Re: Publication proposal discussion summary

On 29 Oct 2007, at 13:57, Jeremy Carroll wrote:

> Bijan Parsia wrote:
>> On 24 Oct 2007, at 15:57, Jeremy Carroll wrote:
>> [snip]
>>> [[
>>>     B) People expressed the desire for more group review before
>>> publishing anything. The consequential benefits haven't been clearly
>>> enumerated to my ken.
>>> ]]
>>>
>>> Jim argued this most effectively - this is the normal W3C WG  
>>> process, and there is nothing so strange about this WG that we  
>>> should behave differently.
>> That's not a consequential benefit. It would be helpful if you  
>> were a little more careful in your reading.
>
> I don't think I could get away with arguing that conventionality is  
> beneficial :)

:)

But you also cannot apply it consistently within the context of the  
group. Wiki's aren't normal (yet). Certainly the proposal that exact  
wording of proposed resolutions (even number of resolutions, even  
that there *may be* a resolution) be in the agenda in advance and a  
cooling off period...these are very very abnormal both in the  
typicality sense and in the normative sense (consider the concept of  
bad standing). Publishing technical documents well before user  
documents is pretty normal. Etc.

That doesn't mean they are wrong or against process (chairs have a  
lot of leeway), but I would strongly appreciate it if we would get  
some better alignment on evidential standards. For example, normality  
isn't an effective argument unless there are no other considerations,  
and even then, it's basically preference. It's fine to register your  
personal preferences, even your strong ones. For the sake of comity,  
I'm often inclined to give in on mere preference, esp. if there are  
no other considerations. Well, as long as I have the feeling that   
*my* preferences and my *real* concerns are also taken seriously.  
Frankly, thus far, it's pretty clear that they aren't, which makes  
the possibility of consensus over controversial things less likely.  
That doesn't mean decisions won't be made of course, just that they  
will be made over objections.

>>> As an example to do with the semantics doc. HP has a view that  
>>> the n-ary datatypes design (which I guess is in that doc) is  
>>> broken. We have stated this on a number of occassions. Currently  
>>> that does not seem to be in an issue list,
>> A simple query of the issue list reveals:
>>     http://code.google.com/p/owl1-1/issues/detail?id=12
>> This is a very minimal piece of research. I.e., go to the owl wg  
>> home page:
>>     http://www.w3.org/2007/OWL/wiki/OWL_Working_Group
>> Under "inputs" click "Issues" then "issues list" to get to:
>>     http://code.google.com/p/owl1-1/issues/list
>> Then search for "n-ary".
>> It would also be helpful if you were a little less bold with  
>> unresearched claims. Esp. ones that suggest that people have been  
>> unresponsive to issues that you've raised.
>
> Sorry - I had looked at the google code issues list, but clearly  
> had difficulties with the mechanisms - I can't recollect what - it  
> seems clear enough now - I think I was searching for "Turner" ...

I don't mind that you were unable to use the software effectively --  
all software is horrible. I do mind the presumptively negative tone,  
something I complained about off list already. Would it be too much  
to ask that, until trust and normal interaction are well established,  
that you be rather more careful in how your words may be taken? I  
pledge to undertake a similar effort both in what I produce and in  
how I interpret.

Similarly, there's a lot of denigration of certain interests  
("academic", comes to mind) and many claims of broad interest or  
representation that seem rather stronger than I have reason to  
currently accept. If one is arguing something controversial, it's  
helpful to present that view with strong evidence. For example, your  
own claim in:

<http://www.w3.org/mid/471F7171.9020605@hpl.hp.com>

"""The OWL Full treatment of a URI used as both a DataProperty and an  
ObjectProperty is that it represents a single property.

This treatment is in OWL 1.0 Full, and extensively deployed (for  
example in the RDF subset of OWL)."""

Has a number of obvious problems, including the fact that the very  
distinction between an object and a data property doesn't exist in RDF 
(S). So there's a *prima facie* difficulty in drawing a useful  
generalization RDFS deployment alone. In particular, I don't think  
that there's a way to *tell* from an RDFS pov whether:
	:s :p :o.
	:s :p "1".
involves one property or two, thus RDFS systems are, prima facie,  
more indifferent to any problems that may arise. Indeed, models  
involving punning can have but a single property in the denotation  
(i.e., a single set of pairs). Renaming is just a simple way to  
implement and to provide intuitions about the entailment behavior.

Cheers,
Bijan.

Received on Monday, 29 October 2007 19:04:24 UTC