Jeremey

      
we've had a number of arguments about section 4


      
danc

      
i'm the issue raiser


      
jeremy

      
i'd like advice for RDF Core instead of beating up the editors


      
jeremy

      
[paraphrases Issuettes Agenda document]

      
have pulled out some issuettes that Bijon found fault with.

      
straw poll:


      
sandro

      
does that mean we do not go back to last call?


      
brian

      
we shouldn't decide based on going back to last call but instead do the right thing


      
danc

      
[something about preference]


      
vote: 

      
accept: 1 -- opposed: 6

      
cut: 7 -- opposed: -2

      
edit: 6 -- opposed: -2 (interpreted as "against spending more time on it")


      
jeremy

      
social conventions are rooted in the URI specification and the registration procedure


      
danc

      
does anyone understand this issue


      
pat

      
it is "who gets to say what a URI denotes?"


      
pps

      
someone gets to say by fiat that this means that


      
jeremy

      
who feels they don't know enough to vote and are not happy abstaining?

      
vote: 0


      
vote: (on Parsia)

      
accept: 6 -- opposed: -4

      
cut: 5 -- opposed: -2

      
edit: 3 -- opposed: -2 (interpreted as "against spending more time on it")


      
vote: (on )

      
accept: 1 -- opposed: -8

      
cut: 8 -- opposed: -2

      
edit: 3 -- opposed: -2 (interpreted as "against spending more time on it")


      
vote: (on asserting RDF)


      
brian

      
asserting that a page is not here is what 4.2 is about


      
jeremy

      
now voting on asserting RDF in the abstract


      
accept: 3 -- opposed: -3

      
cut: 4 -- opposed: -1

      
edit: 2 -- opposed: -0 (interpreted as "against spending more time on it")


      
vote: 4.2 HTTP relationship

      
accept: 2 -- opposed: -3

      
cut: 6 -- opposed: -0

      
edit: 1 -- opposed: -1 (interpreted as "against spending more time on it")


      
vote: 4.4 Mechanical Inference and Social Meaning

      
Jeremy

      
I can logically infer all sorts of conclusions from an RDF graph that aren't asserted by the semantics

      
accept: 3 -- opposed: -7

      
cut: 5 -- opposed: -1

      
edit: 2 -- opposed: -0 (interpreted as "against spending more time on it")


      
vote: 4.2 Siplleriness of Intended Meaning


      
Jeremy

      
This is more philosophical.


      
Pat

      
these comments are about the word "intended". that's a special phrase. could that be edited?


      
Bijan

      
I thought that "intended" was fatal as it implied a lot of machinery. 


      
timbl

      
could you switch it for authoritative? you have used that before.


      
bijan

      
what if you're wrong about the URI spec fixing the meaning? if you look at M&S, the def is in the schema. Most RDF doesn't have schemas. It's better to be quiet than to say it's bad.


      
timbl

      
we're not trying to define a perfect mathmatical system. we can talk about how it's define to work. Generally, most INet specs, HTTP, SMTP, specify how it's supposed to work. I've written schemas for most of what I've written, but that I didn't for a while didn't break it. The system is there and though folks don't always stick to it doesn't mean we shouldn't specify it.


      
bijan

      
if everything you said was false, i don't think everything would be broken. i don't believe your characterization.


      
timbl

      
i think there are phrasing problems. when it says "legal", i say rip it out.


      
bijan

      
i don't think there is a single interpretation for a document


      
murmer

      
and that is provable


      
timbl

      
when there is a specification for h1 in the HTML spec...


      
vote: slipperiness of intended meaning

      
accept: 3 -- opposed: -all

      
cut: 5 -- opposed: -2

      
edit: 6 -- opposed: -0 (interpreted as "against spending more time on it")


      
vote: third party publication


      
accept: 0 -- opposed: -7

      
cut: 8 -- opposed: -1

      
edit: 5 -- opposed: -0 (interpreted as "against spending more time on it")


      
vote:


      
timbl

      
this is an attempt to explain what RDF means in terms what can a person be held to. i imagine "held do", "liable" is just a framework for understanding the spec but bijan's comments are true, it is not in scope to say how the law is involves


      
frank

      
there is nothing like this in HTML spec. this implies that this is somehow subject to different laws.


      
mike

      
do we need a liable markup language


      
jeremy

      
i've expressed A(ccept) as meaning the 


      
timbl

      
this is directly about the combination about the combination of two graphs.


      
bijan

      
i meant to comment on the earlier part of the doucment.


      
jeremy

      
let's change the dicussion to [@@@]

      
could be read as should there be special specs for lawyers?


      
timbl

      
all the RDF spec has to say is "go look up the meanings of the assertions". we don't need assertions about the greater context. this is not to explain about the law. the spec should be clear about what a document means.


      
jeremy

      
there is some explicity text. there is the word "commit".


      
bijan

      
i may not be liable for the entailments of what i said. i could argue an error.


      
timbl

      
you can get out of anything by claiming error.


      
jeremy

      
narrowing question: includes words like "eg. legal"


      
Bernard

      
[@@@]


      
jeremy

      
3 things came up in discussion[@@@]
    
	  
  1. timbl: 
    2. 
	

      


      
danc

      
wide publication?


      
scott peterson

      
existance of mechanisms is valuable to the legal process. they don't define the law, but they are useful. for instance, locks and walls.


      
timbl

      
i think most feel that is true. should that be in the lang spec?


      
scott

      
i don't think it should ignore the fact that it will play a role. we shoudl not exclude the possibility


      
patrick stickler

      
can this go in an aux document?


      
timbl [at white board]

      
when someone makes up a property and uses it, how do we know what it means?

      
my view of web architecture, unformalized, but useful

      
when you have a URI, you look in a registry that points to another spec.

      
that spec points to the DNS spec which allows you to n IP

      
you use the HTTP spec to open a TCP connection and send a request based on that URI

      
you get back a string of bits that have a mime type.

      
that give you a pointer to the RDF spec.

      


      
patrick stickler

      
[re-expresses question]


      
timbl

      
specs says how bits are arranged. rest of this is all context


      
danbir

      
you get a triple set and want to see which are core and which are fluff.


      
timbl

      
if you understand part of it, that's ok.


      
[... lots more dicussion along this line...]


      
timbl

      
Roy and my models differ because it's not tested in the conventional web.

      
Roy maintains philosophically that foo.rdf could be a car.

      
That introduces a lot of inconsistancy for me.


      
red /w yellow stripe

      
why can't it be a car (or piece of hardware)


      
lynn

      


      
danbri

      
you run into the same problem with content negotiation.

      
why can't i give the thermometer in my tivo at home a URL.


      
brian

      
RDF core is trying to write specs that will help some folks build systems and some others use them


      
ericP

      
... flack from topic maps on conflation ...

      
... good practice ...


      
brian

      
i think that this is different from what tim was trying to impose


      
timbl

      
if a document has a hash in it and starts HTTP, it is a conceptual identifier.


      
patrick stickler

      
with conneg, if you have a URI that denotes a car...


      
timbl

      
i disagree with the premis -- no URI for that car


      
pat

      
do you really eman "denote"?


      
timbl

      
i usually use "identify"


      
pat

      
but that is ambiguous


      
lynn

      
what we specify may be different from how it is used. can need to make sure it's not prohibiting common practice


      
frank

      
... not trying to say way document this should go in:

      
we're not trying to specify how people are going to use this.

      
a company can put up an SQL DB with rows and columns and users will derive meaning from them.

      
we're not saying how the law will interpret them, but we can say that laws may exist or may come to exist that hold folks responsible for what they say in RDF.

      
if we can't say this about RDF, we can't say it about OWL.


      
jeremy

      
typically, most folks voted for cutting. editing had less resistance but also less support.

      
easiest thing to do is to cut


      
brian

      
"cut" should be for "this round of specs"

      
c and e are very similar.


      
patrick stickler

      
can it go in another document? as opposed to cut it meaning "throw it away"?


      
danc

      
the coord group could own such a thing

      
if we set up such a mailing list and found an editor, who would participate?

      
vote: 10-ish -- same as who have been participating


      
timbl

      
it is bad in general for a spec to have another spec to help convince you.

      
need not-normative spec


      
jeremy

      
should section 4 be placed in another document?


      
ralph

      
will we have to re-namespace?


      
danc

      
we've done more monkeying than this.


    


    
off to sandro in irc://irc.w3.org:6665/#swarch