Subject: Re: How accurate is Metcalfe's law? (Was: Ximian software)
Date: Fri, 4 Jan 2002 12:22:55 -0500

Stephen J. Turnbull wrote:
> >>>>> "Ben" == Ben Tilly <> writes:
>     Ben> (Of course energy spent only indicates that it was worth at
>     Ben> least that much.  It may have been worth far more.  Find me a
>     Ben> way to estimate that gap though. :-)
> It's called "econometrics," and it's the bread and butter of high-tech
> marketing and the bottom quartile of economics graduate students.

It sounds to me like a wonderful thing to know that is going to be
impossible to come up with useful measurements of.  But if people are
able to be paid to employ their time in (what I consider) a futile
pursuit, the more power to them.

>     Ben> While it is possible for individual connections to be worth a
>     Ben> thousand dollars per hour to the recipient, I think most will
>     Ben> agree that, barring inflation, it will never happen that the
>     Ben> average person will exceed $1000/hour of derived value from
>     Ben> time spent connecting.
> This is quite possibly wrong.  It is certainly arguable that real
> productivity growth will run into a barrier at some point.  However,
> it seems quite plausible to me that someday the average GDP per person
> in some economies will reach $2,000,000/year.  (Assuming that the US
> is currently $40,000, at 1%/year real growth rate that takes less than
> 400 years.  In the New Economy (5%/annum), only 80 years.  :-)

Interesting point.  However you will note that for the next 20 years
that figure will not be reached, no matter how many people we connect
to the Internet.  My argument, amended for significant changes in
overall personal worth, still stands.  Metcalfe's law is not only
wrong, it is ridiculously wrong.

Incidentally long before we hit the required productivity levels for
your scenario, we are likely to hit my nightmare scenario of true AI.
That goes like this.  The history of technology teaches us that when
new technologies are introduced they displace humans, but that work
gets done more efficiently, the humans find other lines of work, and
we wind up better overall.  Old story, seen it dozens of times.

I predict that won't happen if we develop true AI.

Consider what happens when we can manufacture robots that are roughly
equivalent to a human being, for under 3 years salary.  Let us assume
that the knowledge of an existing one can be readily reproduced in new
copies, and that these machines learn new jobs about as fast as a human.

As before, humans get displaced.  But now for every job a human can
learn, the same machines are able to learn more cheaply, and are better
for the employer than a human is.  So people get displaced and stay
displaced.  The worth of a human's work is now capped by effective
competition - the cost of buying a machine.  As the price of that
machine falls, well you see why I call this a nightmare scenario...

This may never happen.  But a back of the envelope estimate I did in
the late 90's said that with Moore's law, in about 2020 computers
should achieve a number of operations roughly equivalent to a human
brain.  (For the computer I used full Pentium instructions, eg a
multiply.  For the brain I used individual neuron firings.  Yeah,
apples to oranges...)  Assuming that it takes a few years for
would-be AI to aquire enough knowledge to be useful, plus assuming
that something like true AI would (Moore holding) be a real possibility
around 2030 (+- big fudge factor).

So this scenario is, in addition to being OT for this list, at least a
few decades off.  But it comes before your scenario, and strongly
suggests that the time of *humans* will never be worth that much. :-)

> Now, I ask you: with productivity increasing at that rate, how much of
> that $2,000,000 is likely to come from tangibles like cars?  The great
> bulk has to be intangibles, ie, Internet-connectivity-based.  And a
> couple years later, your $1000/hour figure is ancient history.

What is the price of a week-long cruise to Mars?  Interstellar travel?
Cellular replacement therapies to extend old age indefinitely?  I
strongly recommend reading "Indistinguishable from Magic" by Robert L.
Forward to get an idea for some of the tangibles which are physically
plausible in decades and centuries to come.  Much nicer than anything we
have, but tremendously expensive.

How many of the tangibles which are important parts of our markets
today were not in use 50, 100, 200 years ago?  Think about when flush
toilets, electric power, cars, televisions, computers, cell-phones, and
so on came into common use.  Should productivity continue to improve as
time continues, then there will be both new tangibles and intangibles.
Until they arrive I don't think we can say what the ratio of values
will be.

> (Note, the above argument uses the following Lemma: "Necessity is the
> Mother of Invention."  Corollary: Anything necessary to make me right
> will be invented.  ;-)

Mind if I borrow this quote? :-)