Subject: Re: near/medium future digital media economics
From: "Ben Tilly" <btilly@gmail.com>
Date: Thu, 18 May 2006 18:58:09 -0700

 Thu, 18 May 2006 18:58:09 -0700
On 5/18/06, Thomas Lord <lord@emf.net> wrote:
> Ben Tilly wrote:
> > [.... (full quote appended) ....]
> Well, speak of the devil :-)
>
> You admit (not to twist your words too badly) that Metcalf may
> apply in "a small number of circumstances."   I would have said
> "some" rather than "small number" but, ok, what kind of circumstances?

Metcalfe's Law (BTW, note the spelling) is going to apply best in
networks where people actively wish to connect with fairly random
strangers.  Very few networks look like this.

One that does is online auctions.  A litmus test to see whether
Metcalfe's Law is close to being correct is that the largest
competitor should be able to price their product at a significant
premium and *still* outcompete the rest.  eBay does this in the online
auction space, demonstrating that online auctions are indeed an
example where Metcalfe's Law is close to correct.

> I suggest that Metcalf applies at least in the case of networks
> comprised of persistent, collaborative documents -- such as
> a distribute, participatory, hyper-library of all digital media
> content.

I suggest that you should look up Benford's Law and explain why it
doesn't apply to this type of content when it does to other kinds.
Because if it *does* apply, then something fairly close to an n log(n)
law will be appropriate for your kind of network.

> Logically this makes some sense because users of the network
> create, at unitary cost, "half-connections" -- contributions to
> the persistent document.   Once created, the resulting non-rival
> half-connection can become a complete connection to another
> user an arbitrary number of times.    In other words, a persistent
> collaborative document makes all of broadcasters, and all of us
> audience for other broadcasters.

This argument, like Metcalfe's original argument, has the implicit
assumption that everyone is equally interested in all potential
connections.  I'm not.  Nor is anyone else.

The result of uneven interest is the natural emergence of power laws
in how much people are interested in others' potential content.  The
result of these power laws is that something like an n log(n) scaling
law naturally emerges.

> We agree that a broadcast network (as a rule of thumb) has a
> value proportionate to the size of the audience.   It's also the case that
> for each audience member, the value of an allocated spectrum is
> at least proportional to the number of broadcasters.  (I would not cap
> the value for an audience member at the limit of their attention because
> feedback among broadcasters means that when an audience member listens
> to one, he is really experiencing a sum of many.)

You're still hiding the same assumption.

As for the feedback point, that is an extremely complex point and to
really understand it you need to differentiate between value at a
point in time and value over time.

Feedback of that form has been going on in various forms for all of
human history.  It naturally results in exponential improvement over
time, with differing exponents in different subject areas.  If a new
medium has more efficient feedback, then what happens is that the
feedback cycle gets faster.  A slight improvement in the feedback
cycle leads to exponential improvement over time.

However at any given moment of time, people do not value the medium
based on that expected eventual exponential over time.  They value it
based on what difference it makes to their lives here and now.  It is
that second value that I'm asserting is n log(n).

Another random note is that if I'm relying on a different medium, then
I do not need to participate in the improved feedback cycle to get
many of its benefits.  What I can do is wait for someone to filter and
summarize the information for me.  Unless the area under discussion is
something that I need to remain on the cutting edge for, this is
sufficient and is a *lot* less work.  (And sometimes I don't even need
that - how much do you need to know about the research on chip design
to take advantage of Moore's Law?)

> Well, if when a communications network is grounded in a persistent
> collaborative document each member is both broadcaster and
> potential receiver of N-1 signals.   Metcalf's N^2 looks quite
> reasonable.
>
> Curiously, while *your* analysis helps to explain certain dot-com
> era failures, the analysis I've just given helps to explain the dot-com
> survivors and the new successes in the Web 2.0 world.

The analysis that you've just given is Metcalfe's original argument,
and suffers from the same deficiencies.  You've just counted
connections and assumed that they all have equal value.  But that
assumption flatly contradicts observations of how people really
network in social situations.

I would say that the successes are founded in different things.  For
instance Amazon's success is IMO largely because they are great at
personalizing their site to each customer.  I've already attributed
eBay's success to the fact that they happen to be in an area where
Metcalfe's Law is close to correct.  (And they are far from the first
marketplace that has benefited from that fact, incidentally.  They are
not even the first online marketplace to so benefit.  The one that
comes to mind for me is Instinet, but I'm sure that there are others.)
 And Google's power derives from the fact that good searching allows
people to come closer to perfectly extracting the value of being on
the network.

> Finally, I would not worry too much about Metcalf's or even
> Reed's law having a threshold at which, with the addition of a
> single member, the value of the economy doubles -- it's the
> way people measure the economy that is wrong.   Rival
> consumables place an absolute upper bound on the size of
> the economy.  Non-rival goods and networks are ways to
> spend surplus and thus their total value is strictly limited.
> With that understanding, it isn't actually all that implausible
> that the addition of a single member to a network might
> double the value *of the surplus*.

Metcalfe's Law has no such threshold.  Reed's law does.  As for your
defence of Reed's Law, well what you're saying is that you think that
it is reasonable that adding one person to the internet could double
everyone's perceived quality of life.  If you think that that's
reasonable, then I don't really know what to tell you.

> This is not to say that at some point, every new member
> has that effect.  The constant factor on these approximations
> matter and we are lucky/happy when a few great minds per
> generation or 10 have such effect.

Oh no.  Reed's Law *explicitly* says that every new member has that
effect.  Which is why it doesn't pass the "smell" test.

Cheers,
Ben