Re: ARENA: New "Tournament" Format

From: carl@gergo.tamu.edu (Carl D. Perkins)

Date: Fri, 01 Oct 1993 02:16:00 +0000

srt@aero.org (Scott 'Dr. Pain' Turner) writes...
>  * No open challenges.  You may challenge any fighter with as
>    many wins as you have; he must accept the fight.

What if he wins a fight (or several) in the turn you challenged him before
you get to him and no longer has the same number of wins?  Will he still fight
the challenge? (I assume so, since the orders are parsed before the turn is
run.)

This also makes life difficult if you miss a turn - like on a weekend, for
example.  Your list of players with the same number of wins will be out of date
and you'll have to waste another turn to do a ROSTER.  Or make a lucky guess.

If you have the highest number of wins, you also get stuck waiting for someone
to catch up to you so you can challenge each other, and anyone can get stuck in
a gap - there just happen to be no other players with 17 wins for a couple of
turns.  The same problem happens on the low end - you loose your first fight
and the character you fought was the only other 0 win character, until another
new character shows up you are stuck.

Unless this means you can challenge any fight with _at_least_ as many wins,
not just an equal number - in which case whoever is listed in any top-3 or
whatever will be getting a lot of bouts.

>  * The ARENA will run for a limited time (4 or more weeks) and a
>    champion will be crowned at the end of that time.

Based on what?  Most wins makes the person who ran every turn, posting lots
of challenges, the most likely victor.  Best ratio sorted in order of
wins makes that 3-0 gladiator have a good chance at winning even if he only
fought in one turn and those were against 3 gladiators played by brand new
players who forgot to wield their weapons. The wins*win_ratio also has a bit
of a problem - since this works out to (wins^2)/(wins+losses) a character with
a lot of fights with a low ratio beats out a character with fewer fights but
a much better ratio: is Bumble (80-210) (wr=.28, w*wr=22.1) really better than
Mr. Skill (32-16) (wr=.67, w*wr=21.3)?  The former lost 2.625 bouts for every
win whereas the latter lost only .5 times per win - it looks like Mr. Skill
really is better than Bumble, but has simply not fought as many bouts,
possibly due to getting a late start or having slow mail turnaround. If
he had played 1 more turn during the time perios, wiunning 1 more bout,
Mr. Skill would come out on top of Bumble in this rating system (with 22.2
points) but Bumbkle would have had to loose 10 more to make it a tie or 11 to
make Mr. Skill the victor.

Makes the player who joined half way though the time period or has a slow
mail turnaround preventing more than 1 turn per day has a distinct
disadvantage for any of those methods except the "best ratio" - which is not
a very good comparison.

What to do about it?  I'm not sure.  Pretty much any method other than an
actual tournament has a problem when the number of bouts is not the same for
everyone.  Closing the game to new characters in the last week, and running
bouts between all possible matches on the last 2 turns might give the skillfull
but low bout count a chance to catch up to the unskillfull but busy.

>  * To make it easier to schedule fights, the ARENA will provide
>    a list of "Suggested Matches" in each turn report.

That's good.  Saves the trouble of having to get a full roster every turn.

--- Carl

References

ARENA: New "Tournament" Format
srt@aero.org (Scott 'Dr. Pain' Turner)
Wed, 29 Sep 1993 04:32:02 +0000

Referenced By

ARENA: New "Tournament" Format
srt@aero.org (Scott 'Dr. Pain' Turner)
Wed, 29 Sep 1993 04:32:02 +0000