Re: call for duelmasters help

From: johnjohn@cco.caltech.edu (John White)

Date: Thu, 29 Oct 1992 01:59:05 +0000

In article <00962C26.CA04F3E0@abby.chem.ucla.edu> ben@abby.chem.ucla.edu
writes:

>Oo! Oo! Someone on my side...
>(below, I mean, not JOhn...)

But that was Dennis...

>>Now, I know we argued about this months ago when I posted that little bit
on
>>skills (John, do you have that anywhere's?), but I really don't think
there'
>>s an actual luck stat per se.  This is because there's enough random
factors
>>in the game already that can make a warrior seem lucky or unlucky.  Also,
a
>>lot of warriors who it might seem are blessed with a high luck factor do
go
>>into slumps eventually.  If they were truly 'lucky' this wouldn't happen.
>>You remember that slasher I kept talking about?  Well she went to the Isle
>>at 19-2, and has gone 1-3 in AD, with an 0-3 in the April tourney.  Also,
>>except for last summer's FtF, she never did well in tourneys period.  Now,
>>if dhe had been blessed with ahigh luck stat, as her regular arena
>>performances would indicate, she'd have gotten at least one TV along the
>>line, in spite of her shitty abilities.  She was just like that while in
>>regular DM.
>
>Exactly my point.  In general, the sample size (i.e. number of fights) is
far
>to small to average out the random factors.  Hence the illusion of 'lucky'
and
>'unlucky' warriors.  One of the problems has to to with the factr that the
>probababilities ARE NOT independent.  Assuming winning warriors learn (on
avg.)
>more skills/fight, the more you win, the more skills, the more you win, the
>more skills, ad infinitum.  The opposite is also true, leading to the
>conclusion, WITHOUT a luck 'stat', that winning warriors tend to win and
learn
>more; while loosers keep loosing...  In this way, a 'Dixie' with WT 11, IF
they
>are good enough to win lots, and 'lucky' enough to get higher thatn average
>skills, they can still be a decent warrior, and even suceeed in ADM.  This
is
>probably not true for WT < 11, however, and the 'minimum' could be 13, or
even
>15.

But Ben, when you make an argument like that, you start from assumptions
that everyone agrees on (or prove your assumptions), then work your way
towards a conclusion.  In this case, I DON'T agree with your assumptions.
I've had plenty of warriors who won consistently and learned horribly.

In addition, I _have_ run 11-13 wit warriors, and no, they DON'T average > 2
skills a fight when they're designed to be dixies.  That's one reason why I
don't dixie anymore.  It isn't worth the effort.

>So, to get back to the original argument I had with John, a bad 13 WT
>warrior may learn 1 skill/turn and take 45 turns to get to ADM, while a
good one
>might learn >2/fight and become a decent warrior.  From this It would seem
>logical to make LU, TP's and WS's with a lower MINIMUM WT, because these
styles
>win a lot.

Heh, heh, heh.  Ben, remind me to show you this post in a year.  Maybe this
is just something that you have to see for yourself.  Warriors like Demon's
slasher, who learn, despite being just a power warrior, are few, and far
between.  Let me just save this post...

And ultimately, nothing much comes of those guys.  All warriors who win
based on luck will top out at some point where people's skill happen to
outdo thier luck.  This happened to Gerald's slasher when it got to AD.  It
happens...

>All this is without assuming a 'luck' or 'learning' stat, or a
>'lenbgth-of-fight' factor (my original propisition), assuming only a small
>sample size, and a Win-learn factor.  This is not a proof, just an opionion
>based on my (and also hersay) observations, using logical arguments.

Wait, so your ultimate warrior would be a 13 wit TP?  I mean, why waste
points on a 21 wit, when all you have to do is make a TP with long fights
that wins a lot, right?  Right?  Wrong.  Well, let me rephrase that.  I
don't buy it.  Do you really?

I mean, I understand that we're trying to talk solely about skill learns
here, but I should point out that generally, warriors with a higher wit end
up with more skills than warriors who raise thier wit higher.  The starting
skills effect.  Want to argue about that?

>John, you also mentioned a 19 WL warrior who missed their first stat train.
I
>was 'told' that the 1st stat train was (WL*5%), giving your guy a 95%
chance.
>Missing this roll once, is unlikely, but haven't you ever fumbled in a
%-based
>game, (usually 5%..)  If on the other hand, if that SAME warrrior missed
his 1st
>train 12 times in a row, I would wonder. (odds = 1/20^12 !!!).  NO evidence
for
>a luck stat here...

Chill.  Again, that was Dennis.  I think his point was along the lines that
unusual things DO occur.  That's all.

>To risk flogging a dead horse, the necessary sample size is inversly
>proportional to the 'narrowness' of the distribution function around the
mean
>(presumed Gaussian).  This distribution (discrete, with values from 0-6,
>representing the number of skill/fight for that specific fighter.) also
varies
>from fight to fight, depending on the relevent variables, whatever they may
be.
>I think, but I'm not a statistician, that these variables, with WT as a
>constant, will always 'widen' this distribution, rather than 'narrow' it.
All
>I'm trying to say, is that I think that the number of variables increases
the
>'streak factor' of a given warrior.  In Chemistry language, the 'streak
factor'
>is a cooperative transition...

Ok, that makes it all clear.

John White
--
John White
johnjohn@cco.caltech.edu
|===============================================================|
|                          What the HELL?!                      |

References

Re: call for duelmasters help
ben@abby.chem.ucla.edu
Wed, 28 Oct 1992 13:52:41 +0000

Referenced By

Re: call for duelmasters help
ben@abby.chem.ucla.edu
Wed, 28 Oct 1992 13:52:41 +0000