High-school debate: Pro
my experience as a judge
About a month ago, my wife forwarded to me from a local mailing list a call for volunteers to help judge in a high school debating tournament. I eagerly expressed interest. Participants in high school debate practice civility and intellectual rigor.
Last week the tournament was held. It was a very positive experience. I also think that a project to train an artificial intelligence model to judge such a debate would be worth attempting.
If I were a college president, I would require every freshman to participate in this sort of debate tournament. If I were in charge of Twitter (or X or whatever), I would require users to participate in this sort of debating activity as a prerequisite for being allowed to post.
As you know, I believe that knowledge comes from social learning, from the give and take between different viewpoints. I emphasize the process of how argument is conducted. My most well-known book, The Three Languages of Politics, reflected on the observation that pundits do not engage in open-minded argument with the other side, but instead use demagoguery and straw-manning to try to close the minds of people on their own side.
How Debating Rules Encourage Social Learning
The rules of high school debate are very consistent with the processes that I have discussed. For example, go back to my essay on Fantasy Intellectual Teams or, more recently, Six Laws of Social Learning.
In a high school debate, a proposition is put forth. In this tournament, the proposition was “The U.S. federal government should invest more in nuclear power.” (I will comment on that proposition later in this essay.) One team argues “pro” and the other team argues “con.”
In the tournament where I judged, there were dozens of teams, with contests going on simultaneously in many classrooms (hence, dozens of judges). Each classroom had one judge and one contest going on in that room.
Going into the tournament, each team must be prepared to argue either side. A coin flip determines who ends up arguing “pro” and who ends up arguing “con.” Each team participated in three rounds, so a team could end up arguing “pro” one round and “con” another. This motivates each team to understand the best arguments of each side.
As judges, we were told that the students were expected to support their arguments with evidence. They could not just argue on the basis of feelings.
The director of the tournament emphasized that as judges we were to focus on the debating process. A key was that each team was supposed to respond to the other team’s points, not just repeat its own points.
How a debate proceeds
In each round, there is one debate per classroom. Each team sends two members into the classroom, where there is one judge. Based on a coin flip, one team gets to choose which side it wants to argue and the other team gets to choose whether to go first or second.
Suppose that the team arguing for “pro” goes first. Team member 1 makes a 4-minute opening statement. Then the first member of the “con” team makes an opening statement.
Next, there are 3 minutes of “cross fire,” in which those two team members are able to ask questions of one another (often, these were rhetorical questions). They are supposed to respond to one another’s questions, not just ignore them and make other arguments.
Next, the second member of the “pro” team makes a 4-minute statement. This is not simply another pre-planned statement, but it is supposed to respond to what the “con” side has been saying. This is followed by the second member of the “con” team making a 4-minute statement that is similarly supposed to be adapted to what the “pro” side has been saying.
This is followed by a 3-minute “cross fire” between the second members of each team.
Next, the first member of each team makes another 2-minute statement, which is supposed to be responsive to what the other side has been saying. This is followed by a 3-minute “cross fire” in which all members participate.
Finally, the second member of each team makes a closing argument, focusing on the key points that have already been made (not introducing any new arguments).
The judge must grade each speaker and choose a winner of the contest. We were given fairly minimal instruction as to how to do this, and my guess is that there is a lot of variation among judges in terms of what pushes their buttons. I expect that students must have to put up with a non-trivial amount of inconsistency in judging.
We also were told to write down specific feedback regarding each speaker. This is supposed to help speakers learn and improve. I tried to do this, but legible handwriting is not my forte.
Could an AI be the judge?
Yes! My experience about a year ago with prompting ChatGPT to grade an op-ed essay according to my “fantasy intellectual teams” criteria was very promising. I was impressed with how quickly the models could be trained to do a decent job of using the desired criteria, although fine tuning them to grade exactly as I would was a struggle. The essay grader gave very useful feedback.
Since then, the models have gotten better, the context windows have gotten wider, and the models have become multi-modal, meaning that they can process voice as well as text. It would take very little effort to teach an AI to implement good judging criteria. I think that the biggest challenge would be getting the AI to understand the different speakers, especially during “cross fire.”
But it is a feasible project. Somebody should apply for a grant to try it.
Speaking of AI, check out this 4-minute AI-generated podcast of my laws of social learning essay.
The topic of nuclear power
When I found out what the topic was going to be, I thought that it put the “con” side at a disadvantage. Investing in nuclear power promises a specific benefit without mentioning the cost. That is why every government spending program sounds good, and why we end up with the bloated government we have. (Should we spend more money on medical research? Who would say no? And then we end up with lots of waste, fraud, and corruption in medical research.)
In practice, every “con” team made a case for investing in wind power rather than nuclear. One “pro” team made the meta point that this was against the rules, because it brought in an extraneous topic. You cannot prepare for a debate if the other side can just bring up any issue, even if it is not directly on the point of the debate.
But I thought it was fair to bring up wind power. The “con” side has be able to be able to compare investing in nuclear power to some alternative. Otherwise, it is just nuclear power or nothing. And raising the wind power alternative is easier than getting into the tricky question of how you do a generic cost-benefit calculation of investing in nuclear power, which is rather much to expect from a high school student.
Also, all three of the “con” teams that I observed brought up the wind power alternative, and since every team had to prepare for each side, I had little sympathy for the “pro” team’s complaint that wind power was a separate issue. They should have been ready for it, and they were.
Maybe the proposition should have been “In the next few years, the government should invest more in nuclear power than in wind power.”
For me, the least persuasive “con” arguments were safety issues. The “pro” side convinced me that modern designs address those issues.
The most persuasive “con” argument was that nuclear plants take a long time to build. This increases the cost and delays the benefits. One of the “con” debaters pointed out that NIMBY makes it very difficult to site nuclear power plants, which contributes to the long timeline. I think that if you delve further into the issue of the long time to build, you would reach the conclusion that government’s leverage would not come from investing money but instead fixing the regulatory process.
This type of debate, called Public Forum, was created as part of a brand synergy exercise by CNN to promote its show Crossfire. In fact, while Public Forum was being trialed by the NFL (National Forensics League, not the other one) it went under the code name name TedTurner Debate and was then called Crossfire for the first official year or two. Its structure is loosely based on the structure of the CNN show.
At that time the majority of debates took place under a style called Policy with about 1/3rd under the Lincoln-Douglas event and very small percentages under some other events like Extemporaneous speaking. The problem with Policy was that it became very enamored with the idea that each and every argument made by each side (Affirmative and Negative sides debating a particular policy resolution) had to be fully addressed and rebutted by the opposition. No matter how inconsequential, if an argument was simply ignored in one round then a judge might consider it a winning argument.
Somewhere along the line someone realized that if they talked faster it would be harder for their opponents to adequately respond to every argument and they would often win. This was called “spreading” because you were basically spreading all your index cards of data and quotations out making it impossible to cover. By the time I came along, one of the principal activities we’d practice at debate practice was speed reading. We would put marbles in our mouths and try to read out loud as fast as possible.
Naturally this took away from time doing research and developing the evidence needed to support our positions. Our coaches would instead order gigantic packets of various arguments, data, and quotations from collegiate debate leagues. A popular one I remember was from the University of Michigan. So, you’d show up to a debate, pull out some index card you’d prepared by cutting out a section from one of these huge debate sets and pasting it to the card, and then speed read as many of these cards as possible for the entire debate. You only needed to know enough about what was on the card to know which argument your were deploying it in response to or which argument chain you were trying to build as the round went on. And again you’re speed reading it in the hopes that the other side doesn’t comprehend or can’t read as fast in response.
As time went on, spreading became both a necessary and hated practice because everyone involved realized that winning in this way removed most of the benefits of debate Arnold mentions here. Winning was about tactics and not about the content or merit of arguments and evidence. When CNN rolls up with Crossfire it seemed like a way out of this mess.
What worries me is the part where Arnold mentions that every single con team took the same position. I worry that this is not because over the year they’ve tested many arguments and picked the most effective. I worry that they are, instead, purchasing content and merely regurgitating it. I don’t know if that is the case here but winning is often seen as more important than learning.
Perhaps an AI judge would be an improvement? If human judges are predictable and all love the wind power argument, then maybe AI offers a way to incentivize a wider variety of arguments.
Did none of the "con" side arguments include "Government getting involved will ruin the process; they shouldn't invest more but should focus on getting out of the way via abolishing regulations." That would have been my tack, mostly because I believe it is true, but also because tactically it accepts almost all of the pro side arguments (nuclear is good, obviously) and instead focuses on why government shouldn't be the one doing it, which, sadly, is probably not what opponents would expect.