[eDebate] Reacting to reactions

repkowil at msu.edu repkowil
Wed May 20 14:13:01 CDT 2009


Ok -- received many emails about the First Round Balloting thing. 

Reactions:

1. No such thing (in this system) as a bad win 

This year, I divided teams in 7 tiers... several teams were in a mythical 8th tier, where defeating them was worth zero positive points. 

But, beating a team that goes 4-4ish at majors (irrespective of venue) was usually worth +1 or +2 points... It is not the same as beating NW FW (worth +7 points), but smaller wins do add up.?

... there were -- however -- some 8th tier wins that Kansas BJ (for example) had that were worth zero additional points. This did not help them... and it did not cost them (this appears to be Mike Davis' question if I understand it correctly).

To be frank, the top 20 teams beat the teams in the mythical 8th tier like 99% of the time... Giving points for such wins would start to become a referendum on raw quantity of debates... Some have backchanneled and recommended "raw number of debates" as a slightly under-estimated criteria. I will consider that -- but that's part of a different post altogether.

.... so, when I say that KU BJ had +226 in 80 debates, it d/n mean that KU only had 80 debates all season... it means that they had 80 debates against the top 7 tiers... I average that and compare it to Cal BP's 66 such debates... 

2. Yes, an ordinal process could be used to break some ties or tweak the system...

... Olney did something like this and I chalk that up to him having more advanced math skills than I do. I think DCH and Stables also have used some aspect of ordinals in their voting... They could comment more effectively on this that I could. But, I think that is an area for improvement in my system.

3. Regional travel

I am a little less concerned about the effect this system will have on regional travel than others. 

As I have understood it in the past, the fear is as follows:

a) there is local tourney that we want a first-round team to attend
b) first round team will pass -- not b/c they need the prep time for the next "major"... and not b/c of school... but b/c there's more mathematical downside (bad losses) than upside (good wins to be had) in the pool...
c) this will hurt the tourney via a bandwagoning effect

However:

a) Having a tier five, six, and seven (as opposed to say 3-4 tiers only for applicant level teams) makes a huge difference. 

Yes, under old systems -- where only "applicant level wins" were considered -- a team could rationally say "there are zero or few other prospective applicants, so why go ?...".... But, this system gives positive points -- sometimes 2 or 3 positive points per win -- for certain non-applicant wins. As explained in the previous post, defeating Michigan LZ (not a an applicant) should be worth something. Wins against that tier of competition are available at a great many regional events. 

Also, LOSING to Michigan LZ is weighted as LESS damaging to the resume than losing to a less-accomplished team. So, if you go to a regional tourney and debate Michigan LZ it counts just as much (or little) as debating them at the Wake tourney. To me, that seems fair to the applicant and to regional travel. 

b) Pragmatically, the bid/applicant level teams are not hurt under this system -- unless they take multiple, multiple losses at a regional event (which should count against you).

These teams (overwhelmingly) don't attend tourneys where zero tier five, six, and seven opponents compete. Thus, they pick up a bunch of +1 and +2 style wins. 

And -- for several reasons -- these bid-level teams tend to lose 0-2 debates at regional tourneys. Thus, their "decent, but not quite 1st round" level wins tend to more-than-offset their 1 or 2 "bad" losses. 

More concretely, an early elim exit (at a non-major) did not really hurt UTD or Oklahoma's bid sheets in this system. That is b/c the "Av elim distance at majors" focuses in on the elim performance *at majors*....?

... so, those losses hurt -- but just as much as losing to any team at any venue or any stage of the tourney (in that it effects the +/- rating only). They do not effect the important (but secondary) elim depth stat at all.

While I am sensitive to supporting regional travel, I do think we need to notice distinctions in the quality of performance at those venues. For instance, UNT CP and USC JL did a better job at non-majors than Whitman, UTD, or Oklahoma... That's not meant to be rude -- it is factual distinction in those resumes. It made it easier for me to vote USC and UNT ahead of those three other teams.

If we want to build-up regional travel, I do caution against a card blanche that says "it can never hurt"... It should help if one does well.. it should not be designed to "hurt"... but "hurt" is contextual... If I reward UNT, I am basically "hurting" a bubble team in the zero-sum-game that is 16 total first-round recipients. 

To me, data is data... win more and build your resume.. lose more and it reflects poorly on one's resume... 

c) This is not meant to be a reaction to ErMo system. 

For one -- I think it is important that voters vote differently and spell out their criteria.

For another, I am confident that ErMo's system "notices" that USC and UNT did better at regional events than Whitman, UTD, or Oklahoma. 

I much more concerned about systems that treat "all losses the same", or ignores poor performaces *because* they are regional performances. I do not know of a voter that presently does that.

4. What about second-rounds ?...

I modify this system greatly for second-round balloting. 

I also think regional results are all-the-more important in the second round process. 

More on this upon request.

all for now,

Will
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.ndtceda.com/pipermail/edebate/attachments/20090520/4f6f854d/attachment.htm 



More information about the Mailman mailing list