Some Discussion on the Economics of Moore's Law
- Rakesh Bhandari
- Carrol Cox
- Rakesh Bhandari
- Doug Henwood
--- from list marxism-international@lists.village.virginia.edu ---
==================
1) Rakesh Bhandari
==================
Date: Sat, 3 May 1997 09:22:58 -0800
From: djones@uclink.berkeley.edu (rakesh bhandari)
Subject: Re: M-I: (POF-1 Notes)
It is possible that Moore's Law is running up against the limits of Marx's
law of the rising weight of fixed capital. Let me quote from William
Greider's latest book (One World Ready or Not: the manic logic of global
capitalism):
"At each stage of advancement, as a rule of thumb, the production cost per
databit falls by more than half while the cost per function to end users
falls with it. Thus, the manufacturing cost was 17 cents per million
databits for the 64M chip in 1995. The cost of manufacturing the same
computing power would fall to 7 cents by 1998, 3 cents by 2001, 1 cent by
2004, half a cent by 2007 and two tenths of a cent by 2010.
"The industrial paradox, however, is that in order to achieve these
advancements in power multiplication and cost reduction, the companies must
assemble increase volumes of investment capital for the research and
development and construction of new factories. Thus for instance, a factory
cost for 64k chips in 1980 cost about $100 million to build, the 256k
factory in 1985 cost $200 million, the 1m factory in 1988 cost $300
million, the 4m factory in 1990 cost $400 million and the 16m factory in
1993 cost $700 billion. A fully automated 64m factoryin 1995 cost 1.2
billion and several new factories have been announced that will cost $2
billion. If the capital curve continues in this manner, companies may soon
be spending $5 billion or more per factory to keep up with Moore's law."
For Greider, as for Joseph Stiglitz (Whither Socialism?), the danger here
is that such costs can only be absorbed in cartelized markets, which will
then empower a few oligopolies to set administered prices. Just through the
threat of accelerating R and D schedules (Stiglitz), along with the rising
capital curve, oligopolies will be able to dramatically raise the barriers
to entry. Upon deeper inspection then Moore's law turns out be Marx's law
of the the concentration and centralization of capital.
Rakesh
==================
2) Carrol Cox
==================
Date: Sat, 3 May 1997 13:52:44 -0500 (CDT)
From: cbcox@rs6000.cmp.ilstu.edu (Carrol Cox)
Subject: Re: M-I: (POF-1 Notes). Query/Queries.
This message from Rakesh raises questions that go beyond my under-
standing of economics, but I want to know whether one speculation
of mine is relevant.
Greider is quoted as claiming that the development of the computer
(following "Moore's Law") involves an immense investment in physical
plant. Now, if I understand Sweezy and Baran's argument in *Monopoly
Capital* (and I think both Sweezy and Magdoff and others still hold
this position), it is precisely such immense demands for new physical
plant that capitalism depends on for survival. S&B's examples were
the building of the railroads and the process of automobilization,
and if I remember correctly, sometime in the last couple decades
Sweezy claimed that electronics (including computerization) would
*not* absorb capital as the railroads and the automobile had done.
Now a number of "ifs." If I remember Sweezy and Baran correctly,
and if their basic argument was correct, *then* (if Greider is correct),
Sweezy, Baran, and Magdoff were *wrong* in respect to the computer,
and computerization will give capitalism a new lease on life just
as the railroads and the automobile did.
Doug? Somebody who knows econ better than I do?
The point about the "rising weight of fixed capital" would be
that factory-building absorbs immense amounts of human labor rather
than (like the computers) replacing that labor.
Carrol
>
> It is possible that Moore's Law is running up against the limits of Marx's
> law of the rising weight of fixed capital. Let me quote from William
> Greider's latest book (One World Ready or Not: the manic logic of global
> capitalism):
>
> "At each stage of advancement, as a rule of thumb, the production cost per
> databit falls by more than half while the cost per function to end users
> falls with it. Thus, the manufacturing cost was 17 cents per million
> databits for the 64M chip in 1995. The cost of manufacturing the same
> computing power would fall to 7 cents by 1998, 3 cents by 2001, 1 cent by
> 2004, half a cent by 2007 and two tenths of a cent by 2010.
>
> "The industrial paradox, however, is that in order to achieve these
> advancements in power multiplication and cost reduction, the companies must
> assemble increase volumes of investment capital for the research and
> development and construction of new factories. Thus for instance, a factory
> cost for 64k chips in 1980 cost about $100 million to build, the 256k
> factory in 1985 cost $200 million, the 1m factory in 1988 cost $300
> million, the 4m factory in 1990 cost $400 million and the 16m factory in
> 1993 cost $700 billion. A fully automated 64m factoryin 1995 cost 1.2
> billion and several new factories have been announced that will cost $2
> billion. If the capital curve continues in this manner, companies may soon
> be spending $5 billion or more per factory to keep up with Moore's law."
>
> For Greider, as for Joseph Stiglitz (Whither Socialism?), the danger here
> is that such costs can only be absorbed in cartelized markets, which will
> then empower a few oligopolies to set administered prices. Just through the
> threat of accelerating R and D schedules (Stiglitz), along with the rising
> capital curve, oligopolies will be able to dramatically raise the barriers
> to entry. Upon deeper inspection then Moore's law turns out be Marx's law
> of the the concentration and centralization of capital.
>
> Rakesh
>
==================
3) Rakesh Bhandari
==================
Date: Sat, 3 May 1997 12:38:33 -0800
From: djones@uclink.berkeley.edu (rakesh bhandari)
Subject: Re: M-I: (POF-1 Notes). Query/Queries.
>
> Greider is quoted as claiming that the development of the computer
>(following "Moore's Law") involves an immense investment in physical
>plant. Now, if I understand Sweezy and Baran's argument in *Monopoly
>Capital* (and I think both Sweezy and Magdoff and others still hold
>this position), it is precisely such immense demands for new physical
>plant that capitalism depends on for survival.
Carrol,
It is interesting that Greider seems to reach the opposite conclusion:
"These few firms already had the power to decide whether the next
generation of DRAMs was technically feasible--and whether it was worth the
price in new capital investment. Conceivably, they may someday determine
that the drive for further innovation was essentially wasteful, consuming
too much *scarce* capital and undercutting estabilished products and
profits. When Moore's Law eventually exhauts its exponential pace, it may
be business economics, not physics, that derails it." p. 181, my emphasis
Of course, Greider provides no explanation either for why output prices may
turn out to be too low to justify new capital investment or for the
scarcity of capital to which he refers; in particular, does not trace it
back to the production process--as do Grossmann, Mattick, Kliman, Daum, Jim
Miller. Ultimately Greider seems to invoke "competition" to explain the
possible derailment of the economy.
Rakesh
==================
4) Doug Henwood
==================
Date: Sat, 3 May 1997 16:10:19 -0500
From: Doug Henwood
Subject: Re: M-I: (POF-1 Notes). Query/Queries.
Carrol Cox wrote:
> Now a number of "ifs." If I remember Sweezy and Baran correctly,
>and if their basic argument was correct, *then* (if Greider is correct),
>Sweezy, Baran, and Magdoff were *wrong* in respect to the computer,
>and computerization will give capitalism a new lease on life just
>as the railroads and the automobile did.
>
> Doug? Somebody who knows econ better than I do?
While I'm not going to deny that computers and their ilk are important,
they hardly occupy the "leading" role that rail and autos once did, at
least in the quantitative sense: they're not kicking capitalism into a new
high-growth, high-accumulation phase (I know folks like Barkley believe
we're in a long-wave upswing, but I don't see any evidence of that yet).
Certainly they're leading to social transformations, though as Robert Solow
famously said, the phenomenon is visible everywhere but in the numbers.
> The point about the "rising weight of fixed capital" would be
>that factory-building absorbs immense amounts of human labor rather
>than (like the computers) replacing that labor.
Or as Marx said, machines presuppose masses of workers.
Doug
--- from list marxism-international@lists.village.virginia.edu ---