RE: [httpRange-14] What do HTTP URIs Identify?

 

-----Original Message-----
From: Tim Berners-Lee [mailto:timbl@w3.org]



On Monday, August 5, 2002, at 11:53 AM, Bullard, Claude L (Len) wrote:-----Original Message----- 

Just for the documentation,  the following was someone else's comment, not mine;   
Possibly Fielding who brought up the situation of the Academie.


(Len:) Where the consumers of those specifications are relatively few in number 

and have comparatively aligned interests this can be made to work (viz. 

the W3C). But the semantic web has considerably grander ambitions and 

the consumers of its specs are (hopefully) considerably more numerous 

and diverse in their interests. Attempting to coral those consumers is 

likely to be about as successful as the Academic Francaise's attempts 

to banish imported anglicisms from French. It would be a shame if the 

W3C ended up looking similarly pompous and preposterous. 


You miss a fundamental difference. The Academie Francaise was 

trying to create a single vocabulary for all communication, and to 

mandate its use throughout France. This flies in the face of the 

way natural language really works. 


[len] I agree.  Somewhat like HTML.  Gencoding has that 

sense of finality that doesn't work in the face of natural 

processes of inner representation of external events with 

a feedback loop between them.  That is the synopsis 

of "chaos is the engine of evolution".

 


1. The Semantic Web will have many vobaularies, and 

all the W3C will standardize is to inftastructure rules 

which make it work - the architecture of HTTP and RDF, etc. 


[len] Yes.  A system that provides a means for interpreters 

to communicate.   But RDF is itself, a vocabulary, so this 

becomes like the linking architecture.  One can do as 

HTTP does and create a syntax and set of processing 

rules for a type of link, or as ISO 10744 did orignally, 

create a set of link types.

 


2. The Semantic Web is a logical system, not a natural language 

system. We are defining it -- engineering it -- in a different 

way to natural language, deliberately, so that it can have 

very consistent properties.


[len] Right.  Systemic definition.  I agree that comparisons 

with natural language and even the mental processes are 

stretch.  One can go to nature for models to emulate and 

the discipline of semiotics attempts that.  Or one can start 

with a model one can be reasonably sure works in the 

medium, and go with that.   The second path has a higher 

probability of getting fielded and maintained.

  


Ah... but where the assets are critical, it is a good 

idea to give teeth to this, to lash the competitors for 

the URI together with knives. 



I don't understand that paragraph, but continuing: 


[len] Some aspects of a system where that system 

has global consequences may require force of law. 

Law is not necessary to define a system, but our 

social constructs are such that it may be needed 

to prescribe outcomes for violations.   The issue 

here will the polity of the representative body that 

can make such prescriptions.   Misuse of a URI 

may be one kind of violation that could require 

a certain prescription.   Willfully fielding systems 

that violate that may require another, and so on.

 


Because of competing organizations for specifications, 

a code specification alone won't do the job, TimBL. 


It simply won't. 


The world is full of people and organizations which 

flout that and always will. For those that see 

specifications for critical systems to be of such 

value that flouting them is dangerous for the polity 

at large, the specification must have the force of 

law and it must have teeth. The W3C is a vendor 

consortium by your own choice and design. That makes 

it the wrong polity to dispense law. 



W3C does not dispense law. All it can do, like any 

group making standards, is to define what a URI 

identifies, what an HTTP message means, etc. 

It is the best body to do that because the US Congress 

and the United Nations are not practitioners in the 

arts of global system engineering. Nor do the 

bodies which do make laws generally do engineering. 


[len] Correct.   That suggests teamwork at the highest 

levels of these organizations.   This is a hard and arduous 

path.  It is short sword work at its most painful, but I don't 

see another way.   Lessig argues or seems to about 

free market vs controlled market in the effect on technology. 

I believe different rules govern different phases of the transitions 

from innovation to law.   Specification is suitable 

for the phases nearest innovation, and standardization 

is suitable for those nearest law.   These are in that sense, 

polarities to be managed, not problems to be solved.  Where 

parts of the Web infrastructure are critical to the proper 

operation of the web, and where these parts have become 

successful to the point that they have global effect over 

the producers and consumers of vital products, then these 

pieces should have the support of law.   It is critical that 

the authorities for these specifications be able to discern 

and enunciate clearly that such a feature has emerged 

from specified systems, and to be present that case to 

those empowered by their representative status to make 

law.   For this, it may be useful to define the role of 

ISO, a body representative of agreements between 

nation states, as that body which makes such presentations.

 


Technical specifications work together. 

If the W3C makes a spec which clearly (enough) 

states what a document (say a P3P privacy policy) 

means, then governments may, in their elected 

authority, make adherence to the spec mandatory. 


[len] They may, but I assert that the documents they 

are citing include the privately owned property of the 

vendors.   This sanctions the tragedy of the commons. 

We should attempt to ensure that where we have such 

documents, that their contents are in the commons even 

if controlled by the document provider.

 

Also, the courts may, in the event that a fraudulent 

misrepresentation has been made in a P3P 

document, refer to the W3C specification, as 

anyone making a P3P policy will typically 

assumed by the courts to be responsible for the 

statement made as interpreted by the spec. 


[len] See above.   It isn't wrong, but it sets a 

precedent for the privatization of the commons.

 So the law and the specifications work together. 

Specs should not be made by lawyers, and 

laws by technical people, but they should be aware 

of each other's fields. 


[len] Yes, but a formal means to do that may be 
useful.   We are seeing more consortia and there 
will be conflicts.   What to do when such a conflict 
escapes the scope of specification and emerges in 
the realm of social conduct is difficult when the 
venue of the conflict is vague or overlapping.  We 
cannot always be precise about this at overlapping 
boundaries, but we can be precise about means 
to arbitrate should negotiations break down between 
parties.   Emergence is simply when two forces 
engage, a control emerges above them.  That is 
both a natural and realistic view of how to solve 
that problem, in my opinion.
 

Isolate out the pieces which are mission critical 

and must have the force of law, then submit these 

specifications to ISO for standardization where the 

mebers are nation-states with the authority granted 

by the people to dispense law "of the people, by 

the people, for the people". This is not poetry, 

philosophical irrelevance, or rhetoric: it is the 

principle which governs and maintains governance 

best, by example, and by historical proof. 


I am sorry to diillusion you, but ISO has no such 

authority granted by the people in general. 


[len] I am not disillusioned.   Often, one has to 

create definitions for a working system; not simply 

live with what the last authority created.  Still, I think 

there is room to do this.   See next response.

 

ISO is a collection of national standards bodies, 

and national standards bodies are effectively industry-based 

consortia. Any claim by ISO to be more representative 

than W3C or IETF would be a matter of great debate. 


[len] They are a collection of standards bodies but 
and the work is performed by the industry based 
consortia, but the *voting* members are nations:
 
"A member body of ISO is the national body "most representative of standardization in its country". Only one such body for each country is accepted for membership of ISO. Member bodies are entitled to participate and exercise full voting rights on any technical committee and policy committee of ISO. " 
 
One might assume a vote by a nation can be construed to be a vote by the bodies represented.   If these do not include the rights of the 
people of the country, then this is something to be considered as an issue to take to ISO and the governments of the members.   One 
can then ensure that the commons is part of that debate. 


You cannot solve it in code, Tim. 


I said that the specifications must define it. 

That is not solving it in code. 

It is the right of W3C (or any other group) 

to define a technical system by making 

specifications. There is an implicit agreement 

that when you use such as system, you 

operate under the terms of the specifications. 


[len] Just so.   The problem here is social.  I suggest a social solution. 


Just as when you fill in a tax form, you operate under the 

form author's definition of what the fields mean. 


I think that this principle should probably more 

widely and clearly stated, or we will see many more attacks 

such as the spammers who claim that they can put 

anything in RDF822 header fields, because 

"internet specifications aren't laws". 

They aren't laws, but they define the meaning of messages, 

and to lie in such a message is fraudulent under existing law. 


[len]  But it is fraud without a prescriptive outcome.


It does have a 

social component, so society must choose. 


Society can chose whether to mandate that everyone 

uses the semantic web -- unlikely. 

Society can chose whether, individual by individual, 

one wants to use the semantic web. 

But once one uses it, the meaning of ones messages 

is deemed to be defined by the specifications. 

And the law should and I hope will uphold that. 


[len] I agree.  And that is something that the lawyers 

must and can find a way to present to the law making 

bodies.   We must find a means to present the issues 

to the lawyers such that the prescription is neutral to 

the source.  This is the concept of isonymy:  equality 

of all before the law.

Thanks very much for this discussion.

len