When Neville Chamberlain returned from Berlin, confident that Hitler had no global ambitions beyond seizing the Sudetenland, he told the nation about his meeting with Hitler:
I got the impression that here was a man who could be relied upon when he had given his word.
Winston Churchill never met Hitler and never wavered from his contrary view, which included that Chamberlain’s visit was 'the stupidest thing that has ever been done'.
This may not be a statistically significant sample size on which to base scientifically rigorous conclusions.
The same cannot be said, however, about an analysis carried out in 2017 by Kleinberg et al of the outcome of 554,689 bail hearings in New York between 2008 and 2013.
In that study (as reported in Malcom Gladwell’s excellent recent book Talking to Strangers (Allen Lane 2019)), the human judges released just over 400,000 defendants. The researchers then built an artificial intelligence system, which, when fed with the same information, nominated its own list of 400,000 defendants to be granted bail. They found that the people on the computer’s list were 25 per cent less likely to commit a crime while awaiting trial.
Gladwell’s theory is that 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'.
While some might view this as a human failing, it is doubtful if society could function if our starting point was that not only should we be suspicious about strangers, but that they positively could not be trusted.
This might suggest that in areas such as bail and final hearings, we should keep the truth-default of the individual out of the process and leave the decision-making to computers. Indeed, I have often heard it said that Blockchain is the future of how legal cases will be determined: provide a sufficiently large database of previous cases and the computer will be more reliable than the individual. The individual (in the form of the judges and also the advocates) can then be excluded altogether from the process.
There are many reasons why this may not be sensible: the algorithm on which the computer system operates may be defective or insufficiently nuanced or it may not have sufficient inputs to cater for the variations that may occur (even if only infrequently); there can be errors or deficiencies in the inputting of the database; there may be errors or deficiencies in the inputting of the particular case (try inputting Carpi rather than Capri into your GPS); and there may be problems with the technology (interrupted power, failure to update, problems with compatibility, bugs and glitches). Perhaps most significantly, an algorithm works on a statistical approach and so cannot guarantee accuracy in each individual case: are we then willing to sacrifice the merits of that individual case to the overall average success of the group of cases?
The fact that the Global Financial Crisis occurred in spite of (or perhaps because of) the sophisticated financial algorithms that the individuals had developed and put in place should give us pause for concern.
In his excellent podcast Cautionary Tales (Pushkin Industries) Tim Harford observes how in the mid-1980s, a group of medical professionals developed a complex computer diagnostic tool for the treatment of acute abdominal pain. Although the tool had an accuracy rate of about two thirds, following its introduction, there was overall a huge and much larger drop in mortality, unnecessary surgery and medical errors.
So what was the additional factor?
Many of us have blindly followed the tracker of an easy-to-use GPS without engaging our senses; and individuals allowed the forecasting of the financial computer models to drive the world on autopilot into the GFC.
The 1980s hospital model, however, was not effortless. It had many fields and inputs and required a significant amount of effort on the part of the individual. It prompted the doctors to engage with the model and to stop and think.
When King Croesus of Lydia consulted the Oracle at Delphi, he was told: 'If King Croesus attacks the Persians, he shall destroy a mighty empire'. He regarded this as an encouraging prophecy, but in attacking the Persians, he only succeeded in destroying his own mighty empire.
Harford describes Alexa, Google and Siri as our modern oracles, to which we might add Wikipedia and Blockchain. As Harford puts it:
Just because you get a good forecast doesn’t mean you are guaranteed to make a good decision…We need to think much harder about what those oracles are telling us.
We should not fear technology and its ability to process which data is obviously far superior to ours, but we must be careful not to trust it blindly.
So what does this have to do with this issue of Bar News on Alternative Dispute Resolution?
For some time I have been remarking on the growing trend of solicitors briefing counsel late and only for the purpose of appearing at trial. I continue to believe that we have a place and can add value across the entire litigation process, but while counsel may be asked to give an earlier advice on prospects, we are often not included in the settlement process (whether through formal mediation or otherwise).
There are three aspects in relation to which the Bar should be promoting the benefits that we can be bring to the ADR process.
The first is the objective and specialist advice that we can give, which represents Harford’s role of the oracle attempting to predict the future. It has always been recognised that counsel’s advice can add value to a case, particularly cases where there are complicated legal issues or where the stakes are high.
There is no reason why this should now change. In the context of a mediation, however, a somewhat more nuanced and sophisticated approach is required.
It is not enough to tell a client that they have 'a good case' and that there is a range of damages if they are successful. We should be able to put figures on our view as to the prospects of success and the likely damages. That enables us to provide the client with a commercial value of their cause of action for the purposes of considering settlement proposals: for instance, a 60% chance in a case worth $2 million represents a commercial value of $1.2 million.
Hugh Stowe has developed a sophisticated settlement model that can incorporate multiple possible outcomes on different aspects of a case with different ultimate results in financial terms, together with the range of outcomes in costs.
Such tools are useful, but we should be careful not to cloak with the appearance of scientific confidence what are in fact personal estimates. For instance, a case with a 50% to 70% chance in a case worth $1 million to $3 million will still have a commercial value of $1.2 million, but it is a different from and less certain case than the example given above. Further, the breadth of any given range will be influenced by each counsel’s own approach to risk. For instance, different counsel may rate the same confidence in success at anywhere between 60% and 90% depending upon their own approach to risk.
Such advice can be, and commonly is, sought in advance without the need for the barrister attending mediation.
The second benefit that counsel can bring is in relation to the dynamics of the settlement process. Solicitors often have to express confidence in a client’s case in order to retain that client, but counsel are expected to express their views in an objective and fearless manner (even at the risk of losing the brief for not having confidence in the case). Furthermore, there can often be a logjam and inflexibility that develops between opposing clients and solicitors, especially as reflected (or cemented) in correspondence; and counsel can be a fresh face that helps to break that impasse.
The third benefit that counsel can bring to the ADR process is in assisting clients to fulfil Harford’s wish that they 'think much harder about what those oracles [i.e. us] are telling [them]'.
A client is entitled to know whether a $1.2 million valuation is based upon a confident evaluation of a figure or the midpoint of a range. If an advice incorporates many different ranges based upon multiple scenarios, it becomes unwieldy, less comprehensible and less useful. This can be amplified by a natural tendency to incorporate caveats and self-protecting phrases.
Attending a mediation enables counsel to engage with the client and the solicitor as to the detail of the advice and the potential scenarios in the context of a developing settlement negotiation. Ultimately a decision on settlement will be driven by the client’s own risk profile, but an offer of $700,000 looks different when compared on the one hand with a firm valuation of $1.2 million (60% chance of $2 million) and on the other with a range of $500,000 to $2.1 million (50% to 70% chance of $1 million to $3 million). These are the sorts of things that can, and should, be discussed at the mediation since they only become relevant as and when an offer is 'in the range'.
As Sir Laurence Street AC KCMG QC used to observe, a mediation is an opportunity to see the other side of the coin, being the case from the other side’s perspective. There can then be a need to revisit, qualify or modify a previously expressed opinion or to consider an alternative scenario in the light of arguments or pieces of evidence that may be raised by the other side.
Harford relates what is termed the illusion of explanatory depth from a psychological study by Rozenblit and Keil in 2002. They asked people to rate their understanding of ordinary household items, such as the flush toilet or a zipper. They were then asked to set out a detailed explanation of how those items worked. Finally, they were asked to re-rate their understanding, at which time it was generally significantly lower.
I suspect that, whether due to stylistic issues in its writing or human nature in the reader, many clients and solicitors do not engage with counsel’s written advice beyond the headline ultimate conclusion and expect that advice to be given with a narrow certainty. It is only by a discussion with us in the context of a developing settlement process that a client can get the best use from our advice in order to be able 'to make a good decision'. We cannot allow ourselves to be pushed out of the ADR process without protest or comment.
1 ‘Should I accept that offer – a practical and comprehensive methodology for settlement claim valuation’, Published in Lexis Nexis, Australia, Alternative Dispute Resolution Law Bulletin, November 2019, pp27-33.