Light & Thought
A collection of Steve Graves' reflections.
Home ยท Tags

Debates and the Problem of Unenforced Truth

Debates are supposed to help us understand what's true.

But that only works if the system is designed to reward truth.

Lately, I've been thinking about how debates are run - and more importantly, how they're evaluated.

In theory, debates are a cooperative system:

But in practice, something has shifted.

Debates have become increasingly adversarial.

And when that happens, the goal changes.

It's no longer: “What is the best answer?”

It becomes: “How do I avoid losing?”

That leads to familiar patterns:

And here's the key problem:

The system doesn't correct for it.

Moderators often move on instead of pointing out: “That question wasn't answered.”

Evasion techniques go unchallenged.
Logical fallacies go unnamed.

And the audience is left to sort it out on their own.

That's a heavy burden - and in an adversarial environment, it often doesn't work.

So what would a better system look like?

First, moderators would enforce structure:

If a question isn't answered, it should be said clearly: “That didn't address the question.”

Second, evasion techniques should be identified in real time:

Not as judgment - but as clarification.

Third, evaluation should shift:

Debate performance shouldn't be judged on:

It should be judged on:

Because debates aren't supposed to reward performance.

They're supposed to help us learn.

If the system rewards evasion, we get more evasion.

If it rewards truth, we get better answers.

That's not about left or right.

It's about whether the system is designed to produce understanding - or just the appearance of it.