I've been toying with an idea of developing two competing theories in parallel in an iterative manner:

  1. writes an initial thesis
  2. does the same
  3. revises their thesis to address 's thesis where it contradicts that of 's
  4. does the same
  5. and so on until both parties feel they have nothing to add

This would be different from adversarial collaboration, as it's commonly understood, in that both sides would work on their own arguments instead of trying to agree on a common summary (which is very hard!).

It's worth emphasising that the idea is not to correspond with the opponent. Instead, one would keep updating one's thesis to meet the challenges presented by the competing thesis so that it remains coherent and stands on its own after every iteration.

I wrote a little ClojureScript library to go with a Pandoc template to facilitate iterative argumentation of this kind. The library makes it easy to navigate between the sides of the argument ensuring that following local links will switch context when appropriate, etc. Additionally, it provides bidirectional links by way of highlighting bits on both sides of the argument. Hot loading is also supported to make writing the document more convenient. Currently it's very much work in progress, of course. I didn't want to commit to too many features at this point not knowing if any of this would be useful.

Would someone here be interested in trying out this sort of thing either with their own "archnemesis" or with me? For starters, I think it would be best to pick a properly contentious topic but not one inciting too much passion.

New to LessWrong?

New Comment
9 comments, sorted by Click to highlight new comments since: Today at 1:03 AM

FWIW I think you are leaving out an important step here: synthesis. The lack of synthesis is my general complaint with these sorts of approaches because it creates a debate rather than a dialectic and lacks a mechanism for convergence without additional constraints. Perhaps that's what you are hoping to point at with "revise" in step 3, but there are many ways to revise something that don't promote convergence over adversarial iterations.

I agree that convergence would be ideal, but I'm quite pessimistic of how often it could be achieved even if our culture of discourse encouraged it a lot more.

Here, the synthesis would of course ultimately rest on the shoulders of the reader. It would be up to them to assess which of the sides has made the better case for their thesis. The iterative nature of the argument would at least ensure that obvious counterarguments aren't left unanswered, or if they are, the reader would have a better chance of noticing it and could then adjust their beliefs accordingly.

Another benefit from an iterative approach of this sort: the reader would have only one self contained argument to consider from both sides instead of having to wade through possibly a long sequence of rebuttals and rejoinders in which the forest is easily lost for the trees.


I would name two communication styles. One being adversarial truth seeking. The other being collaborative truth seeking.

Adversarial: only one of us is right and we have to fight it out. May the best gentleman win.

Collaborative: hmm. We seem to disagree, let's work together to find out why and how and form a better understanding for us both. We agree on truth seeking so let's work together.

They don't like each other very much.

I would certainly like there to be more collaborative truth seeking. But, as said above, I don't see it becoming the norm, certainly not in the short term. Tetlock and Mitchell, for instance, deem it "least feasible when most needed". (1)

I think my concern is "By the time you've gotten people to agree to something like what you've depicted above, you could probably also get them to do actual collaborative truthseeking. Adversarial Truthseeking seems like it'll always be the norm, but that's because even getting people to agree to do something like the above is fairly hard."

No doubt it would be hard to get people to do what's depicted in the post. The conjecture is that in many important instances it would be considerably less hard than collaborative truth seeking. But it's just that: a conjecture. Still, I would think it prudent to explore many different avenues here given how unfruitful debates so often are and how much so often is at stake.


least feasible when most needed.

Yes. There's an element of need and intent in bringing collaborative purpose ones self at all times. And hoping that others do the same. I don't see what more can be done.

This is pretty cool. Instead of just seeing two sides opposed or in agreement, readers get to see a discussion.

One question: Why two theories?

Do you mean as opposed to three or more? That would be less practical, although some fields, like philosophy of mind, certainly have very many competing theories!