[eDebate] A less risky experiment
Mon Nov 5 23:45:00 CST 2007
This suggested experiment from Stefan has the virtue of taking the
guesswork out of the discusison.
Rather than having predictions ( "X will happen" vs "no it won't, Y
will happen), you can actually determine experimentally how the
population of judges translates points from the 30 point to the 100
point scale. These data could then be reported, and you would get an
idea of the skew... given that it is a 3x greater scale, you would
expect a roughly 3-5 point spread for each point, and a nicely curved
distribution. If it is a lot bigger than that, or if there develop
odd distributions, then this could be determined, and people can try
to work on that issue, before tournaments are determined on the new
So simply ask people to double enter points as the first experiment -
on the 30 point scale and the 100 point scale, and developing a
dataset that can be analyzed. And in the meantime, still use the 30
On 11/5/07, Stefan Bauschard <SBauschard at planetdebate.com> wrote:
> How 'bout using the status quo point criteria.
> Next to your 27-29.5 you put what you would give on a 100 point scale.
> The tab room records both scores (only one would need to be recorded
> Comparative results are generated. Judges/coaches could vote online
> (or a general discussion could ensue after the tournament) about what
> is more intuitive.
> Determining speakers in the status quo (particularly say speakers 3-9)
> may be very arbitrary, but I doubt that in most instances most people
> think those are not the 3rd-9th best speakers.
> What team is seed 30 and what team is seed 34 is probably pretty
> arbitrary. However, I'd wager that just about everyone would think
> that in most instances in the status quo seed 32 is better than seed
> The status quo may not be the best, but it does not seem to be entirely broken.
> As many others (recently David G and Scott H) have pointed out, there
> will be radical differences in this first tournament. I see no reason
> why seed 40 could not become seed 32 at Wake under this system. Does
> anyone even seriously doubt that seed 50 could become seed 32?
> So in the short-term there may be *radical* differences in seeding and
> who clears due to a loss of stability in the scale. Over the
> long-term (David G suggests maybe 20 tournaments) this could work
> itself out (assuming the scale recompresses).
> This would only work, however, if other tournaments adopt the 100
> point scale. Has anyone committed to this? If not, what's the value?
> It seems to me that the arg of short-term randomness/instability is
> very hard to answer and that that is likely to discourage adoption by
> My proposal here has the potential of conducting the experiment
> without drammatically increasing randomness of the status quo.
> I like David G's idea, too. It works at identifying more subtle
> differences without generating greater randomness in the interim.
> eDebate mailing list
> eDebate at www.ndtceda.com
More information about the Mailman