From: doug@netcom.com (Doug Merritt)
Date: Sat, 13 Nov 93 12:13:51 PST
To: cypherpunks@toad.com
Subject: Re: Should we oppose the Data Superhighway/NII?
In-Reply-To: <pmetzger@lehman.com>
Message-ID: <199311132012.MAA21160@mail.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain


"Perry E. Metzger" <pmetzger@lehman.com> said:
>Just one question, Doug -- in what sort of "Long Term" do you envision
>individuals needing to be able to send MORE than several thousand
>video signals worth of data simultaneously? Even if you put a camera
>facing every corner of every nook of your house, [...]
>you couldn't exhaust the potential bandwidth of a single fiber.

This is highly misleading. Consider this. The Nyquist limit puts a
fundamental limit on bandwidth; you cannot transmit more information
over a channel than (roughly) the cycles-per-second rate of the carrier.
If a fiber uses optical wavelengths, then that fiber cannot carry more
than a single fully quality analog optical *pixel*. We get a lot more
out of them than that by reducing the quality of the image being sent,
e.g. by sending only 60 frame-samples per second, where each frame-sample
is itself carved up into X * Y discrete pixels, and each pixel has a reduced
dynamic range etc.

My point in saying this is that you're speaking as if current day video
standards are some kind of ultimate load on information transmission,
whereas actually it's just something we've settled for. HDTV will vastly
improve the quality of what we transmit without increasing bandwidth
much, but it is still a far cry from what can be desirable.

60 frames per second makes it impossible to transmit adequate information
about objects moving quickly across the frame of view that are easy to
perceive in person. 1000 frames per second is desirable. 2D images
are less desirable than 3D images. A minimum of about 300 horizontal
views by perhaps 100 vertical views is desirable for 3D viewing. Assume
compression of that 300 * 100 down to a simple factor of 300. Now notice
that depth of field information is desirable for realism (without this
everything is always in focus, good for some things, bad for others).
Let's give that a simple factor of 10.

I'll leave out arguments for increasing e.g. the dynamic range of contrast
and color information, even though they are currently several orders of
magnitude worse than the human eye can perceive.

That all gives us roughly 10 * 300 * 10 = 30,000 times more information
in a single *really* high quality "video" signal than we are currently
accustomed to.

Ultra high quality image transmission like this won't begin to become
significantly widely used for quite a while. But it will happen eventually,
because we'll be able to, and will perceive differences, etc.

> What applications do you envision
>that would require more bandwidth than this, even in twenty or thirty
>years?

Truly high quality video is one answer, even without taking into account
the problems carriers would have in supporting the full bandwidth of
a single fiber in switched networks (combinatorial explosion means that
they can never support every possible connection simultaneously, therefore
switching and multiplexing is here to stay).

The more general answer is to just keep in mind that demand for uses of
technology *always* outstrips the capacity of technology, if it is
affordable. Demand is limited only because of economic issues.
	Doug