# [MUD-Dev] Re: Levels versus Skills, who uses them and when.

J C Lawrence claw at under.engr.sgi.com
Mon Jan 11 18:36:16 New Zealand Daylight Time 1999

```On Tue, 05 Jan 1999 15:31:06 -0800

> I've been giving this a lot of thought lately, because my mud is
> going to use a levelless system. The problem now becomes, A) How
> do we show a clear path of advancement? and B) How can players
> keep track of "where they stand" compared to the other players. I
> was presently thinking on borrowing a system used by a LP mud I
> used to play. They call it an "eval" system (maybe they all do, I
> only found one LP mud I really got addicted to) but the eval
> system works by making a composite number score based on the
> stats and skills a person has over what they started at.

I use what I call "perception stat" model.  For me statistics as
reported to the players are not absolute values, they are perceived
values.  The simple example is:

> score
You are very strong.
> lift rock
You struggle and strain to lift the rock.
> score
You are fairly strong.
Bubba approaches.
> score bubba
Bubba seems fairly strong.
Bubba lifts the rock with ease and throws it down the hill.
> score
You are fairly weak.
> score bubba
Bubba seems quite strong.

The underleing model is that a character has a set of absolute stats
and then a record of how he has seen those stats manifested in
others along with a simplified standard deviation applied to the
distribution (I keep a brief history, a prior arithmetic mean value,
a prior deviation value, and then fudge new figures as I go along).
The result gives the character a sense of a bell curve and where he
lies on that curve.  The *result* of that placement on the curve is
what is reported to the player.

I don't have any concept of "level" per se, but admit its usefulness
as a motivator by providing a series of known thresholds.  I'm not
convinced yet that it is necessary (cf UO and its use titles,
reputation, notoriety etc as level analogues).  Reading the various
UO and DSO fan sites, it seems that the players are rather fond of
the idea that a character has multiple "flavours" each of which is
better/worse to some degree.  This allows more complex and
interesting comparisons to happen, much like the testoterone games
of RL:

I've got a V8!
But I can do nought to 60 in 10 seconds!
But I can do hard corners at 30mph without spinning!
I can beat anybody over the quarter mile!
Yeah, but I'll slaughter you over a 10 mile stretch!

etc.

Were I to do a "level" equivalent I'd have to base it off the
perception model as otherwise I'd expose the mechanics of the stat
engine.

> The problems are numerous, however. Some players will have a lot
> of points in less useful skills, and still be granted eval for
> it. (Or some classes simply have more skills or less powerful
> skills than others)

The assumption of course is that a single scalar can provide a valid
comparitive of different characters.  I dislike this assumption and
the import it has on the game design (that progress thru the game
can be gauged and measured by a single metric) as it encourages
simplistic approaches to problem solutions (eg "when in doubt kill
the nearest thing that moves") and games with little flavour and
variation (cf Ambushbug's Trash).

> This leads to a lot of doctor-type class
> people complaining that a fighter-type of half their eval just
> kicked their ass in the arena.

And just why did a doctor of level X expect to be able to survive
the attack of a fighter of the same level?  Surely they must have
*expected* to be almost slaughtered outright.

> This creates a fundamental power imbalance, since eval is really
> pretty arbitrary, and even fairly easy to manipulate by picking
> up skill levels in select areas.

The use of a single comparitive scalar value encourages the view
that different value sets are really equivalent (eg the doctor and
the fighter with the same "level" stat).

Think of this in data normalisation or even simple entropy terms: If
large data sets can be reduced to smaller data sets without
functional data loss, then the smaller sets can functionally
replace the large sets and the (simpler) computation is just as
valid as the original more complex (more data) calculation.  Ergo
the larger data sets are insignificant and you might as well do
everything with the smaller sets.  Add a little bit of ad infinitum
and you get equations like:

if playerA->level > playerB->level then
playerA.kills(PlayerB)
else
playerB.kills(PlayerA)

Any further calculation or variance really is moot.

> As you gain levels, you gain access to cooler places that less
> famous people can't go. (Select mud areas, clubs, pubs, and maybe
> even stores with nicer items that only cater to those of
> status).

The problem with exclusive areas is that they have small
populations.  MUDs are inherently built on communication -- ya gotta
be able to talk, even in combat (non-talker) MUDs.  There really is
not that much value in an area if there's nobody there to talk to.

--
J C Lawrence                              Internet: claw at kanga.nu
(Contractor)                             Internet: coder at kanga.nu
---------(*)                    Internet: claw at under.engr.sgi.com
...Honorary Member of Clan McFud -- Teamer's Avenging Monolith...

```