Semiotics

I actually wouldn’t mind ordinals; the idea you’d spend hours figuring out distinctions between your 75th and 76th judge is spurious.  If you can’t distinguish between a couple judges, then it doesn’t matter what order you put them in: just put one as 75 and one as 76 and move on with life.  It’s easier in practice than you’d think.  Plus, Tabroom actually has rather nice features built around ordinals; it can auto-generate a pref sheet for you based on your past ratings of a judge as a starting point.

And you’ll pardon me if I casually shrug off the conservative objections to Something New from the group of tabbers whose original schedule for switching over to online tabbing & balloting would have had us trying it out for the first time in early 2016.

But to see the point I’m trying to make about narrower band categories, you start with an ordinal sheet; the “true” sheet of how a given debater prefers judging.  A true ranking of preferences won’t necessarily be a smooth linear progression; it’ll have little clumps.  There’ll be a nest of three judges in the topmost spot, and then maybe a clear 4th, then maybe 3-4 judges tied for 5th, and so on.  But the size and location of those clumps will be random enough that the pref sheet is essentially a gradient.

A tiered system necessarily imposes arbitrary boundaries on that gradient, turning them into bands.  The fewer (and wider) the bands, the more information is lost.  Menick says that when you rate a judge a 2, they’re a 2, and thus are magically mutual with every other 2.  But plonking a label on a judge doesn’t make it so.  The judge could be my favorite 2 and your least favorite 2, in which case the judge isn’t very mutual at all; there’s a lot of slope between our opinions.  The most mutual judge, in fact, may be my favorite 2 and your least favorite 1, being separated only by one notch made more significant by the random chance of the tournament policy.

Look at tournaments that require you to rate 25% of the field a 1 and 25% a 2.  You’ve now rated half your judges in the top two tiers; and you’re going to get all kinds of judge matchups where the gap in preferences may differ by as much 24% of the field.  A 1-2 matchup in that case could be killer.  It’s not uncommon — and not difficult — for such tournaments to put out pairings with all 1-1 matchups, maybe a few 2-2s.  These matchups sometimes really stink; you end up with your opponent’s favoritest judge in the world when it’s the person you held your nose and plunked a 1 because you really needed just one more.  I’ve been on the wrong end of those pairings.

Now take Lex, where 11% were 1, and 11% were 2.  Lex’s 1-2 pairings are a little more precise; they all fall within the top 22% of your pref sheet.  Further, by drawing a line in the middle of that pool of judges, you can minimize their number.  There’s nothing a tournament requiring 25% of the field to rank 1 can do to minimize judge pairings that may be 24% of the field off on people’s actual preferences; it doesn’t have the data, and so doesn’t know that another 1-1 might actually be far more mutual in the debaters’ actual preferences.  Lex didn’t have any matchups 25% of the field off, and minimized the number of pairings that are 11-22% of the field off.   It’s more mutual, by which I mean more reflective of the actual preferences underlying the numbers on the pref sheet, not the fake mutuality of categories, which conflate the signifier and the signified.

I don’t buy the argument that new schools are going to be screwed by not figuring this out; that’s an argument to abandon prefs, not an argument to make them blunt versus narrow.  It’s also non-unique harm; new schools have to figure out theory, framework, impacts, cards, and spreading too.  It’s not the pref sheet keeping rookies out of the bid round.  A new school could get a judge that really favors their style and disfavors mine, and chances are my kid will beat them anyway, if only because I will know the judges’ preferences and the new coach won’t.   “You have to be smart to do that!” is true of nearly everything in debate, and should rather remain so.   Plus, LD is way more open to new schools and disruption than policy is; new schools are competitive on the national circuit all the time.   The answer to this is helping and teaching new schools how debate operates, not changing how debate should operate.  I put my money where my mouth is there, too: I’ve filled out pref sheets for lots of other schools.

In the end the debate is immaterial, I believe.  Whatever the merits, I think the trendline towards greater precision is inevitable.  TRPC limited LD to 6 while Policy could choose 6 or 9; 6 remained the norm in LD but in Policy most tournaments use 9 — in other words, both ended up using the largest number of tiers they could.  College policy, thanks to the good professors Larson and Bruschke, had software that did ordinal prefs years ago, ordinals they have used, with few exceptions.  Now that Tabroom has lifted the arbitrary 6/9 limits, I think high school debate, for high stakes tournaments anyway, will move towards ordinals, no matter what we say or do.  I embrace it, but even if I didn’t, I would think it’s going there anyway.   It’s a natural progression outwards from strikes to prefs, when you think about it; strikes are simply a really inaccurate pref sheet.  So I’d say, practice your ordinals…

…and just wait until I explain the percentile system to you.

Answering innumeracy with data

He still doesn’t understand my point about increased mutuality, but I’ll write that up when it’s not  12:30 AM on a Tuesday.

For now:

  • Percentage of 1-off judges at Lexington:   8.46%
  • Percentage of 1-offs at Columbia:  8.33%

Man, really blew mutuality to shreds with those 9 tiers at Lex, didn’t we?

At Columbia, 12.5% of the VLD pool did not pref; at Lexington it was only 9%, thus making the job harder at Lexington to boot.   As we say in the business, “No Link.”