From: John Conover <john@email.johncon.com>
Subject: Re: vnunet.com [Linux] Linux: long-term commitment or a flash in the pan?
Date: 23 Feb 2000 06:30:32 -0000
John Conover writes:
>
> Gartner Group's take on acceptance of Linux. A bit of
> back-peddling. Note:
>
> "... Linux outside its core areas of ecommerce and web servers ..."
>
> Considering ecommerce as a core application of Linux is a bit
> unusual-ecommerce is owned by Sun and IBM's AIX.
>
> At any rate, their assessment that Linux could be a fluke, and this
> year will tell the difference agrees pretty well with the theory of
> lock-in/stochastic/path-dependency.
>
> If it is assumed that February, 1997, was when the Linux bubble
> started, then there is a 50/50 chance that it will last until mid 2001
> on shear luck alone. Any longer than that, and the chances that it is
> a fluke deteriorate at a 1 / sqrt (t) scenario.
>
> The expansions and contractions of the electronics industry seems to
> follow the 1 / sqrt (t) scenario very well. An analysis, with
> comparative empirical and theoretical graphs is at:
>
> http://www.johncon.com/john/correspondence/981014184454.18095.html
> http://www.johncon.com/john/correspondence/981014210544.18525.html
> http://www.johncon.com/john/correspondence/981014222823.18931.html
> http://www.johncon.com/john/correspondence/981014233807.19309.html
>
> John
>
BTW, as an aside, if the entropic model of industrial markets is
correct, then the probability, P, of the Linux "bubble" continuing
through the third year, and into the forth, would be about 1 / sqrt
(3) = 57.74%.
What that means is that, from information theory, one's portfolio
should contain 2P - 1, or about 15.47% invested in Linux companies,
(or a large company should have 15% of its market share at risk in
Linux.)
Note that it is an iterated game of shuffling money. The 2P - 1
scenario will maximize capital, (or market share,) growth. Having
more, or less, at risk, will result in less growth, in the long run,
(although the scenario can be beat, on sheer luck alone, in the short
run-but not in the long run after the fugitives from the laws of
probability get caught.)
That is why most companies run a product line with 1 / 0.15 = 6
products in it, (more than that, the resources are too diluted-less
than that, the risks are too high, and eventually, the product line
will fail.)
The gross revenue from each of the 6 products will fall into a
log-normal distribution, meaning that 85% of the revenue will come
from 15% of the 6 products, (ie., about one.)
The idea of management is to suck money out of the winners, and put it
into the developing product lines, making the log-normal distribution
Gaussian-which is optimal growth for a company. The reason is that
product life cycles, on average, will follow the industrial average,
of a 4 year cycle, (ie., 1 / sqrt (4) = 0.5, eg., half the cycles are
shorter than four years, and half longer,) and all winners will become
losers.
If you look at the old marketing and CEO adages: run 6-8 products; and
80% of the GR will come from 20% of the product line; and economic
cycles are 4 years long, etc., which all kind of make theoretical
sense, too.
Nice when theoretical frameworks and intuition gained from experience
are in agreement.
John
--
John Conover, john@email.johncon.com, http://www.johncon.com/