A non-column from Bob on Push

Rohit Khare (khare@pest.w3.org)
Mon, 17 Mar 97 14:12:06 -0500

Bob's been having a slow streak; first the Telco Wars, then the hilariously
misconstrued Java-replaces-HTML from last week. Now, this Push is everything
schtick. And a call for Push standards? C'mon... He didn't even expose the
dirty little secret that most push out there is automated pull (even PC Week
got that right, I believe. DanK made this point on FoRK).

Note on WWW6 at the end. I vote he eats it... :-)


PS: "The Internet's first hot protocol technology was Telnet. Then, in short,
came FTP, remote job entry, chat, e-mail, newsgroups, X Windows, network
management, and, in the Internet's 20th year, the World Wide Web." -- Sounds
like the into to a book, say, _The Evolution of Internet Information
Services_... yeah, that's the ticket...


March 17, 1997

When your mother said no pushing, she wasn't talking about this stuff

We've been talking about it for a year or more. And now it's official. The
next big thing on the Internet, according to the current cover of Wired
magazine, is "push."

Here last April, while discussing PointCast (see "_The future of Web news is
shaping up, and it looks like PointCast_," April 22, 1996), I asked if you'd
prefer to pull down your news when you want it or have it pushed at you as it
happens. My answer is both, depending on the news.

Hey, I'm not claiming that pull and push are my buzzwords. I picked them up
from Michael Zissman, then running Lotus. Zissman spoke about not just pushing
information with e-mail, or just pulling it with Web browsers, but having the
best of both with, well, Lotus Notes.

So, what exactly is push? Is it e-mail on steroids? Is it the reason
e-postage will start to make more sense, as messages grow fat with multimedia?
Is push just deferred Web downloads? Is it multimedia newsgroups? Is it
narrowcasting? It is all of those.

Is push new? Let's leave that to historians.

Marimba pushed push over the top. CEO Kim Polese's 100-percent-pure Java
product, Castanet, convinced most everybody, and finally even Wired, that push
is hot, or cool, or maybe bad, whatever Wired people say these days when
something is the cat's meow. See _http://www.marimba.com_.

As a network plumber, I see push, in its many forms, as an inevitable
response to the bogging Internet. Push admits that the Internet's computing
and communication capacities are limited.

Push is, in essence, caching. Were the Internet not bogging, browser pull
would still be get-ting us by. Push software moves information you'll be
looking for later down onto your personal computer when it gets a chance,
hopefully sooner than later. Then, when you are ready, your information will
be on your disk for immediate viewing -- bam! -- instead of trickled down as
you sit and wonder whether you should give up and hit your browser's stop

Push is personalized. The information pushed onto your disk is selected based
on what you are likely to want. Push software collects various profiles of
what you are interested in. Push selects information based on those profiles.
Push is agent technology. Push is data mining and collaborative filtering.

Push is compression. Knowing what's already been selected for pushing down
onto your disk, and what you'll be looking for next, push software sends only
the differences. Even push's relatively small incremental updates can
themselves be compressed. Push not only anticipates but also lightens the

Push is synchronization. Because the Internet is two-way, push software can
know when you've updated information pushed onto your disk. It can know when
you've filled out a pushed form, for example. Changes on your disk can be
reflected back up to push central. In Notes they call this replication.

And push is workflow. Some of the information pushed might require you to
take action. That triggers synchronization, which might push an action on to
someone else. All of that is push. And we'll see what else.

Amid all the excitement about push, however, I do have one little question.
Upon what protocol standards will push be based?

The Internet's first hot protocol technology was Telnet. Then, in short, came
FTP, remote job entry, chat, e-mail, newsgroups, X Windows, network
management, and, in the Internet's 20th year, the World Wide Web. All of these
are based on standards, mostly from the Internet Engineering Task Force
(_http://www.ietf.org_) and the World Wide Web Consortium

I hate to get pushy, but is anybody working on standards for push? Let us all know.

In the meantime, Michael Zissman will receive the Electronic Mail
Association's Lifetime Achievement Award during EMA '97, April 8-10 in
Philadelphia (_http://www.ema.org_). I'd like to be at EMA '97 to help honor
Zissman. I'd love to be at EMA '97's session on e-postage. But, instead, I'll
be keynoting the Sixth International World Wide Web Conference (also known as
WWW6), April 7-12 at Santa Clara, Calif.'s Santa Clara Convention Center
(_http://www6conf.slac.stanford.edu_). There I'll be arguing that I shouldn't
have to eat my _Dec. 4, 1995_, column predicting Internet collapses during

WWW6 will probably not be a lifetime-achievement-award situation for me, but
you're invited to come watch anyway.