Must we have a theory of proof?

• Jurisdiction: South Africa
• Citation: 2003 Acta Juridica 113
• Author: Andrew Paizes
• Published date: 15 August 2019
• Pages: 113-138
• Date: 15 August 2019

Must we have a theory of proof?

ANDREW PAIZES*

University of the Witwatersrand

It is common, when writing a legal textbook, to begin with a general

introductory chapter in which the theoretical foundations of the subject

are expounded and explained. Chapters of this kind usually hold little of

interest for those who are concerned primarily with f‌inding practical

answers to legal problems. They are, typically, ignored by them as being

the province of academics who amuse themselves and each other by

posing and solving riddles that stimulate the mind but have little bearing

on the often prosaic world of legal practice. The question that forms the

title of this chapter would, no doubt, suggest to some, this kind of

exercise. ‘Who cares,’ one can imagine some people thinking, ‘about

competing ‘‘theories of proof?’’’

‘Evidence is lawyers’law. We know what we want to achieve, and we make

sure that our evidentiary rules deliver the results we need. There is little to be

gained by exposing what we do to penetrating analysis or by trying to

construct sophisticated models that rely on metaphysical distinctions, syn-

thetic metaphors or scientif‌ic precision, since most of the things we do when it

comes to legal proof take place at a visceral level where everything rests on

such unscientif‌ic and non-academic notions as intuition, experience, ‘‘gut-

feeling’’and observations on human nature. It is of such things that the world

of forensic fact-f‌inding, the so-called real world, are made, and any attempt to

contain them within a coherent scientif‌ic theory, while it may dignify the

subject and lend it a certain intellectual veneer, are bound to be artif‌icial,

unhelpful and, as a result, tend toward sophistry.’

My purpose in writing this chapter is to show that such views are

harmful and wrong. I hope to show that the selection of one or other

‘theory’ of proof is of the greatest practical importance and that

fundamental every-day problems of proof cannot adequately be addressed

without making such a selection. I propose to indicate, too, what the

implications of the chief competing theories are, and, f‌inally, to put

forward some ideas on how we might go about determining which of the

theories might best serve us in particular cases.

ITHE FIRST PROBLEM: IS THE COURTROOM LIKE A

CASINO?

The unlikely starting point for this exploration of theories of proof is

the casino. Imagine that you wish to play a game of roulette. Imagine a

*BCom LLB PhD (Witwatersrand); Professor of Law, University of the Witwatersrand,

Johannesburg.

113

2003 Acta Juridica 113

© Juta and Company (Pty) Ltd

roulette wheel of 36 numbers, 18 of which are marked in red, the

remaining 18 in black; 18 of which are odd, the remaining 18 even. If you

were to place a single token on a bet that the spin of the wheel will yield,

say, an odd number, it is accepted that your chance of success is one in

two, or, to express this in the language of probability theory,0.5. A bet on

a black number would have the same chance of success. A bet that

combined these outcomes, one that, in other words, is taken on an odd

and black number resulting, would yield a probability of 0.25 (or one in

four), a ratio that is the product of the other two ratios.

That it is entirely appropriate to employ this ‘product rule’in cases of

this kind is plain. It is a simple application of the classical or mathematical

system of probability that has come to be referred to as ‘Pascalian’, after

Blaise Pascal,

1

who developed the early analysis of the probabilities

involved in gambling. But what is much more controversial is whether it

is appropriate to engage in this kind of reasoning in the legal context.

It should be emphasised that the inquiry is not a synthetic one. One is

frequently required, in law, to consider the probability of the truth of an

entire series of propositions taken together where none of the preposi-

tions may be considered, in itself, to be certain. To take but one example:

2

X disappears in mysterious circumstances while hunting in a remote area

of bushveld. He was in the company of Y who returns with an

implausible account of the disappearance. Y had a motive to kill X, who

has not re-appeared in the intervening two years, and Y comes back with

some of X’s valuable personal possessions giving, again, an implausible

account as to how they came into his possession. The area in which X

disappeared is, however, notoriously dangerous, with countless natural

and other hazards, including wild animals, marshes, rivers, f‌loods,

extreme weather conditions, and human occupants hostile to tourists.

The propositions that would have to be established by the prosecution in

a case of murder against Y would include the following:

1.X is dead;

2.Y killed X;

3.The killing was unlawful (and, therefore, carried out in the

absence of defences such as self-defence, compulsion, and neces-

sity); and

4.the killing was intentional.

A probability ratio may be allocated to each of these propositions, none

of which may be taken to be certain, each ratio being determined,

however, on the assumption that the preceding propositions are certain.

1

1623–1662.

2

This example is taken from a murder trial that took place some 15 years ago in Botswana.

It is, to my knowledge, unreported.

114 CRIMINAL JUSTICE IN A NEW SOCIETY

© Juta and Company (Pty) Ltd