On Credibility

I just had a conversation on Skype with my mother and I made an interesting reflection while seemingly lecturing on our attitudes toward great authors. I was talking about the way Edsger Dijkstra was prejudiced and said that I found it too bad because he was otherwise very smart. And then I thought back on the period where I had not realized the prejudices that he had and wondered what I could do against that. I noticed, not for the first time, that, as I read, as the author gains my respect and esteem I tend to give him (or her) the benefit of the doubt. However, people that are always objective throughout their writing probably don't write much so, for the rest, we should remain prudent not to consider everything that is said by any trusted author to be true. The next question that I would expect to ask myself while reading what I wrote is: so, what difference does it make whether an author is trusted (alternatively more trusted) or not (alt. less trusted? Well, for one thing, an author that I trust is one of which I have read many text and it is a little bit less likely that I would be surprised by something he says or write. More importantly, when I hear people talking and I know the particular author is also giving his opinion, I'll give more attention to that author because I trust him. As a corollary, when I have interrogations on topics of which I suspect my author to be interested, I'll first try to find out what his opinion is on the topic. However, it shouldn't mean that I'll be satisfied once I did. As a good computing scientist, I just saw a case where the good behavior to take is ambiguous. If I stress my imagination a lot, I can imagine cases where I trust many authors. What to do? The naive extrapolation is to choose one randomly whenever I need to. It is not very satisfying so I go ahead and look at the next credible solution: to define a more subtle notion of trust for which I can trust author X on topic Z more than I trust author Y on topic Z. If I take one topic at a time for my investigations, this should be fine-grained enough. As a final note, I am not stating that this is what I usually do. It is more like the exposition of a judgment error I do more often than I would like to and a possible solution that I would like to adopt. Simon Hudon Zürich August 19th 2009