it presupposes a real and equal acceptance of risk by all members of
society. At some time, one wants to say, everyone must actually have
agreed to take the same chances. Since clearly there was no such occa-
sion, the principle is unsound. Consider an extreme case: a slaveholder
when confronted by his slaves attempts to justify his position to them by
claiming that, first of all, given the circumstances
of their society, the
institution of slavery is in fact necessary to produce the greatest average
happiness; and secondly, that in the initial contractual situation he would
choose the average principle even at the risk of its subsequently happen-
ing that he is justifiably held a slave. Now offhand we are inclined to
reject the slaveholder’s argument as beside the point, if not outrageous.
One may think that it makes no difference what he would choose. Unless
individuals have actually agreed to a conception of justice subject to real
risks, no one is bound by its requirements.
On the contract view, however, the general form of the slaveholder’s
argument is correct. It would be a mistake for the slaves to retort that his
contentions are irrelevant since there has
been no actual occasion of
choice, no equal sharing of risk as to how things would turn out. The
contract doctrine is purely hypothetical: if a conception of justice would
be agreed to in the original position, its principles are the right ones to
apply. It is no objection that such an understanding has never been nor
ever will be entered into. We cannot have it both ways: we cannot inter-
pret the theory of justice hypothetically when the appropriate occasions
of consent cannot be found to explain individuals’ duties and obligations,
and then insist upon real situations of risk-bearing to throw out principles
of justice that we do not want.
25
Thus in justice as fairness the way to
refute the slaveholder’s argument is to show that the principle he invokes
would be rejected in the original position. We have no alternative but to
exploit the various aspects of this initial situation (on
the favored inter-
pretation) to establish that the balance of reasons favors the two princi-
ples of justice. In the next section I shall start on this task.
The first difficulty with the average principle I have already mentioned
in discussing the maximin rule as a heuristic device for arranging the
arguments favoring the two principles. It concerns the way that a rational
individual is to estimate probabilities. This question arises because there
seem to be no objective grounds in the initial situation for assuming that
one has an equal chance of turning out to be anybody. That is, this
25. I have myself been in error on this matter. See “Constitutional Liberty and the Concept of
Justice,”
Nomos VI: Justice,
ed. C. J. Friedrich and J. W. Chapman (New York, Atherton Press, 1963),
pp. 109–114. I am grateful to O. H. Harman for clarification on this point.
145
28. Difficulties
with the Average Principle
assumption is not founded upon known features of one’s society. In the
early stages of the argument leading to the average principle, the hypo-
thetical newcomer does have some knowledge of his abilities and of the
design of the societies among which he is choosing. The estimates of his
chances are based upon this information. But at the last stage there is
complete ignorance of particular facts (with the exception of those im-
plied by the circumstances of justice). The construction of the individ-
ual’s prospect depends at this stage solely upon the principle of insuffi-
cient reason. This principle is used to assign
probabilities to outcomes in
the absence of any information. When we have no evidence at all, the
possible cases are taken to be equally probable. Thus Laplace reasoned
that when we are drawing from two urns each containing a different ratio
of black to red balls, but we have no information as to which urn we are
faced with, then we should assume initially that the chance of drawing
from each of these urns is the same. The idea is that the state of ignorance
on the basis of which these prior probabilities
are assigned presents the
same sort of problem as the situation where one has a lot of evidence
showing that a particular coin is unbiased. What is distinctive about the
use of the principle is that it enables one to incorporate different kinds of
information within one strictly probabilistic framework and to draw infer-
ences about probabilities even in the absence of knowledge. Prior prob-
abilities however arrived at are part of one theory along with estimates of
chances based on random sampling. The limiting case of no information
does not pose a theoretical problem.
26
As evidence accumulates the prior
probabilities are revised anyway and the principle of insufficient reason at
least insures that no possibilities are excluded from the outset.
Now I shall assume that the parties discount likelihoods arrived at
solely on the basis of this principle. This supposition is plausible in view
of the fundamental importance of the original
agreement and the desire to
have one’s decision appear responsible to one’s descendants who will be
affected by it. We are more reluctant to take great risks for them than for
ourselves; and we are willing to do so only when there is no way to avoid
these uncertainties, or when the probable gains, as estimated by objective
information, are so large that it would appear to them irresponsible to
26. See William Fellner,
Do'stlaringiz bilan baham: 
