Author's note: Despite multiple refinements, this is still rather wordy. Feel free to ask me questions on Mastodon, and that will help me improve this explanation.
Structured debate is a process for maximizing the rationality of a debate. It assumes that the intent of debate is to determine the most likely truth when there is disagreement and that any party may be using invalid reasoning or incorrect facts.
Structured debate provides a system for keeping track of the status of every point and counterpoint that has been raised in a debate. It allows multiple participants to make contributions simultaneously (which would not be possible in a spoken debate), permitting much more thorough fact-checking than in other forms of debate and ensuring that all positions have an opportunity to assert their objections to any given conclusion.
Ideally, the end result would be that everyone could agree on the correct (best, most likely, most reasonable) answer to any given question. In practice, of course, there will be a few people who refuse to recognize illogic (or insist that correct logic must be wrong, probably because it leads to answers they don't like); there are multiple ways of dealing with this conflict. The fact that the debate is reduced to a cascading set of simple premises and conclusions should help minimize unintentional illogic.
In order to be rationally defensible, any claim must be based on one or more premises which are combined into a conclusion via a logical operation (either "if all of X are true, then Y must be true" or "if any of X are true, then Y must be true"). An individual can either agree with the conclusion, or can dispute either the reasoning or the premises. Faulty reasoning is pretty obvious once it's pointed out; faulty evidence is generally based on further evidence-plus-reasoning which may itself be agreeable or not.
Any premise, when disputed, becomes another claim to be examined; this process can be repeated ("drilling down") until a set of claims is reached upon which no further debate seems possible ("drilling down to bedrock").
During the process of drilling down into any given claim, one eventually finds one of the following:
- bad reasoning: the conclusion is not supported by the premises given
- lack of evidence: the logic-tree is unsupported at some critical point, invalidating the ultimate claim
- fundamental agreement: premises that everyone can agree on (plus valid reasoning all the way up) -- the conclusion is valid
- fundamental disagreement: one or more items of logic remain in dispute, with no apparent way to drill down any further
While sentient beings will inevitably be illogical at times, mapping the debate should make it clear (or at least much clearer) to any reasonable observer – that is, any observer who is honestly attempting to seek the truth – where the illogic lies: which conclusions follows logically, and which do not.
- InstaGov's debate mapper module is a software implementation of the structured debate methodology.
- Structured debate is a tool for managing rational debate.
- An argument is a set of assertions that logically draw a conclusion from a set of premises.
- Any argument is in a false state if either the logic or the premises are disputed.
- Arguments disputing another argument is called a counterargument.
- Any counterargument is itself an argument, and may be further disputed.
- An argument is in a true state unless it is countered by one or more counterarguments which are themselves in a true state.
See /rules for more detail.
In theory, a structured debate can be held in an unstructured plaintext discussion environment such as Google+ or Facebook; rules and an example for doing this are here. This format does require significantly more discipline from the participants, however.
- Anyone may start a debate at any time; debate starts with an assertion of fact (the "prime assertion"), preferably defended by supporting arguments.
- Anyone may add counterarguments or additional supporting arguments to this assertion at any time.
- Each supporting or countering argument is itself an assertion, and may be further supported or countered at any time.
- Debate continues indefinitely.
- If new information is discovered which affects the debate, this information can be added immediately, with the potential to immediately change the debate's outcome (and hence the outcomes of any other debates which depend on this one, if others are linked to it; see #Ecosystem Features).
- The design currently presumes that the final state of the prime assertion will eventually settle down to being true or false for long periods of time, as nobody will have any additional points to add; in practicality, decisions based on the debate's outcome will need to be made at specific times, and debates should be held in such a way that the final state will be likely to have settled down by the time any decision must be made.
- There should probably also be a minimum amount of time between the last state-change and the actual decision, so as to negate the effects of last-minute /counterpoint spamming.
Individual debate venues may wish to make changes or additions to the procedural rules in order to accommodate their specific needs and circumstances.
- Linking: In order to negate the need to replay existing debates within new contexts, any given point in a debate can be made dynamically dependent on the outcome of another debate.
- Categorization: any given point within a debate may touch on one or more topics of general interest, and should be findable by anyone exploring that topic. A system for managing crowdsourced topic-tagging is under development.
- Relationships: It may be useful to be able to quantify the nature of a link's relationship with more granularity. One possible relationship:
- A is a generalization of B (= B is a special case of A)
- A may be inferred from B (= B is a premise upon which A is based)
- Types of support-point (this will definitely be needed):
- A is necessary in order for B to be true
- A is sufficient in order for B to be true
Additional features not essential to the basic concept but which makes it more usable:
- Text search: search within a branch for specific text or patterns
- Notifications: users should be able to set a preference indicating that they do (or do not) want to be notified (by any of various methods) when any of the following occurs for any given debate point:
- the point's status changes (from true to false or vice-versa)
- anyone edits the point's text
- the status of any subpoint changes
- anyone adds a new support or counter point
There may be other usability features we will want to include.
- /counterpoint spam: Dishonest participants may repeatedly raise spurious objections solely for the purpose of keeping the correct conclusion in a state of presumed falsehood.
- Issuepedia has an article about SD. Some of the material there should probably be moved or copied over here.
- Note Sakari's comments here.