[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index]
[Thread Index]
[Author Index]
Return to main CEDA-L Archive Page
Why "lose 4" can work
Ok, here goes nothing. In the most friendly fashion possible, I am taking
issue with Professor Larson's numbers on "lose 4" and you are out. I, too,
have been running some simulations using tech and some skills I have
developed in my studies for engineering and computer science degrees. The
thrust of my position is that the rules for pairing the tournament are
the critical determinant of how long this process could take.
If we assume (1) the eventual happening of one final round in which the
winner is the winner and the loser is 2nd regardless of record, and (2) that,
except in rare circumstances (mainly outlined below), a bye is always
awarded to a team on the brink of elimination, I think we can make this
format work in 17 rounds. Yes, this is two more than we will have this
year, and this means that we probably have to squeeze in one more round on
Friday and one more on Monday. I think this is quite defensible in light
of the tradeoffs.
SOME ANSWERS TO PRIOR POSTINGS
I have two caveats before I delve into the analysis.
In my original post, I never claimed to
_solve_ _all_ problems with diversity conflicts (which may be more
accurately handled with mutual preference, though I still think it can
be _helped_ by this system) or devise a _completely_ fair system (however
that could conceivably look). My original post does claim that diversity
is a reality and that a lose-one-and-you're-out format in elims is inherently
unfair. Some reasons include (1) side disparities (aff wins more than neg
in equally matched rounds - that's why we all almost always pick affirmative
when given a choice), (2) a team that wins 8 _can_ lose to a 5-3 team
in a fair round and decision, but the 8-0 team shouldn't be penalized with
elimination, and (3) the 8 win team _can_ lose to the 5-3 team on a highly
questionable and somewhat unfair decision, in which case no one but a very
few would think this was OK. A reduction of unfairness is my goal, not a
complete solution. Think of this as comparative advantage.
As for the educational issues involved, I was/am very taken aback by
Professor Twohy's post. I have always perceived him as a reasonable
person who would like to see quality debate rewarded. Both the current
system and my alternative (or, "this" alternative, I suppose, since this
isn't exactly "mine") seek to educate by providing good rounds to watch
and perform and to reward teams that do well. If anything, those who
wish to see education would perhaps like to see a couple more rounds
and more rounds between teams at the top of the bracket. I don't know
about these issues, since I am not posting about them. I am interested
primarily in fairness.
A second caveat is that I have not run all scenarios. That perhaps should
be done to the extent it is possible before _adopting_ an alternative
system. My point is still that the current system is comparatively unfair,
and we should discuss the need for an alternative. If we decide we need
a new system, then we should indeed research it as thoroughly as possible.
I am not advocating a qualifying system for attendance at Nationals, but if
we had one, this "lose 4 and you're out" becomes even more feasible. If
we don't have one and the tournament does not become appreciably larger,
I still think we can do it.
THE NUMBERS
First, I disagree with Prof. Larson's statement about "high seed" always
wins and "low seed" always wins as the boundary tests. I think the worst
case scenarios vary. In some cases, this may be higher seeds, in some case
lower seeds. As far as I can tell, one bad case scenario is to always
eliminate teams on the brink of elimination when debating teams not on the
brink. If two teams down 3 debate, one is necessarily done after the round.
If one down 2 and one down 3 debate, then the down 2 winning is the worse
scenario. While we have eliminated a team, the down 2 has to still lose
twice more. This requires at least two more rounds and is less determinate.
We don't know how many other teams are in a similar state. On the other hand,
if the down 3 wins, then both are "on the brink" and we know that we can start
eliminating more easily. Gooey enough for you? Yeah, makes T and critique
debates look good. BTW, giving credit where due, the "lower seed winning good"
is the same as Prof. Larson's conclusion, so we partially agree on this. His
work helped quite a bit with me seeing this for some cases.
Some numbers to help. I am running case B from Prof. Larson. It seems a
harder scenario, since more teams have less than 3 losses. I ran the
other scenario and it turns out similarly. My assumption is that we pair
each round hi-lo within brackets. We account for sides as best we can.
My assumption is negatives will win enough to allow us to prevent a team
from having three affs without a neg. Perhaps we don't even care about
sides - it's elims, so maybe we flip if the teams haven't met. In the
first scenario, the lower seed always wins (a good thing according to
Prof. Larson all the time and me some times). Where an odd number of teams
remain, the bye goes to a down 3. Byes are awarded randomly among teams
that have yet to receive one. We are done in 17:
after
round... 8 9 10 11 12 13 14 15 16 17
# of losses
0 1 0 0 0 0 0 0 0 0 0
1 5 4 2 1 0 0 0 0 0 0
2 20 12 8 5 4 2 1 0 0 0
3 54 37 23 17 11 8 5 4 2 1
So, how about the higher seed always wins? This the same IF we assume
that when we get three teams, (1) if we have a scenario with one team
with less than three losses, the other two debate in "semis" and (2) the
winner goes to finals and we have our winner take all round.
after
round... 8 9 10 11 12 13 14 15 16 17
# of losses
0 1 1 1 1 1 1 1 1 1
1 5 2 1 0 0 0 0
2 20 13 7 5 2 1 0
3 54 37 19 9 8 5 4 2 1 finals in 17
If we reverse our assumptions starting in round 14, meaning the lower
seed starts winning, we still get a final round in round 17 with the above
rules. Finally, how about if we reverse in round 13? Still can do. Now
we get
after
round... 13 14 15 16 17
# of losses
0 1 0 0 0 0
1 0 1 0 0 0
2 1 1 2 1 0
3 5 3 2 2 Finals in 17
Yes, the same teams could debate as many as five times by the end of the
tournament with the above scenario. And, if one school had the two down
3's in rule (1) above, the coach would have to choose or make them debate.
In other circumstances, we can make accomodations or set up rules to
handle the situation. But, in my mind, these are better scenarios than
we get with the current system. Someone posted a statement to the effect
that random binary choice is by definition the quickest way to decide a
winner (I think it was Prof Larson, but I'm not sure). Whoever said that
is exactly correct - it is the quickest. But, while s/he did not mean to
imply a randomness in the sense I mean, it does illustrate my point.
Quickest can have some tradeoffs with fairest. I think we should spend
some more time to gain fairness.
CONCLUSIONS
These are not all the scenarios. Nor are these all the rules we would have
to consider. Nor do we eliminate all rounds where one debate is critical.
I happen to think having a "final round" is a good thing - it permits us
to have the "ultimate" round for everyone to watch. That aside, none of this
denies that the process of getting there is not as dependent on the luck
of the draw in one debate in terms of opponent or side or panel. I think
the above analysis also points to the potential feasibility of such a system.
I am also in favor of mutual preference judging. However, these concepts are
really unrelated except that both seek to change the format of the national
tournament and seek to address _some_ similar concerns. I do not see the
need to choose between them or to reject or accept both. Each has its
merits, but I think the "lose 4" method has some additional concerns not
addressed by mutual preference. Going 8-0 should have more rewards than
an "easy" elim draw. A very capable Macalester team once again proved that
theory of 8-0's wrong last year. One could also talk to Kansas State,
William Jewell, Emporia State, San Diego State, and so on.
I am of the opinion that we could have a "lose four" system. However, even
if I am eventually proven wrong, some alternative systems need
exploration. Our organization is not one conducive to the current type
of national tournament. My "vote" is that we seriously pursue some
research of the alternatives.
I am obviously still interested in opinions!!
Bill DeForeest
Gonzaga University
deforeest@gonzaga.edu
Archive created by Jonathan Stanton (jonathan@cs.jhu.edu)
Return to main CEDA-L Archive Page