[crossfire] stat loss and highly specialized players

[Raphaël Quinet]

On Tue, 19 Sep 2006 19:06:18 +0000 (UTC), Alex Schultz <alex_sch at telus.net> wrote:
> Raphaël Quinet <raphael at ...> writes:
> > By the way, I added a link to
> >   http://wiki.metalforge.net/doku.php/dev_todo:exp_table
> 
> A few notes on changing the exp table: To me, the shape of the one on that page 
> you label "Progressive 12%" seems to have what I would consider to be the most 
> reasonable shape for the exp curve, however one question which doesn't seem to 
> have been considered yet, which I440r__ reminded me of by saying something to 
> the meaning of "I wouldn't mind the exp table being scaled up 4x", is do we 
> want the exp table to top out at the same exp value as currently like in the 
> proposed table? I personally think we might want to consider something 
> like "Progressive 15%" in scale, except have the last levels taper upwards like 
> in the "Progressive 12%" one, such that the level 115 exp would be a good bit 
> above what it is currently. I think I may look into making a proposed chart 
> some time.

Yes, the top level (115) could be significantly higher than it is now.  Or we
could use a curve that keeps it roughly at the same level and then adjust some
monsters.  In fact, I don't think that the amount of exp required for levels
beyond 110 is very important to consider because this is not where most
players should be.  Those who want to reach the demigod status will do whatever
is necessary to reach it anyway. :-)

By the way, I made lots of other tests with other curves and I tried to figure
out how these would map to my game experience.  I did not put them all on the
wiki because that would be abusing this nice resource, but maybe I will
consider adding one or two new curves anyway.  One of my favorites is similar
to "Progressive 12%" but uses a 10% slope instead.  It crosses the green curve
(classic exp_table "B") a bit before level 100.

If you want to experiment with exp curves, I can send you my Gnumeric sheet.
I can probably be read by Excel or OpenOffice if you don't have Gnumeric.

> Also, we may want to look at tweaking individual monster exp in some cases. 
> (there are probably many cases that are too generous or don't give enough)

Right.  As I wrote in a previous thread, we should first decide on a new exp
curve and then adjust the "sources of exp" (monsters) according to that curve,
not the other way round.  Using a progressive curve with a constant ratio
between levels should make it much easier to adjust the "sources of exp".

> Another thing, we may want to consider making exp loss for skills something 
> other than the plain percent. Overall should be by percent IMHO, however it may 
> be worth considering a system, which wouldn't hurt low levels skills of a high 
> level character as much (thus making the fact that non-combat skills are hard 
> to gain, and making death hurtfulfor them, less of an issue).

No, this problem is caused by the current exp_table and death_penalty_levels:
it limits the death penalty to 3 levels while the exp_table "B" (current default)
requires only 1% between levels around level 100.  As a result, a player who is
level 100+ in some skills and only about level 10 in some non-combat skills will
only lose 3% of the high-level skills (max 3 levels) but will lose the full 15%
of the other skills.  This is what is wrong: the current system artificially
limits the loss for the skills in which the player has reached a high level, but
still deals the full penalty to the other skills.  If the system would be fair,
there would be no death_penalty_levels limit and the player should lose 10 levels
or more of the high-level skills (with the current exp_table "B").  This may
sound much in terms of levels, but this only represents 15% of the exp.  Then it
becomes obvious that levels are gained (and lost) too easily near the high
levels.

As I wrote several times (do I sound like a broken record yet?), the correct way
to fix this is to use a progressive exp_table and get rid of artificial limit
caused by death_penalty_levels.  If after doing that there are still concerns
about the fact that losing 15% exp hurts too much, then the right thing to do
would be to change the death penalty to 10% or so.

Personally, I think that a 10% death penalty would be better than 15% (at least
as long as there are some "instant death" traps in the game) and an exp_table
that uses a constant ratio of +10% between levels would also be better.  As a
result, each death would cause the loss of approx. 1 level in each skill and
there would not be this unfair difference between high and low levels.

-Raphaël