SMART Letter #65
Of Pachyderms and Visual Acuity
January 3, 2002
SMART Letter #65 -- January 3, 2002
Copyright 2002 by David S. Isenberg
isen.com -- "it really is a rope"
firstname.lastname@example.org -- http://isen.com/ -- 1-888-isen-com
> Of Pachyderms and Visual Acuity
by David S. Isenberg
with comments of various lengths and depths by SMART People
David P. Reed
> Copyright Notice, Administrivia
POLICY ON READER CONTRIBUTIONS:
Write to me. I won't quote you without your explicit
permission. So if you are writing to me for inclusion in
the SMART Letter, *please* say so. I'll probably edit your
writing for brevity and clarity. If you want your remarks
published anonymously, no problem, just ask. And if you
don't want your remarks published, that is default,
OF PACHYDERMS AND VISUAL ACUITY -- ISENBERG AND FRIENDS
FEEL UP THE NETWORK ECONOMY ELEPHANT, WARTS AND ALL
If ideas were races, David P. Reed [email@example.com] would
be a three-minute miler. But since ideas have a 'time to
come', Reed has a timing problem. Or maybe Reed is on time
and everybody else has a timing problem. I wish.
In the early 1980s (before the Intelligent Network was more
than a dull gleam in the eye of the pocket-protected
manager who would soon be assigned to invent AT&T
Marketing) David Reed wrote the original Stupid Network
paper (with Jerome Saltzer and David Clark) that laid out
the argument that communications networks should be as non-
specific as possible because network builders can't begin
to guess what creative uses network users will put them to.
Furthermore, argued Saltzer, Reed and Clark, they
shouldn't. About the same time that some singer put the
bop in the bop-she-bop-a-dop, Saltzer, Reed and Clark put
the slash in TCP/IP.
A couple of weeks ago, David Reed wrote to me to second
Roxane Googin's message. Actually, he firsted it. In 1997
Reed came to the conclusion that bad accounting by the
ILECs would lead to an S&L-like financial crisis. He
wrote, " in mid 1997 . . . I calculated [that] 3/4 of a
trillion dollars as the [ILEC's] accounting overhang at
risk from getting depreciation wrong (it's essentially
confusing 'useful life' with 'competitive life')."
(Incidentally, Reed also mentioned another prediction he
made in 1994, saying, "HFC plants, having been cost-
effective for broadband access and cable distribution
compared to DSL, will run into the end of their 'competive
life' in 2002 or so, and the underlying junk bonds in the
cable industry will not allow them to be refinanced."
Watch out. He's standing by this one seven years later.)
I suspect that much of the financial finagling by telcos
and cablecos to forestall competition is aimed at extending
the competitive life of their plants. If Reed is right
permacession will continue.
Reed's letter continues, "Moore's law is understood in the
computer industry. They don't issue long-term bonds to
finance their capital. But the telecom industry keeps
doing it. As far as I can tell the only reason for this
screwup is 'tradition'. Bond traders can't evaluate a
competitive industry on a Moore's Law curve. But because
people see wires as the asset, the opportunity to move the
real assets to the edges, to the users or others who can
finance them based on incremental value, is lost."
Reed's article, "Accounting in the Age of Moore's Law",
published in 1998, said:
"The basic issue is that computing prices have been
falling exponentially -- 50% every 18 months -- for the
past 30 years and will probably stay on that curve for
another couple of decades. Yet, for convenience,
accounting treats the plunge as a steady, straight line.
All assets, including computers, are depreciated as
though they lose the same small fraction of their value
"Accounting inaccuracies, allowed to accumulate for
years, can lead to spasms of disruptive activity . . .
"To think about accounting, it's useful to return briefly
to first principles. The value of an asset should be its
ability to contribute to future earnings. Predicting
that value is very hard, so companies generally try to
use the replacement cost of, say, a computer as the
asset's appropriate value. But even if companies did a
better job of carrying computers on their books at
replacement cost -- by shifting to exponential
depreciation -- they'd still be missing extremely
"Local phone companies show what can happen when values
are badly wrong. Fundamentally, the local
telecommunications industry is suffering from an
accumulation of technological inefficiency that has been
preserved and covered up by an accounting method that
does not account for innovation. The industry may now
suffer the consequences as competitors position
themselves to make inroads into the local market.
"Here's what has happened so far: For the past 30 years,
Moore's Law has guaranteed that the switches that
compute and process signals have plunged in price. Yet
local phone companies have depreciated switches slowly,
because their monopolistic positions and skill at
regulatory politics gave them the freedom to defer the
pain. As a result, when phone companies look at new ways
of handling communications, they make their calculations
based on balance sheets that assume signal processing is
1,000 times more expensive than it really is. If the
computer industry operated in the same way as the
telecommunications world, we'd all still be using, and
depreciating, IBM 360-model mainframes, which were
introduced more than 30 years ago and which have less
processing power than the typical calculator of today.
"At the same time as this depreciation disparity was
developing, technology made it easier for new
competitors with new, hyper-effective business models to
enter the market. While the phone companies assume that
they need to have switches that provide lots of
intelligence at the core of their networks-to move
calls, handle call waiting, and so on-a new style of
communications has developed that assumes no
intelligence in the network. Instead, this style --
which is the approach that drives the Internet --
assumes the intelligence on the ends of the network, in
my phone and yours. One result is that users can finance
new entrants into this kind of telecommunications
market, by buying intelligent phone-like devices, in
much the way they've financed the progress in the
computer industry over the past two decades by buying
personal computers. New phone companies don't need to
raise risk capital and invest billions of dollars in
central office switches to compete with AT&T.
"The local telephone companies could well have to replace
their old switching technology to keep up with newer
competitors that have a more efficient approach. Such
replacements, while cheap to buy, would entail huge
write-offs, given that existing switches are almost
worthless and yet are carried on the books at fantastic
sums. The central office switches and line cards of the
phone companies could become the nuclear power plants
(the "stranded assets") of the telecommunications
industry. We could wind up with a debacle on the order
of the S&L problem that led to such an enormous federal
bailout -- a bailout that some local phone companies are
beginning to lobby for [remember this was written in
1997 -- David I] under the misguided notion that
politicians will nostalgically support their monopoly
position against new, more competitive entrants."
David P. Reed's entire piece can be found at
How (you might ask yourself) did what Reed and Googin are
saying cause the current recession? Well, to me, it became
obvious that the tele-giants were winning their war against
competition and successfully crippling the emerging
infrastructure of economic growth. It seemed like that to
a lot of other people too. Then the RIAA kneecapped
Napster, and the biggest short-term reason for bandwidth
growth went down in the gutter. Why invest in competition
if it is going to fail? Why invest in bandwidth if the
bandwidth intensive apps are made illegal? Presto, change-
o, disolve in legislation, add a pinch of litigation and
stir in a pint of public relations. Blub, blub, blub.
My AT&T office mate in 1994 said that if AT&T knew what was
good for itself, it would kill the Internet. At the time,
I thought he was right with regard to AT&T's interests --
indeed, I had just written a paper about how the Internet
was threatening AT&T by changing people's communications
Even though I agreed that if AT&T knew what was good for
it, it would kill the Internet, my gut reaction was that
this position was morally reprehensible. Besides I thought
that killing the Internet would be impossible, that the
Internet would route around attempts to kill it. Hah! I
did not know it at the time, but today it is clear; the
Internet only routes around what its owners want it to
route around. Now it looks like the potential victims of
the Internet have begun to understand what my office mate
understood in 1994. It has taken some time -- a short time
in Corporate Time -- for the vested interests threatened by
the Internet to get their act together, but they have. As
a result, today the re-verticalization scenario of SMART
Letter #64 is a real possibility.
Art Kleiner [firstname.lastname@example.org], who wrote the insightful (and
for me, formative) history of corporate culture entitled
_Age of Heretics_ (Doubleday, New York, 1996) -- a book
that I wish Carver "technology solutions" Mead would read -
- wrote that Avoiding Permacession -- SMART Letter #64,
" . . . was probably the most significant and brilliant
SMART letter you've ever done, in my opinion, including
the two scenarios at the end, and it deserves broader
Like I said, my friend Art Kleiner sure is perceptive.
Wait a minute -- maybe he is dissing all my previous work.
Howard Anderson [HAnderson@yankeetek.com] wrote to say that
*he* caused the telecom recession. As he explained it to
The Boston Globe (May 3, 2001):
"Because of Y2K, every corporate technology executive in
the late 1990s all of a sudden had extra money available
to spend that was not charged to his regular account. If
you give a 2-year-old a hammer, everything looks like a
nail. So what did these executives do? They spent with
abandon. They upgraded applications, they improved their
infrastructure. The industry needed an enemy and Y2K was
"At the same time Amazon came on the scene. Every major
corporation - John Hancock, Fleet - was afraid of being
Amazoned. So they spent even more money to Webify their
applications. But now, the Amazon threat is over. Y2K is
over. Y3K is 999 years away. And corporations have gone
back to 1996 technology budget levels, which feels like
a recession after the growth rates of the past few
Myself, I think this is part near-term-hype-recovery phase
of Amara's Law. In other words, the threat of Amazon (and
its descendents) is NOT over.
By the way, if you ever get a chance to see Howard Anderson
do a shtick, don't miss it.
Phil Neches [email@example.com], weighs in:
"If I get the gist of the Smart Letter #64, the
telecommunications industry is leading the the world into
a Japan-like permanent recession, for which the only ways
out are to let the telcos put the world into their old-
fashioned straight-jacket or have the government force a
new-technology solution on a reluctant industry.
Gee whiz, could it get any worse!
[Government doesn't have to lead, but where infrastructure
('outside plant') is involved, government must be
involved. What happens to telecom easements along city
streets? There's no competitive, self-regulating
marketplace there. Are easements private property? And if
so, who gets paid? On what terms? Who enforces
those terms? (Hint: begins with G.) -- David I]
"Let me offer three perspectives on what's happening in
telecom-land, that don't lead to quite such a dismal
assessment of the future.
"First, this could be a replay of the computer industry
in 1984, although on a larger scale and with a slower
time constant. Most of us over the age of 40 remember
the 1980s as the decade in which the microprocessor rose
from being an obscure toy to the main engine of all IT
systems, pushing the mainframe with its 7-year
depreciation schedules and huge costs aside.
"But before the microprocessor outdid the mainframe,
there was a nasty shakeout in microprocessor-based
companies in 1983/4. This came about because VCs backed
what turned out to be lousy business propositions in the
early 1980s. Over 30 PC companies were funded
"Fast-forward 16 years. Venture capital formation is up
a factor of 20 or more ($2B/year then to $40-50B/year
recently). We started a lot more companies, with a lot
more capital each. But the percentage with winning
business plans and the ability to execute them is
probably about the same as ever. It's just that the
dogs are Great Danes instead of Schnauzers, and there
are 10 times more of them.
[Unlike the computer industry, the dogs of infrastructure are
on a leash, held by . . . government. -- David I]
"It took the world 3 quarters to sort out the high-tech
debacle of 1984. Some companies got additional funding,
albeit on lower valuations. Some companies merged.
Many closed. Sound familiar?
"This [current] work-out period is now in its 7th
quarter. Since it's always darkest just before dawn,
I'll risk a prediction that the work-out will largely be
finished sometime in the middle of 2002, after going on
for 10 quarters.
[Or longer, given G-forces :-) -- David I]
"The companies that survive the work-out period will be
smarter, better managed, more focused, and better
targeted. With valuations out of the ionosphere, they
should be able to attract capital to grow. With
interest rates, including long-term rates, at 50-year
lows, we may find companies using debt financing to a
much greater extent than in the 1970s and 1980s.
"Some of the old-line companies may make through. They
won't be the same as before. Industry transformations
usually unseat the broad-based incumbents, forcing them
to also become more focused (or die).
"The lessons for the big telecom companies are clear.
They can go under by sticking to obsolete technology and
business models. They can survive, and even prosper, by
finding something valuable to do in the new world order,
focusing on that niche, and abandon their earlier
pretentions of universality.
"It's clear that the many big US players (Lucent, and the
remaining Baby Bells) haven't got this picture. But
it's also not clear that they are entirely out of time.
"The second perspective comes from relatively recent
findings in mathematical economics relating to
properties of competition in network markets. Some
background is in order.
"In an ordinary market, the marginal cost of a good goes
up with volume. If you grow strawberries, you plant the
most fertile land first, resulting in the lowest cost.
As volume rises, you run out of the most productive
land, and start planting less desirable acreage, with
higher production costs. Good old Econ 101 stuff.
"In a network market, the marginal cost either doesn't
change or even goes down with volume. It cost Uncle Sam
zillions of simoleans to the be first Internet user.
It costs almost nothing to add the zillionth Internet
user. Also, the utility (value) of the good to any user
goes up with volume (more other users to talk to, more
things to do, etc.). Adam Smith didn't imagine that
this could happen.
"It turns out that competition works doesn't work the
same in ordinary markets and network markets. In
ordinary markets, it can be shown (by math that I can't
repeat here) that the market has four wonderful
+ The market is efficient, meaning that over all,
buyers and sellers get the best price possible;
+ The market is voluntary, meaning that buyers and
sellers participate only to the extent that they want
to, free of external compulsion to act against their
own economic interests;
+ The market is balanced, meaning that supply equals
demand, so that distorting surpluses or defecits do
not accumulate, and
+ The market is strategy-proof, meaning that nobody has
an incentive to buy or sell more than they strictly
need (in other words, it doesn't pay to game the
No wonder we like capitalism so much!
"However, network markets don't quite work this way.
Over the last 25 years, mathematical economists have
shown that you can choose any 3 out of 4 desirable
attributes for the market, but you simply cannot get all
four at once. Bummer!
[These are very strong, very specific claims to stand
without the support of citations or, at least a high-level
explanation of the math. I'd really like to be walked
through the logic. Maybe we will hear further from Phil on
this. -- David I]
"Regulation of public utilities (water, electricity,
telephones) began long before any of this math was
understood. Historically, we gave up efficiency to keep
the other three attributes (voluntary, balanced,
"Electricity deregulation as recently practiced in
California gave up the strategy-proof property, in the
hope of getting more efficiency. As each party madly
tried to game the new system, prices spiked then
cratered, leaving a mass of economic and political
carnage in its wake. It isn't clear if anyone is
getting more efficient along the way, at least not yet.
"Microsoft uses its monopoly power to decide who stays or
goes as a supplier, so the market is no longer
voluntary. Critics of the government's effort to break
up or tame Microsoft point out that if Microsoft is
forced to live by rules which make the market voluntary,
it may no longer be efficient.
"My hypothesis is that in the telecom world, we've given
up on balance of supply and demand, most prominently in
wide-area (city-to-city and larger scale) backbone
networks. Build it and they will come was the
"We know what happens when the voluntary goal is relaxed
(monopolies!). We also know what happens when the
efficiency goal is relaxed (old fashioned public utility
regulation, as practiced from around 1895 to 1995).
We're finding out what happens when the strategy-proof
and balance goals are relaxed, and I don't think we are
enamored by the early returns.
"The third perspective I can offer is that the SMART
Letter takes too extreme a view of the change in
economics of telecom. Directionally, the Smart Letter
is right on, but pursuing the argument to an unrealistic
extreme could be leading to some unwarranted
"To illustrate the point, consider the capital cost of
residential and small business service. I'll argue that
as the 50-year trend towards more and more employment in
small versus large businesses continues, this is the
telecom market which will matter most.
"The table below presents the first cost, useful life,
annual, and monthly depreciation for POTS (plain old
Conduit $1,000 40.00 $25.00 $2.08
Cable $200 40.00 $5.00 $0.42
Line Card $300 20.00 $15.00 $1.25
CPE $300 40.00 $7.50 $0.63
Upstream $200 20.00 $10.00 $0.83
Total $2,000 32.00 $62.50 $5.21
"It's easy to see how to make money on $20/month revenue
with these very extended depreciation schedules. Now,
let's see what happens when we decide that we need to
take the depreciation schedules down to the point where
we can see new technology enter the market in our
lifetimes, and continue to do so.
Conduit $1,000 30.00 $33.33 $2.78
Cable $200 5.00 $40.00 $3.33
Line Card $300 5.00 $60.00 $5.00
CPE $300 5.00 $60.00 $5.00
Upstream $200 3.00 $66.67 $5.56
Total $2,000 7.69 $260.00 $21.67
"It looks like it takes almost four times the cash flow,
suggesting that an operator needs to charge about
$80/month to have a viable business. The problem is, no
single service can command that:
Basic Extended Premium
Telephone $20 $35 $50
Television $27 $40 $65
Internet $21 $40 ---
[This is a neat picture that shows the effect of speeded
depreciation, but it does not show the effect of Moore's
Law, the collapsing of services into a single pipe, the
effect of radical simplification of network infrastructure,
the shift of costs to the edge, the decoupling of
connectivity from service or any of the other things that
are, per hypothesis, going on. -- David I]
"But if I can combine any two extended services, or any
premium service with either a basic or an extended
service, I get there.
"Funny thing: I have extended television service and
cable modem internet service, and my bill just rose to
pennies over $80.
"So somebody gets it.
"Further, this somebody is in a position to replace my
cable with something all new by 2005. And this somebody
is not a traditional telco.
[But don't expect the purveyors of yesterday's video
entertainment paradigm to clear the way for us to supercede
them -- the cablecos will never bring us the Internet that
today's off the shelf technology could afford, let alone
tomorrow's. -- David I]
"Postscript. The computer industry went on from the
debacle of 1984 to create over a trillion dollars of new
wealth. Once forced to do the right things, I have
every reason to believe that the telecom industry will
do the same or better. But I don't think that you would
be very happy with a portfolio of 1980 computer industry
leaders today. So why should you be happy with today's
telecom portfolio in 2015?"
[Phil, forgive me for interrupting so often. The above was
fascinating, and great food for thought. I'd love to see
your left hook in response to my jabs above.]
Economist Reuven Brenner [firstname.lastname@example.org]
"I'd like to comment about the similarity you draw
between the Savings and Loans and the telcos.
+ S&Ls were regulated about where they could invest,
and there was also deposit insurance.
+ so this was a bureaucratic 'business' And managers
self-selected. There were not many tough decisions
to make. The not-so-competent managed the business.
+ when they were deregulated the deposit insurance
stayed. But the same managers were suddenly allowed
to invest anywhere - high yield, exotic securities -
name it. The execs were not prepared to make such
decisions. So, unsurprisingly, they made mistakes.
+ in the S&L case, government was the insurer of last
resort, so had to pick up the bill.
"Now where are exactly the similarities with the
+ in the regulated telecom business, execs were
+ but by now the management has changed. Since MCI's
success, you have the Malones, McCaws, Forstmans
etc. allocating capital - so why are so many mistakes
[IMHO, Malone was an opportunist, a financial engineer.
Give him flux, and he will figure out where the flux to
make money. McCaw had one great success -- once again, he
was an opportunist who saw a chance to aggregate vastly
different cellular infrastructures -- his later attempts at
technology coups have yet to succeed. (A cynic could think
of McCaw and Malone as the two pirates who plundered the
foundering AT&T.) Forstman was an investor who got the short
end of the telecom stick when the ILECs defeated competition.
-- David I]
"+ is regulation that keeps telcos from making
investments that would benefit them the problem?
+ or is it that nobody has yet figured out
how to make money from all these technological
innovations or how to build a lasting barrier to entry?
"If the latter is the case then, may be, the best thing
would be to give incentives to experiment, reallocate
capital quickly -- and the best way to achieve that
would be to eliminate capital gains tax. That frees up
locked capital and is the best incentive that I can
think of to experiment with."
[Reuven knew that I'd be most heavily in favor of #4,
technology change. But eliminating the capital gains tax
seems too easy. Some economists (forgive me, Reuven if I
tar you with this old brush -- if the brush doesn't fit,
I'll acquit) . . . some economists see a budget surplus and
say, "Eliminate the capital gains tax!" and then they see a
recession and they say, "Eliminate the capital gains tax,"
and then there is a war and they say, "Eliminate the
capital gains tax." Under what circumstances is the
capital gains tax a good thing, if any? Why not eliminate
the sales tax or the income tax, or the telephone tax?
What about R&D credits? What about tax on static business
models that don't spend at least 10% of their profits on
R&D? What about a government funded institute of network
economics, so we get more network economists who can think
creatively about these problems? OK, OK, these are not
necessarily genius ideas, but whadaya expect for three
minutes of thinking by a failed biologist sitting in front
of a keyboard? -- David I]
Forrester is the thin-report company, the company that
tells executive suite-niks the 'so-what' about the
'so-what' and what to do about it. Jeanne Schaaf
[email@example.com], wrote asking what I thought of
her new Forrester report on the X-Internet. The wha?
The X-Internet is the Forrester story of how the Internet
grows up, gets easier to use, has guaranteed reliability,
thruput, and latency backed by service level agreements,
runs smart SOAPy and XMLized applications that act on data
themselves so we don't have to think about it. According to
Forrester, the telcos are in the best position to build
it. Why, why they MUST build it if it is to be built at
Well, I can't see it.
Guaranteed reliability, thruput, and latency is another way
to say, "Intelligent Network." Once you start telling the
telco how they should shape their pipes to your app, you've
lost the nonspecificity that made the Internet great. And
you quickly lose the ability to separate connectivity from
Then there's the ease-of-use part. Seems to me that ease-
of-use is for device or software manufacturers, not
telephone companies. The last user-friendly device the
telcos built was the Princess Phone. Did you ever try to
use one of those monster business phones that come with a
Definity PBX? It's like Windows without the GUI.
Then there's the XML part -- smart applets and datalets
twittering around the network in perfect harmony like so
many birds foraging in a raspberry bush. Nice vision, but
would the telephone company build XML apps? Would it know
So I asked Schaaf why she thought the telcos were in
such a MUST build position. She wrote back that the telcos
were the only "service providers with the scale, scope, or
financial viability" to do the job. Financial viability?
I remember when AT&T had the best network, the best brand,
the strongest balance sheet. Then it had the best brand
and the strongest balance sheet. Then it had the strongest
balance sheet. Then it had a strategy, which made
$80 Billion in cable assets disappear.
Ah, financial viability is an ephemeral thing. The most
important telecom event in 2001 was that the Incumbent
Local Exchange Carriers lost lines. For the first time,
there were fewer active lines in the United States than the
year before. This is in part because people use their cell
phones as second lines, in part because more people are
connecting to the Internet using persistent non-dial-up
lines, mostly Cable Modems, and in part because fax is
becoming obsolete and/or moving to the Internet. The walls
of the ILEC are caving in -- history will record that Year
2001 was the turning point.
The next Internet will not be built by any monolithic
entity, not by the phone company, not by the U.S.
Government. Instead it will be built by municipalities, by
colleges, by small cable and telephone companies, by
electric districts, by ISPs, by industrial development
corporations, by garden clubs and boy scout troops. The
telephone companies are history. Year 2001 was the turning
point. You heard it here first.
And maybe I'm wrong. (You heard *that* here first, too.)
Happy New Year. May people be smarter and networks be
stupider, and may 2002 be better than 2001 -- David I
COPYRIGHT NOTICE: Redistribution of this document, or any
part of it, is permitted for non-commercial purposes,
provided that the two lines below are reproduced with it:
Copyright 2002 by David S. Isenberg
firstname.lastname@example.org -- http://www.isen.com/ -- 1-888-isen-com
[There are two ways to join the SMART List. The PREFERRED
METHOD is to click on http://isen.com/SMARTreqScript.html
and fill in the form. But you can also send me email saying
who you are, how you see your work connecting to mine,
and why you want to join the SMART List.]
[to quit the SMART List, send a brief "unsubscribe"
message to email@example.com]
[for past SMART Letters, see
David S. Isenberg firstname.lastname@example.org
isen.com, inc. 888-isen-com
-- The brains behind the Stupid Network --