Re: Programming is the Engineering Discipline of the Science that is Mathematics
Date: 11 Jun 2006 22:11:53 -0700
Bob Badour wrote:
> Not having immediate access to Jaynes' book, I had to
> admit that my own knowledge of this comes from a quick
I'm not trying to knock wikipeducation. It's unquestionably a valuable resource. In fact, it's hard for me to imagine returning to pre-net days. What I object to is the somewhat recent and growing phenomenon of people substituting google for their brain. Nothing wrong with using /any resource/ to learn, gather information, check your thinking, etc. But, we should contribute at least some original thinking or analysis when we post (which you BB have done IMO).
My impression was that VC was simply rushing to regurgitate the nearest web-statement that he "thought" "proved me wrong". That is without having actually thought or tried to comprehend my points or any of the resources I referenced. Like he was trying to protect some precious "math ego".
Further, it seems that when this thread began VC was largely ignorant of much of the material now being discussed. Again, nothing wrong with starting off ignorant and learning as you go. At least he is no longer dismissing proofs as "mindless playing with equations" (one of the dumbest things I've heard in a while by the way). No. it's the vociferous way in which he learned, and his inability to admit error and ignorance that was annoying.
I didn't think I was quick to do so. It took a few rounds at least :-) And to me every post is a new post as every day is a new day. At any moment, VC could simply admit his mistakes and prior ignorance, apologize for his vociferous learning, stop the dishonest snipping and outright fabrication of false context, and it would be a clean slate as far as I'm concerned. But, I'm not holding my breath and until he does it's draining to continue communication with him.
> However, it seems his objections are philosophical in
> nature, which I generally find boring and irrelevant. My
> wikipeducation tells me that while there are plenty of
> frequentists and subjectivist bayesians who do not fully
> agree with Jaynes, there are plenty of objectivist
> bayesians who do agree with him.
> As an engineer, I tend to the pragmatic and find myself
> much more sympathetic to the eclectic philosophy.
Then as a fellow engineer, I think you will find Sivia's book most enjoyable and enlightening. It belongs in the family of LSD (Acid) books; being that it is thin, purple, and will expand your mind :-)
(That's not to knock Jaynes' book by the way. His brilliant clear thinking is a delight to explore.)
> Had you been arguing to replace the deductive relational
> model with an inductive model, I would have to reject the
> idea at this time. However, I don't recall you ever making
> any such suggestion.
You are right I didn't suggest that. Having little knowledge and experience with RM I would not presume to make such a suggestion. And from what little I do know of RM, I really don't see myself reaching that conclusion anyhow.
It's strange, my comment was meant simply as an innocuous note of something I personally find interesting.
"And interestingly, one view of logic is ...
So ... perhaps conditional probability theory is the foundation of all :-)"
In light of the whole "one view" and smily thing, the VC attack was ... surprising. I guess it was trying to stroke it's "math ego" and thought it had found a low hanging juicy fruit.
> While I can imagine some fields benefiting from
> conditional probability and while I am aware of some very
> useful applications of it, I think a deductive formalism
> better suits data management. The relational model
> certainly does not prevent anyone from recording
> statements about probabilities.
Absolutely. It's interesting you mention such recording. There are numerous scientific databases that would be far more useful had the designers included uncertainty data. One current example is the Gene Ontology (GO), a database of gene - function - location associations. Unfortunately, there isn't even a simple confidence metric assigned to the associations. Such uncertainty information would be invaluable when analyzing various gene networks etc.
> Probability theory is an interesting topic to me, and it
> is a topic I would like to learn a lot more about. Thank
> you for the suggestions.
I too find probability theory fascinating. At some point, I'd love to hear your impressions of those books.
> However, based on my very limited understanding, one could
> view VC's posts as challenging the idea that probability
> theory makes a good generalization of deductive logic.
If that was the case, it would have been great to hear his thoughts about /better/ generalizations. (Note VC I said /your/ thoughts not the /Internet's/ "thoughts".)
> Thus, it is possible in my mind that you and he are simply
> talking past each other.
Once I thought perhaps we were talking past each other due to different notions of /generalization/. To explore this, I asked VC the simple question: is the gamma function a generalization of the factorial? He dismissed the question as "irrelevant" twice and never answered it.
- Keith --