Subject: Re: How accurate is Metcalfe's law? (Was: Ximian software)
Date: Thu, 3 Jan 2002 10:48:38 -0500

Perry Metzger writes:
> writes:
> > But the value of the net to an individual person is still bottlenecked
> > by that person's ability to enjoy making the connection.  While it is
> > possible for individual connections to be worth a thousand dollars per
> > hour to the recipient, I think most will agree that, barring inflation,
> > it will never happen that the average person will exceed $1000/hour of
> > derived value from time spent connecting.
> Careful there. As the Chicago and especially Austrian economics
> schools mention over and over, the only way to assess monetary
> equivalent value is with what people will pay for something. Net
> connections are worth whatever people are paying for them (typically
> pennies an hour), regardless of the value of the information
> retrieved.

I suspect you missed my basic point.  Metcalfe's "law" says that the
value of a network scales as the square of the number of people who are
connected to it.  I am claiming that this is impossible by pointing out
that this rule implies levels of value per person that are simply
ridiculous.  My point wasn't that access to the net should be worth
$1000/hour, it is that it isn't worth that as can be seen that people
aren't willing to pay it.  In other words I am saying that most people
won't pay for access, therefore we know it isn't worth that.

But even so, any decent economist should also tell you that up front
fees aren't a good way to account costs.  If a business decides to
provide network connectivity, email, etc, its cost for doing that is
not the cost of the network connection.  Instead its cost should
include the cost of keeping it administered, the cost for employee time
spent using the connection for legitimate purposes, and the cost of
employee time for the predictable time people will spend on the net
when they were supposed to be working.

I guarantee you that my employer is paying orders of magnitude more for
the ability to send this message than the cost of the network
connectivity it takes to send it.  The employers of many recipients

> It does not matter that my life is of infinite value to me and that
> without water I will die -- the glass of water in front of me is not
> worth even $1 no matter how much I value my life, unless in odd
> circumstances (say, at an oasis in the desert).

It is true that total value does not matter under normal circumstances.
The value there is unexploitable by legitimate businesses, and is
therefore as irrelevant to their normal life as the energy matter has
by virtue of having mass is in normal life.  (But the rare cases
where even a small fraction of that value can be harnessed is the
economic equivalent of a nuclear bomb.)

However that doesn't mean that it is worthless to attempt to understand
what motivates the perception of value in potential customers.  For one
thing without understanding what they find valuable, you have no idea
what to sell or how to price to get the most out of your audience.

For instance consider one of Odlyzko's points.  When you allow people
to connect to a network, the majority of the value that the people will
get out of it comes from simply connecting to other people.  Therefore
there is more money in selling simple connectivity than in selling
custom content, and people in the business of selling connectivity stand
to make more if they can support a pricing model that encourages
customers to connect at will.  (ie You should try to use a flat rate
pricing model.)  These are non-trivial conclusions with important
consequences for businesses.

Incidentally I keep thinking there is a connection between this and
open source development models.  The idea being that proprietary
software is in the business of content creation, and free software is
content created by network dynamics (the latter eventually surpassing
the former in terms of how much members of the network value it).  If
you push this analogy further the majority of the money to be made is
in providing access to the network.  Access points to this "network"
being employees, consultants, etc who are "connected" to the software
world and can use it to solve non-programmers problems.

I am not sure that the analogy is very accurate, but it predicts what I
already believe so I like it. :-)

> (Attempts to produce other metrics of value, like the Marxian "labor
> theory of value", have been general failures. The Marxian metric would
> value a badly designed and shoddy program that took 300 programmers a
> year to write far above an equivalent program that was beautifully
> written but took only one programmer to write. It also values a ditch
> that takes a week to dig for a man with a shovel far above an
> equivalent ditch produced in an hour by a backhoe. Beware of
> value metrics that are not validated by the intersection of supply and
> demand curves.)

Why do I have this sneaking suspicion that attempts to fully consider
the "true value" of things will lead to calculations that turn out to
be infinity divided by infinity, and when renormalized will give answers
that you can make come out to anything you wish? :-)