Re: Very rough draft of TAG finding on self-describing documents

Elliotte Harold writes:

> There's the hint of a valid principle here, but I think it needs to be 
> fleshed out more and not grounded in the notion of understanding.

Fair enough.  I think many of your comments are on the mark, or at least 
close to it.  My real interest in getting this version out is mostly to 
have a placeholder that will be useful in deciding whether we want to open 
an issue, and if so, with what scope.  I don't consider it to be either 
very carefully worded or carefully reasoned at this point.  Mostly, my 
intention is to indicate the rough outline of a finding that might 
eventually be useful. 

As to the notion of understanding:  I think it does play a role, though I 
think you're right that the current text probably goes too far in implying 
that such understanding is an all or nothing business, when in many 
interesting cases it's a matter of degree.  Indeed, particularly on the 
Web, we benefit from agents like crawlers many of which are built 
specifically to do useful work based on a partial or even in part 
incorrect "understanding" of the documents they retrieve.  Overall, it's 
not my intention to take this toward a very formal notion of information 
transfer or understanding, but rather to try and explain at a commonsense 
level why certain forms of Web content may be more useful, for certain 
purposes, than other forms.  Thanks an any case for the quick response.

Noah

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

Received on Monday, 26 February 2007 14:06:40 UTC