Paul Cockshott wrote:
> It is certainly true that reducing a vector to a scalar reduces
>information, and this is one of the arguments why calculation in kind is
>necessary for socialist planning. On the other hand, there remain situations
>where a scalar measure is necessary:
>
> 1. When allocating budgets to different social priorities it would be
>proceduraly impossible to organise a democratic vote that would allow
>meaningful in kind votes on the allocation of resources for health care,
>education, transport etc.
Joseph:
No, this is a circular argument. You assume implicitly that the only
meaningful way to judge things is via a single scalar measure, and then you
prove that there has to be a single scalar measure, because we need to judge
things.
------------------
Paul:
The point about decisions necessarily involving a scale follows directly from
information theory.
To make a decision between two alternatives requires one bit of information (
Shannon), this
then implies that the simplest decision possible involves a one bit number. If
having made the first
categorical decision we then try and make a second we generate a second less
significant digit.
The finer grows our capacity for decision making then the number of digits will
grow as the log
of this fineness.
This is not a circular argument at all, it draws instead on the most
fundamental theorems of modern information
theory and thermodynamics.
----------------------------------
Joseph:
The issue isn't whether a scalar measure is useful for some purposes. The
issue is whether it's the fundamental or natural unit of economic
calculation.
-------------------
Paul:
The question here is whether Labour is the natural unit of economic
calculation, was Smith right in saying that it was the original currency with
which we purchase our wants and necessities from nature. I think that in a very
deep sense Smith was right here.
---------------------
> On the other hand it is feasible, albeit complicated, to arrange such
>votes in scalar units such as money or labour.
What you seem to be suggesting is that, for example,a figure for a certain
maximum amount of money that can be spent (or labor-hours expended) could be
set. Every option should be given a dollar (or labor-hour figure). And people
are told they can vote for so many options so long as they add up to less
than or equal the total amount of money.
----------------------------------------
paul:
No you are right Joseph that that would not be a practical measure, the proposed
mechanism is here
:http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.154.8769
I believe that a number of other practical mechanisms could be invented.
> There is a bandwidth problem here, the bandwidth of a democratic decision
>making process is comparatively low and can thus only make broad decisions
>whose details have to be delegated to subordinate bodies.
Your implicit assumption here is that broad decisions are decisions that set
the amount of dollars (or labor-hours) to be expended. Then, from the need
for broad decisions, you justify your initial assumption.
-----------------------
Paul
This is just from a hypothesis of how things have to be administered. Consider
health care
and education for a region, my assumption is that the citizens of the region
can not individually
concern themselves with all the details of organising these services. There
need to be committees
or other bodies responsible for the administration of the services. These
committees can not
individually have a call on unlimited resources for the services they provide :
a person can not
at the same instant be working as a teacher and a nurse, though they may do
these in succession.
Thus the committees will need a budget in some sort of units.
How else do you think such public services can be administered.
------------------------
> 2. When attempting at a local level to get at least an approximation for
>which production plan is most likely to fit into the overall social plan of
>production, some sort of multipliers can be of a lot of assistance. The
>proposals which are more expensive in terms of labour are likely, but not
>certain, to be more expensive in terms of overall planning constraints.
>
> Also when looking at the issue mathematically, there are circumstances
>where vectors can be successfully mapped onto scalars with only modest loss
>of information, this arises when the actual information content of the
>vector space is not as high as it could potentially be because of
>correlations between the elements. I have sucessfully built video codecs
>used in mobile video phones on this principle which is described here:
That's a totally different situation. Let's look at a sample economic issue.
Suppose a product requires $10 million worth (or perhaps 10 million labor-
hours worth) of steel and wood and direct labor for its production. Well,
I'll just talk of "units" rather than dollars or labor-hours, in order to
avoid irrelevant considerations here.
So, suppose we know that the product requires 10 million units to be
produced. This doesn't allow us to know either how much steel or how much
wood is used or how much direct labor is used. The one number, 10 million
units, is very convenient, more so than having to keep track of three
numbers, such as 5 million units of steel and 3 million units of wood and 2
million units of direct labor. Or rather, this 10 million unit figure would
be convenient, except that it tells us very little. It is a measurement of
something "unnatural", as Marx would say. Indeed, as Marx did say, about such
figures.
---------------------------------
Paul:
You are right that it innvolves information loss, that is why it has to be
combined with calculation in kind using detailed planning.
Kantorovich type planning can take the in kind considerations into account,
given broad labour budgets and current labour prices one sets the plan ray and
optimises using linear programming.
--------------------
But, you say, redundancy saves the day. OK, let's consider that issue. Could
you give some idea of what redundancy you are talking about? What is the
redundancy that would allow us to deduce from the overall figure of 10
million units how much steel, wood, or direct labor was needed?
>
> Algorithm for the Hierarchical Vector Quantization of Video Data
> Cockshott,W.P. Lambert,R.
> Appeared in:
> IEE Proceedings: Vision, Image and Signal Processing
> Publication Type: Conference Proceedings
> Page Numbers : 222-228
> Publisher: Institute of Electrical Engineers
> Year: 1999
> ISBN/ISSN: 1350-245X
>
> and here www.dcs.gla.ac.uk/~wpc/reports/hvq/algorithmhvq.doc
As long as we're talking about sound, here's another example. Most piano
music is written for two-hands. The sheet music therefore has two streams of
notes, one for the right-hand, and one for the left. The right-hand is
generally, although not always, in the treble cleff, and the left-hand in the
bass cleff.
So we need two streams of notes, one to know what the right-hand is supposed
to play, and one to know what the left-hand is supposed to play. As we move
from right to left, we pass through what is played at each instant of time.
Can those two streams of notes be put together into a single stream of
notes? Why haven't musicians and composers figured out how to do this? Why
haven't they figured out how to represent playing a treble G and a bass D as
simply playing, well, ahem, which note? G? D? The average of the two?
But an economy is much more like an orchestra than a single piano. And an
orchestral score has many streams of notes, not just two. And it is to be
reduced to a single stream of notes? Good luck with that....!
-- Joseph Green
_______________________________________________
pen-l mailing list
[email protected]
https://lists.csuchico.edu/mailman/listinfo/pen-l
The University of Glasgow, charity number SC004401
_______________________________________________
pen-l mailing list
[email protected]
https://lists.csuchico.edu/mailman/listinfo/pen-l
