Solving disagreements is always a difficult task. It ends in quagmires, and sometimes in endless circles. In a philosophical discussion, this is a big problem since nothing gets done at all. There seems to be a definite need for a codified method of discussion.

This method should acknowledge the fact that propositions, in virtually all cases, are based on limited contexts of knowledge, and that disagreements can be subtle, based on importances or incomplete knowledge of the evidence. This implies that disagreements must be quantified and evaluated on the basis of this quantification. Furthermore, it must do this while keeping the flow of argumentation intact.

Such a method has been proposed by Denis Labelle (member of the Sceptiques du Québec), called the REDICO Formula. It consists of posting a number of propositions, with one's numerical evaluation (a simple percentage of approval) of each proposition. Then the other person also evaluates these propositions, and can propose his own.

There are, of course, some rules which facilitate the use of REDICO. First, a discussion based on it should be a straightforward application of its principles - that is, arguing should be discouraged, while it may be done on the sidelines. REDICO is NOT a technique of debate, it is a technique of discussion and resolution of conflicting conclusions.

The best propositions are, of course, the clearest and especially the most measurable ones. For example, "P1 : Leonard Peikoff is an idiot" is not a good proposition, because it is difficult to evaluate. But "P1b : Leonard Peikoff does not understand the difference between certainty and rational confidence" is a better proposition, because it can be measured (albeit with difficulty). And of course a proposition about a quote made by the person would be even easier to measure.

If a participant decides that a given proposition is not clear enough, he may suspend evaluating it - but he must then give his own proposition based on the problem preventing him from evaluate the first. For example, someone could refuse to answer to P1 above and propose P1b instead.

By convention, the person who posts a proposition is the first to evaluate that proposition. The other participant(s) must then evaluate every proposition in turn, and propose their own. Not more than 3 propositions per turn should keep the process from inflating too much, as is its wont, and propositions must follow the logical chain or topic at hand. Evaluations may be revised, but such revisions must be marked clearly.

And of course, one must always remember that complementary propositions add up to 100%.

Here is an example. Person A and person B disagree on confidence and certainty. Person A proposes the following propositions and evaluations :

"P1 : Humans are infallible.
Agreement - A : 1%
P2 : Reason exists because humans need a guide of cognition.
Agreement - A : 99%
P3 : For anyone to be certain of a given proposition, we would need a complete context of knowledge and an infallible mind.
Agreement - A : 90%"

Person B replies with :

"P1 : Humans are infallible.
Agreement - A : 1%, B : 0%. Difference = 1%.
P2 : Reason exists because humans need a guide of cognition.
Agreement - A : 99%, B : 100%. Difference = 1%
P3 : For anyone to be certain of a given proposition, we would need a complete context of knowledge and an infallible mind.
Agreement - A : 70%, B : 0%. Difference = 70%
P4 : I am certain that I exist.
Agreement - B : 100%"

Since P1 and P2 were uncontroversial, there is no need to keep them. Person A could then reply with :

"P3 : For anyone to be certain of a given proposition, we would need a complete context of knowledge and an infallible mind.
Agreement - A : 70%, B : 0%. Difference = 70%
P4 : I am certain that I exist.
Agreement - A : 1%, B : 100%. Difference = 99%
P5 : Profound and axiomatic agreement does not necessarily translate into certainty.
Agreement - A : 85%"

And so on. One can use the letters to track to which chain belongs what proposition.

This method has a number of advantages. It permits to stop tempers from flaring up, and rather analyse a disagreement more clinically. It permits to quantify disagreements which would otherwise be left to our subjective evaluation of the other fellow. It permits to get quickly to the root of a disagreement in conjunction with Rand's Razor (note that the efficiency comes in, not in the presentation of facts, but rather in evaluating deductions).

Despite the claims of its maker, I think REDICO, due to its generality, is a flexible method which can apply to more axis than only true-false (and its corollaries, such as good-evil). It no doubt could be applied, for example, for a comparaison of artistic evaluations. While Rand's Razor would be more difficult to use in this case, it may yield interesting patterns of results.

Suivi