Debate

Debate is an important tool of persuasion and truth-finding. We often look at it as a contest but its main purpose is to clarify thinking. Most debates are informal and unscored. It isn’t really clear who “won” and often both parties come away believing themselves to be the winner. But regardless, the parties also tend to come away with an appreciation for what the other side has to say and why they believe as they do. Debate is an efficient information exchange mechanism.

But while it may enlighten its participants, debate often fails to persuade them or even to establish the truth. Verbal debates, given their informal nature, are generally inconclusive affairs. It would seem we need to rethink debate design and find a mechanism that works better.

The main problem with debate is that it is usually short, verbal, and comes to an end. The quest for truth does not confine itself this way. Just look at the scientific process. In many ways it can be seen as a debate since there are usually promoters and detractors for any new theory. Establishing the truth and persuading the detractors takes years. It is done through academic publications and real experimentation. And it is open-ended. There is no point at which a theory cannot evolve further or be dismantled altogether. Even when we settle on certain “laws” (eg classical mechanics) we leave open the possibility that new experiments might call it into question (eg relativity).

With this in mind, debates should be written and not have short-term time limits imposed on them. Verbal debates are fun to watch but introduce too many confounding factors, such as personality of the debater, speaking style, wit, etc. These may be important characteristics of people, but they don’t really pertain to informed debate. Verbal debate, furthermore, is prone to devolving into tit-for-tat arguments that get lost in the weeds. Writing certainly does not preclude low quality debate but it works against it. Writing also provides the ability to judge the debate in a more dispassionate way.

Written debate formats should allow people to evolve their arguments over time, as their thinking changes and is sharpened by the opponent. This is the purpose of debate and we should enable it through appropriate design. An effective format might be to hold the debate over several levels. The first level would be more of a brainstorming session where arguments are made and responded to. A few exchanges of statement and rebuttal could follow. This level could be unrated, existing as a type of sandbox where debaters can try out ideas without really committing to them. This allows arguments that have no merit to be recognized early and properly discarded. At some point, if the sandbox is successful, the debaters enter a second level where the debate is held again, but this time with the benefit of practice at the first level. A debater who had an effectively rebutted point in the first level can either avoid the point or modify it to meet the objections of his opponent. This second level debate would also go on for a number of exchanges and be rated by the community. More levels of debate could follow where debaters revise their arguments in light of their opponents’ arguments and the ratings they receive from others.

Community involvement in the debate is crucial. It is, after all, the community that will be settling the results of a debate through public policy. It is also the community that can decide whether a debate continues to advance through more levels or has been exhausted and needs to end. For tough issues, the community may decide to leave the debate in an open-ended state or extract certain points from it to act on but otherwise leave the debate to continue.

Debates should be encouraged to get to the essence of their case rather than waste time pursuing tangential points (as so often happens). This is another reason why rating the debates is important. The raters can keep debaters on track or suggest that they take up another topic to debate. The role of raters is flexible. Some raters may want to participate in the debate and form a team with the lead debater. Some may only want to provide advice. Community involvement in debate is generally a good thing and should lead to more democratic decision making.

The multi-level debate format proposed here is only a basic structure. Further rules can be imposed such as the number of exchanges permitted within each level or the type of argument expected (eg opening statement, rebuttal, close). A maximum length might be imposed on each argument although we generally want to allow a great deal of flexibility here, generally in proportion to the complexity of the subject. A fact-checking session might be imposed on all arguments, one in which research is conducted and reported on before debate resumes. This would be another way of allowing debaters to revise their arguments if they prove to be shaky on factual grounds.

Debate Design

Debate Structure

Most of our debate ideas so far would be designed as a tree structure where users can add any number of sub-arguments on the Pro or Con side. The Pro side is colored green and Con side is colored Red, ie:


So far we have implied that the tree can expand without bound but this is not necessarily true, or desirable. A number of options exist which we will touch on here.

Unlimited Debate Tree

In this format, anyone can introduce sub-arguments and the tree the tree can expand without bound:

- Arguments tend to be succinct and of low quality. [Kialo https://www.kialo.com/my], eg, is a good example of this. - Is difficult to visualize in its entirety. Geared toward the content producer rather than the reader.

Limited Debate Tree

Here the tree cuts off at a fixed number of levels and/or sub-arguments per level.

- The number of slots is fixed. Once they are filled, no more can be added. - Has the benefit of easy visualization. - Probably tends toward more writing per argument.

Strict Debate Format

Similar to limited but following an established set of rules, eg NDT (National Debating Tournament), Parliamentary, etc:

• Reproduces oral debating formats (eg opening statement, rebuttal, closing stmt).
• Easily understood by debate enthusiasts.
• Easy to visualize.
• Focuses attention on higher quality arguments.
• Includes rules like no new arguments in the closing statements or rebuttals which keeps the argument organized.
• Has established scoring criteria, although these are tailored to oral debate. For example, if

$S_{init}$ is the Score for an initial/main argument

$S_{rebut}$ is the Score for a rebuttal/reply argument

$S_{cont}$ is the Score for Content

$S_{styl}$ is the Score for Style

$S_{stra}$ is the Score for Strategy

then

$S_{init}=0.4S_{cont}+0.4S_{styl}+0.2S_{stra}$ for $S\leq 80$

$S_{rebut}=0.4S_{cont}+0.4S_{styl}+0.2S_{stra}$ where $S\leq 40$.

So initial/main arguments are scored out of 80 and rebuttal/replies are scored out of 40 where Content, Style, and Strategy are weighted at 40%, 40%, and 20% respectively.

As an example of a strict debate format let's sketch out the Parliamentary debate:


Each side (Pro or Con) has 2 debaters, the leader and the member. The order is as indicated by the arrows although this is obviously for oral debate. In our case, the order could be preserved for an initial round but then more flexibility allowed for changes.

This debate format (like all the others) is simple and is very restricted as to the number of slots allowed. It may be necessary to modify this to allow more slots to engage more participants.

Allowing Changes

For all the formats, the issue of how, and whether, to allow changes needs to be addressed.

Some thoughts on not allowing changes:

• It would have the benefit of simplicity. The arguments that have been made are static. The argument can only get bigger and would simply expand to lower tree levels.
• Favors the unlimited debate structure. Someone can always provide a sub-argument for any argument that gets made.
• Would encourage people to carefully word their arguments up front.
• Any rewording or clarifications would themselves have to be sub-arguments.
• Scoring would get complicated since revisions to arguments made below the original argument would be mixed together.
• Following the argument might be difficult because revisions to arguments that get posted as new sub-arguments would not be clear as to what they really are.
• Has the major disadvantage of not allowing the argument to mature and improve over time.

Some thoughts on allowing changes:

• Promotes the idea of quality argumentation as feedback is obtained and authors have time to reconsider their points.
• Arguments would be more consolidated and less fragmented. We generally want our arguments to be relatively complete essays, not a bunch of single phrases that readers have to piece together.
• Favors the limited debate structure.
• Once authors make changes, other participants will need to be alerted so they can make changes too if necessary (ie to sub-arguments). Scorers will also be alerted so they can revise their scores.
• Owners should be able to concede by marking their arguments as such. This could happen if a sub-argument is made that invalidates their argument. Conceding would leave the argument visible but in an inactive state so other participants know it is no longer being advocated for. New sub-arguments could still be made, especially ones that cause the concession to be revoked.
• A history could be maintained so that the reader can click through a timeline of the argument, as each change is made. Users should be able to do this with the whole tree or a sub-section of the tree, or a single sub-argument.
• UI idea to reduce clutter: Only show the highest scoring arguments / sub-arguments at each level.

I'm generally in favor of users being able to change their arguments. Static arguments will lead to complex looking trees which are hard to read. Besides, it seems unreasonable to lock an argument down once an author has published it. Since we're aiming for high quality, the idea of being able to edit makes sense. Although in verbal debate you can't take back what you said, you should be able to in written debates. Our system should provide for that.

Who gets to debate?

In the unlimited debate discussed above, anyone can participate. However, in any of the limited debate formats there will be a fixed number of "slots" for each debater. How to allocate the slots will be an issue. Various arrangements might exist for doing that:

• First come, first served.
• First come, first served if participant has a score above a certain level. We generally want to encourage the "good" debaters to come forward.
• Competition for each slot. A competitive round takes place with each candidate presents their argument. The highest scorers take the slots.
• Team debate. A group of users is formed which takes on its own identity and is scored separately. Like a debate team in high school or college. The group would have its own trust and private discussion section to hash out arguments collaboratively.
• Delegated debater. A group of users elects a debater who debates on their behalf.
• Anyone can join but low scoring arguments are thrown out. Ie everyone can compete for the slot but only the highest score wins.

For our purposes the best system for assigning debaters is probably to allow for competition for each slot. This gives us the best chance at quality and also gamifies the system to some extent. A variant on this would be allowing losing participants to form a team with the lead debater (the guy who won the slot).

Regardless of who gets to debate, a comment section should be open to all users. Commenters could influence the debater to change their wording, introduce ideas, etc. It would be similar to the comments section in Kialo. Privacy settings could be enabled which allow the section to be used for closed-door discussion by, for instance, debate teams.

Scoring Thoughts

Instead of using a rolled up argument score, discussed previously, we could simplify the system and have everyone rate each argument and sub-argument individually. A mechanism to allow scoring of the entire Pro and Con side of the argument could also be created.

A system to keep track of the time when arguments are made and when scores are given would allow us to track the history of arguments and provide insights, such as which arguments are having the most impact.

In the following hypothetical figure, the entire Pro and Con side score is shown as a function of time. The labeled markers show the argument that was made most recently corresponding to the score.


The markers could also include a group of arguments made recently that would have given rise to the increase/decrease in the score. The ability to click on the argument in the chart and visualize the text might be an added enhancement.

More on scoring of practical arguments

Practical arguments will be important in community policymaking and one way of scoring them is by using the $Cost \over Benefit$ ratio. Another related idea is to assess the probability of the scenario in question. Scorers would then rate the argument for a proposal on its cost, benefits, and probability of occurrence.

It might be easier to think of cost-benefit in terms of return. One simple possibility is:

$R=Benefit-Cost$

Another is the more traditional return on investment:

$$ ROI = {Benefit - Cost \over Cost} $$

These have the advantage of clarity since positive numbers are good and negative numbers are bad. They also do not lead to infinity when the benefit is zero.

Given enough scoring input we could envision a PDF of a scenario under consideration:


Other such PDF's would be generated for each different scenario. Decision makers would then have a ready tool for evaluation purposes.

Along these lines, it would be nice to have a single rolled-up measure of goodness which combines Benefit, Cost, and Probability. Various ways to combine these might be formulated, the simplest of which might be:

$$ G = {{Benefit - Cost} \over Cost}Probability = ROI*Probability $$

Then decision makers can evaluate the highest $G$ proposals since they are presumably both likely and have a high return. In practice, however, this is difficult to do because we don't know the relative importance of each contribution. In our case we are not just considering standard business investments. We also have to consider arguments in which we want to avoid (at any cost) low probability bad outcomes, like nuclear war. If we were to consider direct involvement in Ukraine, for instance, there may be a reasonable probability of success (with high ROI) but a small probability of nuclear war which we want to prevent no matter what. This alone would preclude further consideration of such a policy. For general cases, it seems more reasonable to present return and probability separately and allow decision makers to sort it out.

Gamification

The system should be gamified to provide an additional incentive to participate and produce quality arguments. Arguments will be scored on rhetorical and practical quality as discussed previously. Participants can then also be scored, obviously, from the average of their argument scores, either rhetorical or practical or both. A ranking system will provide ambitious debaters with a clear goal.

Highly ranked debaters can also be given privileges, such as easier access to slots in limited debates (as discussed above), higher voting power, the ability to propose specialized debate formats, etc.

Since the ability to score the entire Pro and Con side of the argument will be available, prominently displaying this score on each debate as it progresses will help gamify it. Easy access and visualization of scores for sub-arguments and individual debaters should also help.

A Reasonableness Score?

Debate, like sporting events, is designed to produce winners and losers. That's part of its appeal and central to gamification. But alot of debate is an attempt to obtain compromise. Our system should enable this by providing a concluding statement that the debaters agree to. No one is forced to do this but if they do, they would be scored on an additional measure, "reasonableness" or "willingness to compromise". A debate between theists and atheists might provide a concluding statement that "Although there is no way to explicitly demonstrate God's existence, everyone's lived experience on this subject should be respected." This doesn't say much but it at least gives users the ability to shake hands and call it a day. Something we don't get alot of in our debates these days. Even a concluding statement that just summarizes the positions is useful in this regard. Incentivizing debate participants to find common ground, if only symbolically, would be useful in reducing polarization and extremism.