In a Folklaw survey of lawyers as to the reasons why people dislike them (quoted in Lawyers Weekly in 2015), the top response was, perhaps unsurprisingly, 'because they do not appreciate our witty and insightful humour and are intimidated by our classical good looks'. Perhaps a more accurate response, however, was 'because it is an adversarial system…The losing person is going to hate not only his own lawyer but also the other side’s'.
There was no discussion, however, of the attitude of litigants to the judge who had decided their case or indeed to the broader legal system. Presumably, the losing litigant will often be dissatisfied with the judge and the system, and things get more complicated when a case goes on appeal. It is likely that where there is a successful appeal, each litigant is left dissatisfied with at least one judge and probably also the system generally.
I have always felt uncomfortable that a minority of judges can hold sway over the result of a case. Thus, in a case ultimately decided in the High Court, a majority of four judges in that Court can hold sway over their three brethren, three intermediate judges and one trial judge, being seven in total. While this primacy of High Court judges is more easily understandable in the context of setting legal precedent and maintaining a coherent system of law, it is more difficult in the area of fact finding.
There is no doubt that one needs appellate review to be available. Even judges are fallible. In spite of the High Court’s recent willingness to engage in detailed reviews of evidence and fact finding, however, (e.g., Robinson Helicopter Company Inc v McDermott (2016) 331 ALR 550, Lee v Lee (2019) 266 CLR 129 and, in a criminal context, Pell v R (2020) 376 ALR 478) the scope of permissible review is still not settled (cf Nationwide News Pty Ltd v Rush (2020) 380 ALR 432 at ) or, if it is settled, it is still not clear to me.
The starting point is the somewhat contradictory propositions that although an appeal from a decision of a single judge is by way of rehearing, it 'is not the occasion to re-run the trial' (Croucher v Cachia (2016) 95 NSWLR 117 at  to ). Thus it is said that an appellant must demonstrate error in the primary judge’s decision, so that, by contrast with the trial, the starting point on an appeal is usually the reasons of the primary judge rather than the evidence.
In relation to findings of fact, appellate courts are 'obliged to conduct a real review of the trial', which requires that they engage in 'weighing conflicting evidence and drawing [their] own inferences and conclusions', but they 'should always bear in mind that [they have] neither seen nor heard the witnesses, and should make due allowance in this respect' (Fox v Percy (2003) 214 CLR 118 at , citing earlier authority).
In Robinson Helicopter, French CJ, Bell, Keane, Nettle and Gordon JJ put the allowance for the advantage of the trial judge thus at :
…a court of appeal should not interfere with a judge’s findings of fact unless they are demonstrated to be wrong by “incontrovertible facts or uncontested testimony”, or they are “glaringly improbable” or “contrary to compelling inferences”.
In Lee v Lee, however, the High Court put this allowance somewhat more narrowly. Thus Bell, Gageler, Nettle and Edelman JJ stated at :
Appellate restraint with respect to interference with a trial judge’s findings unless they are “glaringly improbable” or “contrary to compelling inferences” is as to factual findings which are likely to have been affected by impressions about the credibility and reliability of witnesses formed by the trial judge as a result of seeing and hearing them give their evidence. It includes findings of secondary facts which are based on a combination of these impressions and other inferences from primary facts.
In ABT17 v Minister for Immigration and Border Protection  HCA 24, Kiefel CJ, Bell, Gageler and Keane JJ at  described the deference to be allowed to the findings of the trial judge as follows:
The court of appeal does not, however, have the opportunity of seeing and hearing witnesses give their evidence, or thus the opportunity of making a fully informed assessment of the witnesses’ demeanour. Accordingly, the established position in relation to an appeal by way of rehearing from the judgment of a judge alone is that, where the judge’s decision is affected by his or her impression of the credibility of a witness whom the judge has seen and heard give evidence, the court of appeal must respect the attendant advantages of the judge in assessing the witness‘s credibility.
In the same case, Gordon J at  described the trial judge’s advantage by reference to 'findings based on demeanour'.
In Queensland v Masson (2020) 381 ALR 560 Nettle and Gordon JJ said at :
For present purposes, it is enough to repeat the observations of Gleeson CJ, Gummow and Kirby JJ in Fox v Percy that, at least where the trial judge’s decision might be affected by his or her impression about the credibility of the witness, whom the trial judge sees and hears but the appellate court does not, the appellate court must respect the attendant advantages of the trial judge.
Thus, the restraint and deference for the trial judge’s advantage has been stated variously to apply to cases involving 'credibility', 'credibility and reliability' and 'demeanour', each of which is a distinct although overlapping concept. Thus, for instance, credibility is commonly applied where a witness may be deliberately not telling the truth whereas reliability is often used for a witness who is attempting to tell the truth but may be mistaken.
A judge will often make findings of fact without explicitly stating whether the witness is not to be believed or is mistaken, such as where a document makes some other finding more likely. It is difficult to see why the test for appellate review should be applied differently depending upon whether the trial judge rejected evidence based upon credibility, reliability, demeanour or the balance of probabilities generally, particularly when that basis may be unstated.
The advantage of the trial judge is, as stated in Fox v Percy, having seen or heard a witness, regardless of how that evidence is in fact dealt with.
To view the test for appellate restraint from the point of view of the advantage of the trial judge in having seen or heard a witness rather than by narrow touchstones of 'credibility', 'reliability' or otherwise would suggest that similar appellate restraint ought to be applied in making findings based upon expert evidence (Child and Adolescent Health Service v Mabior  WASCA 151 at ). This would include findings of breach of duty or negligence based upon such evidence, being as they are findings of fact (Swain v Waverley Municipal Council (2005) 220 CLR 517 at  and Vairy v Wyong Shire Council (2005) 223 CLR 422 at ).
In Bauer Media Pty Ltd v Wilson (No 2) (2018) 56 VR 674 at  to , the Victorian Court of Appeal approved a distinction between findings of fact and inferences, approving dicta:
Once the primary facts have been established, however, the question whether particular inferences should be drawn from those established facts is a matter as to which an appellate court is generally in as good a position as a trial court to consider for itself. The strictness with which Robinson Helicopter approaches findings of primary fact is not applicable to purely inferential reasoning.
This is apparently to be contrasted with 'findings of secondary facts which are based on a combination of these impressions and other inferences from primary facts', where, following Lee v Lee, appellate restraint is to be exercised.
The distinction between 'purely inferential reasoning' and inferential reasoning based in part upon primary facts derived from seeing or hearing a witness may not always be easy to draw; and would seem somewhat artificial.
Further, even in the case of 'purely inferential reasoning', the Court in Bauer noted that:
…the appellate court should, however, give respect and weight to the conclusion of the judge, but, having reached its own conclusion, it must give effect to it.
If an appellate court is carrying out a rehearing and is 'in as good a position' as the trial judge to determine an inference, then it is difficult to see why the trial judge’s decision should be given any particular 'respect and weight' other than perhaps out of politeness.
In my view, this is a recognition that the advantages that a trial judge enjoys are not limited to observing demeanour or to seeing and hearing witnesses. The opportunity to assess oral testimony is but one example of those advantages (as discussed by the Full Court of the Federal Court in Jadwan Pty Ltd v Rae & Partners (a firm) (2020) 378 ALR 193 at ). Indeed the High Court recognised this in Fox v Percy at :
These limitations [on the appellate court proceeding on the record] include the disadvantage that the appellate court has when compared with the trial judge in respect of the evaluation of witnesses’ credibility and of the 'feeling' of a case which an appellate court, reading the transcript, cannot always fully share. Furthermore, the appellate court does not typically get taken to, or read, all of the evidence taken at the trial. Commonly, the trial judge therefore has advantages that derive from the obligation at trial to receive and consider the entirety of the evidence and the opportunity, normally over a longer interval, to reflect upon that evidence and to draw conclusions from it, viewed as a whole.
As the Full Court noted in Jadwan at , care must be taken in following 'brief allusions, or necessarily incomplete references to principles in appellate judgments'. Appellate restraint of the type described in Robinson Helicopter should therefore not be confined to findings based upon 'credibility', 'reliability', 'demeanour' or some other such similar phrase, but rather by reference to the advantages enjoyed by the trial judge, which may vary on a case by case basis but will generally include facts in relation to which a judge saw or heard a witness give evidence.
Is seeing or hearing a witness, though, really such an advantage?
In an often-cited passage from Watson v Foxman (1995) 49 NSWLR 315 at 319, McLelland CJ in Eq referred to the fallibility of human memory increasing over time:
…particularly where disputes or litigation intervene, and the processes of memory are overlaid, often subconsciously, by perceptions or self-interest as well as conscious consideration of what should have been said or could have been said. All too often what is actually remembered is little more than an impression from which plausible details are then, again often subconsciously, constructed.
In Fox v Percy, the High Court was alive to the impact of this fallibility upon appellate deference to findings of trial judges, noting at  (per Gleeson CJ, Gummow and Kirby JJ) 'the dangers of too readily drawing conclusions about truthfulness and reliability solely or mainly from the appearance of witnesses' and at :
Further, in recent years, judges have become more aware of scientific research that has cast doubt on the ability of judges (or anyone else) to tell truth from falsehood accurately on the basis of such appearances.
In his recent book How to Make the World Add Up (Hachette Australia, 2020), economist, journalist and broadcaster Tim Harford talks about how our emotions and partisanship can, and often do, shape our reactions to asserted facts and our beliefs more than any logic or objective reasoning. We have a tendency to adopt 'motivated reasoning' and thus more information can simply provide us with more ammunition to reach the conclusion we already hoped (even if subconsciously) to reach. As Moliere once wrote: 'A learned fool is more foolish than an ignorant one'.
There is a particular danger of preconceptions and prejudices, particularly if unacknowledged, influencing demeanour based findings. Further, a preliminary view formed without hearing any evidence or a particular view formed of one witness may provide motivated reasoning in viewing the evidence that then follows.
In his recent book Talking to Strangers (Allen Lane 2019), Malcolm Gladwell describes how in assessing strangers, humans generally default to truth and only stop believing 'when our doubts and misgivings rise to the point where we can no longer explain them away'.
He describes how, with catastrophic consequences, US authorities ignored obvious signs of counterintelligence officer Ana Montes being a double agent. He gives as similar examples the convicted paedophile Jerry Sandusky and the convicted financier Bernie Madoff.
As Gladwell puts it, 'We are bad lie detectors in those situations when the person we’re judging is mismatched.'
This principle, however, is not limited to believing liars who seem sincere, but extends to honest people who look suspicious. Thus, he uses the example of Amanda Knox, who was convicted of killing her friend and only later cleared. He describes how Italian authorities ignored overwhelming evidence pointing elsewhere, because she was not acting how they expected a grieving friend to act. Lindy Chamberlain would probably be a similar example.
So there is a real tension between, on the one hand, a view that a trial judge possesses an advantage from having seen or heard a witness; and, on the other, the difficulty in forming any reliable view based upon the demeanour of a witness, particularly given the part that subconscious prejudices and 'motivated reasoning' may play.
This might suggest that appellate review should not give any particular weight to findings of the primary judge, howsoever based. There could then be a real and unlimited rehearing, albeit on the record.
This would be to constitute an appeal as a re-run of the trial and, apart from anything else, the system could probably not survive such an imposition. Further, there is no reason to think that appellate judges should be any better at making findings of fact than trial judges. Indeed it seems to be generally accepted that the trial process does have some advantages in relation to fact finding, even if this is only from the 'feeling' of a case.
Appellate courts are well-able to recognise their limitations in reviewing factual findings of trial judges. It is artificial to require them to characterise the nature of the factual finding (eg based upon credibility) or to explore the advantages that the trial judge had in any particular case (eg seeing a witness or the feeling of the case) and determine the impact of those advantages upon a particular finding.
A sensible and pragmatic response is called for, even if this is in fact a compromise between the requirement that an appeal be a rehearing but not a re-run of the trial. Appellate courts should not interfere with any finding of fact unless persuaded that it is plainly (or decisively, clearly or glaringly) wrong. While there can be argument upon how the test should be framed, it should be a single test for all findings of fact, requiring general appellate restraint.
There is a real tension between, on the one hand, a view that a trial judge possesses an advantage from having seen or heard a witness; and, on the other, the difficulty in forming any reliable view based upon the demeanour of a witness...