Re: Oh, this is making *me* sneezy and teary...

Gerald Oskoboiny (gerald@impressive.net)
Mon, 20 Dec 1999 05:44:46 -0500


On Fri, Dec 17, 1999 at 08:29:36PM -0800, Adam Rifkin -4K wrote:
> Back in the early days (say 1993-1994 when Yahoo was rising out of the
> primordial ooze), I never thought I would see the day when $6 million
> was "nothing money" that people like Jim Barksdale and Brad Silverberg
> would throw at companies like TellMe:
> http://www.redherring.com/insider/1999/1217/vc-tellme.html

Hey, cool, I'm glad someone is finally doing this!

Tellme has been secretive about its plans until recently.
"We're switching from stealth mode to low-altitude mode,"
says CEO and cofounder Mike McCue. The company announced plans
to build an interactive media service, allowing people to
access the Internet via simple spoken commands given over the
telephone. "Not just data phones but any phone," says Mr. McCue.

Reminds me of something I wrote three and a half years ago:
(I wonder if I'll eventually get to cite this as prior art
against patents from TellMe ;)

---------------------------------------------------------------------------
Subject: Device-independent pages (was ACCESS THE INTERNET WITHOUT A COMPUTER)
To: hwg-main@hwg.org
From: Gerald Oskoboiny <gerald@cs.ualberta.ca>
Date: Sun, 10 Mar 1996 18:47:52 -0700 (MST)
In-Reply-To: <v01530505ad692b94dec8@[205.216.154.8]>
from "Howard Owens" at Mar 10, 96 05:02:00 pm

Howard Owens writes:

> This appeared on another mailing list I belong to. I thought it might be
> of interest to this group. Obviously, a website that isn't Lynx friendly
> isn't going to be interesting to phone browsers.
:
> >[...] Web-On-Call Voice Browser lets anyone obtain
> >Web documents using only a touch-tone phone, cellular phone, or
> >fax machine.

Excellent point! This once again illustrates the benefit of HTML's
"structure-based" markup approach when compared to a page description
language, and why we as HTML authors should try to use structure-based
markup effectively rather than trying to use HTML to do page layout.

We haven't even begun to discover all the possible ways we'll be
navigating information a few years from now. Recently there was a
Web browser released for the Apple Newton: how many of us know how
our pages will look with that? If you mark up your pages according
to their structure, you don't *need* to check them with every new
browsing platform that becomes available: you just trust the browser
to display your work appropriately given its unique needs and abilities.

Getting back to this "phone-based browser": I checked out the site a
few days ago and was a bit disappointed to see (if I remember correctly)
that you need to purchase some extra software to run on your Web server,
and that this voice-based browsing only works with documents residing
on servers that are running the extra software.

I think it's just a matter of time before we're able to retrieve *any*
Web document using a phone. (I can't count the number of times I've
forgotten someone's phone number and have known *exactly* how to get
it from their Web site, but couldn't because I wasn't near a terminal.)

Below is an excerpt from an e-mail conversation I had with a friend
a few months ago:

[ me ]
| > > Five years from now, I'll be able to pick up any pay phone in
| > > North America, insert a quarter and ask a question about anything.
| > > A supercomputer with an index of the Web's knowledge will figure
| > > out the answer to my question and speak it to me over the phone.

[ him ]
| > The talking yellow pages already exist. If you are referring to a
| > system that actually understands natural language, such as
| > "Hello computer. I'd like to know a bit about Bond's shoe size; for the
| > earliest Bond I mean.", then you'll have to wait far more than 5 years.

[ me ]
| Because of the voice-recognition stuff, you mean? Okay, so they'll have
| some minimum-wage-earning operator sitting there entering whatever I say
| into a computer. Big whoop. An extra 10 seconds, maybe.
|
| Regarding the search itself, there are already good "concept search"
| programs out there that return half-decent results. (like excite.com
| or infoseek.com, which are available from my hotlist.html page).
| It might be a while until we can ask a complicated question and get
| a single answer, but if we ask something like "What is James Bond's
| shoe size?", and the computer says:
|
| Press "1" if you want me to read the section entitled "Shoe Size"
| from the "James Bond -- Statistics" page maintained at the University
| of Manchester;
|
| Press "2" if you want me to read the Internet Movie Database's entry
| on "Trivia" for "Goldfinger", which mentions "shoe size";
|
| Press "3" if you want to hear a list of relevent pages available from
| the James Bond Page maintained by Turner Broadcasting;
|
| So what if it takes an extra minute or two to wade through some menus
| over the phone? It's still pretty durn advanced and useful technology,
| and makes backgrounds, blinking, and other layout-centric markup much
| less compelling.
|
| Or do you doubt that the information content itself will be available in
| five years? (i.e., the answer to the question "What is James Bond's
| shoe size?", in any form). In five years, full-text transcripts of every
| movie ever made will be available on the Net: surely one of the many
| Bond movies makes reference to his shoe size?
|
| > And when such a computer system does come along, I uspect it will be
| > powerful enough to "browse" pages that do not conform to strictly "standard"
| > HTML (which must constitute a good 3% of the net).
|
| But it won't be. Much of the meaning of a page is lost unless it's
| appropriately marked up with tags for headings, paragraphs, lists, etc.
| You can accomplish something that *looks* like a heading using
| <FONT SIZE="+3">Some Heading</FONT>, but that means something much
| different than a real <H2>-based heading, because the <FONT> approach
| doesn't say "we're starting a new section here". (And the "Press 1"
| example I gave above is only possible because presumably there is a
| page somewhere that has a <TITLE> of "James Bond -- Statistics" and
| an <H2> or <H3> that says "Shoe Size". If this page is made with
| layout-oriented markup, such a query is not possible.

I hope this message is considered on-topic: I'm hoping it will help to
illustrate some of the many advantages of structure-based markup. This
doesn't mean pages all have to be boring, only that they should still
make sense when all the layout-specific niceties are removed.

Gerald

-- 
Gerald Oskoboiny  <gerald@cs.ualberta.ca>  http://ugweb.cs.ualberta.ca/~gerald/

HURL: HTML Writers Guild mailing list archives, maintained by Gerald Oskoboiny.

A permanent URL for this message is: http://www.hwg.org/msgid/96Mar10.184836-0700_mst.138805-2+41@amisk.cs.ualberta.ca

[ note: this page uses my ugly HURL software which basically hasn't updated its look since late 1994 or so. ]

-- 
Gerald Oskoboiny <gerald@impressive.net>
http://impressive.net/people/gerald/