[SCL] Issues and proposals (details)

[John F. Sowa]

Jim,

I completely agree with the following point:

 > One thing we've been learning in the RDFS/OWL world is that just
 > putting things into XML doesn't help much - the key is identifying
 > some real world "use cases," focusing on their needs, and being
 > compatible with the languages and tools they use.

That has always been true, and it always will be true.  I spent
30 years working at IBM, and I've had those lessons drummed into
my head many times, painfully and often.

But your response to my point missed the point:

JS> However, I don't believe that we should sacrifice a good
 > solid design for CL/SCL in order to perpetuate poor design
 > decisions by people who didn't understand the issues.

JH> which may be why none of the proposed web logics has found much
 > traction on the web yet -- it's worth noting that purity of heart
 > by us AI folks is fine and good, but there's a lot more of "them"
 > out there than us, and we need to make things inoperate.

Everbody appreciates good design decisions -- "them" and "us".
They aren't stupid -- they're just ignorant.  Ignorance is
curable by education -- but stupidity is forever.  There were
a lot of ignorant people at IBM, and they made many ignorant
decisions, such as the adoption of EBCDIC instead of ASCII
because EBCDIC required less circuitry in the card reader.
They are the same kind of people who designed RDF.

And what I was recommending maintains all the compatability and
interoperability, and it supports every "use case" that can be
dreamed up for RDF.  It simply avoids perpetuating a dead-end
syntax.  Following is my recommendation:

  1. SCL has a notation-independent semantics, which can be
     expressed in many different concrete syntaxes.

  2. Among the concrete syntaxes are KIF, CGIF, and many
     different versions of XML-based notations.

  3. So we can support several different XML-based notations.

     (a) A truly ugly one that is based on the current RDF.

     (b) A clean, but verbose version that follows the lines
         that Chris M. suggested.

     (c) More compact versions, which allow KIF, CGIF, and any
         other SCL-compatible language to be supported with tags
         such as <logic language="KIF">...</logic>.

  4. All of these notations are compatible, interoperable, and
     intertranslatable.  They support exactly the same use cases.
     Idiots can use (a), ande cognoscenti can use (c).  End users
     won't be able to tell the difference, except in performance
     and reliability.

John