Roughly, the idea behind the statement that any argument should stand on its own merit and should be considered without prejudice is a sympathetic one. The goal of the people who advocate the 'just the facts'-approach is based in their abhorrence of prejudice. It is however also based on the a priori assumption that the merit of any argument could be contained inside the argument. In the case of pure mathematical logic, this might undoubtedly be the case (however, there are deeper philosophical issues that even touch the foundations of mathematics, but I shall not go into those here), but in the real world, where political, cultural and technical issues are at stake, the assumption that the merit of any argument could be contained inside the argument is often questionable.
Let's start with a real world example. Suppose we have a mailing list on issues of copyright law and the Internet. This mailing list is not moderated, therefore we cannot (as with official periodicals) trust the editor to have made a selection for us. Now suppose, some 'never before seen scheme' is proposed to handle some particular problem. We now take three examples. In case (a) the scheme is proposed by Joe A. Lunatic, someone who is known to the reader to have proposed several completely crackpot ideas in the past and who has at all times in the past been proven ill-informed and illogical. In case (b) the scheme is proposed by the well-known Professor of law and informatics Prof. R. V. Trustworthy, who at all times in the past has been proven to know what he is talking about. In case (c) the scheme is proposed by an anonymous person, who could be anyone, including Joe A. Lunatic and Prof. R. V. Trustworthy.
Now suppose, one reads the article and finds the reasoning in it compelling (case I). In case (b) there is corroboration between our judgement of the ideas and our judgement of source. In case (a) there is no corroboration between our judgement of the ideas and our judgement of the source. And in case (c) there is no judgement of the source alltogether. In case we find the reasoning faulty (case II), the picture is reversed.
This still does not mean that we could not evaluate the scheme on its own merits. But in practical terms, in case I-b we will trust our initial judgement more, as we will in case II-a. In cases I-a and II-b we will be prompted to re-evaluate, or in normal language: if the Professor is proposing this nutty scheme, maybe it is I, the reader, who hasn't understood it and it isn't nutty at all.
Now, again, this still does not mean that the system could not work or work better without these connections between context and content. In case (c), we do not have this particular context. But in practice, it will often work so. We choose our newspaper because we know that the messages inside are in a known context to us and we can therefore evaluate them better. The same goes for above cases (a) and (b). Of course, it is possible that Joe A. Lunatic will have his brilliant idea and that Prof. R. V. Trustworthy will have a terrible off-day, but since it is impractical to take the time to scrutinize both to the full extent, we choose - because of the limited time available - to make an educated guess as the most effective and efficient route to knowledge. So, even if the arguments could stand on itself, we would out of reasons of efficiency like to have some context to be able to judge and interpret. It is of course true, that the danger of using the context is to become uncritical of Prof. R. V. Trustworth and dismiss beforehand everything Joe A. Lunatic has to offer. It is however not a necessary result of having the context.
In short: prejudice is a mechanism we use for a more efficient handling of new ideas. On the other hand, extreme prejudice is counterproductive.
Note another example of such a context. Professional magazines or scientific magazines offer editorial services, making this selection on merit for us. For efficiency reasons we use that mechanism as it is impossible to read everything yourself. The result is that inside a professional periodical, we tend to look less to the context of 'where did this originate' as we trust the selection that has been made. We have a prejudice in favour of the articles that have been selected by a good editor.
Suppose somebody writes his ideas and illustrates them with some reasoning by analogy. This reasoning contains the use of a chair. We can question the definition of a chair. What is a chair? We say: it has four legs. We are countered with the fact that there are also chairs with another number of legs. So we say, a chair has three or more legs. But then we enter a room where a large box is surrounded by smaller boxes and we know immediately what a person means by saying: go sit on a chair. So we say, a chair is something you sit on which is not a stool. But then we make a chair out of balsa wood. If you sit on it, it will break. You cannot sit on it, still, we will recognize it as a chair. So even, you cannot define a chair as something you can sit on. Etc.
Even with something that seems as easy and straightforward as a chair, trying to pin a definition down is (nearly) impossible. The further away you go from pure logic, the more 'unformalizable' the concepts involved get.
This problem lies at the heart of a series of philosophical problems that are also the reason why AI on digital computers has failed so far. Early AI failed, because when one tries to make a list of propositions which should define a concept (like a chair), there is a exponential (and often combinatorial) explosion of options, because concepts are defined in terms of other concepts and real underlying basic concepts have not been found. You find yourself in an infinite regress of rules (rules to govern the use of rules) or contexts (contexts to govern the implementation of contexts). This infinite regress is the result of the attempt to divide the concept in a list of separate propositions.
Now, if I state something, like the paragraph on the chair above, an interpretation of the validity includes things I have not stated, because my argument is not of infinite length and therefore I will stop at some level, assuming your interpretation of the subject at hand is the same as mine. Again, this might be the case for pure logic, but for anything in the real world it is just an assumption. Therefore, if you know my background you have extra information to judge my argument. If you know your philosophy and you know I know my philosophy, you might be able to interpret the argument above including possible fallacies better. If I am Joe A. Lunatic, you will not only for efficiency reasons dismiss my arguments earlier, you will also interpret them differently. If I am Prof. R. V. Trustworthy, you will also interpret them differently, because you will assume several things about me. Including things that result from knowing my connections. So if I say 'A' and you know a minor counterargument could be 'B', but from my connections you may assume I am familiar with 'B' (e.g. because it was proposed by a committee of which I was a member) and say 'A' nonetheless, you might (try to) evaluate 'A' in a different light and conclude that I have stated something different than what you thought without knowing the connection. Of course, I could have made the refutation of 'B' part of 'A', but that is part of the infinite regress.
Now, of course, part of this may be the result of you having read statements by Joe A. Lunatic and Prof. R. V. Trustworthy before. So we could have called them X. Anonymous and Y. Anonymous and state that I don't have to know who is stating something (and if he or she is a Professor), as long as previous statements are under the same anonymous identity so that the statements can be linked, so that previous statements form the context for interpreting the current one. This is of course true, but this reasoning has two conseqences. First, by linking the to previous statements of the person who states something you are exactly doing that what was supposed to be unnecessary: creating a form of identity outside of the arguments themselves. What is the identity of someone for you other than the experiences you have had with that someone over time? The second consequence is that you don't want to be influenced by judgement of others. Someone may become known and famous in his area of expertise, and thus become the influential Prof. R. V. Trustworthy because other have often judged this person to be of interest.
Even the essay mentioned above does use this kind of argument, when it says things like "Professional logicians recognize two necessary and jointly sufficient [...]" or "Tarsky's definition of truth", where the reasoning is supposed to be believed because it comes from 'professional logicians' or Tarsky. In my opinion, there is nothing wrong with such sentences, but the fact that something comes from 'professional logicians', is of the same order as knowing who has written the essay, it adds to the context of the statement.
Conclusion: the idea that all statements should be considered in their own right and without the bias of knowing who has authored the statement does have merit, especially as a warning against extreme prejudice. However, removing context from any argument will make it less intelligeable, and knowing who has authored a statement may be part of that context. The main obstacle (which is a consequence of this essential role for contexts) is however a practical one: without the context, many selection mechanisms that are essential for efficiency (and that prevent the regresses mentioned above) will not work.
Draft - 29/9/96, Not checked for errors.
Last modified: 20 Jun 1998
This document was generated with Orb v1.3 for OS/2.