[whatwg] RDFa statement consistency

[Kristof Zelechovski]

Having alternate interleaved content streams is a new concept on the web; it
has happened before but only to replaced elements that are guests to the
document but not to the main text.  Returning to Ben's example, the content
verifier should verify that an element claimed to claimed to contain the
author of the resource is recognizable as such to the human reader who does
not have immediate access to the metadata.  Of course, it should also check
that the entities in question exist and can have the property specified, but
that is a technical problem that is much easier to solve.
It is easier to generate description from metadata than to check the
consistency afterwards IMHO but you insist on not doing it.
Do you think that HTML5 should allow arbitrary experimentation under the
banner "Let us just do it and we shall see?"  I suppose you do not think so
either.  RDFa without any means of consistency validation, as incomplete and
imperfect as they have to be, is not mature enough.
I do not think we need to have a consistency checker for the natural
language text in the Web page, for the following reasons:
1. A natural language does not require repeating the same information in two
different ways.
2. All natural language text is, roughly speaking, visible in display mode
so any inconsistencies can be easily recognized; whereas the alternative
information streams are not; they are for different readers to read.

-----Original Message-----
From: whatwg-bounces at lists.whatwg.org
[mailto:whatwg-bounces at lists.whatwg.org] On Behalf Of Manu Sporny
Sent: Friday, August 29, 2008 6:46 AM
To: whatwg at lists.whatwg.org
Subject: Re: [whatwg] RDFa statement consistency

Kristof Zelechovski wrote:
> HTML5 is too crucial as a technology to allow arbitrary experimentation. 

Please refrain from making wildly opinionated and loaded comments such
as this without logically backing up your argument Kristof. Many on this
list and off this list would view a number of HTML5 features as
"arbitrary experimentation".

"Experimentation" is fine and "arbitrary" is a matter of opinion when
applied in broad strokes. The important thing is to discuss the benefits
and drawbacks of each decision without resorting to wording such as yours.

> I
> would rather wait for a consistency checker to exist, at least
approximately
> and conceptually, before having alternate content streams in HTML.  Maybe
> that is just me.

Yes, it does seems to be just you that is making this argument that the
Web must be completely consistent at all times. Perhaps if you could
outline exactly what your consistency checker would check, then we could
make some progress on whether or not it is achievable. Do you think that
we should also have a consistency checker for natural language used in a
web page? Is your idea of a valid consistency checker to solve the
Natural Language Processing (NLP) problem first?