Re: Netscape & New HTML

Brian Behlendorf (brian@wired.com)
Fri, 21 Oct 1994 22:47:52 -0700 (PDT)


Date: Fri, 21 Oct 1994 22:47:52 -0700 (PDT)
From: Brian Behlendorf <brian@wired.com>
To: Dylan Northrup <northrup@chuma.cas.usf.edu>
Cc: Multiple recipients of list <www-html@www0.cern.ch>
Subject: Re: Netscape & New HTML 
In-Reply-To: <Pine.SUN.3.91.941021170313.8522A-100000@chuma>
Message-Id: <Pine.BSI.3.91.941021213311.26918j-100000@get.wired.com>

On Fri, 21 Oct 1994, Dylan Northrup wrote:
> Before I start to reply, let me just state this:
> 
> 	I am glad that Mcom did what they did (even if they went about it 
> 	the wrong way).  They implemented features that people wanted.  If 
> 	they'd have waited for a standards committee to complete the HTML 2.0 
> 	spec I doubt Netscape would have been released anytime before next year 
> 	if then.

Okay, and before I go on:

The pace of development in public WWW technology has indeed been 
agonizingly slow.  Why? It seems to me that this stuff took off so fast 
it knocked the wind out of a lot of us, and it was all we could do just 
to tread water.  When only 5000 people knew about it, it was a lot easier 
to experiment with it, and have those experiments rolled into the 
canonical release.  Many of those experiments worked marvelously - many 
didn't, yet are a part of our system by sheer force of momentum.  

I'm sure many others have a better idea why certain things that should 
have happened didn't, some can speak publicly about it, others can't.
Yeah, you could say I'm disappointed.  But to some extent it's a 
chicken-and-egg situation - no one's experimenting, so no one's talking
about the proposals, and because no one's talking about the proposals,
no one's experimenting, etc.  The fact that many of the premiere browser
authors left to work for one company might be significant, it might not 
be, who knows.  The fact that NCSA and CERN have been glacial about even
experimenting might be an issue.  Without the cycle of experimentation no 
progress is being made, because no one has a clue how well a part of that 
proposal might work.  

> 	The Netscape extensions are not set in stone (they've only 
> 	released a beta).  They might change if there was a intelligent, 
> 	reasonable suggestion for the change.  

But will they, after some of their customers have been weaned on the 
promises the tags make and have lots of documents which look really ugly 
without them?  Seriously, this is a one-way street.  

>	I'm quite sure if the 
> 	HTML3 group would make some decisions and submit a final draft 
> 	the guys at netscape would be more than happy to look at it.  
> 	Perhaps then the HTML3 guys will finally get around to doing some 
> 	work on HTML3, ne?

check out 

http://info.cern.ch/hypertext/WWW/MarkUp/HTMLPlus/htmlplus_1.html

It looks like a pretty complete proposal if you ask me.  Last_modifieds
are around April or May, while it looks like it's been around since 
November 8th of last year.  The fact that there's been very little 
discussion about it is much more troubling than the fact that there's
no browser for it yet.  The fact that Mcom chose not to implement any
of it as an experiment isn't as troubling as the fact that they didn't
even express any opinion about it at all.  Folx, this just ain't gonna 
work if no one communicates.  And yeah, I still have problems with it - 
I'm working on a list of things I'd do differently, and will submit it 
when ready. 

> Now, on to the responses:
> 
> On Fri, 21 Oct 1994, Brian Behlendorf wrote:
> 
> ["HTML is big, PostScript is small.  Let's just make HTML more powerful 
> and use it" paragraph deleted]
> 
> > I disagree.  If you start dangling the carrot of page-layout to document 
> > authors, they won't be happy until they have complete, *complete* control 
> > of how it looks.  Adding tags every once in awhile to approach a page 
> > layout language just isn't going to work - give them <center> now, give 
> > them <font> later, give them <pitch> and <indent> and <color> after that,
> > and pretty soon you have something that functions a whole lot like 
> > PostScript.  
> 
> I believe you are right that some people (servers) will want to have 
> complete control over how their documents are presented however there 
> will also be people (clients) who will have the fnal say about how they 
> view the file ("I hate 'san serif'.  Mothra, my WWW viewer automatically 
> changes all the <FONT TYPE="san serif"> to <FONT TYPE="arial">.")

In total agreement - Hoken's proposal for style sheets addresses that 
very nicely.  

> > HTML
> > could have hundreds of tags, and hopefully the browsers would be smart enough
> > to fill in the details when needed.  Though I think we'd have an easier time
> > going the other direction, putting semantic information into PostScript or
> > PDF, but whatever.  
> 
> Which would make obselete any HTML documents currently written.  I don't 
> want to go back and rewrite my pages again.  But, I agree this is not 
> really the issue.

Wrong, it wouldn't make it obsolete, since text/html; version=2.0 would 
still be a supported standard.  

> > And these people will then clamor for more page layout
> > possibilities, and gradually the semantic power of HTML will fall by the
> > wayside.  I mean, who's going to say <H1>Very Important Line</H1> when they
> > can say <font size=10>Very Important Line</font>?  Who's going to say
> > <address>brian@wired.com</address> when they can write <italics sans
> > serif><font size=3>brian@wired.com</font></italics>?  Who will care, as long
> > as it looks right on "their browser"?
> 
> You show me a Web page that really uses the semantic power of HTML as 
> opposed to page layout and I'll grant you that its importance.  
> However, I've not found one in my travels across this widely woven web.  I 
> (and almost everyone else I've personally talked to) uses HTMl for it's 
> layout possibilities only (i.e. <li> instead of using tabs).

I've tried to do a reasonably good job on our site.  <li>'s only used for 
lists, <address> is used for any place an address is cited, <code> used 
whenever program code is shown, if there were a <copyright> we use it at 
the bottom, we use <blockquote> a fair amount when one article quotes 
another.  I digress and use <b> when I could use <strong> at times, but 
whatever.  If I had known about it earlier I probably would have put
in <meta type="author" content="Joe Author"> type headers in the <HEAD>, 
but when I have a few spare weeks I'd like to put that in.  There will be 
a few scars of battle in the stuff we will be putting up soon, but even
there we use semantic markup when it's needed.  Nick could probably 
comment on the usefulness of semantic markup in searching - i.e., have a 
search algorithm that assigns words in <H1> tags a relative weight of 10, 
in <h2> a weight of 5, in <h3> a weight of three, etc., compared to 
normal text.  

I would also argue that any page that wasn't properly semantically marked 
up can't be guaranteed to look attractive on the wide variety of 
platforms and browsers out there.  But then again some people don't think 
there will be a huge variety of browsers in the near future.

> People are familiar with using a WYSIWYG word processors where you use 
> fonts, tabs, and spaces (not headers, lists, and breaks).  It's only 
> natural that as the Web made it's way into the public, people would try 
> to reshape it into what's familiar to them ("What do you mean it doesn't 
> let me put two spaces between my sentences?").  Most people are worried 
> about the general appearence of their document, not the logical style.

Unfortunately, yes.  

Does anyone have a URL to a really good resource on the net that explains 
in nontechnical terms why semantic markup is a good thing?  The best I 
can say right now is that it expresses ideas at a much higher level than 
page layout languages do.  And because it's at that higher level, you can 
do a lot more with it, it's more reusable, it's more portable, it can be 
transmogrified into something completely different yet still convey its 
ideas.  In an information space where the amount of information present is
just overwhelming, as the internet has become (and it will only get 
worse), being able to deal with and navigate among documents semantically 
is essential.

I dunno, maybe I'm alone in this, in which case I'll shut up and let HTML
become whatever people want it to become, and consider it opportunity
lost.  

	Brian