[Web4lib] FW: Cave Day Announcement

Sun Apr 15 14:09:40 EDT 2007

Aaron,

A number of years ago I gave a talk at SLA  titled "The Web in the Blink of
an Eye."  The premise was to ask would the Web be fast enough so that
clicking on the next page would result in the page being painted as fast as
your eye blinks.

Foolishly, I said at one point during the talk that I didn't actually know
how fast an eye blinks.  Very foolishly I said this in front of a crowd of a
few hundred librarians.  Never confess your ignorance to hundreds of
librarians.  :-)  When I got home, I had an inbox full of citations to
scholarly research as to the average duration of an eyeblink.

My talk back then cited Amdahl's Law, which says that if you make part of a
computer infinitely fast, it doesn't matter if other parts remain slower.
That law still applies today, and it's still true.

Tomorrow when I give a talk on Detroit's main NPR station, WDET, I will use
a dedicated ISDN line.  ISDN offers pathetic speeds -- 64 K or more
depending on how you count.  But ISDN guarantees performance at the rated
speed.

This is not a minor point.

Circa 1997 when I arranged a broadcast of a Doobie Brothers concert from a
field on Michigan State's campus, we set up an ISDN line through a manhole
cover on Munn Field into then-Ameritech (cum SBC cum AT&T) .  (We worked
with Audionet.com, later sold to Yahoo, making its CEO, Mark Cuban, a
billionaire.)

Even today, if I was doing a live broadcast from Michigan State I'd choose a
dedicated line with guaranteed quality of service.  In essence, that's what
WDET has asked for tomorrow.  We have a gigabit or better pipe on Michigan
Lambda Rail yet we choose a 64K or 128K pipe because it has the elusive QoS
that the Internet community has promised for a decade and never delivered.

Interstate highways theoretically offer Autobahn speeds, but we don't drive
150 miles per hour on real roads.  France just demonstrated how a train can
go 350 mph but that was a demonstration.  When you cite the burst rate of a
medium you mislead people.  A very real example is a university classroom
that supports all Wi-Fi standards.  Fill it with 50 MBA students who expect
802.11g (theoretically 54 megabits/sec) but it falls back to 802.11b speed
(11 meg/sec) if a single person has a legacy Wi-Fi card.

In short: you can't measure your broadband performance by the rate of your
fastest link,

This is an important, even vital, point.  Promising patrons a 1M or 2M or 5M
of 1G link, without considering the entire fabric of the network, is a false
promise.

/rich

PS -- See some of you at Computers in Libraries in Arlington, where no doubt
folks will tell me about other times I've blinked...

On 4/12/07, Dobbs, Aaron <AWDobbs at ship.edu> wrote:
>
> Hi Rich,
>
> Thanks for pointing out my combination of high-speed and broadband --
> most of the lay-people (non-computer/information scientists) I know use
> the two terms interchangeably (obviously, I do myself) -- these terms do
> have pretty narrow definitions that are useful to digiterati of course.
> Sorry for forgetting my audience :)
>
> My "worldwide definition of broadband" is my loose combination of
> apparent national averages in well connected countries (Japan being the
> fastest-connected at 5+ MBps, see: http://speedmatters.org/ala for some
> fun comparisons.
>
> My (academic) library & campus connects at an average of 3MBps midday,
> my house (nominally charged for 3 MBps via Comcast - for *way* too much
> money) averages about 1.7 MBps in the evening.
>
> Lastly, I suspect the FCC defined the 200KBps as high-speed so we could
> look good in the worldwide self-reported internet connectivity
> statistics lists -- I mean, who would want to admit to the US public
> that we are now 5-10 years *behind* most of the rest of the developed
> world in terms of connectivity?
>
> ALA's Office for Information Technology (OITP) has done a far better job
> of terms definition than I did in my casual riffing on a fairly broadly
> held view amongst the general public.  See:
> http://www.ii.fsu.edu/plinternet_reports.cfm for the ALA/Gates
> Foundation Public Library Connectivity Study.
>
> -Aaron
> :-)'
>
> -----Original Message-----
> From: Richard Wiggins [mailto:richard.wiggins at gmail.com]
> Sent: Thursday, April 12, 2007 3:37 AM
> To: Dobbs, Aaron
> Cc: web4lib at webjunction.org
> Subject: Re: [Web4lib] FW: Cave Day Announcement
>
> Aaron,
>
> Sounds like a cool demonstration and application...
>
> With all due respect, you're conflating two terms -- "high speed" and
> "broadband," and you're actually overstating the FCC's definition of
> "high speed" Internet access.
>
> Also, would you please cite an authoritative source as to the "world
> definition of broadband"?  A former boss defines "broadband" as
> "whatever is twice as fast as what I have on my desk."
>
> The FCC has caught considerable flak for its relatively low bar for what
> might be called high-speed.  Their definition basically called for
> something faster than ISDN in one direction, presumably download.  But
> many people, including myself, felt that a goal of universal ISDN speeds
> to homes and businesses was a worthwhile and achievable target.  Mitch
> Kapor argued for it.  And years later, in 2007, we don't have it.
>
> If the ALA is doing a study and proposing standards and goals, they
> should define their terms carefully, for today, and for 5 and 10 years
> out.
>
> Following is the FCC definition of "high speed"
>
> /rich
>
> FEDERAL COMMUNICATIONS COMMISSION RELEASES DATA ON
>
> HIGH-SPEED SERVICES FOR INTERNET ACCESS
>
> High-Speed Connections to the Internet Increased 34% During 2004 for a
> Total of 38 Million
>
> Lines in Service
>
> Washington, D.C. - The Federal Communications Commission (FCC) today
> released
>
> new data on high-speed connections to the Internet in the United States.
> Twice a year, facilitiesbased
>
> broadband providers must report the number of high-speed connections in
> service pursuant
>
> to the FCC's local competition and broadband data gathering program (FCC
> Form 477).
>
> For reporting purposes,
>
> high-speed lines are connections that deliver services at speeds
>
> exceeding 200 kilobits per second (kbps) in at least one direction,
> while
>
> advanced services lines
>
> are connections that deliver services at speeds exceeding 200 kbps in
> both directions.
>
>
> /rich
>
> On 4/11/07, Dobbs, Aaron <AWDobbs at ship.edu> wrote:
>
>        Question: High speed internet connectivity in public libraries?!
> Why on
>        earth would *they* need *that*?!
>
>        Answer: See the Cave Day announcement below for yet another good
> example
>        supporting high speed internet in public libraries.
>
>        (High-speed in this case means the world definition of broadband
>        (2+MBps) vs the U.S. legislative/regulatory definition
> (744KBps).  I2
>        would be nice, but an I2 connection is not necessary for the
> program
>        below.)
>
>        -Aaron
>        :-)'
>
>
>
>
>