Oral argument in the Article 50 litigation

The government's case got a lot of attention, and not necessarily the good kind

Today oral argument in the Article 50 litigation (R (on the application of Miller and Dos Santos) v Secretary of State for Exiting the European Union and associated references) came to an end.

The case is not the longest to have been heard by the Supreme Court. In 2012, the court dealt with two rather technical cases in taxation and insurance which ran for six and eight days respectively. But Miller is certainly the most important case the Supreme Court has decided since it was created, and the most assiduously followed. I can no think of no other case which has given rise to Carnwath/doge memes:

The Supreme Court has, exceptionally, made full same-day transcripts of oral argument available on its website. Over four days, eleven judges and thirteen counsel spoke almost a million words, or (to use the traditional measurement) around two-fifths of a Chilcot.

Not everyone spoke to the same extent. Here's a bar-chart showing the number of words spoken by the different individuals involved in the case. I've coloured individuals according to whether they were

• acting for the government of the United Kingdom (blue),
• acting for the original plaintiffs and other non-governmental interest parties (red)
• acting forvarious intervenors (green),
• judges (purple)

The chart shows the great weight placed on James Eadie QC. Eadie is used to it — he has made more appearances in the Supreme Court than any other barrister.

The total number of words spoken by barristers appearing for the different applicants (Miller, Dos Santos, Agnew and others, and McCord) was not very much less than the total number of words spoken by the different barristers acting for the UK government. Yet whereas David Pannick QC, acting for Gina Miller, was able to rely on other lawyers to amplify points made in his submission, Eadie was not able to do the same, and the arguments of Lord Keen of Elie focused understandably on devolution issues.

Four judges accounted for almost three quarters of the words addressed to counsel. I am not sure this indicates anything in particular. Some judges tend not to be particularly voluble, whilst others (like Lord Neuberger) are required to take a more active role and keep proceedings moving along.

(The picture would look more exaggerated if I counted only the number of interventions, in which case Lord Neuberger would dominate. It might be possible to restrict the analysis just to interventions of a certain length — but at this point one might as well just count words).

What is perhaps more interesting is to look at where judges' comments went. The graph below shows, for each judge, the number of words spoken to James Eadie and Lord Keen (on the horizontal axis) against the number of words spoken to all other lawyers (on the vertical axis). Judges south-east of the dashed diagonal line spent more time talking to government lawyers; judges north-west of the line spent more time talking to all other lawyers.

I think it's generally true that more is asked of the lawyers appearing for the appellant (in this case, the government), but I don't think that general tendency is strong enough to stop us from suggesting that the government's case got quite a lot of attention, and not always the good kind.

You might reasonably ask whether any of this matters, and whether it's even worth noting whether one side receives more questions than the other side.

It matters because previous research has suggested that the side which receives more questions is more likely to lose. This research is based entirely (as far as I know) on oral argument before the Supreme Court of the United States.

The reasoning behind this claim is simple. Before oral argument judges have strong prior beliefs about how best to decide the case. Judges ask questions of counsel in order to highlight weaknesses in their argument. (Judges also ask questions of counsel in order to signal their position to their colleagues). Judges who believe that the case should be decided in favour of the appellants will ask lots of questions of the respondents, and vice versa.

The relative number of questions asked of counsel for the appellant and the respondent therefore indicate how each judge in a case will decide the case; and (as a result) how the court itself will decide. In this reasoning, the actual answer given to the questions is irrelevant. Both questions and the outcome flow from judges’ prior beliefs.

It’s not clear whether this reasoning holds anywhere outside of the United States Supreme Court. In particular, it seems unlikely that it would extend to the UK Supreme Court. This is so for two reasons.

First, oral argument before the UKSC is much, much longer than oral argument before SCOTUS. The latter is ordinarily limited to an hour. The arguments in Miller have gone on for four days. If oral argument is much longer, it becomes more difficult to argue either that judges’ questions emerge from judges’ prior beliefs and a desire to signal positions from colleagues, rather than emerging in response to statements made by counsel; or that barristers’ answers to questions can be ignored as having no effect on the outcome.

Second, judges on the UKSC are less likely to have strong prior beliefs about how best to decide cases. Analysis of the notebooks left by Justices of the Supreme Court of the United States suggest that justices switch their vote (from affirming to reversing, or vice versa) around seven percent of the time. Alan Paterson’s analysis of the notebooks left by Lord Reid and Lord Bingham (which admittedly dwell on more important cases) suggests that Law Lords switched their vote (from allowing to dismissing an appeal, or vice versa) between 15 and 20% of the time.

The signalling role of questions is therefore likely to be less important in the UKSC compared to SCOTUS. But might questions still point to the likely outcome because of what they tell us about the quality of the arguments before the court? One possibility is that judges principally react to statements which seem to them to be wrong. A case which is constructed entirely on the basis of questionable claims will invite more questions.

Conversely, a case which is brilliant and unimpeachable in every respect will
receive far fewer questions. Accordingly, the side which receives more questions is more likely to lose, but for different reasons than those originally put forward to explain the pattern found in the United States.

This makes sense intuitively, but there are three broad categories of
counterexamples. First, judges may ask friendly questions. Second, judges may ask questions not because the argument is weak, but because it is novel, and because they are interested. Third, judges may hold back from questioning, not because the argument is strong, but because they are worried about dominating. As David Allen Green has noted, the fact that judges at a certain point stopped asked questions of James Eadie in his reply may be a source of concern, because that may mean they've already found a weak point in the argument.

There are, therefore, reasons to be skeptical about this kind of analysis. I put it forward not necessarily because I believe in it, but rather because it seems to me to be a hypothesis worth testing. Testing the hypothesis properly requires the analysis of many cases. After the tremendous success of crowd-funded litigation, is it time to crowd-fund transcription of oral argument?