, each capable of being learnt independently of each
>other (but possibly allowing "similar" skills to be obtained cheaply).
>A slightly coarser system might group weapons and spells into a dozen
>or so classes, but still allowing each to be learnt individually. A
>suitable set of magic types might be:

[list cut]

Some muds use a concept called a skill tree or skill web; a structure
which shows how skills are related to each other, and which can allow
skills to influence each other. For example:

Combat
+-Melee Combat
+-Slashing/Chopping Weapons
| +-Slashing Swords
| | +-short sword
| | +-long sword
| +-two-handed sword
|
+-Axes
+-hand axe
+-battle axe

(of course, this is only a partial table)

A simple way to use such a table is to allow skills to default to the
meta-skills above them, and to make a meta-skill be equal to the value of,
say, one-quarter the highest skill directly under it. Thus, raising one's
long sword skill to 100 would give one a skill of 25 in Slashing Sword
(which can be used with any slashing sword), a skill of 6 in
Slashing/Chopping Weapons, and a skill of 1 in Melee Combat. (Of course,
more complicated methods of transferring skill points are possible -- for
example, one's skill in a meta-skill could be the average of the levels
of the skills below it.)

Meta-skills, of course, can't be directly learned or trained -- they only
exist for the purpose of defaults.

A more sophisticated version of this is what's called a skill web -- a
graph with skills as nodes, and weighted links between them (possibly
directed). Thus, "long sword" might connect to "short sword" with a weight
of .25, but to "two-handed sword" with a weight of .1, and to "rapier" with
a weight of only .05). Meta-skills can still exist in such a system, to
simplify the graph.

>By letting a skill only be improved by use, training or learning with
>a teacher, you not only get a more realistic system but also make
>people consider if the want to invest time in using/training a skill
>at low levels to gain an advantage that may only be worthwhile when it
>reaches high levels. If, for example, you are good with an axe, then
>you may consider twice before starting training with swords as it will
>take a while before you skill in that will be comparable to your axe
>skill.

With skill trees/webs, more considerations come in -- for example, it
may take a while to learn to use those swords effectively if you're already
skilled with an axe, but you may find that your training with an axe has
prepared you well for learning to use a pick.

This can also encourage skill specialization more -- if you already know a
lot of magical skills, learning more magical skills will be cheaper than
branching out into combat skills, because the basics transfer... like it's
usually easier for a physicist to switch over to doing mathematics than to
switch to doing art history.

>As for cheating by copying a save-file, there is really nothing you
>can do on a system where the user can make a complete file-system
>backup, so you might as well give up on trying to avoid that.

Well, there are some things you could do on most such systems (like have
code which checks to make sure the file is in the same i-node or disk
sector still), but why would you want to? I personally dislike games
which make me redo everything just because of a simple slip-up. If you
want to control how players can save, just set things up so they can only
save at certain points in the game... that can give a similar effect
without being as annoying to the user.

--
|\ _,,,---,,_ Travis S. Casey <efi...@io.com>
ZZzz /,`.-'`' -. ;-;;,_ No one agrees with me. Not even me.
|,4- ) )-,_..;\ ( `'-' Keeper of the rec.games.design FAQ:
'---''(_/--' `-'\_) http://www.io.com/~efindel/design.html

Peter Knutsen

unread,
Apr 27, 1998, 3:00:00 AM4/27/98
to

LucFrench wrote:
>
> I've been thinking about building a Roguelike recently, and a point came to
> mind; namely, I've played Daggerfall; I liked the concept, but the system was
> buggy and holeish.
>
> So, how would one go about designing a "clean" classless CRPG? Especially a
> Roguelike, where YASDs are common, and backing up your saves is techinally
> cheating.

I'm working closely with Laurence Withers
<lwit...@lwithers.demon.co.uk>
on a game called LW-angband (it's supposed to be a variant of angband,
but
with so many changes that I'd say it's a whole new game)

I sent Laurence some material, some of which is clean (he's hosting that
material on his website, it's the FFRE stuff). You can use that material
for non-commercial purposes.

Last month I sent Laurence some sketchy notes (because I was pressed for
time)
which I've asked him not to publish, this is more detailed ideas, but
they
are inconsistent and contain typos, and so on...

With regards to the "backing up your saves" problem, I suggested to
Laurence
that the player should have something called "luck points".
FFRE is originally designed for face-to-face roleplaying, but can easily
be
adapted to a computer game. In face-to-face, the player can spend a luck
point
to re-roll any roll which affects him or her. How fast the luck points
are
regained determines the realism-level (fast regain=low realism, slow
regain=
good realism).
The number of luck points are determined basically by the Faith
statistic of the
player, but there is also a Luck Bonus which is independent of Faith
(I'm not
100% certain how Laurence will implement this, probably as an advantage)
Cuteness should also give more Luck (in a Star Wars game, Ewoks would be
considered cute. In a Tolkien setting, hobbits would, and so on.)

Oh - in a computer game, Luck points aren't used to reroll dice, but
each time
you save your position, you must pay one Luck point. They are regained
slowly
(how slowly should probably be determined by the difficulty level,
selected by
the player).

This is the link to the LW-Angband site, Laurence is doing the
programming, I'm
just supplying lots of ideas on the game mechanics.

http://www.lwithers.demon.co.uk/lwangband/


> Thanks
> Luc French

You're welcome to use my FFRE system for any purpose that does not
involve money.
It's 100% skill-based, and there's no AD&D smell left from the standard
Angband
games.

--
Peter Knutsen

0 new messages