[whatwg] Should <address> be more general-purpose?
ddailey at zoominternet.net
Tue Feb 27 19:42:58 PST 2007
Benjamin Hawkes-Lewis and Simon Pieters are having a discussion that I
understand (at last... at least, sort of, or at least ... I think I do) .
The discussion concerns the meaning of the word "address" and the tag
<address>. How much of the meaning of the word should reflect (for good or
for ill) in the <tag>.
I gather there may be ... for want of a better word ... "context" here:
things like what the proposed use of <address> was and historical arguments
and worst case scenarios and cost benefit analyses and best practices and
schema, rubrics and editorial policy. Accordingly, via any of said
contextual precedents, my remarks may be summarily (and, I hope, smugly)
address as in:
(1) : my address is 914 Chamomile Row, Herbtown, General Nostalgia
(2) : http://www.nobodyhere.com/justme/me.here
(4): Dept. of Departmental Subdivisions, Universe of Indivisibles,
lucubration -- case #
(1) is sort of like a unique locator within some physical space
-- often used to contact a person or corporation; sometimes used to
verify an identity
(2) is a URL/URI. is a unique identifier within the Internet
-- generally used to retrieve information, usage cases can become
(3) is sort of like (2) except for oddities.
(4) -- used to verify identity (for purposes of certainty in
attribution), or to contact a person.
Example (2) is handled in various ways in HTML -- usually, I think, as the
value of an attribute (or object property):
* : value=url -- (the "value" attribute is just a string and has no
We may imagine, with little difficulty, a schema (perpetrating a minimum of
linguistic injustice) that handles these cases.
The others: (1) (3) and (4), on first glance, would seem to require some
sort of semantic tag living outside HTML but rather in some OWL-like space.
But then, what does "address" really mean in this modern era?
The term "address" probably has taken on a host of specialized meanings
(within the technological realm) that are slightly different than the above
uses -- some technology induced; others not.
To what extent can those who define "standards" allow themselves to be
influenced by "meanings" that humans bring with them? Should the humans
learn to abide by the standards, or should the standards adapt to extant
human usage (all bolluxed as it seems to be) ?
Let's look at one of the simplest uses of "address": use case (1)
a unique locator within some physical space -- (1) : my address is
914 Chamomile Row, Herbtown, General Nostalgia 3.14159
Lots of relativity surrounds the way we provide unique locators in physical
space. Google Maps give us two obvious, but correlated, ways: street address
and (x.y) pairs of the form (latitude, longitude). Both rely on technology.
Street addresses only exist in places for which postal authorities have
already consented on a system of unique identifiers. Latitude longitude
pairs work well for areas that have been mapped (or have good cell phone
reception). Lots of places on earth remain indeterminate: Mines, jungles,
and the places where aircraft and submarines go. UPS probably does not
deliver in the Mariana Trench.
So suppose we step away from the technology for a moment and ask, what do we
mean by an address?
: what is your address? == translate == : how may I contact
: what is your address? == translate == : how may I trust you?
How does that differ from
: what is Marilyn's address? == translate == : how may I contact
: what is Marilyn's address? == translate == : how may I trust
In the second case "how may I trust her" coming from a third party is quite
different than "how may I trust you" coming from the second person.
What humans (including those who invent internets) mean by "address" can be
pretty complicated. If we, who invent internets, were to pretend that we
know better what an "address" is than those who for several centuries have
used, defined, and inhabited "addresses" then we will have made a new
language that humans will not only refuse to learn, but which they will be
compelled, by sheer moral indignation, to refuse to learn.
In experimental psychology in the US in the 1950's behaviorism was rampant.
Folks were fond of making "operational definitions". The classic example
was B.F. Skinner defining "thinking" as its behavioral correlate: movements
of the vocal chords (which often accompanied people's unvocalized rehearsal
of thoughts they might later speak). The movements were measurable; the
thoughts were not.
Some reviled the attempt to reduce "thought" to the "thought-related
behavior" which happened, at time t, to be measurable by technology
Tech(t) -- If, for two times t and s, if t>s, then the "modern" sense of
progress would imply that Tech(t) is "better than" Tech(s). The
epistemological conundrums were rather fun for the philosophers of the
1960's, who made good sport of it in their journals. Nevertheless, the
psychologists still got more money from NSF. The AI researchers got to laugh
loudest, but they still have to pay for their music downloads.
> Simon Pieters wrote:
>> Do UAs need to know the scope of the <address>? What could they do
>> with this information? (If it is important, then we could use a class
>> or a new attribute for this IMHO.)
> Using <address> in this way has been difficult since it's hard for
> agents to infer document structure from current tag soup (I'm currently
> grappling with the mess in my Hypertextuality extension where I'm trying
> to find the relevant permalink for articles and posts). With it's
> heading parsing rules and <article> element, (X)HTML5 should at least
> make this easier for documents explicitly authored to its specification.
> That there should be a way of expressing the scope of contact
> information is more important than the more technical question of
> whether it's in an attribute or element or registered class name.
> Obviously specifying an element or attribute is preferable, as then UAs
> would be substantially more likely to do something with it.
>> <address> has been around forever. Yet no UA has done anything useful
>> its semantics as far as I know. That suggests to me that the use-case is
>> not a real-world one.
> I tend to think the relationship between "real-world" utility and HTML
> elements is mostly the other way round. Elements become widely useful
> because user agents happen to make use of them (or, more often, invent
> them in the first place, q.v. canvas); agent developers don't
> necessarily recognize the utility of new elements in external
> specifications, however. In fact, individual developers are often
> entirely unaware of their very existence.
>> Isn't it better to make <address> more general so that its semantics
>> is more like how most authors use it so that it becomes a convenient
>> styling hook for authors?
>> I don't think it's a good idea to invent a new element when the use-case
>> is so weak that most authors don't bother using it and no UA have
>> implemented anything useful with it. I'd rather drop <address>
> I don't follow. You seem to be asserting both that "most authors" misuse
> <address> to mean any contact info /and/ that "most authors" have no use
> for an element like <contactinfo> that is actually for "any contact
> With regards to the practical utility of <address>, I think this is
> bound up with the whole matter of the web's highly immature techniques
> and technologies of citation. The age of print took a while to sought
> out it's techniques and technologies too. The fact that it took such
> time does not mean that there was no use-case for citation.
> An <author> element might kill several of these birds with one stone.
> Benjamin Hawkes-Lewis
More information about the whatwg