Rethinking how to present disagreements, political and otherwise

"Debate me, bro!" is probably not the answer

It’s all the rage now at universities and elsewhere to have “civil discourse” events where (usually) two people on opposite sides of something have an exchange where they model being civil with each other. The idea behind this is that students can watch two people “disagree better,” and that will somehow lower the harshness of our society and its conflicts. Having tried a few of these things, I’ve come to the conclusion, while sometimes entertaining, it’s a pointless exercise. But I do have some ideas of other ways to accomplish the goal that show some promise.

When I first became editor-in-chief of Science, I gamely showed up for these things. The first was to debate “Scientocracy” (i.e., big government messing up science) at the Cato Institute, which was pre-pandemic and long before I really knew a lot about these issues. It was civil, and I actually had fun. The people I debated (which was basically everyone there) were good-natured and I got a lot of props for showing up. You can watch the whole thing here:

Let’s just say those were simpler times.

As the vitriol started spewing during the pandemic and beyond, and I started building up a record of opinion columns on various things, I got more invitations. I mostly didn’t do them, but I did agree to help my alma mater, The University of North Carolina at Chapel Hill, with their Program on Public Discourse, which is a pretty standard one of these where people on two sides engage in a moderated discussion and then congratulate each other on how civil they were, regardless of how they act before or after. The one I did was with Luana Maroja at Williams College who is a biology professor and part of the “heterodox” movement that believes that science is independent of who works on it and that affirmative action is disrespectful to the people it is applied to. You can watch it here:

I disagree with her and the essence of my argument is in a piece called “It Matters Who Does Science,” which has been widely circulated and even used as the first assignment in an intro biology lab at a major university.

My discussion with Dr. Maroja was civil and all the stakeholders at UNC thought that was really great, but afterwards, she posted a piece on Substack that attacked me personally and made all of the less civil points that she would have made, anyway. So what was the point of that? We were nice to each other for an hour and nothing changed.

My second try was to gamely attend the annual meeting of the Heterodox Academy, which is a group that believes and advocates for a lot of these same things, mainly that so-called “woke” politics are destroying higher education and journalism and, well, pretty much everything. Their leader is Jon Haidt, and every time he got the microphone at the meeting, a hush went over the room. I served on a panel with Rich Lowry of the National Review, Batya Ungar-Sargon of Newsweek, and Matt Yglesias of Slow Boring, which is one of the most successful Substacks. Thankfully, the video of this was not posted, because I was clearly outgunned by much more skilled debaters. When COVID was brought up, Lowry and Ungar-Sargon rattled off a set of conservative talking points so fast and with such skill; this technique is known as the “Gish gallop” after a creationist who was really good at doing this. When they got to me, there was no way in the time I had that I could have ticked through everything they said and explain why the scientific evidence says otherwise.

And then I also tried to play the game with critics on Twitter for years. It was a great way to get more followers because my fans loved it, but I came to the conclusion that it wasn’t getting anyone anywhere because it just became a contest of who could out-snark whom. In retrospect, this didn’t help anything, and I’ve since retired from Twitter. I highly recommend it: I even apologized for one of my less constructive tweets under oath in Congress. Of course, my critics said I got off because I was afraid to debate them. I clearly wasn't afraid given my time on the site, just decided it wasn’t getting anywhere in terms of moving substantive evidence into the discussion.

So where does that leave us in terms of getting resolutions and advancing dialogue? Here are some things that I think are promising.

The first is that instead of having debates, we could have people appear separately in the same forum. That gives each side the opportunity to present their evidence without being interrupted and with less emphasis on the style with which it is presented. I teach a class at GWU on disagreements in science, and academic freedom is very important. I had two prominent advocates for academic freedom from each side speak to the class on separate days. This worked great. The students were exposed to both arguments and asked great questions of each. They also found more areas of overlap than they expected, and we were able to discuss that when both speakers were done.

The second thing I found really promising is this show on PBS, Deadlock. (Disclosure: I’m on the board of PBS.) This show has prominent people who disagree about politics solving a problem of election integrity together. What makes it great is that people do end up agreeing with each other about a few things when they are forced to work together on a problem. This was so much more interesting and moved the needle much farther than a one-on-one debate. Not even close. Here’s the episode:

Finally, there is the role of AI. A few weeks ago, we published a paper showing that a customized large language model was better at lowering belief in conspiracy theories than humans. In my accompanying editorial, I quoted one of the authors as saying that the evidence and counterarguments were “doing the work” of lowering the beliefs. The reason that the LLM was better than humans is that it had access to more of the evidence and didn’t get worn down (i.e., not susceptible to the Gish gallop).

This week, we’re out with another paper from Google showing that LLMs are better at moderating democratic debates than humans and leading to common ground, because the LLM is better at summarizing both arguments in a way that the participants are more likely to agree with. In other words, humans are always going to put their own views into summarizing both arguments in a way that skews the playing field. I would say this is true in all of the human-moderated “debates” that I have participated in.

So all of this leads me to a point where I have no interest in participating (or holding for my students) head-to-head debates or panels with multiple people disagreeing. That just becomes a contest of who is a better debater, and most people who watch it end up thinking the person they agreed with at the outset “won” the debate. This format puts the emphasis on style and debating skill rather than the evidence. I do think it’s good to have two sides presented separately and uninterrupted where the evidence can do the work. I think creative formats like that in Deadlock have a lot of potential. And I think trying things with LLMs looks quite promising.

And trading snark on Twitter/X? Maybe not really moving the needle.