Fw: Column-eatin' season

Rohit Khare (khare@www10.w3.org)
Wed, 15 Jan 1997 11:09:55 -0500

Just for all y'all who were curious how much of a suck-up I can be... :-)

Seriously, I've been real quiet on FoRK because of the incredible
grad-school-app pressure and my own procrastination. It's killer, though,
because this time, I *really* desperately want to shift over to the PhD and
the stakes seem really high.

The upside of starting this late is that I can't waste the 20x effort I
spent two years ago preparing 1/2"-thick binders of application info :-)

Rohit Khare

> From: Rohit Khare <khare@w3.org>
> To: metcalfe@infoworld.com
> Subject: Column-eatin' season
> Date: Wednesday, January 15, 1997 11:06 AM
> Hi --
> I'm the guy who originally suggested you eat your 'Net collapse column at
> WWW4 in December 1995. Not a stroke of originality, mind you -- the
> was John Perry Barlow; us NeXTWORLD staffers made him eat his 1993
> prediction column that NeXT would fail (with salt and pepper, mind you,
> we're not inhumane). Still, just thought I'd indulge myself by letting
> know where the consequences you so narrowly avoided last month came
> In return, I wanted to offer some kudos for your 1997 agenda-setting
> on challenges facing the Internet. In particular:
> > Internet plumbing that needs fixing in 1997 includes the underlying
> > protocol (TCP/IP), the DNS, and the Web's HTTP. Too many
> > routes is the problem with TCP/IP. Robust directory services are
> > needed to enhance DNS. And HTTP needs to open many fewer
> > short TCP connections
> At least I can report that W3C is making progress on the last one, just
> teaching people how to use HTTP/1.1 correctly:
> http://www.w3.org/pub/WWW/Protocols/HTTP/Performance/Pipeline.html
> "This work shows a significant performance gain by using HTTP/1.1
> persistent connections and pipelining. This can decrease the number of
> packets by a factor of 5 and yield a factor of 2 in speed. "
> But I'm wondering how we'll make progress on the rest. Scratch that, I'm
> wondering how *I* can help make progress on the rest.
> I joined Tim Berners-Lee's startup consortium right out of Caltech to
> develop the next generation of Web technology. Now, though, I want to
> further out and do some fundamental research on how we're going to lick
> next massive increase in scale, mobility, and bandwidth -- for a whole
> range of core protocols, not just HTTP. This seems like a very fruitful
> area for PhD research, but I keep coming up short on places to embark on
> it.
> It seems like many of the best Net protocol designers, especially above
> wires-and-bits layer, are outside academia. MIT LCS is one notable
> exception, but even after working here two years, I think it's a real
> nut to crack. What are some of the degree-granting labs you think are
> adressing some of these issues?
> Any other advice for those of us seeking to follow in those research
> PhD-startup company-publishing mogul steps? :-)
> Thanks,
> Rohit Khare
> http://xent.w3.org/FoRK-archive
> ---
> Rohit Khare -- World Wide Web Consortium -- Technical Staff
> w: 617/253-5884 -- f: 617/258-5999 -- h: 617/491-5030
> NE43-344, MIT LCS, 545 Tech Square, Cambridge, MA 02139