[SCL] Again: new version of SCL draft attached

[pat hayes]

>Hi,
>
>pat hayes wrote:
>
>Let me suggest a compromise. Agreed about the 
>split, but not about exactly where it goes. Let 
>us first define a single 'kernel' language with 
>one (abstract) syntax and one model theory.  (We 
>can do this without using Greek letters, even 
>quite formally, using the general style of XML 
>Schematon 
>http://xml.com/pub/a/2003/11/12/schematron.html.) 
>This admits extensions which can considered as 
>syntactic sugar, in the sense that they can be 
>transcribed in the kernel without loss of 
>meaning; and it allows sub-languages which are 
>simply the kernel with some syntax classes 
>omitted and maybe some extra syntactic 
>restrictions imposed.  Once stated for the 
>kernel, the MT never needs to be re-stated: it 
>is fixed. 
>
>OK, fine with me.

Ok, thanks for the feedback.

>
>Is the current SCL core abstract syntax and MT OK or not
>as this kernel language?

I think it needs to be simplified/modified. I am 
working on that right now, whenever I can. I'll 
try hard to get a draft done by the end of the 
week.

>
>If yes, please:
>
>A) state so
>B) reformulate Chris presentation in the
>    terminology of the core SCL, building
>    everything up in a detailed and
>    understandable way with examples
>
>If not, please:
>
>A) write a new core SCL
>B) do the same as (B) for the 'yes' case.

Yes, I am trying to do that, at least in outline.

>
>>One sublanguage of central importance, which we
>should define as early in the document as 
>possible, is a conventional KR/FOL syntax, which 
>is probably going to be . Then we should 
>describe how to embed the kernel into that 
>syntax, by the use of an appropriate holds/app 
>translation for the relational quantifications 
>and the use of an explicit lists ontology for 
>variadic relations and sequence quantifiers.
>
>Exactly. We have the sublanguage. Please look into
>this, accept or write a new one, and then embed.
>
>Please, let us not make more divisions than are 
>required. Chris has been putting valiant work 
>into writing a unified technical specification.
>
>Which is otherwise fine, but as it is presented now, I am
>incapable of fully understanding the presentation.
>
>Considering myself a typical audience for the SCL spec,
>I draw a conclusion based on analogy that most
>of the audience will not understand either.

Point taken. But there really are two issues 
which we need to try to keep separate: the 
forbidding mathematical complexity of the 
presentation, and the complexity of the 
underlying ideas. Im far more worried about the 
latter than the former.

Im particularly worried by the possible 
consequences of a perception that identifies 
familiarity of presentation with underlying 
simplicity of the overall language. The things 
that are hard to get formally right in the 
larger language ore often apparently minor 
matters or even nonexistent in more familiar 
presentations of sublanguages. That is why I am 
reluctant to proceed in the way you suggest, at 
this stage, by focussing on your document in 
detail: even if get the 'core' document written, 
until we have a single coherent picture of the 
full language, we can have no confidence that we 
have this simple stuff technically right.

>Look: I am not just complaining that it is hard
>to understand. I have a written a lot of structured,
>hopefully understandable chapters presenting simple
>stuff, which IMHO could be perceived as core,
>with the hope that we can continue building
>fancier stuff on this core.
>
>I have seen no serious comments or suggestions
>on these chapters. There is currently nothing
>I can do. Only Chris and me are writing material for
>publication, I cannot understand what Chris writes,
>Chris does not comment or modify anything I write.

I share your frustration. I think there are also 
divergences on background assumptions and style 
that we have not previously encountered but which 
are causing some current trouble.

>Other people are giving nice advice, but not about
>the text Chris or me have written, but rather
>about interesting issues in general.
>
>IMHO we are badly stuck right now.
>
>In a letter on history you write:
>
>>  This was roughly when we changed the name from 
>>SKIF to CL. Thus, the  > original, rather 
>>limited, aims of the original group had become 
>>rather
>>  broadened and even in a sense messianic, causing further rifts within
>>  the CL group between the pragmaticists and the wild-eyed theoreticians
>>  (Chris and me, mostly).
>
>As I understand from your writing, both SKIF and CL got nowhere, because
>of the failure to agree on the common core and the growing
>focus on messianic ideas.

Well, the key problem was that the group was in 
fact less unified in its presumptions and goals 
than it thought it was. That may be our problem 
here also, and I must take the blame for not 
taking more care to ensure that we all were in 
basic agreement. I think it is very important 
that we proceed by finding basic agreement first. 
If writing documents turns out to be a process of 
discovering disagreements, then let us tackle the 
issues that the writing process has uncovered, 
rather than defend the documents; as until we 
have more agreement, the documents are only a 
mechanism for our internal discussions. Right now 
they are not even performing that role as we seem 
to be unable to understand one another's 
documents. That is one reason why I do not want 
us to be rushed by an externally imposed deadline.

>Don't you see that happening for SCL in the same way?

See above. I will try to produce something 
readable, if sketchy, reasonably quickly. Of 
course, then we will have three draft documents 
:-)

Pat

>We could avoid this by taking the choices and steps
>outlined above here. Concrete small steps have to
>be taken, all written down in the beginning of this
>letter.
>
>Regards,
>          Tanel Tammet


-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes at ihmc.us       http://www.ihmc.us/users/phayes