Sunday, February 14, 2010

Structure vs. Content

Gala Contemplating the Mediterranean Sea. Salvador Dali

I have been playing around in my mind for quite a while with a theme that seems to recur in many contexts. I still have not figured out a way to spell it out exactly or what it actually means and I am sure I have seen variations on the subject on other places (probably in D.Hofstadter's G.E.B). Maybe his post will help me figure it out...

I am speaking of the Structure vs. Content duality. What is the difference between the structure of, let's say, a piece of text and it's content? Other names for this duality might include:
  • essence vs. form
  • code vs. data
  • syntax vs. semantics
  • Deontological vs. consequentialist ethics.
I would like to gather some examples of this theme and maybe define it better.

It recently occurred to me in a post by R.J.Lipton where he suggested an automatic program to predict acceptance chance of academic papers to some theoretical computer science conference. The point was to make both the job of the committee and of the the authors easier by automating the process of paper submission.
Though not spelled out explicitly, it seems that a paper has two 'layers' - one is the structural layer. This might include the sectioning of the paper so it includes "introduction" and "conclusion" or "discussion" sections. It might include the element of sufficient reference to previous work etc...
The other layer is the "essence" layer. That is the "crux" of the paper. What it actually wants to say (present a new proof to some conjecture for example).
So to rephrase the goal - the suggestion is a two-tier submission process. The first stage which analyses the structure of the paper and is done by a machine, and a second stage which analyses the content of the paper which is done by a human committee.

Sounds reasonable enough. I started to wander though if there is clear distinction between the two layers? The suspicion is raised by such remarks made by Lipton (he refers to the paper-checking program as "site"):
It is a stretch to expect the site to check the correctness of proofs in the papers, but I feel it should be at least feasible to build a site that measures the “hotness” or relevance of the paper
Well, is it easier to check the "hotness" or "relevance" of a paper than the correctness of the proofs? probably ... But it seems to me there is a scale of difficulty there rather than a phase transition. Structure lying in the lower end and essence in the upper end.

No comments:

Post a Comment